ARCHIVES
OF BIOCHEMISTRY
Vol. 293, No. 1, February
AND
BIOPHYSICS
14, pp. 79-84,1992
The Interaction between Retinoic Acid and the Transforming Growth Factors-P in Calf Articular Cartilage Organ Cultures Teresa
1. Morales1
and Anita
B. Roberts
Bone Research Branch, National Institute of Dental Research and the Laboratory of Chemopreuention, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892
Received September 4, 1991, and in revised form October 18, 1991
In calf articular cartilage organ cultures, retinoic acid depressed proteoglycan anabolism to levels -10% of control values and increased their catabolism - 14-fold at concentrations of 1 X lo-’ and 1 X lo-’ M, respectively, leading to a severe depletion of this component from the extracellular matrix (95% loss in 3 weeks). These effects were powerfully antagonized by maximal levels of transforming growth factors+ (TGF-Bs) 1, 2, and 3, leading to preservation of matrix components. At a concentration of 1 X lOma M retinoic acid, the TGF-&I restored anabolism to control levels and lowered catabolic rates >3-fold. While the TGF-@s increased protein synthesis 2- to 3-fold over controls, retinoic acid alone did not change protein synthesis, as determined by incorporation of 13H]serine. Nevertheless, retinoic acid effectively antagonized the stimulation of protein synthesis by TGF+3 and restored control levels of synthesis at 1 X lo-’ M. Analysis of proteins, labeled using [3H]serine and [35S]sulfate as precursors, by SDS-PAGE revealed that large molecular weight proteins (>lOO kDa) were not detectable in retinoic-acid-treated cultures, but treatment with the TGF-@s restored these components in coincubation cultures, again supporting the antagonistic role of the polypeptide effecters on retinoid action. Treatment of the cultures with retinoic acid elevated levels of TGF-fl2 synthesis, but not TGF+l. While the role of the newly synthesized TGF-/32 in the set of events elicited by retinoic acid in articular cartilage is unclear, the results establish an intrinsic metabolic link between the isoprenoid and TGF-/3 in articular cartilage. We propose that the retinoids and TGF-@s are integral parts of a regulatory network that controls homeostasis, resorption, or growth, depending on their relative contributions. Q 1992 Academic Press, Inc.
’ To whom correspondence
should be addressed.
0003.9861/92 $3.00 Copyright 0 1992 by Academic Press, All rights of reproduction in any form
The uncoupling of anabolic and catabolic pathways in articular cartilage is characteristic of degenerative joint disease, or osteoarthritis (1). While it is well documented that progressive loss of matrix proteoglycans (aggrecans) from articular cartilage correlates with disease severity and with loss of the resilient, protective function of this tissue on the subchondral bone, the regulation of metabolic homeostasis in articular cartilage is incompletely understood (2). Previous work with articular cartilage organ cultures has established that a variety of effecters, added exogenously, can modulate the homeostatic balance of the tissue (2). Among these, excessive levels of vitamin A (retinol or retinoic acid) lead to progressive loss of aggrecans from the extracellular matrix (3,4). Interest in this finding has been rekindled with the discovery of specific nuclear binding proteins for retinoic acid, which interact with regulatory sequences in target genes to control their transcription (for reviews, see Refs. (5,6)). As in other tissues, the regulation of retinoid activity may be a vital component in the intrinsic metabolic regulation of cartilage matrix physiology. In contrast to the catabolic effects of retinoids on artitular cartilage, addition of transforming growth factor/3 (TGF$Q2 to cartilage organ cultures increases synthesis of aggrecans and diminishes their loss, leading to maintenance of the tissue’s homeostatic balance (7-9). Recent data suggest that in the epithelium, the action of these two agents, one an isoprenoid which acts through a set of nuclear receptors and the other a secreted peptide signaling through a set of cell membrane receptors, might be linked (10, 11). ’ Abbreviations used: TGF-/3, transforming growth factor-& DMEM, Dulbecco’s modified Eagle’s medium; Chaps, 3-[(3cholamidopropyl)dimethylammonio]propanesulfonic acid; PMSF, phenylmethylsulfonyl fluoride; o-phe, o-phenanthroline; N-EM, N-ethylmaleimide; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis. 79
Inc. reserved.
80
MORALES
AND
The present investigation tests the effect of physiological levels of retinoic acid on the articular cartilage matrix and explores the putative relationships between retinoic acid and TGF-P in this tissue. We present evidence that (a) physiological levels of retinoic acid induce pronounced changes in the proteoglycan matrix without changing overall levels of protein synthesis, (b) this effect is antagonized by the closely related TGF-/3 isoforms, TGF@l, TGF-@2, and TGF-P3, and (c) retinoic acid selectively induces synthesis of TGF-P2 but not TGF-@l in articular cartilage. This establishes for the first time a metabolic link between retinoic acid and TGF-P in a skeletal tissue. We suggest that these two agents are integral parts of a regulatory network that controls homeostasis, growth, or resorption in articular cartilage, depending on their relative contributions. METHODS Tissue Culture and Labeling Procedures Articular cartilage was dissected from the metacarpalphalangeal joints of calves -4-6 months old (slaughterhouse estimate) and placed in organ culture (12,13). Briefly, for each experiment, cartilage slices were pooled from the two joints of one calf, distributed into capped culture vials, and cultured in the presence of 15 vol (ml/g wet weight tissue) of Dulhecco’s modified Eagle’s medium (DMEM) containing 4.5 g/liter glucose and 0.58 g/liter L-glutamine for 7-10 days. The tissue was distributed in portions of -100-150 mg wet weight into individual wells of 24-well Costar plates for the initiation of experimental treatments All-trans retinoic acid was from Hoffmann LaRoche Inc. or Sigma Chemical Co. Porcine TGF-j31 and TGF-82 were from R & D Systems. Recombinant chicken TGF-03 (14) was purified to homogeneity by three successive HPLC steps consisting of Cl8 reverse-phase, TSK-CM ionexchange, and Cl4 reverse-phase chromatography. Sixty-fold concentrated solutions of either one of the TGF-@s or retinoic acid were added to the appropriate experimental wells containing basal medium (DMEM + 0.1% bovine serum albumin, Sigma Chemical Co., type V, crystalline). The experimental groups were as follows: (a) control cultures were maintained in basal medium, (b) retinoic-acid-treated samples were cultured in basal medium + the indicated molar concentration of all-trans retinoic acid, lo-“’ to 10e6 M, (c) TGF-@-treated samples were maintained in basal medium + saturating levels of either TGF-/31(10 rig/ml), TGFj32 (10 rig/ml), or TGF-j33 (1.5 rig/ml), and (d) retinoic acid + TGF-P cultures were maintained in basal medium containing these two effecters at the maximal concentration of TGF-fi and the indicated concentration of retinoic acid. For all reported results, n represents the number of experiments using tissues from different animals.
Measurements
of Proteoglycan
Metabolism
Methods have been described and validated (12, 13). Briefly, [36S]sulfate (20 &i/ml) was used to pulse the tissue and selectively label the glycosaminoglycans. For anabolic measurements, the glycosaminoglycans were extracted with alkali (0.5 N NaOH) following the pulse, and incorporated radiolabel was assessed following chromatography on Pharmacia PDlO columns (12, 13). For catabolic measurements, the %S-labeled tissue was washed in basal medium for 2 days (37°C) to remove unincorporated isotope and chased for at least 7 days under the appropriate treatment conditions. Medium was collected and changed daily. The total ?S activity in each sample was assessed by adding the cumulative daily release into the medium (following the wash) and the residual matrix activity (NaOH extract); the % of the total ?!l-glycosaminoglycan released daily was then determined.
ROBERTS
General Protein and Proteoglycan Synthesis
Core Protein
1. Protein synthesis. Experiments were as described above with the following exceptions. The labeling medium contained 20 &i/ml [35S]sulfate (carrier-free) + 60 &i/ml [3H]serine (92 Ci/mmol; both precursors were from ICN Biomedicals). Intact proteoglycans and other proteins were extracted from the tissue in chaotropic buffer (15) (0.1 M Tris maleate, pH 5.8, containing 4 M guanidine HCl, 0.5% 3-[(3-cholamidopropyl)dimethylammonio]propanesulfonic acid (Chaps), and the following proteinase inhibitors: 7 ag/ml each of pepstatin A and leupeptin, 1 mM phenylmethylsulfonyl fluoride (PMSF), 5 mM iv-ethylmaleimide (N-EM), and 1 mM o-phenanthroline (o-phe)). The residue was dispersed by proteinase K treatment (Sigma Chemical Co., 1 mg/ ml, 60°C -18 h). The incorporated ?S and 3H activity was assessed as described above. Protein synthesis was expressed as the incorporated 3H activity in the conditioned medium + guanidine HCl extract t proteinase K extract of each sample. For control experiments, portions of the 4 M guanidine HCl extracts and of the conditioned media were dialyzed against dilute urea buffer (6 M) at 4’C and applied on DEAE-Fast Flow columns (Sigma Chemical Co.), -500 pg chondroitin sulfate/ml gel. These DEAE columns were equilibrated and eluted with urea buffer (0.05 M sodium acetate buffer, pH 6.0, containing 8 M urea, 0.3 M NaCI, and proteinase inhibitors at l/10 the concentration described above). Proteoglycans, selectively retained by these columns, were eluted by 1.0 M NaCl in urea buffer (15). The collagen content of the tissue was stable during all the experimental treatments (not shown, 7) and was therefore used as a normalization factor for all anabolic measurements. Collagen (hydroxyproline) and chondroitin sulfate were assayed as described (13, 16, 17). 2. Electrophoresis of 3H-labeledproteins. 3H-proteins in the 4 M guanidine HCl extracts (m-70-80% of the total in the tissue and conditioned medium) were fractionated by chromatography on DEAE columns as indicated above. The 0.3 M NaCl eluenta were extensively dialyzed against 4 mM HCl containing 0.1 pg/ml each of pepstatin A and leupeptin and 0.1 mM each of o-phe and PMSF (4°C). Portions of the dialyzates were dried in the speed vacuum centrifuge, resuspended in Tris-SDS buffer (Seprasol I, Enprotech Co.) containing 0.7 M @-mercaptoethanol and boiled for 5 min. SDS-PAGE was carried out on 4-20% minigels (Enprotech Co.) and fluorography was performed using Kodak X-omat AR film.
Synthesis of TGF-@I and TGF-@2 Cartilage cultures were treated as indicated and then labeled with 250 &i/ml [35S]cysteine (ICN Biomedicals, Inc., 960 Ci/mmol) in basal medium with or without retinoic acid for -16 h. The immunoprecipitation procedures have been described (9, 18). Briefly, portions of each conditioned medium and 4 M guanidine extract of the tissue were immunoprecipitated using specific antibodies against TGF-@l or TGF-j32. Immunoprecipitates were dissolved in 100 ~1 Seprasol I by boiling for 5 min and subjected to SDS-PAGE.
RESULTS A. Proteoglycan
Metabolism
Articular cartilage organ cultures maintained under basal conditions (without serum and/or added growth factors) decreased their rates of synthesis of proteoglycans to a relatively stable background during the first week in culture (not shown, 9, 12), but remained phenotypically stable (8,12). The experimental treatments were therefore initiated following the first week of culture (Methods). Treatments were for 7-9 days, since previous studies had indicated a lag time of 6 days for expression of the full
RETINOIC
ACID
AND
TRANSFORMING
GROWTH
FACTORS-/l
IN CALF
ARTICULAR
CARTILAGE
81
effect of TGF-fl on proteoglycan biosynthesis. However, 330 t -90% of the maximal response to retinoic acid was observed within 2 days (not shown). 250 L 1. Proteoglycan anubolism. Figure 1 illustrates the dose-response curve for all-trans retinoic acid added alone to the basal cultures and in combination with TGF-fil and TGF-02. The data show that retinoic acid is a powerful inhibitor of proteoglycan synthesis, lowering rates to - 10% of basal levels at the saturating dose of 1 X lo-* M, with an ED5,, of 1 X lo-” M. By contrast, in previous work we demonstrated a dosedependent, equipotent ability of TGF-Pl and TGF-/32 to increase levels of proteoglycan synthesis, with saturation at 5-10 rig/ml (7-9). In the present set of experiments, the average stimulation of proteoglycan synthesis by maximal levels of TGF-fis 1 and 2 was 290 t- 56% and 270 + 67% (&SE, n = 6 and n = 5, respectively; n = number of experiments using pooled tissues from different I I I , I animals). We have now extended the analysis of the TGFI Basal 10-10 10-9 10-8 10.’ fl isoforms to include TGF-63 in selected experiments. This isoform increased synthesis of proteoglycans, an average of 190 f 36% (&SE, n = 3) at 1.5 rig/ml, found to Molar Concentrations of All-Trans Retinoic Acid be a saturating concentration for this isoform (EDso = FIG. 1. The effect of retinoic acid and TGF-P on proteoglycan syn0.5 rig/ml). Addition of maximal levels of either TGF-Pl or TGF- thesis in calf articular cartilage organ cultures. The vertical bars on data point represent the standard error of the mean for three to fl2 to cultures containing retinoic acid antagonized the each four representative experiments using tissue from different animals, or inhibitory effect of the retinoid. At concentrations of ret- the range for two experiments. TGF-@s 1 and 2 were added to a final inoic acid of 1 X lOmaM or less, the TGF-@s maintained concentration of 10 rig/ml. synthetic rates at levels above basal controls (Fig. 1). At concentrations of retinoic acid greater than 1 X lo-’ M, cultures treated with 1 X lop8 M retinoic acid. The total the TGF-0s reduced the inhibitory effect of the retinoid chondroitin sulfate mass left in the tissue at this time but the overall synthetic rate was suppressed compared was 14% of initial values. This is consistent with the seto basal levels. TGF-83 also antagonized the effect of retvere catabolic depletion from the tissue and extremely inoic acid (1 X lo-’ M retinoic acid + 1.5 rig/ml TGF-83 low anabolic rates. The cultures treated with 1 X 10-s M = 80 -t 39% of basal level (&SE, n = 3), compared to 13 retinoic acid + TGF-01 (10 rig/ml) retained 55% of the + 5% with retinoic acid alone in the same set of experi?S activity and 68% of the chondroitin sulfate, a finding ments). consistent with the partial antagonism by TGF-P of the 2. Proteoglycan catabolism. The average catabolic effect of this concentration of retinoic acid on both carates of basal cultures were low (the tlj2, i.e., time for the tabolism and anabolism of proteoglycans. Similar trends release of 50% of the labeled proteoglycans from the were observed in a longer experiment. Table I shows the matrix, was 40 -t 21 days (&SE, n = 4).3 There was a chondroitin sulfate content of a group of experiments in dose-dependent increase in catabolism (tl,zbasal/tll,exwhich tissues were maintained for 7-9 days under various perimental) in the presence of retinoic acid, at concen- treatments, and also supports the metabolic studies. trations ranging from 1 X 10-l’ M to the maximum at 1 X 10e6M. The EDso was at -1 X lo-* M (Fig. 2A). B. Synthesis of Other Proteins Addition of TGF-/I to retinoic-acid-containing cultures The following set of experiments were undertaken to again antagonized the effect of the retinoid, decreasing (a) determine whether the effects of retinoic acid on procatabolic rates (Fig. 2A). This is also illustrated for a longteoglycan metabolism were due to a generalized metabolic term experiment in Fig. 2B (14 days of chase shown). effect, (b) assess interactions of retinoids and TGF-P at Following 14 days of chase in this experiment, only 5% the level of overall rates of protein synthesis, and (c) define of the incorporated matrix 35Sactivity remained in the specific effects of the two effecters on particular proteins in order to increase understanding of their overall impact 3 The basal rates for this set of cultures were much lower than those on the molecular architecture of the tissue. of the tissues previously examined (t,,, = 40 days vs 6.5 days). Possibly, 1. Anabolic rates. The cartilage organ cultures were this is related to the developmental age and/or to uncertainties about the breed of bovines. treated as indicated above (proteoglycan anabolism) ex-
MORALES
AND
ROBERTS TABLE
I
Total Chondroitin Sulfate Tissue Content Retinoic
Acid Alone
Experiment 1
Sample
2
3
4
Average f SE
/?
& 0
R”nokA;zE Basal TGF$l TGF-82 TGF-/33 Retinoic acid 1 X lo-” M 1 X lo+ M 1 X lo-* M 1 X lo-' M 1 X lo-* M + 1 X lo-’ M + 1 X lo-’ M + 1 X lo-’ M + 1 X lo-' M +
10-g 10” 10-7 10-E 10-10 Molar Concentration of All-Tram Retinoic Acid
% of basal 100 91 105 100
100 99 96 99
100 93 80 102 a7
87
81 @2 /33 ,81 p2
100 112
83 64 51 103 95 88
25
84 24 72 76 65 57 56
76 68 N.D.
31 94
99+ 8 94 + 10 lOOk 1 a7 83 51 f 38 f 86 + 80 f 76+ 57 56
24 14 13 11 11
Note. Cultures were treated for 7-9 days as indicated. Chondroitin sulfate assays of guanidine HCI and proteinase K extracts were as described (17). Values were normalized to the collagen content of the tissue. Similar trends were observed when hexuronic acid (19) was measured (not shown). N.D., not determined. I,,
1
2
3
I
I
4
5
I
I
I
I
6 7 8 9 Days in Culture
I
10
I
I
11 12
I
I
13
14
FIG. 2. The effect of retinoic acid and TGF-8 on proteoglycan catabolism. Articular cartilage slices were labeled with [35S]sulfate as indicated under Methods and then chased in the presence or absence of effecters as indicated. The tllz was defined as the time required for 50% of the radiolabel to be released from the matrix (Methods). A shows the fold increase in catabolism and the vertical lines represent the range for the average of typical experiments. The TGF-Bs were added at the maximal concentration shown to elicit maximal effects (10 rig/ml of TGF-/Is 1 and 2 and 1.5 rig/ml of TGF-03). B exemplifies the kinetics of daily release for samples treated with 1 X 10-s M retinoic acid and/ or TGF-@l (10 rig/ml).
on the chondrocytes at the concentrations tested and is in agreement with previous observations that following treatment of bovine explants with retinoic acid, its removal from the treatment medium relieves the inhibition
p= 3
3 2
cept that all samples were double labeled with [3H]serine and [35S]sulfate (Methods). Control experiments, in which proteoglycans were separated from other proteins by DEAE chromatography, showed that in all cases 40% of the 3H label was incorporated into proteoglycans.4 The effect of treatment protocols on the total 3H-protein synthesis is shown in Fig. 3. Retinoic acid, at concentrations between 1 X lo-” and 1 X 10d7 M, did not change levels of protein synthesis as compared to basal cultures. Separate experiments showed that incorporation of another radiolabel, [35S]cysteine, into protein was not inhibited by 1 X 10e7 M retinoic acid (not shown). These results suggest that the retinoid does not have a cytotoxic effect ’ There was a -70% inhibition of 3H-proteoglycan synthesis by retinoic acid as determined in the guanidine HCl extracts, in reasonable agreement with a -90% inhibition of 36S-glycosaminoglycan synthesis.
.c .g
2@)-
2 E $
9
looRetmoic’Acid I
Basal
I
lo-‘0
Alone 1
I
10-s
10-7
T&3 Alone Molar Concentration of All-Trans Retinoic Acid
FIG. 3. The effect of retinoic acid and TGF-@ on protein synthesis. Cartilage organ cultures were labeled with [3H]serine and [?S]sulfate (Methods). Incorporation of ‘H label into macromolecular components was used to estimate protein synthesis (Methods). The vertical bars represent the range for duplicate experiments. The TGF-6s were added as indicated in Fig. 2.
RETINOIC
ACID
AND
TRANSFORMING
GROWTH
of proteoglycan synthesis and restores anabolic levels to those of control cultures (20, 21). TGF-0s 1 and 2 increased 3H-protein synthesis - threefold in the presence or absence of 1 X 10-l’ M retinoic acid (Fig. 3); TGF-P3 increased synthesis - twofold with or without the retinoid (not shown). At higher concentrations, retinoic acid again antagonized the effect of the TGF-8 isoforms on protein synthesis. 2. SDS-PAGE of 3H-proteins. In general, the pattern of major proteins synthesized following stimulation by TGF-fls 1,2, and 3 was comparable to the pattern of proteins synthesized on the day of tissue explantation (Day 0) and on the corresponding day of culture in basal samples (Fig. 4). However, it is worth noting the following exceptions: (a) a protein of -200 kDa was synthesized on Day 0 but was not detected in basal cultures. TGF-/3 treatment increased the relative prominence of this protein. (b) A second protein of -100 kDa was stimulated over other proteins following treatment with TGF-& Neither of these proteins was sensitive to collagenase treatment (not shown). Proteins of >lOO kDa were not prominent in extracts of cultures treated with retinoic acid alone, but following treatment with both retinoic acid and TGF-/3, accumulation of the large components was restored. C. Induction
of TGF-(32 Synthesis
by Retinoic Acid
Figure 5, top, shows the fluorogram of immunoprecipitates obtained using anti-TGF-01 antibodies (Methods). Synthesis of TGF+l in basal and retinoic-acid-treated
FIG. 4. Pattern of proteins synthesized under retinoic acid and/or TGF-fl treatment. Cartilage organ cultures were treated as indicated for the experiment shown in Fig. 3 and under Methods. The ‘H-labeled proteins from the guanidine HCl extracts of each sample were purified by DEAE chromatography, concentrated by speed vacuum centrifugation, and applied on 4-20% SDS minigels (Methods). Electrophoresis was run under reducing conditions. Approximately 30,000 dpm of ‘H activity was applied per lane. * Shows the proteins of -200 and -100 kDa that were consistently increased in TGF-b-treated cultures.
FACTORS-B
IN CALF
ARTICULAR
83
CARTILAGE
TGF-fil TGF-/32
I BWi.4
I + Retinoic Basal Acid Blocked
I Basal
I ‘251-TGF-P + Retinoic
w-5
Day 2
Day 7
FIG. 6. SDS-PAGE of immunoprecipitates of cartilage extracts using antibodies against TBF-0. Aliquots containing lo-15 million dpm of [?S]cysteine were used for immunoprecipitation of the guanidine extracts of the tissue. Equivalent portions of each immunoprecipitate (Methods) were electrophoresed under nonreducing conditions on lo-20% SDS minigels. The first lane shows molecular weight markers, with the major bands from top to bottom being ovalbumin (43 kDa) and carbonic anhydrase (29 kDa).
samples was comparable on corresponding days. On the other hand, synthesis of TGF-/32 was significantly increased in cultures treated with retinoic acid for 2 and 7 days of culture, -8- and 25-fold, respectively, as determined by densitometry (band area/mg tissue [wet weight]) (this method was validated for cartilage extracts in Ref. (9)). The increase in TGF-/32 expression was also observed when the conditioned culture media were analyzed (not shown). This indicates a selective increase in the induction of TGF-fi2 by retinoic acid. DISCUSSION In this paper we demonstrate that physiological levels of retinoic acid (1 X lOpa to 1 X 10-l’ M) have significant effects on both the anabolism and the catabolism of proteoglycans in articular cartilage. Further, we show for the first time a powerful antagonism between the TGF-fls and retinoic acid. This antagonism occurred at several levels: (a) proteoglycan anabolism, (b) proteoglycan catabolism, and (c) general protein synthesis. The pattern of labeled proteins produced in the presence of retinoic acid was deficient in large proteins (> -100 kDa), but culture with retinoic acid and TGF-0 led to their restoration, The calf articular cartilage organ cultures examined in this study synthesize two major proteoglycan classes under basal conditions: approximately 80% of the total proteoglycan is of the large molecular weight, chondroitin sulfatelkeratan sulfate aggrecan species and the rest is a population of smaller, interstitial proteoglycans which includes decorin and biglycan (8). TGF-01 stimulates synthesis of aggrecans and interstitial proteoglycans (8), with biglycan being selectively stimulated in the latter population (Morales, unpublished observations). The assembly and role of aggrecans in cartilage is well defined; many proteoglycan monomers bind firmly to a central strand of hyaluronan, forming polyanionic complexes which determine the resiliency of the tissue. Interestingly,
84
MORALES
AND
TGF-@l increases synthesis of these stable complexes, while retinoic acid, at 1 X 1O-8M, completely suppresses synthesis of aggrecans (Morales, unpublished observations). The question as to the overall action of the two effecters, separately and combined, on the precise structure of the collagen-proteoglycan matrix is worthy of detailed examination and will be published separately. The finding that retinoic acid treatment induces synthesis of TGF-/32 is consistent with observations of this effect in epithelial cells and tissues (10, 11). In primary cultures of isolated keratinocytes, retinoic acid induced a -loo-fold increase in TGF+2; effects of retinoic acid on inhibiting cell growth were shown to be mediated, in part by TGF-P (10). In cartilage, the effects are more complex, since the retinoic-acid-induced TGF-82 does not have the same action on the tissue as exogenous TGF-82. At present, it is not possible to assessthe amount of newly synthesized TGF-P2 from the incorporated radioactivity in the immunoprecipitates, and thus the possibility that the levels of TGF-P2 induced by retinoic acid are too low to have a significant effect on the system cannot be ruled out. Further, articular cartilage contains large stores of inactive TGF-fl (9), of which -1540% is TGF-82, and it is possible that the newly synthesized TGF-@2 enters this pool. While there are many plausible explanations for the relative ineffectiveness of the retinoic-acid-induced TGF-@2 that require further analysis, it is noteworthy that the present results demonstrate the existence of an intrinsic metabolic link between these two effector systems in articular cartilage. Our observation that physiological levels of retinoic acid counteract the effects of a polypeptide effector, TGF-8, believed to be critically important to cartilage function, coupled to the recent demonstration of nuclear receptors for retinoic acid (5,6), points to the potential importance of retinoids in the pathophysiology of articular cartilage. Studies of the complement of retinoic acid receptors in articular cartilage are now critical, as are efforts to define whether isoprenoid antagonists can be modeled and effectively used to diminish cartilage resorption under experimental and/or pathological conditions. ACKNOWLEDGMENTS We thank Drs. Michael B. Sporn (Laboratory of Chemoprevention, NCI, NIH) and L. Stefan Lohmander (University of Lund, Lund,
ROBERTS Sweden) for their review of this paper and their valuable comments. T.M. acknowledges the careful technical assistance of Mrs. Kathryn Smith. Thanks also to my editorial assistant, Ms. Kelly DeGraff, for her competent and proficient help in preparing the manuscript.
REFERENCES 1. Mankin, H. J., Brandt, K. D., and Shulman, L. E. (Eds.) (1986) J. Rheumatol. 13, 1126-1160. 2. Morales, T. I., and Hascall, V. C. (1989) Arthr. and Rheum. 32, 1197-1201. 3. Caputo, C. B., Sygowski, L. A., Wolanin, D. J., Patton, S. P., Caccese, R. G., Shaw, A., Roberts, R. A., and DiPasquale, G. (1987) J. Phurmucol. Exp. Ther. 240,460-465. 4. Campbell, M. A., and Handley, C. J. (1987) Arch. Biochem. Biophys. 268,143-l%. 5. Wolf, G. (1990) J. N&r. Biochem. 1, 284-289. 6. Sporn, M. B., and Roberta, A. B. (1991) Mol. Endocrirwl.
5, l-7.
7. Morales, T. I., and Roberts, A. B. (1988) J. BioL Chem. 263,12,82812,831. 8. Morales, T. I. (1991) Arch. Biochem. Biophys. 286,99-106. 9. Morales, T. I., Joyce, M. E., Soble, M. E., Danielpour, D., and Roberts, A. B. (1991) Arch. Biochem. Biophys. 288,397-405. 10. Glick, A. B., Flanders, K. C., Danielpour, D., Yuspa, S. H., and Sporn, M. B. (1989) Cell Regul. 1,87-89. 11. Glick, A. B., McCune, B. K., Abdulkaren, N., Flanders, K. C., Lamadue, J. A., Smith, J. M., and Sporn, M. B. (1991) Development
111,1081-1085. 12. Hascall, V. C., Handley, C. J., Mcquillan, D. J., Hascall, G. K., Robinson, H. C., and Lowther, D. A. (1983) Arch. Bbchem. Biophys. 224,206223. 13. Morales, T. I., Wahl, L. M., and Hascall, V. C. (1984) J. Bill. Chem. 259,6720-6729. 14. Roberta, A. B., Kondaiah, P., Rosa, F., Watanabe, S., Good, P., Danielpour, D., Roche, N. S., Rebbert, M. L., Dawid, I. B., and Sporn, M. B. (1990) Growth Factors 3, 277-286. 15. Morales, T. I., and Hascall, V. C. (1989) Conn. Tiss. Res. 19, 255275. 16. Woessner, J. F. (1961) Arch. B&hem. Biophys. 93,440-447. 17. Farndale, R. W., Sayer, C. S., and Barrett, Res. 9.247-248.
A. J. (1982) Corm. Tiss.
18. Robey, P. G., Young, M. F., Flanders, K. C., Roche, N. S., Kondarah, P., Reddi, A. H., Termine, J. D., Sporn, M. B., and Roberts, A. B. (1987) J. Cell Biol. 106, 457-463. 19. Bitter,
T., and Muir, H. (1962) Anal. Biochem. 4, 330-334.
20. Vosburgh, F., and Hascall, V. C. (1984) Ann. N. Y. Acad. Sci., 157159. 21. Campbell, M. A., and Handley, C. J. (1987) Arch. B&hem. Biophys. 263,462-474.
OF BIOCHEMISTRY
Vol. 293, No. 1, February
AND
BIOPHYSICS
14, pp. 79-84,1992
The Interaction between Retinoic Acid and the Transforming Growth Factors-P in Calf Articular Cartilage Organ Cultures Teresa
1. Morales1
and Anita
B. Roberts
Bone Research Branch, National Institute of Dental Research and the Laboratory of Chemopreuention, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892
Received September 4, 1991, and in revised form October 18, 1991
In calf articular cartilage organ cultures, retinoic acid depressed proteoglycan anabolism to levels -10% of control values and increased their catabolism - 14-fold at concentrations of 1 X lo-’ and 1 X lo-’ M, respectively, leading to a severe depletion of this component from the extracellular matrix (95% loss in 3 weeks). These effects were powerfully antagonized by maximal levels of transforming growth factors+ (TGF-Bs) 1, 2, and 3, leading to preservation of matrix components. At a concentration of 1 X lOma M retinoic acid, the TGF-&I restored anabolism to control levels and lowered catabolic rates >3-fold. While the TGF-@s increased protein synthesis 2- to 3-fold over controls, retinoic acid alone did not change protein synthesis, as determined by incorporation of 13H]serine. Nevertheless, retinoic acid effectively antagonized the stimulation of protein synthesis by TGF+3 and restored control levels of synthesis at 1 X lo-’ M. Analysis of proteins, labeled using [3H]serine and [35S]sulfate as precursors, by SDS-PAGE revealed that large molecular weight proteins (>lOO kDa) were not detectable in retinoic-acid-treated cultures, but treatment with the TGF-@s restored these components in coincubation cultures, again supporting the antagonistic role of the polypeptide effecters on retinoid action. Treatment of the cultures with retinoic acid elevated levels of TGF-fl2 synthesis, but not TGF+l. While the role of the newly synthesized TGF-/32 in the set of events elicited by retinoic acid in articular cartilage is unclear, the results establish an intrinsic metabolic link between the isoprenoid and TGF-/3 in articular cartilage. We propose that the retinoids and TGF-@s are integral parts of a regulatory network that controls homeostasis, resorption, or growth, depending on their relative contributions. Q 1992 Academic Press, Inc.
’ To whom correspondence
should be addressed.
0003.9861/92 $3.00 Copyright 0 1992 by Academic Press, All rights of reproduction in any form
The uncoupling of anabolic and catabolic pathways in articular cartilage is characteristic of degenerative joint disease, or osteoarthritis (1). While it is well documented that progressive loss of matrix proteoglycans (aggrecans) from articular cartilage correlates with disease severity and with loss of the resilient, protective function of this tissue on the subchondral bone, the regulation of metabolic homeostasis in articular cartilage is incompletely understood (2). Previous work with articular cartilage organ cultures has established that a variety of effecters, added exogenously, can modulate the homeostatic balance of the tissue (2). Among these, excessive levels of vitamin A (retinol or retinoic acid) lead to progressive loss of aggrecans from the extracellular matrix (3,4). Interest in this finding has been rekindled with the discovery of specific nuclear binding proteins for retinoic acid, which interact with regulatory sequences in target genes to control their transcription (for reviews, see Refs. (5,6)). As in other tissues, the regulation of retinoid activity may be a vital component in the intrinsic metabolic regulation of cartilage matrix physiology. In contrast to the catabolic effects of retinoids on artitular cartilage, addition of transforming growth factor/3 (TGF$Q2 to cartilage organ cultures increases synthesis of aggrecans and diminishes their loss, leading to maintenance of the tissue’s homeostatic balance (7-9). Recent data suggest that in the epithelium, the action of these two agents, one an isoprenoid which acts through a set of nuclear receptors and the other a secreted peptide signaling through a set of cell membrane receptors, might be linked (10, 11). ’ Abbreviations used: TGF-/3, transforming growth factor-& DMEM, Dulbecco’s modified Eagle’s medium; Chaps, 3-[(3cholamidopropyl)dimethylammonio]propanesulfonic acid; PMSF, phenylmethylsulfonyl fluoride; o-phe, o-phenanthroline; N-EM, N-ethylmaleimide; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis. 79
Inc. reserved.
80
MORALES
AND
The present investigation tests the effect of physiological levels of retinoic acid on the articular cartilage matrix and explores the putative relationships between retinoic acid and TGF-P in this tissue. We present evidence that (a) physiological levels of retinoic acid induce pronounced changes in the proteoglycan matrix without changing overall levels of protein synthesis, (b) this effect is antagonized by the closely related TGF-/3 isoforms, TGF@l, TGF-@2, and TGF-P3, and (c) retinoic acid selectively induces synthesis of TGF-P2 but not TGF-@l in articular cartilage. This establishes for the first time a metabolic link between retinoic acid and TGF-P in a skeletal tissue. We suggest that these two agents are integral parts of a regulatory network that controls homeostasis, growth, or resorption in articular cartilage, depending on their relative contributions. METHODS Tissue Culture and Labeling Procedures Articular cartilage was dissected from the metacarpalphalangeal joints of calves -4-6 months old (slaughterhouse estimate) and placed in organ culture (12,13). Briefly, for each experiment, cartilage slices were pooled from the two joints of one calf, distributed into capped culture vials, and cultured in the presence of 15 vol (ml/g wet weight tissue) of Dulhecco’s modified Eagle’s medium (DMEM) containing 4.5 g/liter glucose and 0.58 g/liter L-glutamine for 7-10 days. The tissue was distributed in portions of -100-150 mg wet weight into individual wells of 24-well Costar plates for the initiation of experimental treatments All-trans retinoic acid was from Hoffmann LaRoche Inc. or Sigma Chemical Co. Porcine TGF-j31 and TGF-82 were from R & D Systems. Recombinant chicken TGF-03 (14) was purified to homogeneity by three successive HPLC steps consisting of Cl8 reverse-phase, TSK-CM ionexchange, and Cl4 reverse-phase chromatography. Sixty-fold concentrated solutions of either one of the TGF-@s or retinoic acid were added to the appropriate experimental wells containing basal medium (DMEM + 0.1% bovine serum albumin, Sigma Chemical Co., type V, crystalline). The experimental groups were as follows: (a) control cultures were maintained in basal medium, (b) retinoic-acid-treated samples were cultured in basal medium + the indicated molar concentration of all-trans retinoic acid, lo-“’ to 10e6 M, (c) TGF-@-treated samples were maintained in basal medium + saturating levels of either TGF-/31(10 rig/ml), TGFj32 (10 rig/ml), or TGF-j33 (1.5 rig/ml), and (d) retinoic acid + TGF-P cultures were maintained in basal medium containing these two effecters at the maximal concentration of TGF-fi and the indicated concentration of retinoic acid. For all reported results, n represents the number of experiments using tissues from different animals.
Measurements
of Proteoglycan
Metabolism
Methods have been described and validated (12, 13). Briefly, [36S]sulfate (20 &i/ml) was used to pulse the tissue and selectively label the glycosaminoglycans. For anabolic measurements, the glycosaminoglycans were extracted with alkali (0.5 N NaOH) following the pulse, and incorporated radiolabel was assessed following chromatography on Pharmacia PDlO columns (12, 13). For catabolic measurements, the %S-labeled tissue was washed in basal medium for 2 days (37°C) to remove unincorporated isotope and chased for at least 7 days under the appropriate treatment conditions. Medium was collected and changed daily. The total ?S activity in each sample was assessed by adding the cumulative daily release into the medium (following the wash) and the residual matrix activity (NaOH extract); the % of the total ?!l-glycosaminoglycan released daily was then determined.
ROBERTS
General Protein and Proteoglycan Synthesis
Core Protein
1. Protein synthesis. Experiments were as described above with the following exceptions. The labeling medium contained 20 &i/ml [35S]sulfate (carrier-free) + 60 &i/ml [3H]serine (92 Ci/mmol; both precursors were from ICN Biomedicals). Intact proteoglycans and other proteins were extracted from the tissue in chaotropic buffer (15) (0.1 M Tris maleate, pH 5.8, containing 4 M guanidine HCl, 0.5% 3-[(3-cholamidopropyl)dimethylammonio]propanesulfonic acid (Chaps), and the following proteinase inhibitors: 7 ag/ml each of pepstatin A and leupeptin, 1 mM phenylmethylsulfonyl fluoride (PMSF), 5 mM iv-ethylmaleimide (N-EM), and 1 mM o-phenanthroline (o-phe)). The residue was dispersed by proteinase K treatment (Sigma Chemical Co., 1 mg/ ml, 60°C -18 h). The incorporated ?S and 3H activity was assessed as described above. Protein synthesis was expressed as the incorporated 3H activity in the conditioned medium + guanidine HCl extract t proteinase K extract of each sample. For control experiments, portions of the 4 M guanidine HCl extracts and of the conditioned media were dialyzed against dilute urea buffer (6 M) at 4’C and applied on DEAE-Fast Flow columns (Sigma Chemical Co.), -500 pg chondroitin sulfate/ml gel. These DEAE columns were equilibrated and eluted with urea buffer (0.05 M sodium acetate buffer, pH 6.0, containing 8 M urea, 0.3 M NaCI, and proteinase inhibitors at l/10 the concentration described above). Proteoglycans, selectively retained by these columns, were eluted by 1.0 M NaCl in urea buffer (15). The collagen content of the tissue was stable during all the experimental treatments (not shown, 7) and was therefore used as a normalization factor for all anabolic measurements. Collagen (hydroxyproline) and chondroitin sulfate were assayed as described (13, 16, 17). 2. Electrophoresis of 3H-labeledproteins. 3H-proteins in the 4 M guanidine HCl extracts (m-70-80% of the total in the tissue and conditioned medium) were fractionated by chromatography on DEAE columns as indicated above. The 0.3 M NaCl eluenta were extensively dialyzed against 4 mM HCl containing 0.1 pg/ml each of pepstatin A and leupeptin and 0.1 mM each of o-phe and PMSF (4°C). Portions of the dialyzates were dried in the speed vacuum centrifuge, resuspended in Tris-SDS buffer (Seprasol I, Enprotech Co.) containing 0.7 M @-mercaptoethanol and boiled for 5 min. SDS-PAGE was carried out on 4-20% minigels (Enprotech Co.) and fluorography was performed using Kodak X-omat AR film.
Synthesis of TGF-@I and TGF-@2 Cartilage cultures were treated as indicated and then labeled with 250 &i/ml [35S]cysteine (ICN Biomedicals, Inc., 960 Ci/mmol) in basal medium with or without retinoic acid for -16 h. The immunoprecipitation procedures have been described (9, 18). Briefly, portions of each conditioned medium and 4 M guanidine extract of the tissue were immunoprecipitated using specific antibodies against TGF-@l or TGF-j32. Immunoprecipitates were dissolved in 100 ~1 Seprasol I by boiling for 5 min and subjected to SDS-PAGE.
RESULTS A. Proteoglycan
Metabolism
Articular cartilage organ cultures maintained under basal conditions (without serum and/or added growth factors) decreased their rates of synthesis of proteoglycans to a relatively stable background during the first week in culture (not shown, 9, 12), but remained phenotypically stable (8,12). The experimental treatments were therefore initiated following the first week of culture (Methods). Treatments were for 7-9 days, since previous studies had indicated a lag time of 6 days for expression of the full
RETINOIC
ACID
AND
TRANSFORMING
GROWTH
FACTORS-/l
IN CALF
ARTICULAR
CARTILAGE
81
effect of TGF-fl on proteoglycan biosynthesis. However, 330 t -90% of the maximal response to retinoic acid was observed within 2 days (not shown). 250 L 1. Proteoglycan anubolism. Figure 1 illustrates the dose-response curve for all-trans retinoic acid added alone to the basal cultures and in combination with TGF-fil and TGF-02. The data show that retinoic acid is a powerful inhibitor of proteoglycan synthesis, lowering rates to - 10% of basal levels at the saturating dose of 1 X lo-* M, with an ED5,, of 1 X lo-” M. By contrast, in previous work we demonstrated a dosedependent, equipotent ability of TGF-Pl and TGF-/32 to increase levels of proteoglycan synthesis, with saturation at 5-10 rig/ml (7-9). In the present set of experiments, the average stimulation of proteoglycan synthesis by maximal levels of TGF-fis 1 and 2 was 290 t- 56% and 270 + 67% (&SE, n = 6 and n = 5, respectively; n = number of experiments using pooled tissues from different I I I , I animals). We have now extended the analysis of the TGFI Basal 10-10 10-9 10-8 10.’ fl isoforms to include TGF-63 in selected experiments. This isoform increased synthesis of proteoglycans, an average of 190 f 36% (&SE, n = 3) at 1.5 rig/ml, found to Molar Concentrations of All-Trans Retinoic Acid be a saturating concentration for this isoform (EDso = FIG. 1. The effect of retinoic acid and TGF-P on proteoglycan syn0.5 rig/ml). Addition of maximal levels of either TGF-Pl or TGF- thesis in calf articular cartilage organ cultures. The vertical bars on data point represent the standard error of the mean for three to fl2 to cultures containing retinoic acid antagonized the each four representative experiments using tissue from different animals, or inhibitory effect of the retinoid. At concentrations of ret- the range for two experiments. TGF-@s 1 and 2 were added to a final inoic acid of 1 X lOmaM or less, the TGF-@s maintained concentration of 10 rig/ml. synthetic rates at levels above basal controls (Fig. 1). At concentrations of retinoic acid greater than 1 X lo-’ M, cultures treated with 1 X lop8 M retinoic acid. The total the TGF-0s reduced the inhibitory effect of the retinoid chondroitin sulfate mass left in the tissue at this time but the overall synthetic rate was suppressed compared was 14% of initial values. This is consistent with the seto basal levels. TGF-83 also antagonized the effect of retvere catabolic depletion from the tissue and extremely inoic acid (1 X lo-’ M retinoic acid + 1.5 rig/ml TGF-83 low anabolic rates. The cultures treated with 1 X 10-s M = 80 -t 39% of basal level (&SE, n = 3), compared to 13 retinoic acid + TGF-01 (10 rig/ml) retained 55% of the + 5% with retinoic acid alone in the same set of experi?S activity and 68% of the chondroitin sulfate, a finding ments). consistent with the partial antagonism by TGF-P of the 2. Proteoglycan catabolism. The average catabolic effect of this concentration of retinoic acid on both carates of basal cultures were low (the tlj2, i.e., time for the tabolism and anabolism of proteoglycans. Similar trends release of 50% of the labeled proteoglycans from the were observed in a longer experiment. Table I shows the matrix, was 40 -t 21 days (&SE, n = 4).3 There was a chondroitin sulfate content of a group of experiments in dose-dependent increase in catabolism (tl,zbasal/tll,exwhich tissues were maintained for 7-9 days under various perimental) in the presence of retinoic acid, at concen- treatments, and also supports the metabolic studies. trations ranging from 1 X 10-l’ M to the maximum at 1 X 10e6M. The EDso was at -1 X lo-* M (Fig. 2A). B. Synthesis of Other Proteins Addition of TGF-/I to retinoic-acid-containing cultures The following set of experiments were undertaken to again antagonized the effect of the retinoid, decreasing (a) determine whether the effects of retinoic acid on procatabolic rates (Fig. 2A). This is also illustrated for a longteoglycan metabolism were due to a generalized metabolic term experiment in Fig. 2B (14 days of chase shown). effect, (b) assess interactions of retinoids and TGF-P at Following 14 days of chase in this experiment, only 5% the level of overall rates of protein synthesis, and (c) define of the incorporated matrix 35Sactivity remained in the specific effects of the two effecters on particular proteins in order to increase understanding of their overall impact 3 The basal rates for this set of cultures were much lower than those on the molecular architecture of the tissue. of the tissues previously examined (t,,, = 40 days vs 6.5 days). Possibly, 1. Anabolic rates. The cartilage organ cultures were this is related to the developmental age and/or to uncertainties about the breed of bovines. treated as indicated above (proteoglycan anabolism) ex-
MORALES
AND
ROBERTS TABLE
I
Total Chondroitin Sulfate Tissue Content Retinoic
Acid Alone
Experiment 1
Sample
2
3
4
Average f SE
/?
& 0
R”nokA;zE Basal TGF$l TGF-82 TGF-/33 Retinoic acid 1 X lo-” M 1 X lo+ M 1 X lo-* M 1 X lo-' M 1 X lo-* M + 1 X lo-’ M + 1 X lo-’ M + 1 X lo-’ M + 1 X lo-' M +
10-g 10” 10-7 10-E 10-10 Molar Concentration of All-Tram Retinoic Acid
% of basal 100 91 105 100
100 99 96 99
100 93 80 102 a7
87
81 @2 /33 ,81 p2
100 112
83 64 51 103 95 88
25
84 24 72 76 65 57 56
76 68 N.D.
31 94
99+ 8 94 + 10 lOOk 1 a7 83 51 f 38 f 86 + 80 f 76+ 57 56
24 14 13 11 11
Note. Cultures were treated for 7-9 days as indicated. Chondroitin sulfate assays of guanidine HCI and proteinase K extracts were as described (17). Values were normalized to the collagen content of the tissue. Similar trends were observed when hexuronic acid (19) was measured (not shown). N.D., not determined. I,,
1
2
3
I
I
4
5
I
I
I
I
6 7 8 9 Days in Culture
I
10
I
I
11 12
I
I
13
14
FIG. 2. The effect of retinoic acid and TGF-8 on proteoglycan catabolism. Articular cartilage slices were labeled with [35S]sulfate as indicated under Methods and then chased in the presence or absence of effecters as indicated. The tllz was defined as the time required for 50% of the radiolabel to be released from the matrix (Methods). A shows the fold increase in catabolism and the vertical lines represent the range for the average of typical experiments. The TGF-Bs were added at the maximal concentration shown to elicit maximal effects (10 rig/ml of TGF-/Is 1 and 2 and 1.5 rig/ml of TGF-03). B exemplifies the kinetics of daily release for samples treated with 1 X 10-s M retinoic acid and/ or TGF-@l (10 rig/ml).
on the chondrocytes at the concentrations tested and is in agreement with previous observations that following treatment of bovine explants with retinoic acid, its removal from the treatment medium relieves the inhibition
p= 3
3 2
cept that all samples were double labeled with [3H]serine and [35S]sulfate (Methods). Control experiments, in which proteoglycans were separated from other proteins by DEAE chromatography, showed that in all cases 40% of the 3H label was incorporated into proteoglycans.4 The effect of treatment protocols on the total 3H-protein synthesis is shown in Fig. 3. Retinoic acid, at concentrations between 1 X lo-” and 1 X 10d7 M, did not change levels of protein synthesis as compared to basal cultures. Separate experiments showed that incorporation of another radiolabel, [35S]cysteine, into protein was not inhibited by 1 X 10e7 M retinoic acid (not shown). These results suggest that the retinoid does not have a cytotoxic effect ’ There was a -70% inhibition of 3H-proteoglycan synthesis by retinoic acid as determined in the guanidine HCl extracts, in reasonable agreement with a -90% inhibition of 36S-glycosaminoglycan synthesis.
.c .g
2@)-
2 E $
9
looRetmoic’Acid I
Basal
I
lo-‘0
Alone 1
I
10-s
10-7
T&3 Alone Molar Concentration of All-Trans Retinoic Acid
FIG. 3. The effect of retinoic acid and TGF-@ on protein synthesis. Cartilage organ cultures were labeled with [3H]serine and [?S]sulfate (Methods). Incorporation of ‘H label into macromolecular components was used to estimate protein synthesis (Methods). The vertical bars represent the range for duplicate experiments. The TGF-6s were added as indicated in Fig. 2.
RETINOIC
ACID
AND
TRANSFORMING
GROWTH
of proteoglycan synthesis and restores anabolic levels to those of control cultures (20, 21). TGF-0s 1 and 2 increased 3H-protein synthesis - threefold in the presence or absence of 1 X 10-l’ M retinoic acid (Fig. 3); TGF-P3 increased synthesis - twofold with or without the retinoid (not shown). At higher concentrations, retinoic acid again antagonized the effect of the TGF-8 isoforms on protein synthesis. 2. SDS-PAGE of 3H-proteins. In general, the pattern of major proteins synthesized following stimulation by TGF-fls 1,2, and 3 was comparable to the pattern of proteins synthesized on the day of tissue explantation (Day 0) and on the corresponding day of culture in basal samples (Fig. 4). However, it is worth noting the following exceptions: (a) a protein of -200 kDa was synthesized on Day 0 but was not detected in basal cultures. TGF-/3 treatment increased the relative prominence of this protein. (b) A second protein of -100 kDa was stimulated over other proteins following treatment with TGF-& Neither of these proteins was sensitive to collagenase treatment (not shown). Proteins of >lOO kDa were not prominent in extracts of cultures treated with retinoic acid alone, but following treatment with both retinoic acid and TGF-/3, accumulation of the large components was restored. C. Induction
of TGF-(32 Synthesis
by Retinoic Acid
Figure 5, top, shows the fluorogram of immunoprecipitates obtained using anti-TGF-01 antibodies (Methods). Synthesis of TGF+l in basal and retinoic-acid-treated
FIG. 4. Pattern of proteins synthesized under retinoic acid and/or TGF-fl treatment. Cartilage organ cultures were treated as indicated for the experiment shown in Fig. 3 and under Methods. The ‘H-labeled proteins from the guanidine HCl extracts of each sample were purified by DEAE chromatography, concentrated by speed vacuum centrifugation, and applied on 4-20% SDS minigels (Methods). Electrophoresis was run under reducing conditions. Approximately 30,000 dpm of ‘H activity was applied per lane. * Shows the proteins of -200 and -100 kDa that were consistently increased in TGF-b-treated cultures.
FACTORS-B
IN CALF
ARTICULAR
83
CARTILAGE
TGF-fil TGF-/32
I BWi.4
I + Retinoic Basal Acid Blocked
I Basal
I ‘251-TGF-P + Retinoic
w-5
Day 2
Day 7
FIG. 6. SDS-PAGE of immunoprecipitates of cartilage extracts using antibodies against TBF-0. Aliquots containing lo-15 million dpm of [?S]cysteine were used for immunoprecipitation of the guanidine extracts of the tissue. Equivalent portions of each immunoprecipitate (Methods) were electrophoresed under nonreducing conditions on lo-20% SDS minigels. The first lane shows molecular weight markers, with the major bands from top to bottom being ovalbumin (43 kDa) and carbonic anhydrase (29 kDa).
samples was comparable on corresponding days. On the other hand, synthesis of TGF-/32 was significantly increased in cultures treated with retinoic acid for 2 and 7 days of culture, -8- and 25-fold, respectively, as determined by densitometry (band area/mg tissue [wet weight]) (this method was validated for cartilage extracts in Ref. (9)). The increase in TGF-/32 expression was also observed when the conditioned culture media were analyzed (not shown). This indicates a selective increase in the induction of TGF-fi2 by retinoic acid. DISCUSSION In this paper we demonstrate that physiological levels of retinoic acid (1 X lOpa to 1 X 10-l’ M) have significant effects on both the anabolism and the catabolism of proteoglycans in articular cartilage. Further, we show for the first time a powerful antagonism between the TGF-fls and retinoic acid. This antagonism occurred at several levels: (a) proteoglycan anabolism, (b) proteoglycan catabolism, and (c) general protein synthesis. The pattern of labeled proteins produced in the presence of retinoic acid was deficient in large proteins (> -100 kDa), but culture with retinoic acid and TGF-0 led to their restoration, The calf articular cartilage organ cultures examined in this study synthesize two major proteoglycan classes under basal conditions: approximately 80% of the total proteoglycan is of the large molecular weight, chondroitin sulfatelkeratan sulfate aggrecan species and the rest is a population of smaller, interstitial proteoglycans which includes decorin and biglycan (8). TGF-01 stimulates synthesis of aggrecans and interstitial proteoglycans (8), with biglycan being selectively stimulated in the latter population (Morales, unpublished observations). The assembly and role of aggrecans in cartilage is well defined; many proteoglycan monomers bind firmly to a central strand of hyaluronan, forming polyanionic complexes which determine the resiliency of the tissue. Interestingly,
84
MORALES
AND
TGF-@l increases synthesis of these stable complexes, while retinoic acid, at 1 X 1O-8M, completely suppresses synthesis of aggrecans (Morales, unpublished observations). The question as to the overall action of the two effecters, separately and combined, on the precise structure of the collagen-proteoglycan matrix is worthy of detailed examination and will be published separately. The finding that retinoic acid treatment induces synthesis of TGF-/32 is consistent with observations of this effect in epithelial cells and tissues (10, 11). In primary cultures of isolated keratinocytes, retinoic acid induced a -loo-fold increase in TGF+2; effects of retinoic acid on inhibiting cell growth were shown to be mediated, in part by TGF-P (10). In cartilage, the effects are more complex, since the retinoic-acid-induced TGF-82 does not have the same action on the tissue as exogenous TGF-82. At present, it is not possible to assessthe amount of newly synthesized TGF-P2 from the incorporated radioactivity in the immunoprecipitates, and thus the possibility that the levels of TGF-P2 induced by retinoic acid are too low to have a significant effect on the system cannot be ruled out. Further, articular cartilage contains large stores of inactive TGF-fl (9), of which -1540% is TGF-82, and it is possible that the newly synthesized TGF-@2 enters this pool. While there are many plausible explanations for the relative ineffectiveness of the retinoic-acid-induced TGF-@2 that require further analysis, it is noteworthy that the present results demonstrate the existence of an intrinsic metabolic link between these two effector systems in articular cartilage. Our observation that physiological levels of retinoic acid counteract the effects of a polypeptide effector, TGF-8, believed to be critically important to cartilage function, coupled to the recent demonstration of nuclear receptors for retinoic acid (5,6), points to the potential importance of retinoids in the pathophysiology of articular cartilage. Studies of the complement of retinoic acid receptors in articular cartilage are now critical, as are efforts to define whether isoprenoid antagonists can be modeled and effectively used to diminish cartilage resorption under experimental and/or pathological conditions. ACKNOWLEDGMENTS We thank Drs. Michael B. Sporn (Laboratory of Chemoprevention, NCI, NIH) and L. Stefan Lohmander (University of Lund, Lund,
ROBERTS Sweden) for their review of this paper and their valuable comments. T.M. acknowledges the careful technical assistance of Mrs. Kathryn Smith. Thanks also to my editorial assistant, Ms. Kelly DeGraff, for her competent and proficient help in preparing the manuscript.
REFERENCES 1. Mankin, H. J., Brandt, K. D., and Shulman, L. E. (Eds.) (1986) J. Rheumatol. 13, 1126-1160. 2. Morales, T. I., and Hascall, V. C. (1989) Arthr. and Rheum. 32, 1197-1201. 3. Caputo, C. B., Sygowski, L. A., Wolanin, D. J., Patton, S. P., Caccese, R. G., Shaw, A., Roberts, R. A., and DiPasquale, G. (1987) J. Phurmucol. Exp. Ther. 240,460-465. 4. Campbell, M. A., and Handley, C. J. (1987) Arch. Biochem. Biophys. 268,143-l%. 5. Wolf, G. (1990) J. N&r. Biochem. 1, 284-289. 6. Sporn, M. B., and Roberta, A. B. (1991) Mol. Endocrirwl.
5, l-7.
7. Morales, T. I., and Roberts, A. B. (1988) J. BioL Chem. 263,12,82812,831. 8. Morales, T. I. (1991) Arch. Biochem. Biophys. 286,99-106. 9. Morales, T. I., Joyce, M. E., Soble, M. E., Danielpour, D., and Roberts, A. B. (1991) Arch. Biochem. Biophys. 288,397-405. 10. Glick, A. B., Flanders, K. C., Danielpour, D., Yuspa, S. H., and Sporn, M. B. (1989) Cell Regul. 1,87-89. 11. Glick, A. B., McCune, B. K., Abdulkaren, N., Flanders, K. C., Lamadue, J. A., Smith, J. M., and Sporn, M. B. (1991) Development
111,1081-1085. 12. Hascall, V. C., Handley, C. J., Mcquillan, D. J., Hascall, G. K., Robinson, H. C., and Lowther, D. A. (1983) Arch. Bbchem. Biophys. 224,206223. 13. Morales, T. I., Wahl, L. M., and Hascall, V. C. (1984) J. Bill. Chem. 259,6720-6729. 14. Roberta, A. B., Kondaiah, P., Rosa, F., Watanabe, S., Good, P., Danielpour, D., Roche, N. S., Rebbert, M. L., Dawid, I. B., and Sporn, M. B. (1990) Growth Factors 3, 277-286. 15. Morales, T. I., and Hascall, V. C. (1989) Conn. Tiss. Res. 19, 255275. 16. Woessner, J. F. (1961) Arch. B&hem. Biophys. 93,440-447. 17. Farndale, R. W., Sayer, C. S., and Barrett, Res. 9.247-248.
A. J. (1982) Corm. Tiss.
18. Robey, P. G., Young, M. F., Flanders, K. C., Roche, N. S., Kondarah, P., Reddi, A. H., Termine, J. D., Sporn, M. B., and Roberts, A. B. (1987) J. Cell Biol. 106, 457-463. 19. Bitter,
T., and Muir, H. (1962) Anal. Biochem. 4, 330-334.
20. Vosburgh, F., and Hascall, V. C. (1984) Ann. N. Y. Acad. Sci., 157159. 21. Campbell, M. A., and Handley, C. J. (1987) Arch. B&hem. Biophys. 263,462-474.
