Doxycycline inhibits type X collagen synthesis in avian hypertrophic chondrocyte cultures.

Doxycycline, a member of the tetracycline family, has been shown to reduce a type X collagen epitope as detected by immunohistochemistry with a monoclonal antibody in an avian explant culture system (). It was also shown to decrease collagenase and gelatinase activities and thus matrix degradation. This study investigates the effect of doxycycline on type X collagen synthesis in monolayer cultures of hypertrophic chondrocytes. Protein synthesis was evaluated by radioisotopic labeling during doxycycline, tetracycline, or minocycline treatment. Radiolabeled proteins were analyzed by gel electrophoresis, and total collagen was quantitated by hydroxyproline analysis. Additionally, the synthesis of type X collagen was measured by immunoprecipitation. Doxycycline was found to inhibit type X production more effectively than either of the other tetracyclines at comparable dose levels. Furthermore, type X collagen was inhibited more than other collagens, non-collagenous proteins and proteoglycans, with maximal inhibition at 80 microg/ml and an IC50 of 7 microg/ml. This inhibition by doxycycline was specific for type X collagen at 10 microg/ml, and the pattern was distinct from cycloheximide, a recognized inhibitor of protein translation. This suppression of type X collagen could not be overcome by excess extracellular calcium, conditions that have been demonstrated to induce synthesis of this protein (2).

Several of the tetracycline derivatives have demonstrated the ability to reduce collagenase and gelatinase activity associated with degradation of tissue matrices (1,(3)(4)(5)(6). Initial observations of these inhibitory activities were made in gingival fluids of patients who were prescribed minocycline (7). Further investigations demonstrated that the activities were present not only in gingival secretions but also in synovial fluids of patients taking the antibiotic (8). In a canine model of osteoarthritis, Yu et al. (9) showed that doxycycline not only inhibited collagenase activity and collagen turnover, but also acted to ameliorate the associated damage to the bone. This inhibitory activity appears to be independent of its antibacterial properties, since a chemically modified tetracycline (CMT-1), retains this inhibitory ability, though it is not an active antimicrobial agent (3,5).
Doxycycline is a semi-synthetic derivative of tetracycline that demonstrates a broad range of antibacterial activity. The antibacterial action of these compounds, particularly the early derivatives, chlortetracycline, oxytetracycline, and tetracycline, has been studied and reviewed extensively (10 -14). Early work in bacteria found that all three of these agents inhibit protein synthesis. This is due to binding of the 30 S ribosomal mRNA complex at the donor site (A site) in prokaryotes (15). The tetracycline backbone consists of a polycyclic ring structure with multiple side groups that lend it a variety of activities. The oxygens of the A, B, and C rings at positions 11,12 and the carboxamide group at position 2 all appear to play a role in metal complexation (16,17). This is apparently responsible for much of the antimicrobial activity, for changes in any of these side chains result in loss of antibiotic activity (18). Doxycycline has a chemical structure identical to the parent compound, tetracycline, except that it contains a hydroxyl group in the 5 rather than in the 6 position.
Type X collagen is a short, homotrimeric collagen that is composed of 59-kDa chains. It is produced by hypertrophic chondrocytes that are associated with bone growth and development (19 -23). Type X collagen is not made by less mature chondrocytes, nor is it synthesized by the chondrocytes which occupy the uncalcified regions of the adult articular cartilage matrix (24,25). Thus, this collagen is exclusively produced at the bone-cartilage interface and may be considered a specific marker for hypertrophic chondrocytes.
Although the function of type X collagen is not known, its location suggests it is important in the organization of the growth plate cartilage. An example of this importance may be seen in individuals afflicted with Schmid metaphyseal chondrodysplasia. In this disease, affected individuals have a shorter stature and disorganization of the growth plates (26). Genetic analyses of these patients revealed defects in the type X collagen gene (27)(28)(29). The most recent of these studies by Chan et al. (29) identified a mutation in the non-helical carboxyl domain of type X collagen in one patient. A full-length transcript of the defective gene was generated by polymerase chain reaction and translated in vitro. The resulting collagen chains were unable to assemble in triple helical molecules and were secreted inefficiently. Thus the gene defect directly resulted in the production of dysfunctional type X collagen molecules. Other evidence for the importance of type X collagen function lies in transgenic mice experiments (30). The introduction of a type X collagen transgene containing deletions resulted in disorganized growth plates and deformed skeletons of the affected mice. However, other investigators using type X collagen null mice, in which the gene was completely deleted, observed no phenotypic changes in the animals (31). This might suggest that there are differences in the importance of type X collagen between species, at least in humans and mice.
In addition to type X collagen, the hypertrophic stage of chondrocyte differentiation is associated with the production of highly active matrix metalloproteinases (32)(33)(34)(35). These enzymes contribute to the rapid degradation of cartilage during the process of endochondral ossification. Therefore, the differentiation of chondrocytes to this hypertrophic stage is likely to be detrimental to cartilage tissue. Such may be the case in osteoarthritis where type X collagen has been identified in the diseased cartilage (24,25,36). The synthesis of type X collagen by articular chondrocytes in osteoarthritis indicates that these chondrocytes have become hypertrophic and may be contributing to the degradation of the cartilage tissue. Any agents which interfere with this differentiated state may then be beneficial to the osteoarthritic patient.
Our recent investigations using an avian tibial explant culture system showed that doxycycline might be a useful agent to interfere with chondrocyte hypertrophy, since it altered the detection of type X collagen epitopes and decreased the matrix metalloproteinase activities in the cultures (1). Consequently, the destruction of the cartilage tissue in this system was effectively blocked. It was unclear if the reduced antibody staining for type X collagen was due to proteolytic loss or masking of the type X collagen epitope or to a reduction of protein synthesis.
The present studies investigate the ability of doxycycline to inhibit type X collagen synthesis in monolayer cultures of avian hypertrophic chondrocytes. Tetracycline and minocycline were also examined for comparison in this regard. Additionally, other proteins were examined to determine the specificity for inhibition of type X collagen. Additional experiments utilizing excess calcium ions were also performed to evaluate chelation as a potential mechanism by which doxycycline may alter type X collagen production.

EXPERIMENTAL PROCEDURES
Cell Cultures-Hypertrophic chondrocytes were obtained from the cartilage beneath the bony sheath of 12-day embryonic chick tibiae. The tissue was enzymatically digested to yield single cell suspensions, plated at subconfluent densities and allowed to recover overnight with 20% bovine calf serum in Dulbecco's modified Eagle medium (Life Technologies, Inc. number 31600-0083, containing: 2 g/liter glucose; 100 units/ml penicillin; 0.1 mg/ml streptomycin; 2.5 g/ml amphotericin B) (2). The following day, cells were treated with 1 mg/ml of hyaluronidase for 1 h to facilitate attachment of nonadherent cells (2). Cells were then rinsed with Dulbecco's modified Eagle medium and treated with either doxycycline, minocycline, or tetracycline hydrochloride (Sigma) at doses of 5, 10, 20, 40, 80, and 160 g/ml in the presence of 50 g/ml ascorbate; 100 g/ml ␤-aminopropionitrile; 0.5% bovine calf serum in Dulbecco's modified Eagle medium and appropriate radiolabel for 24 h. Comparison cultures contained cycloheximide (0.01-0.1 g/ml), EGTA (80 M), or additional calcium chloride (10 mM) to yield a final calcium ion concentration of 11.8 mM.
Protein Synthesis-To monitor overall protein synthesis, cultures were radiolabeled with 50 Ci/ml of L-[2,3,4,5-3 H]proline (Amersham Corp., 101 Ci/mmol) or 25 Ci/ml [ 35 S]methionine (ICN, translabel, 1183 Ci/mmol) for 24 h in the presence of either doxycycline, minocycline, or tetracycline hydrochloride (Sigma) as described above. The medium was removed and the extracellular matrix proteins extracted with 0.15 M phosphate buffer, pH 7.6, for 2 h at 4°C. The remaining cell layer consisting of intracellular and membrane proteins was solubilized in 1% Triton X-100 (20). Proteinase inhibitors were added to each fraction (10 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 5 mM N-ethylmaleimide) prior to dialysis against Tris buffer (50 mM Tris, 0.4 mM NaCl, and 5 mM EDTA, pH 7.5) to remove excess unincorporated radiolabel. Radiolabel incorporation into nondialyzable proteins was determined by liquid scintillation counting for each fraction. Total radiolabel incorporation into cultures was determined by summing the radiolabel incorporation into each fraction (media, phosphate, and Triton). Additionally, aliquots of samples containing equal counts/min and dithiothreitol were subjected to 7% SDS-polyacrylamide gel electrophoresis (37) followed by fluorography using Amplify (Amersham, United Kingdom (UK)) to visualize radiolabeled proteins.
Collagen Analysis-Total collagen synthesis was measured from the [ 3 H]hydroxyproline content in each fraction of the cultures. The radiolabeled hydroxyproline and proline contents in each fraction were determined by cation exchange high pressure liquid chromatography. Briefly, aliquots of each fraction from [ 3 H]proline-labeled samples were dialyzed against 50 mM acetic acid and then hydrolyzed at 108°C for 18 h in 6 N HCl. Samples were then lyophilized to dryness and resuspended in 0.1 M citrate buffer. [ 3 H]Proline and [ 3 H]hydroxyproline were separated on a cation exchange column (Vydac, number 401TP104; 10 m, 4.6 ϫ 250 mm) using an isocratic buffer consisting of 92% 0.1 M citric acid, 8% 0.1 M trisodium citrate. Total collagen in each culture was calculated by summing the total counts/min attributable to hydroxyproline in each fraction (media, phosphate extract, and Triton extract). [ 3 H]Type X collagen was quantitated by immunoprecipitation with monoclonal antibody (X-AC9) coupled to Sepharose beads (2,38). Approximately 75,000 -100,000 cpm from each fraction were incubated with antibody beads (50 l) and bound [ 3 H]type X collagen released with 4 M guanidine hydrochloride following centrifugation and several washes with incubation buffer (1 M NaCl, 50 mM EDTA, 50 mM Tris, 1 mg/ml Chaps, 1 pH 7.5). Both the supernatants containing incubation buffer and the guanidine rinse were subjected to liquid scintillation counting to determine the percent of [ 3 H]type X collagen in each sample. Total type X collagen in each culture was determined by summing the [ 3 H]type X collagen in each fraction (media, phosphate extract, and Triton extract). Other collagens were estimated by subtracting the [ 3 H]type X collagen from total [ 3 H]collagen. A proline hydroxylation rate of 40% was used for these calculations (39).
Proteoglycan Synthesis-Cultures were incubated in the presence of 25 Ci/ml 35 SO 4 (ICN, 1.277 Ci/ml carrier free) for 24 h to determine radiolabel incorporation into proteoglycans. The medium was removed and the cell layer extracted with 4 M guanidine hydrochloride. Aliquots were then subjected to either PD-10 chromatography (40) or Alcian blue precipitation (41) to determine incorporation of radiolabel into large molecular weight proteoglycans.

RESULTS
Control Cultures-Hypertrophic chondrocyte cultures radiolabeled with [ 3 H]proline maintained their polygonal shape and produced collagens characteristic of the growth plate. Types II and X were the major collagen products found upon electrophoresis of radiolabeled proteins. The [ 3 H]collagen represented approximately 60% of the total proline incorporation as determined by hydroxyproline analysis. Type X collagen was detected in all fractions (media, phosphate extract, and triton extract); however, the majority was present in the culture medium (94%).
Protein Synthesis-Maximal reduction of [ 3 H]proline incorporation into all fractions (media, phosphate extract, and Triton extract) of treated cultures occurred with a doxycycline dose of 80 g/ml (Fig. 1). Total [ 3 H]proline incorporation into the medium and phosphate fractions was almost completely inhibited at the maximal dose, whereas the Triton fraction was only reduced by approximately 50%. Doxycycline also inhibited total [ 3 H]proline incorporation by 50% (IC 50 ), at 30 g/ml. For comparison, the effects of tetracycline and minocycline on [ 3 H]proline incorporation were also examined (Fig. 2). Neither tetracycline nor minocycline significantly affected [ 3 H]proline incorporation until higher doses were attained, 40 and 160 g/ml, respectively. Even at these relatively high levels, total [ 3 H]proline incorporation could only be inhibited to 50% of control levels, whereas doxycycline was able to inhibit incorporation by 80%. Because [ 3 H]proline incorporation reflects collagen synthesis more than other proteins, another radiolabel, [ 35 S]methionine, was also used to monitor protein synthesis. In contrast to the radiolabeled proline incorporation, the total incorporation rate of [ 35 S]methionine was not affected until a dosage of approximately 40 g/ml was reached (Fig. 3).
Collagen Analysis-Quantification of type X collagen in [ 3 H]proline labeled samples shows type X collagen to be inhibited approximately 30% at 5 g/ml, 70% at 10 g/ml, and nearly 100% at 80 g/ml of doxycycline (Fig. 3). Gel electrophoresis of [ 35 S]methionine-labeled proteins synthesized in the presence of doxycycline revealed a dramatic decrease in type X collagen at 5-10 g/ml (Fig. 4). By subtracting the [ 3 H]type X collagen counts/min from the total [ 3 H]collagen determined from hydroxyproline analysis, the synthesis of collagens other than type X collagen was determined. In addition, Fig. 3 also shows that like type X collagen, other collagens were also inhibited by doxycycline. The inhibition, however, was less at the dose which dramatically affects type X collagen (10 g/ml). This can also be seen upon examination of the electrophoresis pattern of [ 35 S]methionine-labeled types II and X collagens (Fig. 4, lane 3). When cycloheximide (Fig. 4, lane 9) was used to inhibit protein synthesis (ϳ30% [ 35 S]Met incorporation) at comparable levels to doxycycline, all protein bands appeared to be equally decreased. This is strikingly different from the pattern observed with doxycycline treatment where collagen bands, particularly type X, are significantly reduced (Fig. 4,   lane 5). This suggests that the inhibition of type X collagen synthesis is somewhat selective and that doxycycline does not act like cycloheximide by blocking translational activity nonspecifically. 35 SO 4 Incorporation-The pattern observed for 35 SO 4 incorporation into proteoglycans was similar to that observed for [ 35 S]methionine incorporation where inhibition was not reached until a dose of 40 g/ml doxycycline (Fig. 3). Thus, although proteoglycans are a major extracellular matrix product of chondrocytes, they do not appear to be inhibited at concentrations that inhibit collagen synthesis, particularly type X collagen. This observation again demonstrates the selectivity of doxycycline inhibition of type X collagen synthesis at low concentrations.
Calcium Additions-Since the tetracyclines are known to act as chelating agents (42,43), excess calcium ions in the form of calcium chloride were added to the culture medium to test if elevated concentrations of this divalent cation might overcome ]methionine-labeled medium proteins from cultures treated with 0, 5, 10, 20, 40, 80, and 100 g/ml doxycycline, respectively. Lanes 8 and 9 represent radiolabeled medium from cultures treated with cylcoheximide, 0.1 and 0.01 g/ml, respectively. Lane 10 represents [ 3 H]proline-labeled collagen types II: procollagen (p␣1(II)), procollagen containing the carboxyl terminus (pC␣1(II)), procollagen containing the amino terminus (pN␣1(II)) and fully processed type II collagen (␣1(II)) and type X standards. the inhibitory effect of doxycycline. A higher calcium concentration (11.8 mM) stimulated the synthesis of type X collagen in control cultures (compare lanes 1 and 2, Fig. 5) without doxycycline as reported previously (2). However, fluorograms of medium from doxycycline-treated cultures show that type X collagen release into this fraction was inhibited at 40 g/ml (lane 3), even with a 125-fold molar excess of calcium ions over doxycycline (lane 4). Neither tetracycline (lanes 5 and 6) nor EGTA (lanes 7 and 8) had the same effect on type X collagen at molar concentrations equivalent to doxycycline. These results suggest that doxycycline did not inhibit type X synthesis by chelating extracellular calcium. The higher intensity of type II collagen apparent in this fluorogram (lane 3) compared with Fig. 4 (lane 5) is due to the increased proportion of type II collagen in the proline-labeled medium fraction when type X collagen is lost and equal cpm are loaded in each lane. DISCUSSION The results of this study demonstrate that doxycycline inhibits type X collagen synthesis by hypertrophic chondrocytes in monolayer cultures. This inhibition occurs at a relatively low dose where the synthesis of other proteins is apparently not affected. Neither the parent compound tetracycline nor minocycline, another tetracycline derivative, were able to inhibit macromolecular synthesis in hypertrophic chondrocytes at dose levels comparable with doxycycline. Doxycycline also inhibited type X collagen synthesis more specifically than cycloheximide, a recognized inhibitor of protein translation. Although the mechanism by which this inhibition occurs is not clear, it does not appear to be reversed by excess calcium ions, nor is it mimicked by the calcium chelator EGTA.
The most specific inhibitory effect observed occurs at a relatively low dose (5-10 g/ml), where type X collagen synthesis is affected. The production of this protein in the hypertrophic chondrocyte population is inhibited almost exclusively at this dose level. As the doxycycline concentration is increased, type II collagen begins to be affected, and at still higher doses (40 -80 g/ml), the synthesis of other proteins is affected, including proteoglycans. Our results are consistent with the serum-free explant culture studies in which doxycycline treatment reduced the detection of the type X collagen epitope in the cartilage matrix (1). This work would suggest that the observed reduction in type X collagen epitope was most likely due to the inhibition of protein synthesis rather than proteolytic loss or masking of the type X collagen epitope or to inefficient secre-tion of the protein.
Another display of selectivity becomes evident when two other structurally similar tetracyclines were tested. Doxycycline is much more potent than either tetracycline or minocycline at inhibiting type X collagen synthesis. It is interesting that tetracycline and doxycycline would have such different effects on hypertrophic chondrocytes when their chemical structures are so similar. It is possible that the change in the position of the hydroxyl group in doxycycline lends more specificity in binding, so that its action is more direct, or it may be due to the increased lipophilicity which the change imparts (18). The increase in antibacterial activities against Gramnegative bacteria correlates well with increasing lipophilicity of the derivative (18). Studies on the metabolism and disposition of doxycycline in mammalian systems have shown that it is absorbed more readily and has a longer half-life in the body than the less lipophilic parent compound, tetracycline (44). Although a greater lipophilicity may allow greater penetration of doxycycline into hypertrophic chondrocytes, and possibly increase its action, it does not explain the increased potency observed over minocycline, one of the most lipophilic tetracycline derivatives. It is surprising that minocycline does not appear to inhibit type X collagen in this system, as it has been demonstrated to be one of the more active derivatives that can inhibit the metalloproteinases, collagenase, and gelatinase both in vitro (8) and in vivo (4,8).
It is likely that doxycycline can elicit a variety of different responses depending on which cells are treated and the dosage employed. One example may be seen in epithelial keratinocytes where low doses (5 g/ml) of doxycycline decreased levels of gelatinase A mRNA (45). However, much higher concentrations (50 g/ml) were required to reduce purified enzyme activity from these cultures. Doxycycline also inhibited growth in endothelial cells (46) and suppressed several neutrophil functions (47) in the same dose range that inhibited type X collagen synthesis in hypertrophic chondrocytes. Other tetracyclines, particularly minocycline, have been used to restore osteoblast (48) and osteoclast (49,50) structures and functions in a streptozotocin-induced diabetic rat model. Several investigators have suggested that the metal binding capacity, particularly with calcium, may play a role in these different responses (3,47,51).
It is intriguing to speculate that the mechanism by which doxycycline inhibits type X collagen synthesis may be mediated through calcium. It has been documented that type X collagen binds calcium (52) and that its synthesis can be modulated by extracellular calcium concentrations (2). In addition to type X collagen, the synthesis of other cartilage collagens was reduced by doxycycline treatment. Mollenhauer et al. (53) have shown that collagen synthesis in chondrocyte cultures is also dependent on the concentration of calcium in the culture medium. Furthermore, Clark et al. (54) have shown that the Ca 2ϩ -ATPase inhibitor thapsigargin alters the intracellular and endoplasmic reticulum calcium concentrations, which results in selective inhibition of collagen synthesis. However, in our system the addition of excess calcium ions could not overcome the inhibition of type X collagen. It seems unlikely then that the doxycycline action on type X collagen occurs via a chelation of extracellular calcium.
Clearly there is more work to be done investigating the mechanisms by which the tetracyclines exert their effects on cartilage. It would also be of interest to investigate the different potencies of the tetracycline analogs, which will help define their mechanisms of action. Clinically these findings are important as they demonstrate that type X collagen can be inhibited at physiological levels of doxycycline found in both serum  7 and 8) in the presence of [ 3 H]proline for 24 h as described under "Experimental Procedures." Lane 9 contains [ 3 H]proline-labeled collagen types II: procollagen (p␣1(II)), procollagen containing the carboxyl terminus (pC␣1(II)), procollagen containing the amino terminus (pN␣1(II)), and fully processed type II collagen (␣1(II)) and type X collagen standards. and in synovial fluid (9). These same levels have also been shown to inhibit the metalloproteinases associated with the arthritic disease state. Since approval for doxycycline has been sought for clinical trials, it becomes important to understand the mechanism by which it inhibits type X collagen synthesis and any other effects on chondrocyte metabolism.