EXPERIMENTAL
CELL
RESEARCH
195,
38-46
(1991)
Retinoic Acid Treatment Induces Type X Collagen Gene Expression in Cultured Chick Chondrocytes MAURIZIO
PACIFICI,’
ELEANOR
of Anatomy/Histology,
Department
School
B. GOLDEN,
of Dental
Medicine,
MASAHIRO linkers&
The vitamin A derivative retinoic acid (RA) is widely thought to be involved in cartilage development, but its precise roles and mechanisms of action in this complex process remain unclear. We have tested the hypothesis that RA is involved in chondrocyte maturation during endochondral ossification and, in particular, is an inducer of maturation-associated traits such as type X collagen and alkaline phosphatase. Immature chondrocytes isolated from the caudal region of Day 19 chick embryo sterna were seeded in secondary monolayer cultures and treated either with a high dose (100 nM) or with physiological doses (lo-35 nM) of RA for up to 3 days. We found that after an initial lag of about 24 h, physiological doses of RA indeed induced type X collagen gene expression in the immature cells. This induction was not accompanied by obvious changes in expression of the type II collagen and large aggregating proteoglycan core protein genes. As revealed by immunocytochemistry, 30-35% of the cells in cultures treated with RA for 3 days were engaged in type X collagen production. Interestingly, these cells were relatively similar in size to chondrocytes in which no type X collagen was detected, suggesting that chondrocytes can initiate type X collagen production independent of cell hypertrophy. RA treatment also led to increased alkaline phosphatase activity occurring as early as 24 h after the start of treatment. The data in this study indicate that RA may have a role in endochondral ossification as an inducer/promoter of maturation-associated ( 1991 Academic traits during chondrocyte maturation. Press,
Inc.
INTRODUCTION
The vitamin A derivative retinoic acid (RA) has long been known to affect both cartilage development and function. Early studies found that high doses of retinoids given during pregnancy are teratogenic and cause severe limb and craniofacial defects involving abnormal cartilage and skeletal development [l, 21. Similar high doses block chondrogenesis in mesenchymal cell cul’ To whom
reprint
requests
should
he addressed.
0014.4827/91 $3.00 Copyright IE 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
IWAMOTO,
of Pennsyluania,
AND SHERRILL Philadelphia,
L. ADAMS
Pennsyluania
19104-6003
tures [3,4] and inhibit normal phenotypic expression in cultured chondrocytes [5,6]. Interestingly, recent studies have found that RA at physiological concentrations may actually be involved in normal cartilage development. A gradient of RA occurs naturally in the developing limb, from 20 nM at the anterior edge to 50 nM at the posterior edge [7, 81. Exogenous RA applied to the anterior limb edge matching the concentration naturally present in the posterior edge induces a mirrorimage duplication of the limb skeletal elements, thus acting as a skeletal morphogen [9, lo]. When embryonic mouse limb buds in organ culture are treated with low doses of RA, total cartilage mass increases over control values [ 111. Similar physiological doses of RA stimulate chondrogenesis in limb mesenchymal cell cultures [ 121 and proliferation in growth plate chondrocyte cultures [13]. Thus, although hypervitaminosis A is deleterious to cartilage development and function, physiological levels of RA may promote these processes. The precise roles RA plays in cartilage development and function, however, remain unclear. This issue is complicated by the fact that after its emergence during embryogenesis, cartilage undergoes a further complex developmental process of maturation as it takes part in endochondral ossification. During maturation, the newly emerged, resting, immature chondrocytes first develop into proliferating chondrocytes. Whereas the resting chondrocytes are small flat cells producing very little matrix, the proliferating chondrocytes are very active, replicate readily, and produce large quantities of matrix components such as type II collagen and the aggregating proteoglycan [141. As proliferation dwindles, the cells mature into large round hypertrophic chondrocytes, initiate production of a unique collagen, type X, and increase alkaline phosphatase activity and release of matrix vesicles. The abundant matrix surrounding the hypertrophic chondrocytes then undergoes mineralization [15-191. To assesswhether the complex process of maturation is influenced by RA, we recently carried out experiments with maturing cultures of chick embryo vertebral chondrocytes [20]. Cells isolated from Day 12 chick embryos were allowed to initiate the process of maturation by growth for about 2 weeks in permissive monolayer
RETINOIC
ACID
AND
CHONDROCYTE
culture conditions. Maturation was assessed by active cell replication, increase in average cell size, and initiation of type X collagen gene expression. Cultures were then treated with a high dose of RA (300 nn/l) and expression of five extracellular matrix genes, including the maturation-specific type X collagen gene, was assessed. We found that. expression of genes encoding typical chondrocyte macromolecules such as type II collagen and aggregating proteoglycan was nearly shut off by 48 h of RA treatment [21-231. Strikingly, this decrease was accompanied by a threefold increase in type X collagen gene expression. These differential changes in gene expression are quite similar to those normally accompanying chondrocyte maturation in viuo [ 181, thus leading us to conclude that RA may have a role in maturation, in particular, in induction of type X collagen gene expression. In this report, we have tested this hypothesis using cultures of immature chondrocytes isolated from the caudal region of chick sternum and treated with both high and physiological doses of RA.
Fn XI
39
MATURATION
-
205
-
116
c
II i IX [ x-
MATERIALS
AND
METHODS
Chondrocyte cultures. Primary cultures of immature sternal chondrocytes isolated from the caudal region of Day 19 chick embryo sternum [l’i] were prepared and grown for 6-7 days as described [24]. Medium consisted of high glucose Dulbecco’s MEM containing 10% fetal bovine serum (Hyclone Labs.), 2 mM L-glutamine, and 50 U/ml each of penicillin and streptomycin. Pure populations of floating chondrocytes harvested from the primary cultures were treated with 0.25% trypsin and replated into secondary monolayer cultures at 3 x lo”/60 mm tissue culture dish or 10 X lo6 cells/100 mm dish. Fresh 100X stock solutions of all-trans RA (Sigma) were prepared in 95% ethanol daily and used under safety light conditions. Control cultures received 10 pi/ml of 95% ethanol. Medium was changed daily. Whole cellular RNA isolated by the RNA isolation and analysis. guanidine isothiocynate method [25] was denatured by glyoxalation, electrophoresed on agarose gels, transferred to Hybond-N membranes (Amersham) by capillary blotting, and hybridized to nicktranslated 32P-labeled cDNA probes. Blots were washed at high stringency (0.1X SSPE at 65°C) and exposed to Kodak films at ~70°C for various lengths of time; quantitative analyses were carried out by video digitization [26]. When necessary, blots were dehybridized at 9O’C in 0.05X SSPE for 5 min and hybridized to a different probe. A final rehybridization was carried out using an 18s ribosomal RNA probe [27] to control for variations in RNA concentrations, gel loading, and transfer efficiency. The plasmids used were the chick type X collagen cDNA pDLrl0 [28], a 197.bp subclone of pYN3116 [29]; pCs2, a plasmid containing a 1.2.kb cDNA representing a portion of the 3’ end of chicken al(I1) collagen [30]; pCPG.1, a plasmid containing a 1.6.kb cDNA coding for a portion of the C-terminal end of the chick cartilage aggregating keratan sulfate/chondroitin sulfate-rich proteoglycan core protein (201. Antiserum to type X collagen. An antiserum to chick type X collagen was prepared during the course of the present study to analyze type X collagen production by immunoprecipitation and immunocytochemistry. Type X collagen was isolated from the culture medium of chick sternum cephalic chondrocyte cultures as described [31]. Collagenous proteins recovered from the medium after two rounds of precipitation with 30% ammonium sulfate were salt-fractionated at acid pH. The material precipitated by 2 M NaCl in 0.5 M acetic acid consisted primarily of type X collagen. This was assessed by SDS-poly-
FIG. 1. Fluorogram of an SDS-polyacrylamide gel to analyze immunoprecipitated macromolecules. [3H]proline-labeled collagens secreted by cultured mature chondrocytes (lane l), immature chondrocytes (lane 51, and fibroblasts (lane 9) were reacted with Sepharoseprotein A beads left uncoated (no serum) (lanes 2, 6 and IO), beads coated with preimmune serum (lanes 3, 7, and ll), or beads coated with immune serum (lanes 4, 8, and 12). The immunoprecipitated material was then solubilized and analyzed by electrophoresis and fluorography. Fn, fibronectin; XI, subunits of type XI collagen; II, mature and precursor forms of type II collagen (note in lane 9 that type I collagen and its precursors migrate to a similar electrophoretic position); IX, two of the three subunits of type IX collagen; X, the single subunit of type X collagen.
acrylamide gel electrophoresis and partial amino acid sequencing. For the latter, the protein was attached to a Sequelar(TM)-DITC membrane and subjected to 15 cycles of solid-phase automated N-terminal sequencing by Edman degradation on a Milligen/Biosearch 6600 Prosequencer with on-line reverse phase chromatographic analysis of the PTH amino acid from each cycle. PTH-amino acids were detected at 269 and 313 nm and quantified by comparison to 100 pmol of PTHamino acid standards. The sequence obtained, Ser-Asp-Gly-TyrPhe-Ser-Glu-Arg-Tyr-Gln-X-Gln-Ser-Ser-Ile. is in agreement with the cDNA-predicted sequence [32]; this probably also includes a lysine residue in position “X” which could not be detected by our procedure since the protein was coupled to the DITC-membrane through this amino acid. The protein was separated on preparative 6% SDS-polyacrylamide gels under reducing conditions, visualized by incubation of the gel in 0.25 M KC1 [33], and electroeluted. Approximately 200 pg of the gelpurified protein was mixed with 1 ml of complete Freund’s adjuvant and injected intradermally in a rabbit (preimmune serum from the same rabbit served as a control). A booster injection with 100 fig of protein in incomplete adjuvant was performed after 3 weeks and blood was collected 1 week later. The specificity of the antiserum was established by immunoprecipitation and ELISA assays. In immunoprecipitation assays, cultures of mature cephalic sternal chondrocytes (synthesizing types II, IX, X, and XI collagen; Fig. 1,
40
PACIFIC1
lane l), immature caudal sternal chondrocytes (synthesizing types II, IX, and XI collagen; Fig. 1, lane 5), and tendon fibroblasts (synthesizing mainly type I collagen; Fig. 1, lane 9) were pulse-labeled with [sH]proline for 24 h and the radiolabeled collagens secreted into the culture medium were recovered by ammonium sulfate precipitation [34]. After solubilization in immunoprecipitation buffer C [35], aliquots of the collagen samples were mixed with uncoated Sepharoseprotein A beads, beads coated with preimmune serum, or beads coated with immune serum (25 ~1 serum/50 ~1 of beads), and then incubated for 1.5 h at 4°C on a shaker. The beads were recovered by centrifugation and rinsed five times; the immunoprecipitated material was solubilized in Laemmli sample buffer (361 and analyzed by electrophoresis on a 6% SDS-polyacrylamide gel and by Auorography. As expected, the uncoated beads did not precipitate any substantial amount of the various collagen types (Fig. 1, lanes 2, 6, and 10) though they did precipitate a small amount of fibronectin (usually present in ammonium sulfate-precipitated medium samples). Similar results were obtained with the preimmune serum (Fig. 1, lanes 3. 7 and 11). In contrast, the immune serum precipitated exclusively type X collagen, which was present only in the mature chondrocyte sample (Fig. 1, lanes 4; cfr. with lanes 8 and 12). Identical results were obtained when the cell layer samples were used instead of medium samples; however, the amount of type X collagen present was quite low (not shown). The specificity of the immune serum for type X collagen was confirmed by ELISA assays. Types II, IX, and XI collagen were isolated from 10 dozen Day 17 chick embryo sterna by ion exchange chromatography [37]; type I collagen and its precursors were isolated from the medium of chick tendon fibroblast cultures by two rounds of ammonium sulfate precipitation. The collagen preparations were examined by electrophoresis to assess purity. ELISA assays were carried out as described [38]. Immunolon 96-well plates were precoated overnight in 0.1 M carbonate buffer, pH 9.6, at 16°C with 0.5 @g/well of type X, I, II, IX, or XI collagen. After rinsing, plates were incubated with various dilutions of immune serum in PBS containing :3% BSA and 0.05% Tween 20, and bound antibodies were quantified at 450 nm after immunoperoxidase reaction and color development for 30 min. Clearly, the immune serum reacted only with type X collagen and did so in a dilution-dependent manner (Fig. 2A). For indirect ELISA assays, the Immunolon plates were precoated overnight with 0.5 rg/well of type X collagen. Aliquots of the immune serum diluted 1:3000 (a dilution producing about 1 OD of color development in direct assays; see Fig. 2A) were preincubated at 16°C overnight with the indicated amounts of types X, I, II, IX, or XI collagen and were then added to the plates precoated with type X collagen. Bound antibodies were quantified as above. As shown in Fig. 2B, only preincubation with type X collagen inhibited interaction of the immune serum with bound type X collagen and did so in a concentration-dependent manner. Thus, both the immunoprecipitation and ELISA assays showed that the immune serum was specific for type X collagen and did not cross-react at appreciable levels with the other collagen types tested. Type X collagen analysis. Control and RA-treated cultures were labeled with 25 j&i/ml of [5-sH]proline during the last 16 h of incubation in the presence of 50 pg/ml ascorbic acid and 100 gg/ml b-aminopropionitrile. The collagenous proteins secreted into the culture medium were recovered by ammonium sulfate precipitation, solubilized in immunoprecipitation buffer C [35], and processed for immunoprecipitation using Sepharose-protein A beads precoated with type X collagen antiserum (25 ~1 serum/50 ~1 of beads) as above. For immunocytochemistry, control and RA-treated chondrocytes were detached from the tissue culture plates by incubation for 3 min in 0.1% trypsin and 0.02% EDTA in calcium and magnesium-free saline. After recovery by centrifugation, they were allowed to adhere for 10 min to 35 mm tissue culture plates which had been precoated with 0.05% polylysine. Cells were fixed with 70% ethanol and processed for immunofluorescence [35] using a 1:200 dilution of type X
ET
AL. A 1.2 OX 0 I, II, IX or XI
1.0 I
I
I
I
1/32,000
I
I
1116,000 1/8000 Antiserum dilution
l/4000
J
l/2000
B
0
0.01
0.10 pglmlsoluble
1.00 antigen
10.00
100.00
FIG. 2. (A) Direct ELISA assays to test reactivity of the antiserum raised against type X collagen with various collagen types. Wells coated with 0.5 pg of types X, I, II, IX, or XI collagen were incubated with the indicated dilutions of antiserum, and bound antibodies were quantified by immunoperoxidase, color development, and absorbance at 450 nm. (B) Indirect ELISA assays to further test the specificity of the antiserum. Aliquots of the antiserum diluted I:3000 were incubated with increasing concentrations of types X, I, II, IX, or XI collagen (soluble antigens). The samples were then added to wells precoated with 0.5 +g of type X collagen (bound antigen), and bound antibodies were quantified as above.
collagen antiserum. When indicated, cultures were stained with type II collagen antibodies (kindly provided by Dr. T. Linsenmayer). Bound antibodies were localized by reaction with rhodamine-conjugated goat anti-rabbit antibodies (Cappel) and viewed under epifluorescence Alkaline phosphatase (APase) assays. APase activity associated with the cell layer was measured by a modification of the method of Bessey et al. [39] usingp-nitrophenyl phosphate (pNP) as a substrate. Cells were scraped into the culture medium with a rubber policeman, recovered by centrifugation, and then resuspended in ice-cold TBS (0.9% NaCl in 3 mM Tris-HCl, pH 7.4). One-half of each cell suspension was centrifuged for 1 min in a microfuge, and the resulting cell pellets were solubilized in 0.9% NaCl and 0.2% Triton X-100. Sam-
RETINOIC Al
2
3
4
ACID
AND
CHONDROCYTE
5
XB lBS-
FIG. 3. Autoradiograms of a northern blot to determine the levels of type X collagen RNA. (A) Whole cellular RNAs from control cells (lanes 1 and 5) and cells treated for 1, 2, and 3 days with 100 nM RA (lanes 2, 3, and 4, respectively) were separated by agarose gel electrophoresis, blotted, and hybridized to 32P-labeled type X collagen cDNA pDLrl0. Note the time-dependent induction of type X collagen gene expression by RA treatment. (B) Blot was dehybridized and hybridized to “‘P-labeled 1% ribosomal probe.
ples were clarified by centrifugation for 5 min and the supernatants were mixed with 1 vol of 1 M Tris-HCl (pH 9.0) containing 1 mA4 pNP and 1 mM MgCl,. The reaction was stopped by addition of 0.25 vol of 1 N NaOH, and hydrolysis of pNP was monitored as change in absorbance at 410 nm. The remaining half of each cell suspension was used to determine DNA content by a fluorometric procedure [40]. APase activity in the culture medium was determined by modifications of previous methods [41]. The medium samples were clarified by centrifugation at 500g and then filtered through Millipore 0.45 pm nitrocellulose filters which had been presoaked with ice-cold TBS. Filters were rinsed twice with TBS and material trapped on the filters was solubilized in 2% Zwittergent 3-12 (Calbiochem). APase activity in the solubilized material was determined as above. More than 80% of the total APase activity present in medium samples was recovered with this method as compared to the conventional ultracentrifugation method [42].
RESULTS
41
MATURATION
treated for 1 day only exhibited little or no detectable type X collagen RNA (Fig. 3A, lane 2); cells treated for 2 and 3 days contained increasing amounts of this RNA. Rehybridization of the blot in Fig. 3A to a 32P-labeled 18s ribosomal RNA clone [27] showed that the lanes contained comparable amounts of cellular RNA (Fig. 3B, lanes l-5). Thus, treatment with 100 nM RA can clearly induce type X collagen gene expression in immature sternal chondrocytes. The 24-h period preceding major changes in gene expression has been observed previously during RA treatment of chondrocytes [6,20]. We next asked whether RA at physiological doseswas capable of exerting a similar effect and, if so, whether this was accompanied by changes in gene expression for other chondrocyte-characteristic matrix components. Secondary chondrocyte cultures were treated for 2 and 3 days with 10 or 35 nM RA, concentrations in the range of those occurring naturally in the chick embryo [7, 81. Companion cultures were treated with 100 nM RA for comparison within the same experiment. As revealed by Northern blot analysis of total cellular RNA (Fig. 4), physiological doses of RA indeed induced type X collagen gene expression and did so in a both temporal- and dose-dependent manner. Thus, cells treated with 10 nM RA for 2 days contained a significant amount of type X collagen RNA (Fig. 4A, lane 2) as compared to the barely detectable levels in untreated cells (Fig. 4A, lane 1); this amount increased further by Day 3 (Fig. 4A, lane 5). In cells treated with 35 nM RA the amount of type X collagen RNA was one- to twofold higher than in cells treated with 10 nM RA at either time point (Fig. 4A, lanes 3 and 6). Interestingly, the amount of this RNA in
Type X Collagen Gene Expression On the basis of our previous study in which treatment with high doses of RA had increased type X collagen gene expression in maturing vertebral chondrocyte cultures [20], we performed experiments to determine whether similar high doses were able to initiate type X collagen gene expression in immature chondrocytes (that is, cells in which type X collagen gene expression could not yet be detected). The floating sternal chondrocytes harvested from primary cultures were treated with trypsin to induce cell attachment and replated into secondary cultures on tissue culture dishes. They were then treated with 100 nM RA for 1, 2 and 3 days and harvested simultaneously with control untreated cultures. Whole cellular RNA was isolated, processed for Northern blot analysis, and hybridized to 32P-labeled chick type X collagen cDNA pDLrl0. The resulting autoradiogram showed that control, untreated cells did not contain detectable amounts of type X collagen mRNA (Fig. 3A, lane 1 and 5), thus demonstrating that they were still immature. In contrast, cells treated with RA showed a time-dependent increase in type X collagen gene expression. Cells
Al234567
xB II C PG -
RA (nM)
FIG. 4. Autoradiograms of a northern blot to determine the effects of RA on chondrocyte gene expression. RNAs from control cells (lane 1) and cells treated for 2 days (lanes 2-4) and 3 days (lanes 5-7) with the indicated doses of RA were electrophoresed, blotted, and hybridized to different 32P-labeled probes: (A) type X collagen cDNA pDLrl0; (B) type II collagen cDNA pCs2; (C) the cartilage aggregating proteoglycan (PG) cDNA pCPG.1; and (D) 18s ribosomal probe.
42
PACIFIC1
cells treated with the nonphysiologic dose of RA (100 nM) did not increase further (Fig. 4A, lanes 4 and 7), probably reflecting an interference of this high RA dose with normal phenotypic expression [5, 6, 20-231. Blots were rehybridized to 32P-labeled type II collagen cDNA clone pCs2 (Fig. 4B) and the aggregating proteoglycan cDNA clone pCPG.1 (Fig. 4C) to determine whether the initiation of type X collagen gene expression induced by physiologic doses of RA was accompanied by changes in gene expression of these two typical chondrocyte macromolecules. Interestingly, expression of these genes was not altered by the physiologic doses of RA after 2 or 3 days (Figs. 4B and 4C, lanes 2,3,5, and 6, respectively) compared to control cells (Figs. 4B and 4C, lane 1). Expression of both genes, however, was decreased about 50% by treatment with 100 nM RA (Figs. 4B and 4C, lanes 4 and 7). Rehybridization of these blots to the 32P-labeled 18s ribosomal probe showed that the lanes contained comparable amounts of cellular RNA (Fig. 4D). Clearly, physiologic doses of RA induce type X collagen gene expression in immature chondrocytes; this change is not accompanied by major changes in expression of either type II collagen or the aggregating proteoglycan genes.
ET
AL.
12345
kDa -
205
-
116
%
x-
0
10
35
10
35
RA(nM) FIG. 5. Fluorogram of an SDS-polyacrylamide gel to analyze type X collagen production. Control chondrocytes (lane 1) and chondrocytes treated for 2 days (lanes 2 and 3) and 3 days (4 and 5) with the indicated doses of RA were pulsed with [3H]proline for 16 h. Collagenous proteins secreted into the medium were recovered by ammonium sulfate precipitation, subjected to immunoprecipitation with the type X collagen antiserum, and analyzed by electrophoresis and Auorography. X, the single subunit of type X collagen.
Type X Collagen Synthesis We next asked whether the induction of type X collagen RNA during RA treatment was accompanied by type X collagen synthesis and whether type X collagen gene expression was induced in every chondrocyte or a fraction of them. Secondary chondrocyte cultures were treated for 2 and 3 days with 10 or 35 nM RA and labeled during the last 24 h of culture with [3H]proline along with untreated cultures. The radiolabeled collagenous proteins secreted into the culture medium were recovered by ammonium sulfate precipitation, processed for immunoprecipitation, and analyzed by SDS-gel electrophoresis and fluorography. While, as expected, the control cultures did not synthesize appreciable amounts of type X collagen (Fig. 5, lane l), the RA-treated cells produced this collagen in amounts dependent on both length and dose of treatment (Fig. 5, lanes 2-5). Cells treated for 3 days with 10 or 35 nM RA synthesized more type X collagen (Fig. 5, lanes 4 and 5) than cells treated for 2 days (Fig. 5, lanes 2 and 3); likewise, cells treated with 35 nM RA synthesized more protein at either time point than cells treated with 10 nM RA. The data complement well those on type X collagen RNA levels. To determine the proportion of cells induced to synthesize type X collagen by RA treatment, cultures were treated with 35 nM RA for 1,2, and 3 days. Control and treated cultures were then detached simultaneously from their respective tissue culture dishes, allowed to
adhere for 10 min to polylysine-coated dishes, and processed for immunocytochemistry using the type X collagen antiserum. This procedure allowed the chondrocytes to assume a round cell shape before they were fixed and processed for immunofluorescence, enabling us to minimize equivocal results due to varying cell morphology and thickness. As expected, most of the control untreated chondrocytes failed to stain positively with the type X collagen antiserum; occasionally, one immunopositive cell was observed (Figs. 6A and 6B, arrow). No major change in the number of type X collagen-producing cells occurred after 1 day of treatment (Figs. 6C and 6D). After 2 and 3 days of treatment, however, many immunopositive cells were present (Figs. 6E and 6F, and 6G and 6H, respectively), amounting to approximately 30-35% of the cell population by Day 3 of treatment. When companion cultures were stained with type II collagen antibodies, all the cells present exhibited positive immunostaining (Figs. 7A and 7B), thus showing that the entire cell population was composed of functional chondrocytes. Alkaline Phosphatase Activity In addition to initiating production of type X collagen, mature chondrocytes increase synthesis of alkaline phosphatase (APase), an enzyme involved in cartilage mineralization [15]. We thus asked whether RA treat-
RETINOIC
ACID
AND
CHONDROCYTE
MATURATION
43
FIG. 6. Phase t A, C, E, and G) and immunofluorescence (B, D, F, and H) micrographs of chondrocytes treated with 35 nMRA for up to 3 days and then processed for immunofluorescence with the type X collagen antiserum. A and B, control culture; C and D, 1 day of treatment; E and F, 2 days of treatment; and G and H, 3 days of treatment. Arrows point to immunopositive cells viewed under phase and fluorescence microscopy; note that the average size of the positive cell is similar to that of surrounding cells that do not stain positively. Bar, 150 pm.
ment of immature chondrocytes would lead to increased APase activity, in addition to inducing type X collagen gene expression. Secondary chondrocyte cultures were treated with 10,35, and 100 nM RA for 1,2, and 3 days.
At each time point analyzed, about 90% of the enzymatic activity was recovered in the cell layer compartment and the remaining 10% in the culture medium. RA treatment led to increased cell-layer-associated APase
44
PACIFIC1
ET
AL.
Within 3 days of treatment, there occurs a coordinate increase in type X collagen RNA levels, type X collagen synthesis, and the percentage of type X collagen-producing cells in these cultures. The treatment also leads to increased APase activity occurring as early as 24 h from the start of treatment. These changes in phenotypic expression are not accompanied by major changes in gene expression of type II collagen and the large aggregating proteoglycan (see below), suggesting that they represent specific changes in gene expression. The data in this study strengthen our conclusion [20] that RA is involved in cartilage maturation as an inducer or promoter of traits typical of mature chondrocytes. The mechanisms by which RA may exert this role remain to be elucidated. The RA effects on cell phenotype are thought to be mediated by nuclear RA receptors (RARs) which bind to specific regulatory sequences (RA responsive elements) and bring about changes in gene expression (43, 441. In preliminary experiments using a mouse gamma RAR cDNA (kindly provided by Dr. P. Chambon) [45,46], we have established that mature sternal chondrocytes contain gamma RAR transcripts (Pacifici, unpublished). Therefore, it is possible that the RA effects on type X collagen gene expression and alkaline phosphatase activity described above may also be mediated by gamma RAR in chondrocytes. The type X collagen gene, and its regulatory sequences in particular, have not yet been analyzed in great detail [32]. Our data predict that RA responsive elements may be present in the gene and may be involved in the
FIG. 7. Phase (A) and immunofluorescence (B) micrographs of chondrocytes stained with type II collagen antibodies. Note that every cell present in the culture produces this collagen type. Bar, 30 pm.
activity in a time and dose-dependent fashion (Fig. 8). While 10 nM RA increased APase activity only after 3 days of treatment, 35 nM RA was effective at each time point analyzed. Interestingly, 100 nM RA increased APase activity during the first 2 days of treatment but was followed by a decrease in activity, probably reflecting the adverse effects of this high RA dose on the chondrocyte phenotype. Clearly, RA treatment leads to increased APase activity in immature sternal chondrocytes. Because this can be observed at 24 h of treatment, it appears to precede the induction of type X collagen synthesis, which is most prominent at 48 h of treatment. DISCUSSION
The data in this study show that RA at physiological doses is a rapid inducer of maturation-associated traits in cultures of immature caudal sternal chondrocytes.
RA
treatment
1 day
2 days
3 days
FIG. 8. Effects of RA treatment on APase activity. Chondrocytes grown on 3%mm dishes were treated with the indicated concentrations of RA for 1, 2 and 3 days. APase activity associated with the cell layer was determined using p-nitrophenyl phosphate (pNP) as a substrate. Data are expressed as nanomoles of pNP hydrolysis per 30 min per microgram DNA. *Significantly different from control values, P > 0.05.
RETINOIC
ACID
AND
CHONDROCYTE
changes in type X collagen gene expression during RA treatment described here. The Northern blot analysis shown in Fig. 4 indicates that proteoglycan gene expression is not markedly altered by treatment of immature chondrocytes withphysiological doses of RA. However, previous studies had found that proteoglycan synthesis measured by radiosulfate incorporation decreases 50-60% in cultured rabbit growth plate [13] or rat costal chondrocytes [47] treated with 10 nh4 RA for 3 to 4 days. Other parameters such as polygonal morphology or type II collagen production were not altered by this RA treatment [ 131. The seemingly different effects of physiological RA doses on proteoglycan expression in this and previous studies could reflect species-specific differences. Alternatively, they could be due to the fact that proteoglycan production measured by radiosulfate incorporation is probably more sensitive than Northern blot analysis. Lastly, this RA treatment may not greatly affect expression of proteoglycan core protein mRNA but could still lead to decreased radiosulfate incorporation by interfering with other steps involved in proteoglycan biosynthesis and secretion. Our immunofluorescence data (Fig. 6) demonstrate that only 30-35% of the cells appear to be sensitive to RA treatment and activate type X collagen production. Two possibilities can account for this phenomenon. The first is that the secondary cultures used in these experiments may cont.ain chondrocytes able to respond to RA and chondrocytes unable to do so. For example, some cells may contain RARs while other cells do not. The latter cells may require further growth in culture before they acquire the ability to produce RARs and respond to RA treatment. If this is so, this possibility would imply that sensitivity to RA treatment is maturation-dependent and changes during chondrocyte maturation [ 131. The alternative possibility is that the cultures contain chondrocytes that differ in their sensitivity to RA treatment, such that higher doses of RA are necessary to trigger their response. These possibilities are being tested in ongoing experiments. A close examination of Fig. 6 also reveals that the chondrocytes engaged in type X collagen production in both control and RA-treated cultures are similar in size to the chondrocytes that do not synthesize this collagen at detectable levels. This observation would seem to be in sharp contradiction with the widely accepted view that type X collagen is produced only by hypertrophic chondrocytes. Our data, however, are in full agreement with those in a recent study using in situ hybridization in which type X collagen transcripts were found to be already present in late proliferating, nonhypertrophic, oval-shaped chondrocytes in chicken tibia1 growth cartilage [ 181. We have confirmed this observation and detected by immunohistochemistry significant amounts of type X collagen in late proliferating chondrocytes and
45
MATURATION
their surrounding matrix (unpublished observations). Together, the data indicate that cell hypertrophy and initiation of type X collagen production are not coordinately regulated [18, 241 and that maturing chondrocytes can initiate type X collagen synthesis before they have completed the process of cell hypertrophy. If left untreated, the sternal chondrocyte cultures used in this study would have eventually activated type X collagen gene expression at readily detectable levels; however, they would have required approximately 4 additional weeks of culture to do so [24,48]. One interpretation of this phenomenon is that chondrocytes can progress toward maturity in either an RA-dependent or RA-independent manner: RA would promote this process but is not essential. An alternative possibility is that other factors or nutrients could substitute for RA [49, 501. One such nutrient could obviously be retinol, which is normally present in serum. Lastly, it may be that several exogenous cues (including RA) participate in the induction and promotion of chondrocyte maturation and gene expression of maturation-associated traits both in vitro and in ho. The rate at which chondrocytes would progress toward maturity could depend on whether all such cues are present, whether they are present in optimal concentrations, and perhaps more importantly, whether the chondrocytes have acquired the ability to respond to them. We thank Drs. I’. Leboy, Y. Ninomiya, B. R. Olsen, and M. Young for their generous gifts of plasmids; Dr. T. Linsenmayer for the type II collagen antibodies; and Dr. W. Abrams for help with amino acid sequencing. This study was supported by grants from the National Institutes of Health. REFERENCES 1.
Kochhar,
D. M. (1967)
Teratology
7, 2899298.
2. Tamarin,
A., Ctawley, A., Lee, J., and Tickle, Exp. Morphol. 84, 105-123.
bryol.
I Hassell, ‘3 c’fI1
Res.
4. Pacifici, Cell Res.
J. R., Pennypacker,
J. P., and Lewis,
C. (1984)
J. Em-
C. A. (1978)
Exp.
112, 409-417. M.,
Cossu.
G., Molinaro,
M.,
and Tato,
F. (1980)
Exp.
129, 469-474.
CELL
RESEARCH
195,
38-46
(1991)
Retinoic Acid Treatment Induces Type X Collagen Gene Expression in Cultured Chick Chondrocytes MAURIZIO
PACIFICI,’
ELEANOR
of Anatomy/Histology,
Department
School
B. GOLDEN,
of Dental
Medicine,
MASAHIRO linkers&
The vitamin A derivative retinoic acid (RA) is widely thought to be involved in cartilage development, but its precise roles and mechanisms of action in this complex process remain unclear. We have tested the hypothesis that RA is involved in chondrocyte maturation during endochondral ossification and, in particular, is an inducer of maturation-associated traits such as type X collagen and alkaline phosphatase. Immature chondrocytes isolated from the caudal region of Day 19 chick embryo sterna were seeded in secondary monolayer cultures and treated either with a high dose (100 nM) or with physiological doses (lo-35 nM) of RA for up to 3 days. We found that after an initial lag of about 24 h, physiological doses of RA indeed induced type X collagen gene expression in the immature cells. This induction was not accompanied by obvious changes in expression of the type II collagen and large aggregating proteoglycan core protein genes. As revealed by immunocytochemistry, 30-35% of the cells in cultures treated with RA for 3 days were engaged in type X collagen production. Interestingly, these cells were relatively similar in size to chondrocytes in which no type X collagen was detected, suggesting that chondrocytes can initiate type X collagen production independent of cell hypertrophy. RA treatment also led to increased alkaline phosphatase activity occurring as early as 24 h after the start of treatment. The data in this study indicate that RA may have a role in endochondral ossification as an inducer/promoter of maturation-associated ( 1991 Academic traits during chondrocyte maturation. Press,
Inc.
INTRODUCTION
The vitamin A derivative retinoic acid (RA) has long been known to affect both cartilage development and function. Early studies found that high doses of retinoids given during pregnancy are teratogenic and cause severe limb and craniofacial defects involving abnormal cartilage and skeletal development [l, 21. Similar high doses block chondrogenesis in mesenchymal cell cul’ To whom
reprint
requests
should
he addressed.
0014.4827/91 $3.00 Copyright IE 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
IWAMOTO,
of Pennsyluania,
AND SHERRILL Philadelphia,
L. ADAMS
Pennsyluania
19104-6003
tures [3,4] and inhibit normal phenotypic expression in cultured chondrocytes [5,6]. Interestingly, recent studies have found that RA at physiological concentrations may actually be involved in normal cartilage development. A gradient of RA occurs naturally in the developing limb, from 20 nM at the anterior edge to 50 nM at the posterior edge [7, 81. Exogenous RA applied to the anterior limb edge matching the concentration naturally present in the posterior edge induces a mirrorimage duplication of the limb skeletal elements, thus acting as a skeletal morphogen [9, lo]. When embryonic mouse limb buds in organ culture are treated with low doses of RA, total cartilage mass increases over control values [ 111. Similar physiological doses of RA stimulate chondrogenesis in limb mesenchymal cell cultures [ 121 and proliferation in growth plate chondrocyte cultures [13]. Thus, although hypervitaminosis A is deleterious to cartilage development and function, physiological levels of RA may promote these processes. The precise roles RA plays in cartilage development and function, however, remain unclear. This issue is complicated by the fact that after its emergence during embryogenesis, cartilage undergoes a further complex developmental process of maturation as it takes part in endochondral ossification. During maturation, the newly emerged, resting, immature chondrocytes first develop into proliferating chondrocytes. Whereas the resting chondrocytes are small flat cells producing very little matrix, the proliferating chondrocytes are very active, replicate readily, and produce large quantities of matrix components such as type II collagen and the aggregating proteoglycan [141. As proliferation dwindles, the cells mature into large round hypertrophic chondrocytes, initiate production of a unique collagen, type X, and increase alkaline phosphatase activity and release of matrix vesicles. The abundant matrix surrounding the hypertrophic chondrocytes then undergoes mineralization [15-191. To assesswhether the complex process of maturation is influenced by RA, we recently carried out experiments with maturing cultures of chick embryo vertebral chondrocytes [20]. Cells isolated from Day 12 chick embryos were allowed to initiate the process of maturation by growth for about 2 weeks in permissive monolayer
RETINOIC
ACID
AND
CHONDROCYTE
culture conditions. Maturation was assessed by active cell replication, increase in average cell size, and initiation of type X collagen gene expression. Cultures were then treated with a high dose of RA (300 nn/l) and expression of five extracellular matrix genes, including the maturation-specific type X collagen gene, was assessed. We found that. expression of genes encoding typical chondrocyte macromolecules such as type II collagen and aggregating proteoglycan was nearly shut off by 48 h of RA treatment [21-231. Strikingly, this decrease was accompanied by a threefold increase in type X collagen gene expression. These differential changes in gene expression are quite similar to those normally accompanying chondrocyte maturation in viuo [ 181, thus leading us to conclude that RA may have a role in maturation, in particular, in induction of type X collagen gene expression. In this report, we have tested this hypothesis using cultures of immature chondrocytes isolated from the caudal region of chick sternum and treated with both high and physiological doses of RA.
Fn XI
39
MATURATION
-
205
-
116
c
II i IX [ x-
MATERIALS
AND
METHODS
Chondrocyte cultures. Primary cultures of immature sternal chondrocytes isolated from the caudal region of Day 19 chick embryo sternum [l’i] were prepared and grown for 6-7 days as described [24]. Medium consisted of high glucose Dulbecco’s MEM containing 10% fetal bovine serum (Hyclone Labs.), 2 mM L-glutamine, and 50 U/ml each of penicillin and streptomycin. Pure populations of floating chondrocytes harvested from the primary cultures were treated with 0.25% trypsin and replated into secondary monolayer cultures at 3 x lo”/60 mm tissue culture dish or 10 X lo6 cells/100 mm dish. Fresh 100X stock solutions of all-trans RA (Sigma) were prepared in 95% ethanol daily and used under safety light conditions. Control cultures received 10 pi/ml of 95% ethanol. Medium was changed daily. Whole cellular RNA isolated by the RNA isolation and analysis. guanidine isothiocynate method [25] was denatured by glyoxalation, electrophoresed on agarose gels, transferred to Hybond-N membranes (Amersham) by capillary blotting, and hybridized to nicktranslated 32P-labeled cDNA probes. Blots were washed at high stringency (0.1X SSPE at 65°C) and exposed to Kodak films at ~70°C for various lengths of time; quantitative analyses were carried out by video digitization [26]. When necessary, blots were dehybridized at 9O’C in 0.05X SSPE for 5 min and hybridized to a different probe. A final rehybridization was carried out using an 18s ribosomal RNA probe [27] to control for variations in RNA concentrations, gel loading, and transfer efficiency. The plasmids used were the chick type X collagen cDNA pDLrl0 [28], a 197.bp subclone of pYN3116 [29]; pCs2, a plasmid containing a 1.2.kb cDNA representing a portion of the 3’ end of chicken al(I1) collagen [30]; pCPG.1, a plasmid containing a 1.6.kb cDNA coding for a portion of the C-terminal end of the chick cartilage aggregating keratan sulfate/chondroitin sulfate-rich proteoglycan core protein (201. Antiserum to type X collagen. An antiserum to chick type X collagen was prepared during the course of the present study to analyze type X collagen production by immunoprecipitation and immunocytochemistry. Type X collagen was isolated from the culture medium of chick sternum cephalic chondrocyte cultures as described [31]. Collagenous proteins recovered from the medium after two rounds of precipitation with 30% ammonium sulfate were salt-fractionated at acid pH. The material precipitated by 2 M NaCl in 0.5 M acetic acid consisted primarily of type X collagen. This was assessed by SDS-poly-
FIG. 1. Fluorogram of an SDS-polyacrylamide gel to analyze immunoprecipitated macromolecules. [3H]proline-labeled collagens secreted by cultured mature chondrocytes (lane l), immature chondrocytes (lane 51, and fibroblasts (lane 9) were reacted with Sepharoseprotein A beads left uncoated (no serum) (lanes 2, 6 and IO), beads coated with preimmune serum (lanes 3, 7, and ll), or beads coated with immune serum (lanes 4, 8, and 12). The immunoprecipitated material was then solubilized and analyzed by electrophoresis and fluorography. Fn, fibronectin; XI, subunits of type XI collagen; II, mature and precursor forms of type II collagen (note in lane 9 that type I collagen and its precursors migrate to a similar electrophoretic position); IX, two of the three subunits of type IX collagen; X, the single subunit of type X collagen.
acrylamide gel electrophoresis and partial amino acid sequencing. For the latter, the protein was attached to a Sequelar(TM)-DITC membrane and subjected to 15 cycles of solid-phase automated N-terminal sequencing by Edman degradation on a Milligen/Biosearch 6600 Prosequencer with on-line reverse phase chromatographic analysis of the PTH amino acid from each cycle. PTH-amino acids were detected at 269 and 313 nm and quantified by comparison to 100 pmol of PTHamino acid standards. The sequence obtained, Ser-Asp-Gly-TyrPhe-Ser-Glu-Arg-Tyr-Gln-X-Gln-Ser-Ser-Ile. is in agreement with the cDNA-predicted sequence [32]; this probably also includes a lysine residue in position “X” which could not be detected by our procedure since the protein was coupled to the DITC-membrane through this amino acid. The protein was separated on preparative 6% SDS-polyacrylamide gels under reducing conditions, visualized by incubation of the gel in 0.25 M KC1 [33], and electroeluted. Approximately 200 pg of the gelpurified protein was mixed with 1 ml of complete Freund’s adjuvant and injected intradermally in a rabbit (preimmune serum from the same rabbit served as a control). A booster injection with 100 fig of protein in incomplete adjuvant was performed after 3 weeks and blood was collected 1 week later. The specificity of the antiserum was established by immunoprecipitation and ELISA assays. In immunoprecipitation assays, cultures of mature cephalic sternal chondrocytes (synthesizing types II, IX, X, and XI collagen; Fig. 1,
40
PACIFIC1
lane l), immature caudal sternal chondrocytes (synthesizing types II, IX, and XI collagen; Fig. 1, lane 5), and tendon fibroblasts (synthesizing mainly type I collagen; Fig. 1, lane 9) were pulse-labeled with [sH]proline for 24 h and the radiolabeled collagens secreted into the culture medium were recovered by ammonium sulfate precipitation [34]. After solubilization in immunoprecipitation buffer C [35], aliquots of the collagen samples were mixed with uncoated Sepharoseprotein A beads, beads coated with preimmune serum, or beads coated with immune serum (25 ~1 serum/50 ~1 of beads), and then incubated for 1.5 h at 4°C on a shaker. The beads were recovered by centrifugation and rinsed five times; the immunoprecipitated material was solubilized in Laemmli sample buffer (361 and analyzed by electrophoresis on a 6% SDS-polyacrylamide gel and by Auorography. As expected, the uncoated beads did not precipitate any substantial amount of the various collagen types (Fig. 1, lanes 2, 6, and 10) though they did precipitate a small amount of fibronectin (usually present in ammonium sulfate-precipitated medium samples). Similar results were obtained with the preimmune serum (Fig. 1, lanes 3. 7 and 11). In contrast, the immune serum precipitated exclusively type X collagen, which was present only in the mature chondrocyte sample (Fig. 1, lanes 4; cfr. with lanes 8 and 12). Identical results were obtained when the cell layer samples were used instead of medium samples; however, the amount of type X collagen present was quite low (not shown). The specificity of the immune serum for type X collagen was confirmed by ELISA assays. Types II, IX, and XI collagen were isolated from 10 dozen Day 17 chick embryo sterna by ion exchange chromatography [37]; type I collagen and its precursors were isolated from the medium of chick tendon fibroblast cultures by two rounds of ammonium sulfate precipitation. The collagen preparations were examined by electrophoresis to assess purity. ELISA assays were carried out as described [38]. Immunolon 96-well plates were precoated overnight in 0.1 M carbonate buffer, pH 9.6, at 16°C with 0.5 @g/well of type X, I, II, IX, or XI collagen. After rinsing, plates were incubated with various dilutions of immune serum in PBS containing :3% BSA and 0.05% Tween 20, and bound antibodies were quantified at 450 nm after immunoperoxidase reaction and color development for 30 min. Clearly, the immune serum reacted only with type X collagen and did so in a dilution-dependent manner (Fig. 2A). For indirect ELISA assays, the Immunolon plates were precoated overnight with 0.5 rg/well of type X collagen. Aliquots of the immune serum diluted 1:3000 (a dilution producing about 1 OD of color development in direct assays; see Fig. 2A) were preincubated at 16°C overnight with the indicated amounts of types X, I, II, IX, or XI collagen and were then added to the plates precoated with type X collagen. Bound antibodies were quantified as above. As shown in Fig. 2B, only preincubation with type X collagen inhibited interaction of the immune serum with bound type X collagen and did so in a concentration-dependent manner. Thus, both the immunoprecipitation and ELISA assays showed that the immune serum was specific for type X collagen and did not cross-react at appreciable levels with the other collagen types tested. Type X collagen analysis. Control and RA-treated cultures were labeled with 25 j&i/ml of [5-sH]proline during the last 16 h of incubation in the presence of 50 pg/ml ascorbic acid and 100 gg/ml b-aminopropionitrile. The collagenous proteins secreted into the culture medium were recovered by ammonium sulfate precipitation, solubilized in immunoprecipitation buffer C [35], and processed for immunoprecipitation using Sepharose-protein A beads precoated with type X collagen antiserum (25 ~1 serum/50 ~1 of beads) as above. For immunocytochemistry, control and RA-treated chondrocytes were detached from the tissue culture plates by incubation for 3 min in 0.1% trypsin and 0.02% EDTA in calcium and magnesium-free saline. After recovery by centrifugation, they were allowed to adhere for 10 min to 35 mm tissue culture plates which had been precoated with 0.05% polylysine. Cells were fixed with 70% ethanol and processed for immunofluorescence [35] using a 1:200 dilution of type X
ET
AL. A 1.2 OX 0 I, II, IX or XI
1.0 I
I
I
I
1/32,000
I
I
1116,000 1/8000 Antiserum dilution
l/4000
J
l/2000
B
0
0.01
0.10 pglmlsoluble
1.00 antigen
10.00
100.00
FIG. 2. (A) Direct ELISA assays to test reactivity of the antiserum raised against type X collagen with various collagen types. Wells coated with 0.5 pg of types X, I, II, IX, or XI collagen were incubated with the indicated dilutions of antiserum, and bound antibodies were quantified by immunoperoxidase, color development, and absorbance at 450 nm. (B) Indirect ELISA assays to further test the specificity of the antiserum. Aliquots of the antiserum diluted I:3000 were incubated with increasing concentrations of types X, I, II, IX, or XI collagen (soluble antigens). The samples were then added to wells precoated with 0.5 +g of type X collagen (bound antigen), and bound antibodies were quantified as above.
collagen antiserum. When indicated, cultures were stained with type II collagen antibodies (kindly provided by Dr. T. Linsenmayer). Bound antibodies were localized by reaction with rhodamine-conjugated goat anti-rabbit antibodies (Cappel) and viewed under epifluorescence Alkaline phosphatase (APase) assays. APase activity associated with the cell layer was measured by a modification of the method of Bessey et al. [39] usingp-nitrophenyl phosphate (pNP) as a substrate. Cells were scraped into the culture medium with a rubber policeman, recovered by centrifugation, and then resuspended in ice-cold TBS (0.9% NaCl in 3 mM Tris-HCl, pH 7.4). One-half of each cell suspension was centrifuged for 1 min in a microfuge, and the resulting cell pellets were solubilized in 0.9% NaCl and 0.2% Triton X-100. Sam-
RETINOIC Al
2
3
4
ACID
AND
CHONDROCYTE
5
XB lBS-
FIG. 3. Autoradiograms of a northern blot to determine the levels of type X collagen RNA. (A) Whole cellular RNAs from control cells (lanes 1 and 5) and cells treated for 1, 2, and 3 days with 100 nM RA (lanes 2, 3, and 4, respectively) were separated by agarose gel electrophoresis, blotted, and hybridized to 32P-labeled type X collagen cDNA pDLrl0. Note the time-dependent induction of type X collagen gene expression by RA treatment. (B) Blot was dehybridized and hybridized to “‘P-labeled 1% ribosomal probe.
ples were clarified by centrifugation for 5 min and the supernatants were mixed with 1 vol of 1 M Tris-HCl (pH 9.0) containing 1 mA4 pNP and 1 mM MgCl,. The reaction was stopped by addition of 0.25 vol of 1 N NaOH, and hydrolysis of pNP was monitored as change in absorbance at 410 nm. The remaining half of each cell suspension was used to determine DNA content by a fluorometric procedure [40]. APase activity in the culture medium was determined by modifications of previous methods [41]. The medium samples were clarified by centrifugation at 500g and then filtered through Millipore 0.45 pm nitrocellulose filters which had been presoaked with ice-cold TBS. Filters were rinsed twice with TBS and material trapped on the filters was solubilized in 2% Zwittergent 3-12 (Calbiochem). APase activity in the solubilized material was determined as above. More than 80% of the total APase activity present in medium samples was recovered with this method as compared to the conventional ultracentrifugation method [42].
RESULTS
41
MATURATION
treated for 1 day only exhibited little or no detectable type X collagen RNA (Fig. 3A, lane 2); cells treated for 2 and 3 days contained increasing amounts of this RNA. Rehybridization of the blot in Fig. 3A to a 32P-labeled 18s ribosomal RNA clone [27] showed that the lanes contained comparable amounts of cellular RNA (Fig. 3B, lanes l-5). Thus, treatment with 100 nM RA can clearly induce type X collagen gene expression in immature sternal chondrocytes. The 24-h period preceding major changes in gene expression has been observed previously during RA treatment of chondrocytes [6,20]. We next asked whether RA at physiological doseswas capable of exerting a similar effect and, if so, whether this was accompanied by changes in gene expression for other chondrocyte-characteristic matrix components. Secondary chondrocyte cultures were treated for 2 and 3 days with 10 or 35 nM RA, concentrations in the range of those occurring naturally in the chick embryo [7, 81. Companion cultures were treated with 100 nM RA for comparison within the same experiment. As revealed by Northern blot analysis of total cellular RNA (Fig. 4), physiological doses of RA indeed induced type X collagen gene expression and did so in a both temporal- and dose-dependent manner. Thus, cells treated with 10 nM RA for 2 days contained a significant amount of type X collagen RNA (Fig. 4A, lane 2) as compared to the barely detectable levels in untreated cells (Fig. 4A, lane 1); this amount increased further by Day 3 (Fig. 4A, lane 5). In cells treated with 35 nM RA the amount of type X collagen RNA was one- to twofold higher than in cells treated with 10 nM RA at either time point (Fig. 4A, lanes 3 and 6). Interestingly, the amount of this RNA in
Type X Collagen Gene Expression On the basis of our previous study in which treatment with high doses of RA had increased type X collagen gene expression in maturing vertebral chondrocyte cultures [20], we performed experiments to determine whether similar high doses were able to initiate type X collagen gene expression in immature chondrocytes (that is, cells in which type X collagen gene expression could not yet be detected). The floating sternal chondrocytes harvested from primary cultures were treated with trypsin to induce cell attachment and replated into secondary cultures on tissue culture dishes. They were then treated with 100 nM RA for 1, 2 and 3 days and harvested simultaneously with control untreated cultures. Whole cellular RNA was isolated, processed for Northern blot analysis, and hybridized to 32P-labeled chick type X collagen cDNA pDLrl0. The resulting autoradiogram showed that control, untreated cells did not contain detectable amounts of type X collagen mRNA (Fig. 3A, lane 1 and 5), thus demonstrating that they were still immature. In contrast, cells treated with RA showed a time-dependent increase in type X collagen gene expression. Cells
Al234567
xB II C PG -
RA (nM)
FIG. 4. Autoradiograms of a northern blot to determine the effects of RA on chondrocyte gene expression. RNAs from control cells (lane 1) and cells treated for 2 days (lanes 2-4) and 3 days (lanes 5-7) with the indicated doses of RA were electrophoresed, blotted, and hybridized to different 32P-labeled probes: (A) type X collagen cDNA pDLrl0; (B) type II collagen cDNA pCs2; (C) the cartilage aggregating proteoglycan (PG) cDNA pCPG.1; and (D) 18s ribosomal probe.
42
PACIFIC1
cells treated with the nonphysiologic dose of RA (100 nM) did not increase further (Fig. 4A, lanes 4 and 7), probably reflecting an interference of this high RA dose with normal phenotypic expression [5, 6, 20-231. Blots were rehybridized to 32P-labeled type II collagen cDNA clone pCs2 (Fig. 4B) and the aggregating proteoglycan cDNA clone pCPG.1 (Fig. 4C) to determine whether the initiation of type X collagen gene expression induced by physiologic doses of RA was accompanied by changes in gene expression of these two typical chondrocyte macromolecules. Interestingly, expression of these genes was not altered by the physiologic doses of RA after 2 or 3 days (Figs. 4B and 4C, lanes 2,3,5, and 6, respectively) compared to control cells (Figs. 4B and 4C, lane 1). Expression of both genes, however, was decreased about 50% by treatment with 100 nM RA (Figs. 4B and 4C, lanes 4 and 7). Rehybridization of these blots to the 32P-labeled 18s ribosomal probe showed that the lanes contained comparable amounts of cellular RNA (Fig. 4D). Clearly, physiologic doses of RA induce type X collagen gene expression in immature chondrocytes; this change is not accompanied by major changes in expression of either type II collagen or the aggregating proteoglycan genes.
ET
AL.
12345
kDa -
205
-
116
%
x-
0
10
35
10
35
RA(nM) FIG. 5. Fluorogram of an SDS-polyacrylamide gel to analyze type X collagen production. Control chondrocytes (lane 1) and chondrocytes treated for 2 days (lanes 2 and 3) and 3 days (4 and 5) with the indicated doses of RA were pulsed with [3H]proline for 16 h. Collagenous proteins secreted into the medium were recovered by ammonium sulfate precipitation, subjected to immunoprecipitation with the type X collagen antiserum, and analyzed by electrophoresis and Auorography. X, the single subunit of type X collagen.
Type X Collagen Synthesis We next asked whether the induction of type X collagen RNA during RA treatment was accompanied by type X collagen synthesis and whether type X collagen gene expression was induced in every chondrocyte or a fraction of them. Secondary chondrocyte cultures were treated for 2 and 3 days with 10 or 35 nM RA and labeled during the last 24 h of culture with [3H]proline along with untreated cultures. The radiolabeled collagenous proteins secreted into the culture medium were recovered by ammonium sulfate precipitation, processed for immunoprecipitation, and analyzed by SDS-gel electrophoresis and fluorography. While, as expected, the control cultures did not synthesize appreciable amounts of type X collagen (Fig. 5, lane l), the RA-treated cells produced this collagen in amounts dependent on both length and dose of treatment (Fig. 5, lanes 2-5). Cells treated for 3 days with 10 or 35 nM RA synthesized more type X collagen (Fig. 5, lanes 4 and 5) than cells treated for 2 days (Fig. 5, lanes 2 and 3); likewise, cells treated with 35 nM RA synthesized more protein at either time point than cells treated with 10 nM RA. The data complement well those on type X collagen RNA levels. To determine the proportion of cells induced to synthesize type X collagen by RA treatment, cultures were treated with 35 nM RA for 1,2, and 3 days. Control and treated cultures were then detached simultaneously from their respective tissue culture dishes, allowed to
adhere for 10 min to polylysine-coated dishes, and processed for immunocytochemistry using the type X collagen antiserum. This procedure allowed the chondrocytes to assume a round cell shape before they were fixed and processed for immunofluorescence, enabling us to minimize equivocal results due to varying cell morphology and thickness. As expected, most of the control untreated chondrocytes failed to stain positively with the type X collagen antiserum; occasionally, one immunopositive cell was observed (Figs. 6A and 6B, arrow). No major change in the number of type X collagen-producing cells occurred after 1 day of treatment (Figs. 6C and 6D). After 2 and 3 days of treatment, however, many immunopositive cells were present (Figs. 6E and 6F, and 6G and 6H, respectively), amounting to approximately 30-35% of the cell population by Day 3 of treatment. When companion cultures were stained with type II collagen antibodies, all the cells present exhibited positive immunostaining (Figs. 7A and 7B), thus showing that the entire cell population was composed of functional chondrocytes. Alkaline Phosphatase Activity In addition to initiating production of type X collagen, mature chondrocytes increase synthesis of alkaline phosphatase (APase), an enzyme involved in cartilage mineralization [15]. We thus asked whether RA treat-
RETINOIC
ACID
AND
CHONDROCYTE
MATURATION
43
FIG. 6. Phase t A, C, E, and G) and immunofluorescence (B, D, F, and H) micrographs of chondrocytes treated with 35 nMRA for up to 3 days and then processed for immunofluorescence with the type X collagen antiserum. A and B, control culture; C and D, 1 day of treatment; E and F, 2 days of treatment; and G and H, 3 days of treatment. Arrows point to immunopositive cells viewed under phase and fluorescence microscopy; note that the average size of the positive cell is similar to that of surrounding cells that do not stain positively. Bar, 150 pm.
ment of immature chondrocytes would lead to increased APase activity, in addition to inducing type X collagen gene expression. Secondary chondrocyte cultures were treated with 10,35, and 100 nM RA for 1,2, and 3 days.
At each time point analyzed, about 90% of the enzymatic activity was recovered in the cell layer compartment and the remaining 10% in the culture medium. RA treatment led to increased cell-layer-associated APase
44
PACIFIC1
ET
AL.
Within 3 days of treatment, there occurs a coordinate increase in type X collagen RNA levels, type X collagen synthesis, and the percentage of type X collagen-producing cells in these cultures. The treatment also leads to increased APase activity occurring as early as 24 h from the start of treatment. These changes in phenotypic expression are not accompanied by major changes in gene expression of type II collagen and the large aggregating proteoglycan (see below), suggesting that they represent specific changes in gene expression. The data in this study strengthen our conclusion [20] that RA is involved in cartilage maturation as an inducer or promoter of traits typical of mature chondrocytes. The mechanisms by which RA may exert this role remain to be elucidated. The RA effects on cell phenotype are thought to be mediated by nuclear RA receptors (RARs) which bind to specific regulatory sequences (RA responsive elements) and bring about changes in gene expression (43, 441. In preliminary experiments using a mouse gamma RAR cDNA (kindly provided by Dr. P. Chambon) [45,46], we have established that mature sternal chondrocytes contain gamma RAR transcripts (Pacifici, unpublished). Therefore, it is possible that the RA effects on type X collagen gene expression and alkaline phosphatase activity described above may also be mediated by gamma RAR in chondrocytes. The type X collagen gene, and its regulatory sequences in particular, have not yet been analyzed in great detail [32]. Our data predict that RA responsive elements may be present in the gene and may be involved in the
FIG. 7. Phase (A) and immunofluorescence (B) micrographs of chondrocytes stained with type II collagen antibodies. Note that every cell present in the culture produces this collagen type. Bar, 30 pm.
activity in a time and dose-dependent fashion (Fig. 8). While 10 nM RA increased APase activity only after 3 days of treatment, 35 nM RA was effective at each time point analyzed. Interestingly, 100 nM RA increased APase activity during the first 2 days of treatment but was followed by a decrease in activity, probably reflecting the adverse effects of this high RA dose on the chondrocyte phenotype. Clearly, RA treatment leads to increased APase activity in immature sternal chondrocytes. Because this can be observed at 24 h of treatment, it appears to precede the induction of type X collagen synthesis, which is most prominent at 48 h of treatment. DISCUSSION
The data in this study show that RA at physiological doses is a rapid inducer of maturation-associated traits in cultures of immature caudal sternal chondrocytes.
RA
treatment
1 day
2 days
3 days
FIG. 8. Effects of RA treatment on APase activity. Chondrocytes grown on 3%mm dishes were treated with the indicated concentrations of RA for 1, 2 and 3 days. APase activity associated with the cell layer was determined using p-nitrophenyl phosphate (pNP) as a substrate. Data are expressed as nanomoles of pNP hydrolysis per 30 min per microgram DNA. *Significantly different from control values, P > 0.05.
RETINOIC
ACID
AND
CHONDROCYTE
changes in type X collagen gene expression during RA treatment described here. The Northern blot analysis shown in Fig. 4 indicates that proteoglycan gene expression is not markedly altered by treatment of immature chondrocytes withphysiological doses of RA. However, previous studies had found that proteoglycan synthesis measured by radiosulfate incorporation decreases 50-60% in cultured rabbit growth plate [13] or rat costal chondrocytes [47] treated with 10 nh4 RA for 3 to 4 days. Other parameters such as polygonal morphology or type II collagen production were not altered by this RA treatment [ 131. The seemingly different effects of physiological RA doses on proteoglycan expression in this and previous studies could reflect species-specific differences. Alternatively, they could be due to the fact that proteoglycan production measured by radiosulfate incorporation is probably more sensitive than Northern blot analysis. Lastly, this RA treatment may not greatly affect expression of proteoglycan core protein mRNA but could still lead to decreased radiosulfate incorporation by interfering with other steps involved in proteoglycan biosynthesis and secretion. Our immunofluorescence data (Fig. 6) demonstrate that only 30-35% of the cells appear to be sensitive to RA treatment and activate type X collagen production. Two possibilities can account for this phenomenon. The first is that the secondary cultures used in these experiments may cont.ain chondrocytes able to respond to RA and chondrocytes unable to do so. For example, some cells may contain RARs while other cells do not. The latter cells may require further growth in culture before they acquire the ability to produce RARs and respond to RA treatment. If this is so, this possibility would imply that sensitivity to RA treatment is maturation-dependent and changes during chondrocyte maturation [ 131. The alternative possibility is that the cultures contain chondrocytes that differ in their sensitivity to RA treatment, such that higher doses of RA are necessary to trigger their response. These possibilities are being tested in ongoing experiments. A close examination of Fig. 6 also reveals that the chondrocytes engaged in type X collagen production in both control and RA-treated cultures are similar in size to the chondrocytes that do not synthesize this collagen at detectable levels. This observation would seem to be in sharp contradiction with the widely accepted view that type X collagen is produced only by hypertrophic chondrocytes. Our data, however, are in full agreement with those in a recent study using in situ hybridization in which type X collagen transcripts were found to be already present in late proliferating, nonhypertrophic, oval-shaped chondrocytes in chicken tibia1 growth cartilage [ 181. We have confirmed this observation and detected by immunohistochemistry significant amounts of type X collagen in late proliferating chondrocytes and
45
MATURATION
their surrounding matrix (unpublished observations). Together, the data indicate that cell hypertrophy and initiation of type X collagen production are not coordinately regulated [18, 241 and that maturing chondrocytes can initiate type X collagen synthesis before they have completed the process of cell hypertrophy. If left untreated, the sternal chondrocyte cultures used in this study would have eventually activated type X collagen gene expression at readily detectable levels; however, they would have required approximately 4 additional weeks of culture to do so [24,48]. One interpretation of this phenomenon is that chondrocytes can progress toward maturity in either an RA-dependent or RA-independent manner: RA would promote this process but is not essential. An alternative possibility is that other factors or nutrients could substitute for RA [49, 501. One such nutrient could obviously be retinol, which is normally present in serum. Lastly, it may be that several exogenous cues (including RA) participate in the induction and promotion of chondrocyte maturation and gene expression of maturation-associated traits both in vitro and in ho. The rate at which chondrocytes would progress toward maturity could depend on whether all such cues are present, whether they are present in optimal concentrations, and perhaps more importantly, whether the chondrocytes have acquired the ability to respond to them. We thank Drs. I’. Leboy, Y. Ninomiya, B. R. Olsen, and M. Young for their generous gifts of plasmids; Dr. T. Linsenmayer for the type II collagen antibodies; and Dr. W. Abrams for help with amino acid sequencing. This study was supported by grants from the National Institutes of Health. REFERENCES 1.
Kochhar,
D. M. (1967)
Teratology
7, 2899298.
2. Tamarin,
A., Ctawley, A., Lee, J., and Tickle, Exp. Morphol. 84, 105-123.
bryol.
I Hassell, ‘3 c’fI1
Res.
4. Pacifici, Cell Res.
J. R., Pennypacker,
J. P., and Lewis,
C. (1984)
J. Em-
C. A. (1978)
Exp.
112, 409-417. M.,
Cossu.
G., Molinaro,
M.,
and Tato,
F. (1980)
Exp.
129, 469-474.
