Cortisol Increases Interstitial Collagenase Expression in Osteoblasts by Post-transcriptional Mechanisms

Glucocorticoids regulate both bone formation and bone resorption. In osteoblasts, they inhibit type I collagen synthesis; however, there is limited information about their effects on interstitial collagenase, the enzyme that degrades type I collagen. We used primary cultures of osteoblast-enriched cells from fetal rat calvariae (Ob cells) to study the effects of cortisol on collagenase expression. Northern blot analysis showed that cortisol increased collagenase transcript levels in a dose- and time-dependent manner, which was paralleled by an increase in immunoreactive metalloproteinase in the culture medium. Cortisol increased the half-life of collagenase mRNA from 6 to 12 h in transcription-arrested Ob cells. In contrast, cortisol modestly decreased collagenase gene transcription after 24 h of treatment. The up-regulation of collagenase by cortisol is osteoblast-specific, since the glucocorticoid decreased phorbol 12-myristate 13-acetate-induced collagenase mRNA expression in rat fibroblasts, a result that agrees with other studies of collagenase gene regulation in fibroblastic cells. In conclusion, cortisol increases interstitial collagenase transcript levels by post-transcriptional mechanisms in osteoblastic cells. Our data demonstrate that glucocorticoids regulate collagenase gene expression in a novel tissue-specific manner, further highlighting the differences in gene regulation between osteoblastic and fibroblastic cells.

Glucocorticoids have marked effects on bone metabolism, regulating bone formation and bone resorption (1). In vitro studies have shown that glucocorticoids have complex effects on osteoblast gene expression, and these effects are dependent on the stage of osteoblast growth and differentiation and on the cell model and culture conditions used (2). Glucocorticoids induce cells of the osteoblastic lineage to differentiate into mature cells expressing the osteoblastic phenotype (3)(4)(5). However, their inhibitory actions on multiple aspects of osteoblastic function have a major impact on bone mass. Glucocorticoids inhibit cell replication, depleting a cell population capable of synthesizing bone collagen, and they inhibit ␣1(I) collagen expression by transcriptional and post-transcriptional mechanisms (6,7).
Additionally, glucocorticoids regulate bone collagen degradation, although the effects have varied with the models and culture conditions used (1). Recently, glucocorticoids were shown to increase interstitial collagenase (matrix metalloproteinase 1) transcript levels in osteoblast cultures (8). This effect is observed only in osteoblasts; indeed, glucocorticoids inhibit transcription of interstitial collagenase in nonskeletal fibroblasts (9 -12). This suggests novel tissue-specific regulation of collagenase by glucocorticoids. Matrix metalloproteinases and their inhibitors are considered active participants in the degradation of osteoid, and interstitial collagenases are the only proteases known to initiate the degradation of type I collagen at neutral pH (13). Thus, up-regulation of osteoblast collagenase may play a role in the bone loss associated with pathological glucocorticoid excess. However, the mechanisms by which glucocorticoids stimulate the expression of collagenase in osteoblasts are unknown.
Investigations of the regulation of interstitial collagenase are critical for understanding the role glucocorticoids play in bone remodeling. Since glucocorticoids differentially regulate collagenase transcripts in osteoblasts and fibroblasts, determining the mechanisms by which this occurs will contribute to our knowledge of cell type-specific gene regulation. In this study, we examined the mechanisms of action of cortisol on interstitial collagenase synthesis in cultures of osteoblast-enriched cells from fetal rat calvariae (Ob cells) 1

MATERIALS AND METHODS
Culture Technique-The culture method used was described in detail previously (14). Parietal bones were obtained from 22-day-old fetal rats immediately after the mothers were sacrificed by blunt trauma to the nuchal area. This project was approved by the Institutional Animal Care and Use Committee of Saint Francis Hospital and Medical Center. Cells were obtained by five sequential digestions of the parietal bone, using bacterial collagenase (CLS II, Worthington Biochemical, Freehold, NJ). Cell populations harvested from the third to the fifth digestions were cultured as a pool at a density of ϳ10,000 cells/cm 2 and have been previously shown to have osteoblastic characteristics (14). In one set of experiments, fibroblastic cells derived from collagenase digestion of skin from 22-day-old fetal rats were cultured and tested (9). Cells were cultured in Dulbecco's modified Eagle's medium supplemented with nonessential amino acids (Life Technologies, Inc.) and 10% fetal bovine serum (HyClone, Logan, UT) (14). Except for the nuclear run-off assay, primary cultures of Ob cells were used in all experiments, whereas skin fibroblasts were used after three or four passages. At confluence the cells were rinsed and transferred to serum-free medium for 24 h and then exposed to test or control medium in the absence of serum for 2-48 h. In experiments lasting longer than 24 h, the medium was replaced with freshly prepared test and control solutions. Cortisol, cycloheximide, phorbol 12-myristate 13-acetate (PMA) and 5,6-dichlorobenzimidazole riboside (DRB) (all from Sigma) were dissolved in absolute ethanol, and at dilutions of Ͻ1:10,000 an equal amount of ethanol was added to control cultures.
Nuclear Run-off Assay-Subconfluent cultures of Ob cells were treated with trypsin, harvested, subcultured at a 1:6 dilution, and allowed to grow to confluence as described previously (20). These first passage Ob cells retain the osteoblastic phenotype and respond to cortisol in a manner similar to that of primary cultures (7). At confluence, cells were serum-deprived and treated for 2-24 h, and nuclei were isolated by Dounce homogenization in a Tris-Cl buffer containing 0.5% Nonidet P-40. Nascent transcripts were labeled by incubation of nuclei in a reaction buffer containing 500 M each ATP, GTP, and CTP, 150 units of RNAsin (Promega), and 250 Ci of [ 32 P]UTP (3000 Ci/mM, DuPont) (modified from Ref. 21). RNA was isolated by treatment with DNase I and proteinase K, followed by ethanol precipitation. Linearized plasmid DNA containing ϳ1 g of cDNA was immobilized onto Gene-Screen Plus by slot blotting according to the manufacturer's directions (DuPont). The plasmid vector pGL2-Basic (Promega) was used as a control for nonspecific hybridization, and cDNA for rat ␣1(I) collagen was used as a positive control (22). Equal counts per minute of [ 32 P]RNA from each sample were hybridized to cDNA using the same conditions as for Northern blot analysis and were visualized by autoradiography. The nuclear run-off assay shown is representative of three experiments.
Transient Transfections and Reporter Gene Assays-A NotI/XhoI rat genomic DNA fragment containing 2.1 kb of a recently cloned rat interstitial collagenase promoter 2 was used to drive expression of the luciferase gene in the vector pGL2-Basic. A construct containing the cytomegalovirus promoter driven ␤-galactosidase gene (pCMV␤-Gal, Clontech, Palo Alto, CA) was used to control for transfection efficiency. Ob cells were cultured to approximately 80% confluence and were transiently co-transfected with the collagenase promoter-luciferase construct and pCMV␤-Gal by calcium phosphate/DNA coprecipitation, followed by glycerol shock as described (23)(24)(25). Cells were washed, serum-deprived, and treated for 6 h with control medium or 1 M cortisol. Cell lysates were made using 1 ϫ Reporter Lysis Buffer (Promega), and luciferase activity was measured by injecting luciferase assay reagent (Promega) into a portion of the cell lysate and counting photons using an Optocomp luminometer (MGM Instruments, Hamden, CT) according to the manufacturer's instructions. ␤-galactosidase activity was measured by incubating cell lysates with the chemiluminescent substrate for ␤-galactosidase 3-(4-methoxyspiro[1,2-dioxetane-3,2Ј-tricyclo-[3.3.l.l. 3,7 ]decan]-yl)phenyl-␤-D-galactopyranoside (Galacton; Tropix, Bedford, MA) with the modifications described (24). Luciferase activity was normalized to ␤-galactosidase activity to control for slight variations in transfection efficiency.
Collagenase Immunoassay-Immunoreactive collagenase was determined by an enzyme-linked immunosorbant assay, as described previously (26, 27). Briefly, samples were incubated with antiserum to rat interstitial collagenase and added to the wells of microtiter plates coated with 0.25 g pure rat uterine interstitial collagenase. The plates were incubated overnight at 4°C and washed, and the surface-bound antibody was quantitated by the addition of goat antirabbit ␥-globulin conjugated with alkaline phosphatase. The bound alkaline phosphatase-dependent formation of p-nitrophenol was then determined spectrophotometrically. This assay detects collagenase at concentrations as low as 1 ng/ml of culture medium.
Western Immunoblot Analysis-Ob cells were cultured as described previously and media were stored at Ϫ80°C after the addition of polyoxyethylene sorbitan monolaurate (Tween 20, Pierce) to a final concentration of 0.1%. Proteins were separated by polyacrylamide gel electrophoresis under denaturing, nonreducing conditions and transferred to Immobilon P membranes (Millipore, Bedford, MA) (26). After blocking with 2% bovine serum albumin, the membranes were exposed to a 1:1000 dilution of rabbit antisera raised against rat collagenase (27,28) and then to goat anti-rabbit IgG antisera conjugated to horseradish peroxidase. The blots were washed and developed with a horseradish peroxidase chemiluminescence detection reagent (DuPont), visualized by autoradiography on DuPont Reflection film employing Reflection intensifying screens, and analyzed by densitometry. The Western blot shown is representative of five cultures.
Statistical Analysis-Statistical differences were calculated by analysis of variance, and post hoc examination was performed by the Ryan-Einot-Gabriel-Welsh F test (29,30). Slopes were analyzed by the method of Sokal and Rohlf (31).

RESULTS
Continuous treatment of Ob cells with 1 M cortisol caused a time-dependent increase in collagenase steady state transcripts. Northern blot analysis showed that treatment with cortisol increased collagenase transcripts 2-fold after 8 h, and 7-fold after 16 and 24 h (Fig. 1). This stimulatory effect on collagenase transcripts was sustained for at least 48 h (not shown), and cortisol treatment did not modify the abundance of the 1.9-kb glyceraldehyde-3-phosphate dehydrogenase transcript. The basal unstimulated level of collagenase transcripts decreased with time in culture, possibly due to the inhibitory effect of endogenous insulin-like growth factor I. 3 The cortisolmediated increase in collagenase mRNA levels was dose-dependent and was observed at 0.1 and 1 M cortisol (Fig. 2). In parallel with its effects on collagenase transcripts, cortisol increased levels of the protein in the culture medium of Ob cells treated for 24 h. The concentration of immunoreactive metalloproteinase was below the limit of detection in control cultures but was increased to 44 Ϯ 6 ng/ml (mean Ϯ S.E.; n ϭ 4) in cultures treated with 1 M cortisol. Western blot analysis showed a 5-fold increase in a ϳ57-kDa immunoreactive protein in the medium of cells treated with cortisol (Fig. 3). This band had the same mobility as the rat uterine procollagenase standard, while the lower molecular weight cross-reactive band, which was not regulated by cortisol treatment, may represent another species of metalloproteinase.
To determine if the effect of cortisol on collagenase transcripts was dependent on protein synthesis, confluent cultures of Ob cells were treated with cortisol in the presence or absence of 3.6 M cycloheximide, a dose known to inhibit protein synthesis in Ob cells by at least 85% (32). After 24 h of treatment, cycloheximide alone superinduced collagenase mRNA levels, suggesting that it may stabilize the transcript (33,34). Cotreatment with cortisol reduced the superinduction of collagenase by cycloheximide (Fig. 4).
To determine if cortisol modified the stability of collagenase mRNA in Ob cells, the RNA polymerase II-specific inhibitor DRB was used to arrest transcription, and the decay of collagenase mRNA was monitored by Northern blot analysis (35,36). Serum-deprived confluent cultures of Ob cells were exposed to control medium or to 1 M cortisol for 4 h and then treated with 72 M DRB for up to 12 h in the absence or presence of cortisol at 1 M. In transcription-arrested Ob cells, the half-life of collagenase mRNA was approximately 6 h, and cortisol increased the half-life of the transcript to approximately 12 h (Fig. 5, left panel). A similar increase in collagenase transcript stability was observed in Ob cells treated with cortisol for 12 h prior to the addition of DRB (Fig. 5, right  panel). In both experiments, the slope for the collagenase mRNA decay in the cortisol-treated cells was significantly different from control (31). The decay of glyceraldehyde-3-phosphate dehydrogenase transcripts was the same in control and cortisol treated cultures (not shown). To determine if cortisol modified transcription of the collagenase gene, nuclear run-off assays were performed on nuclei from Ob cells treated with 1 M cortisol for 2, 6, or 24 h. The levels of ␣1(I) collagen gene transcription were used as a control, since cortisol has been shown to decrease transcription of this gene (7). Cortisol did not alter transcription of the collagenase gene after 2 h (not shown) or 6 h (Fig. 6), although it did decrease ␣1(I) collagen gene transcription. After 24 h of treatment, cortisol caused a small decrease in transcription from the collagenase gene.
In human and rabbit fibroblastic cells, glucocorticoids antagonize the induction of collagenase by the protein kinase C agonist PMA (9 -12). To determine if up-regulation of collagenase by cortisol in rat osteoblasts was a species-specific or a cell type-specific phenomenon, rat skin fibroblasts were treated for 6 and 24 h with 0.1 M PMA in the presence or absence of cortisol at 1 M (Fig. 7). Collagenase transcripts in untreated cells were almost undetectable, but treatment with PMA dramatically increased collagenase mRNA after 6 h. Co-treatment with cortisol antagonized this effect, and cortisol alone did not increase collagenase mRNA, documenting the specific nature of the cortisol effect in Ob cells. The ability of cortisol to augment Confluent cultures were serumdeprived and exposed to cortisol or to control medium for 4 h (left panel) or 12 h (right panel) prior to the addition of 72 M DRB. At selected times after the addition of DRB, total RNA from control (E) or cortisol (q) treated cultures was subjected to Northern blot analysis with 32 Plabeled rat collagenase cDNA. Collagenase mRNA was visualized by autoradiography and quantitated by densitometry. Values are mean Ϯ S.E. for three cultures. In the left panel, the slope for DRB ϭ Ϫ0.052, and the slope for DRB ϩ cortisol ϭ Ϫ0.022. These values were significantly different, p Ͻ 0.01. In the right panel, the slope for DRB ϭ Ϫ0.054, and the slope for DRB ϩ cortisol ϭ Ϫ0.029. These values were significantly different, p Ͻ 0.05. or antagonize the PMA induction of collagenase transcripts in Ob cells also was tested (Fig. 8). After 2 h of treatment, PMA at 0.1 M increased collagenase transcripts, while cortisol at 1 M was modestly inhibitory. After 6 h, PMA increased collagenase mRNA levels 20-fold, and co-treatment with cortisol antagonized this effect by approximately 40%. Treatment with PMA for 24 h down-regulated osteoblast collagenase transcript levels to approximately 20% of the untreated control. In contrast, cortisol increased collagenase mRNA by 5-fold at 24 h, and co-treatment with PMA decreased this effect by 60 -80%.
To further characterize the effects of PMA and cortisol on transcription of the collagenase gene in osteoblasts, Ob cells were transiently transfected with a construct containing a 2.1-kb fragment of the rat collagenase promoter driving expression of the reporter gene luciferase. Treatment of transfected cells with cortisol alone for 6 h caused a 25% decreased in luciferase activity (p Ͻ 0.05) (Fig. 9). Treatment of transfected cells with 0.1 M PMA increased luciferase activity 2-fold, and co-treatment with 1 M cortisol antagonized this effect. DISCUSSION Glucocorticoids have significant effects on bone remodeling. Previous work demonstrated that they inhibit ␣1(I) collagen synthesis by transcriptional and post-transcriptional mechanisms, but there is limited information about their effects on collagen degradation and collagenase expression (1,7). In the present study we demonstrated that cortisol causes a time-and dose-dependent stimulation of interstitial collagenase transcripts in cultures of Ob cells, which was paralleled by increased levels of immunoreactive collagenase in the culture medium. These effects were observed at doses of cortisol that modify parameters of osteoblastic differentiated function and at concentrations that were only slightly higher than physiological serum levels of cortisol (3-6). The same doses of cortisol modestly decreased transcripts for tissue inhibitor of metallo- 1-kb fragment of the rat collagenase promoter was used to drive expression of the luciferase gene in the promoterless reporter plasmid pGL2-Basic. Ob cells were transiently co-transfected with collagenase promoter-luciferase plasmid and plasmid containing the cytomegalovirus promoter driving expression of the ␤-galactosidase gene. Transfected cells were treated for 6 h with control medium (C) or with medium containing glucocorticoid (GC), PMA, or PMA plus glucocorticoid (PMA ϩ GC). Luciferase activity was normalized to ␤-galactosidase activity to control for slight differences in transfection efficiency. These results are representative of three experiments. *, significantly different from C, p Ͻ 0.01; **, significantly different from PMA, p Ͻ 0.01. proteinases (TIMP) 1 and did not affect the expression of TIMPs 2 and 3, 3 suggesting that up-regulation of collagenase by glucocorticoids is important in increased extracellular matrix degradation.
Experiments using the RNA polymerase II inhibitor DRB demonstrated that cortisol stabilized collagenase mRNA in transcriptionally arrested Ob cells. In contrast, nuclear run-off assays and transient transfection of a rat interstitial collagenase promoter-reporter gene construct showed that collagenase gene transcription was slightly decreased by cortisol. These results indicate that cortisol increases collagenase expression by increasing transcript stability. The rat collagenase mRNA has a 1.2-kilobase 3Ј-untranslated region that is AU-rich and contains three repeats of the sequence AUUUA (17). In short lived mRNAs, such AU-rich sequences play a role in regulating transcript stability (reviewed in Ref. 37). The human and rabbit collagenase mRNAs contain repeats of AUUUA, and mutation of these motifs in the human transcript increases the stability of the RNA (38). It is probable that the glucocorticoidmediated increase in Ob cell collagenase transcripts involves proteins interacting with such AU-rich regions of the mRNA (39).
Cortisol antagonized the induction of collagenase promoter activity by PMA in Ob cells. These effects are similar to those observed in fibroblasts (10 -12), suggesting that osteoblasts and fibroblasts share a common mechanism for the regulation of collagenase transcription. Like the human interstitial collagenase gene, which is transcriptionally regulated by PMA and glucocorticoids, the rat collagenase promoter contains a TPAresponsive element (10 -12, 40, 41). Components of the AP-1 complex can interact with the TPA-responsive element, and glucocorticoid repression of collagenase gene transcription in fibroblastic cells can be mediated through antagonism of AP-1 (10 -12).
Collagenase gene transcription and mRNA levels were downregulated in Ob cells treated for 24 h with PMA. A nuclear run-off assay showed that after 24 h treatment with PMA, collagenase gene transcription was barely detectable. 4 This is most likely due to depletion of protein kinase C activity following prolonged exposure to phorbol esters. In cells treated with PMA and cortisol for 24 h, the level of collagenase mRNA was intermediate between that of cortisol alone and PMA alone. This suggests that the effects of cortisol and PMA may be additive at this time point and that cortisol and PMA increase collagenase transcripts in Ob cells by distinct mechanisms.
The expression of collagenase in osteoblasts is up-regulated by a number of osteoresorptive agents, including parathyroid hormone, glucocorticoids and prostaglandins (8,40,42). The role of osteoblast-derived interstitial collagenase in the bone compartment is currently being explored, and there is increasing evidence for the coupling of osteoblastic function with osteoclastic bone resorption (reviewed in Ref. 43). Localized extracellular matrix degradation by osteoblast collagenase may provide a means for activated osteoclasts to adsorb to target bone surfaces. Collagenase may also play a role in regulating the availability of bone growth factors. For example, localized matrix degradation may release growth factors sequestered in the matrix, which may stimulate or inhibit osteoblastic function or may activate or be chemotactic for osteoclasts (44,45). Insulin-like growth factors are among the most prevalent growth factors secreted by bone cells, and they stimulate the differentiated function of the osteoblast (45). Their activity can be modulated by insulin-like growth factor binding proteins, the abundance of which can be regulated by proteases, includ-ing Ca 2ϩ -dependent serine proteases and metalloproteinases (46 -49). A study characterizing a parathyroid hormone-regulated receptor for collagenase on a rat osteosarcoma cell line showed that this receptor is responsible for clearance of collagenase from the cellular environment (50). This suggests that the osteoblast maintains a tight control over collagenase activity and that promiscuous expression of collagenase would be undesirable. The importance of appropriately regulated osteoblast collagenase also is suggested by the development of collagenase-expressing bone tumors in transgenic mice overexpressing c-fos (51).
In conclusion, cortisol causes a cell type-specific increase in interstitial collagenase expression in osteoblasts, which is mediated by post-transcriptional mechanisms. These results further highlight the differences in gene regulation between osteoblasts and fibroblasts, two cell types that arise from a common precursor.