Regulation of collagen deposition and lysyl oxidase by tumor necrosis factor-α in osteoblasts

Tumor necrosis factor-α (TNF-α) inhibits osteoblast function in vitro by inhibiting collagen deposition. Studies generally support that TNF-α does not inhibit collagen biosynthesis by osteoblasts but that collagen deposition is in some way diminished. The study investigated TNF-α regulation of biosynthetic enzymes and proteins crucial for posttranslational extracellular collagen maturation in osteoblasts including procollagen C-proteinases, procollagen C-proteinase enhancer, and lysyl oxidase. The working hypothesis is that such regulation could inhibit collagen deposition by osteoblasts. We report that in phenotypically normal MC3T3-E1 osteoblasts, TNF-α decreases collagen deposition without decreasing collagen mRNA levels or procollagen protein synthesis. Analyses of the cell layers revealed that TNF-α diminished the levels of mature collagen cross-links, pyridinoline and deoxypyridinoline. Further analyses revealed that the mRNA expression for lysyl oxidase, the determining enzyme required for collagen cross-linking, is down-regulated by TNF-α in a concentration- and time-dependent manner by up to 50%. The decrease was accompanied by a significant reduction of lysyl oxidase protein levels and enzyme activity. By contrast, Northern and Western blotting studies revealed that procollagen C-proteinases bone morphogenic protein-1 and mammalians Tolloid and procollagen C-proteinase enhancer were expressed in MC3T3-E1 cells and not down-regulated. The data together demonstrate that TNF-α does not inhibit collagen synthesis but does inhibit the expression and activity of lysyl oxidase in osteoblasts, thereby contributing to perturbed collagen cross-linking and accumulation. These studies identify a novel mechanism in which proinflammatory cytokine modulation of an extracellular biosynthetic enzyme plays a determining role in the control of collagen accumulation by osteoblasts.

TNF-␣ 1 is an inflammatory cytokine produced primarily by monocytes and macrophages and also by a variety of mesen-chymal cells. TNF-␣ levels are elevated in various bone disorders such as rheumatoid arthritis, osteoporosis, and periodontitis (1)(2)(3)(4). In bone tissue, TNF-␣ inhibits osteoblast function and increases osteoclastogenesis, thus favoring net matrix destruction (5,6) and the collagenous matrix structure is disrupted by TNF-␣ (7)(8)(9). Reports indicate that TNF-␣ somehow inhibits collagen deposition while having little effect on collagen synthesis, although the mechanisms that contribute to this phenomenon have not been elucidated (6,7).
Type I collagen is the major structural protein in the extracellular matrix of bone tissue. Normal collagen structure and the balance between production, deposition, and degradation of collagen are important in the development and maintenance of skeletal tissue (1). Collagen biosynthesis is a multistep process that involves intracellular posttranslational modifications, assembly of procollagen chains, secretion, extracellular processing, and cross-linking to form a mature functional matrix (10). The mechanisms that control collagen deposition in osteoblasts are not well understood but seem likely to include the regulation of extracellular collagen-biosynthetic enzymes and proteins (11). Moreover, TNF-␣ regulation of enzymes and proteins crucial for posttranslational extracellular collagen maturation such as procollagen C-proteinases and the crosslinking enzyme lysyl oxidase have not been studied directly to our knowledge.
This study demonstrates that TNF-␣ down-regulates lysyl oxidase expression in phenotypically normal osteoblasts. TNF-␣ inhibited lysyl oxidase steady-state mRNA expression by up to 50%, and this was time-and concentration-dependent. The decrease in lysyl oxidase mRNA levels was accompanied by a significant reduction of lysyl oxidase protein and enzyme activity. Although collagen synthesis was not inhibited, less collagen accumulated in the extracellular matrix in the presence of TNF-␣ and the deposited collagen contained diminished levels of mature cross-links, pyridinolines (Pyd) and deoxypyridinolines (Dpd). By contrast, BMP-1, mTLD, and their biological enhancer, PCPE, were not regulated by this cytokine.
These findings identify a novel mechanism in which downregulation of an extracellular collagen biosynthetic enzyme by TNF-␣ contributes to the deposition of abnormal collagen more likely to be degraded. This mechanism could contribute to net bone resorption that occurs in vivo as a consequence of inflammation.

Cell Layer Hydroxyproline and Collagen Cross-links-200,000
MC3T3-E1 cells were plated on 100-mm tissue cultures plates and grown until confluency. Cells then were grown in differentiation medium (␣-MEM) supplemented with 10% fetal bovine serum, 50 g/ml ascorbate, and 10 mM ␤-glycerophosphate with or without 10 ng/ml or 100 ng/ml TNF-␣ for 12 days. Cell layers were hydrolyzed in 6 N HCl at 110°C for 24 h in sealed glass ampules. Hydrolysates were vacuum-dried and then used to determine hydroxyproline levels by colorimetric assays (25) and for mature type I collagen cross-links Pyd and Dpd using the Metra PYD enzyme immunoassay kit (Quidel, San Diego, CA). The cross-link assay is a competitive enzyme-linked immunoassay in which Pyd and Dpd in hydrolyzed cell extracts compete with Pyd coated on a microtiter plate for soluble monoclonal anti-pyridinium cross-link antibody conjugated to alkaline phosphatase. Following a washing step, the substrate p-nitrophenyl phosphate was added for 60 min. The reaction then was stopped by adding 1 N NaOH to each well, and the optical density was measured at 405 nm. The Pyd plus Dpd concentration in each sample was determined from a standard curve run on the same microtiter plate.
Collagen Synthesis Assay-Subconfluent MC3T3-E1 cells were treated with 0, 10, or 100 ng/ml TNF-␣ for 24 h. Cells then were incubated for 2 h with serum-and proline-free ␣-MEM containing 10 Ci/ml [2,3-3 H] proline in presence and absence of TNF-␣. The cell layers were harvested in 0.05 M Tris-HCl, pH 7.6, on ice and homogenized (PowerGen 125, Fisher Scientific, Morris Plains, NJ). Radioactive protein was separated from unincorporated tritiated proline by precipitation with 5% trichloroacetic acid at 4°C. Residual trichloroacetic acid then was removed by 3:1 ethanol/ether washes. The remaining pellet was air-dried and redissolved in 1 N NaOH at 37°C for 5 min. 500 l of incubation buffer (60 mM HEPES, pH 7.2, 0.25 mM CaCl 2 , 1.25 mM N-ethylmaleimide) was added to each sample. 100 l of the sample in the incubation buffer was mixed with 5 ml of scintillation fluid, and total incorporated tritiated proline was measured in a liquid scintillation spectrometer. 200 l of the sample in the incubation buffer was incubated with 25 g/ml purified bacterial collagenase (Worthington Biochem, Freehold, NJ) for 90 min at 37°C under agitation (26). The reaction was stopped with 10% trichloroacetic acid, 0.5% tannic acid on ice and then centrifuged at 400 ϫ g for 5 min. The supernatant, which contained the collagen-derived peptides, was transferred to counting vials with 5 ml of scintillation fluid, and the radioactivity was measured as above. Data for collagen synthesis were normalized to the DNA content.
Regulation Studies-MC3T3-E1 cells were maintained in ␣-MEM supplemented with 10% fetal bovine serum, 1% nonessential amino acids, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were cultivated for no more than three passages from one set of frozen cell stocks. For experiments, 350,000 cells were plated on a 100-mm tissue culture dish and grown to 70% confluence at 37°C in a fully humidified atmosphere of 5% CO 2 in air. Medium then was changed to serum-free containing 0.1% BSA, 50 g/ml ascorbate, and 10 mM ␤-glycerophosphate. After 24 h, cells were then re-fed with the same serum-free medium with or without TNF-␣ as indicated. TNF-␣ concentrations up to 100 ng/ml were not cytotoxic to MC3T3-E1 cells as determined by a trypan blue exclusion assay for cell viability (27) and crystal violet analyses for cell layer DNA content (28).
RNA Isolation and Northern Analysis-Total RNA was extracted by RNeasy mini kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. 10 g of denatured RNA/lane was electrophoresed on 1% agarose/formaldehyde gels and transferred to GeneScreen TM nylon membranes. cDNA probes were radiolabeled with 3000 Ci/mmol [␣-32 P]dCTP by random priming (29). Membranes were prehybridized for 30 min and then hybridized (ExpressHyb TM , Clontech) for 1 h at 68°C with probes for mouse lysyl oxidase (30), BMP-1/mTLD (31), and PCPE (32). For normalization, blots were stripped and rehybridized with a radiolabeled 18 S rRNA probe (33). Blots were washed in 2ϫ SSC, 0.05% SDS for 30 -40 min at room temperature and then twice in 0.1ϫ SSC, 0.1% SDS for 40 min at 50°C. Washed membranes then were exposed to Kodak X-Omat AR film with intensifying screens at Ϫ80°C. Autoradiograms were quantitated by the scanning densitometry of autoradiograms exposed for varying lengths of time.
Western Blot Analysis-Subconfluent cells were treated with or without TNF-␣ in serum-free ␣-MEM cell culture medium for 24 h. After cells were re-fed with fresh medium plus or minus TNF-␣, the cell layers were harvested after 24 h, washed in phosphate-buffered saline, and scraped in 500 l of sample buffer (0.1 M Tris-HCl, 4% SDS, 10% glycerol, 5% ␤-mercaptoethanol). Media samples were collected (20 ml) and concentrated to 1 ml (10 kDa, Centricon, Amicon, Bedford, MA) and suspended in 500 l of sample buffer. Protein concentration in the samples was measured using a fluorometric protein assay (Nano Orange, Molecular Probes, Eugene, OR). 2.5 g of protein of the media and 25 g of protein of the cell lysates were subjected to 10% SDS-PAGE and transferred to polyvinylidene difluoride transfer membranes (PerkinElmer Life Sciences) by electroblotting in 25 mM Tris, 192 mM glycine, and 10% methanol. Blots were blocked in BlockHen II (Aves Lab Inc., Tigard, OR) for 1 h and incubated with the primary antibody in phosphate-buffered saline, 0.1% Tween 20 for 2 h. After washing three times with phosphate-buffered saline, blots were incubated for 1 h with alkaline phosphatase-conjugated secondary antibody. Western Blue substrate (Promega, Madison, WI) was applied for detection. A BMP-1-specific antibody was raised in rabbit against a synthetic peptide (GRPHQLKFRVQKRNRTPQC) corresponding to the unique carboxyl-terminal sequence of human BMP-1. The cysteine residue was added to allow coupling to ovalbumin using the Imject maleimideactivated immunogen kit (Pierce). An IgG fraction was prepared from the immune serum by ammonium sulfate precipitation and used for detection of BMP-1. PCPE antibody was described previously (34). Lysyl oxidase antibody was described previously (35). Recombinant human TNF-␣ was obtained from PeproTech Inc. (Rocky Hill, NJ). ␣-MEM, Dulbecco's phosphate-buffered saline, fetal bovine serum, nonessential amino acids, trypsin-EDTA solution, penicillin, streptomycin, BSA fraction V, ascorbic acid, and ␤-glycerophosphate were purchased from Sigma. Culture plates and multiwell culture dishes were from Costar (Cambridge, MA). Murine calvarial osteoblast MC3TC-E1 subclone 14 (CRL-2594) was obtained from the American Type Culture Collection (Manassas, VA).
Lysyl Oxidase Activity Assay-200,000 cells were plated on a 100-mm tissue culture dish and grown to 80% visual confluence. Phenol red-free and serum-free ␣-MEM containing 0.1% BSA, 50 g/ml ascorbate, and 10 mM ␤-glycerophosphate was added for 24 h. Cells then were re-fed with fresh medium plus or minus TNF-␣ for 24 h. Lysyl oxidase enzyme activity was measured in the conditioned medium by a fluorometric assay (36). In this assay, lysyl oxidase oxidatively deaminates alkyl monoamines generating hydrogen peroxide. Horseradish peroxidasecatalyzed oxidation of N-acetyl-3,7-dihydroxyphenoxazine (Amplex Red) by hydrogen peroxide produces fluorescent resorufin, which was measured using a Hitachi F-2000 fluorescence spectrometer with excitation and emission wavelengths at 563 and 587 nm, respectively. Parallel assays were prepared with 500 M ␤-aminoproprionitrile fumarate to completely inhibit the activity of lysyl oxidase, and the difference in emission intensity was recorded. The amount of hydrogen peroxide produced by the action of lysyl oxidase was determined by comparing fluorescence changes to a standard plot relating fluorescence change to nanomoles of hydrogen peroxide. A fluorometric assay was performed to determine DNA content in the cells/100-mm tissue culture plate (37). Osteoblasts were scraped in 0.01% Triton-X, and cell lysates then were homogenized and hydrolyzed with 1 M perchloric acid at 70°C for 20 min to liberate deoxyribose. After the reaction with mdiaminobenzoic acid, the amount of liberated deoxyribose was measured by fluorescence spectrophotometer at excitation and emission wavelengths of 405 and 520 nm, respectively.

TNF-␣ Inhibits Collagen
Accumulation-Osteoblasts deposit abundant amounts of collagen as they differentiate in culture to ultimately form a bone-like extracellular matrix. Previous data reveal that maximum collagen accumulation in the cell layer of MC3T3-E1 cells occurs between days 12 and 15 (11). To understand the regulation of extracellular collagen-biosynthetic enzymes by TNF-␣, we first investigated how TNF-␣ controls collagen deposition in MC3T3-E1 cells under these culture conditions. Therefore, MC3T3-E1 cells were cultured in ascorbate-and ␤-glycerophosphate-supplemented medium for 12 days in the presence and absence of 10 and 100 ng/ml TNF-␣. Hydroxyproline levels as an index of collagen deposition then were measured in hydrolyzed total cell layers. The results in Fig. 1A show that in the presence of TNF-␣ cell layer collagen accumulation was inhibited and was dose-dependent and significant (p Ͻ 0.01). Cell layer hydroxyproline levels in cultures treated with 100 ng/ml TNF-␣ were reduced to 34% of untreated control cultures (Fig. 1A).
TNF-␣ Does Not Alter Collagen Synthesis-Since hydroxyproline analysis indicated that TNF-␣ inhibits collagen accumulation, we further studied the effects of TNF-␣ on new collagen synthesis in MC3T3-E1 cells. MC3T3-E1 cells were treated for 24 h with or without TNF-␣ and then pulsed with [2,3-3 H]proline for 2 h. A previous study (38) indicates that labeling periods of 1-2 h are optimal for measuring the synthesis of newly formed procollagen. Thus, collagen synthesis was determined in the cell layer by measuring the amount of tritiated proline incorporated into collagenase-sensitive protein. As shown in Fig. 1B, the exposure of MC3T3-E1 cells to TNF-␣ did not affect collagen synthesis significantly, although a trend toward decreased collagen synthesis at 100 ng/ml TNF-␣ was observed that was not significant statistically (p ϭ 0.515). Normalization to DNA levels/plate did not change these findings, because DNA levels were the same in all of the treatment groups: 564.63 Ϯ 71.22 g/plate (mean Ϯ S.D.).
TNF-␣ Does Not Alter Steady-state Collagen mRNA Levels-We next determined the effects of 0, 10, and 100 ng/ml TNF-␣ treatment of MC3T3-E1 cells for 24 h on steady-state ␣ 1 -type I collagen mRNA levels in MC3T3-E1 cells. Total RNA was isolated and subjected to Northern blot analyses. Two of the three experiments showed no regulation of ␣ 1 -type I collagen steadystate mRNA levels by TNF-␣ (Fig. 1C). However, one experiment revealed a down-regulation of 15% at TNF-␣ concentrations of 100 ng/ml. These data show that TNF-␣ does not decrease steady-state collagen, type I mRNA expression significantly in MC3T3-E1 cell cultures, although a tendency toward decreased mRNA levels was observed at 100 ng/ml.

TNF-␣ Decreases Total Mature Cross-link Formation-
The data indicate that TNF-␣ down-regulates collagen accumulation but hardly affects collagen synthesis and steady-state collagen type I mRNA levels. This finding strongly suggests that TNF-␣ primarily controls extracellular collagen modifications rather than procollagen biosynthesis. Therefore, we wished to further examine the collagenous matrix deposited under the influence of TNF-␣ to investigate whether cross-linking might be inhibited. MC3T3-E1 cells were cultured in ascorbate-and ␤-glycerophosphate-supplemented cell culture medium for 12 days in the presence and absence of 10 and 100 ng/ml TNF-␣. The amount of mature cross-links Pyd and Dpd then was assessed in the total cell layer by enzyme immunoassay as described under "Experimental Procedures." TNF-␣ inhibited collagen cross-link levels in a dose-dependent manner (Fig. 1D). These data indicate that TNF-␣ reduced mature collagen crosslinks to levels of 25% found in control cultures. This is a greater reduction compared with that found for TNF-␣-dependent reduction of total collagen deposition as determined above in Fig.  1A. This finding points to the possibility that lysyl oxidase-dependent cross-linking could be a primary target of TNF-␣.

TNF-␣ Reduces Lysyl Oxidase Steady-state mRNA Levels, Protein Expression, and Enzyme Activity in MC3T3-E1
Cells-As TNF-␣ decreases mature collagen cross-link levels, we wished to assess whether TNF-␣ regulates lysyl oxidase. MC3T3-E1 cells were treated in serum-free ␣-MEM supplemented with 0.1% BSA in the absence or presence of TNF-␣ for varying periods of time. Total RNA was isolated and subjected to Northern blot analyses. Results presented in Fig. 2, A and B, revealed that treatment of osteoblasts with TNF-␣ for 24 h inhibits steady-state lysyl oxidase mRNA levels and is dose-dependent. Diminished lysyl oxidase mRNA levels were observed at 10 and 100 ng/ml TNF-␣ compared with the untreated control (Fig. 2B). In multiple experiments, down-regulation of lysyl oxidase mRNA levels by TNF-␣ ranged between 20 and 50% compared with untreated control cultures. Time-dependent regulation of steady-state lysyl oxidase mRNA levels was tested as well. The data indicate that 10 ng/ml TNF-␣ downregulated lysyl oxidase mRNA levels slowly and was first ob-

FIG. 1. Effect of TNF-␣ in MC3T3-E1 cells on collagen accumulation (A), collagen synthesis (B), collagen mRNA levels (C), and mature collagen cross-link levels (D).
In A, confluent cell cultures were treated with 10 or 100 ng/ml TNF-␣ for 12 days. Total cell extracts were hydrolyzed and vacuum-dried, and cellular hydroxyproline levels in micrograms were measured by colorimetric assay. Data are presented as the means Ϯ S.D. of two experiments (n ϭ 3 cultures/group). *, p Ͻ 0.01 versus control using an unpaired Student's t test assuming equal variances. In B, subconfluent MC3T3-E1 cells were treated for 24 h with 10 or 100 ng/ml TNF-␣ and then pulsed with 10 Ci/ml (60) proline for 2 h. Total radiolabeled cell layer protein and collagenasedigestible protein (CDP) were determined in counts/min by liquid scintillation counter. Shown is the ratio between CDP and total amount of incorporated [ 3 H]proline per culture plate. Data are the means Ϯ S.D. of two experiments (n ϭ 4 cultures/group). In C, subconfluent MC3T3-E1 cells were treated for 24 h in serum-free ␣-MEM (0.1% BSA) with 0, 10, or 100 ng/ml TNF-␣. Ten micrograms of total RNA was analyzed by Northern analyses for ␣ 1 -type I collagen (COL1A1) mRNA and 18 S rRNA as described under "Experimental Procedures." Data shown are from one of three experiments with the same findings. In D, 200,000 cells were plated, and confluent cell layers were allowed to differentiate in the constant presence of 10 or 100 ng/ml TNF-␣ for 12 days in complete medium containing ascorbate and ␤-glycerol phosphate. Hydrolyzed total cell layers were measured for Pyd and Dpd levels in nanomole/culture plate by enzyme immunoassay. Data are presented as the means Ϯ S.D. of two experiments (n ϭ 3 cultures/ group). *, p Ͻ 0.05 versus control using an unpaired Student's t test assuming equal variances. served after 16 h and reached its maximum effect by 24 h of treatment (Fig. 3, A and B). The extended treatment of cells for up to 48 h caused no further down-regulation of lysyl oxidase (data not shown).
We next examined whether lysyl oxidase protein levels are regulated by TNF-␣. Cells were treated with serum-free medium with or without TNF-␣ for 24 h and then re-fed with or without TNF-␣ for another 24 h. Media samples then were subjected to Western blot analysis with an antibody against lysyl oxidase. Fig. 4 shows that 100 ng/ml TNF-␣ treatment of cells decreased the amount of the mature 32-kDa lysyl oxidase found in conditioned medium. The data in Fig. 5 present the effect of TNF-␣ on lysyl oxidase enzyme activity. Lysyl oxidase enzyme activity was decreased significantly and was dose-dependent.
Procollagen C-proteinases process fibrillar procollagens by cleaving the carboxyl-terminal propeptide. The 50-kDa prolysyl oxidase is processed and activated by procollagen C-proteinases to generate the mature 32-kDa enzyme. Therefore, we assessed whether BMP-1/mTLD steady-state mRNA levels are regulated by TNF-␣. As determined in three independent experiments, TNF-␣ does not regulate BMP-1 and mTLD gene expression (Fig. 3A). Procollagen C-proteinase activity against procollagens is enhanced by PCPE, which is found at relatively high levels in the extracellular matrix of connective tissues. Northern analyses revealed that TNF-␣ does not affect PCPE mRNA levels (Fig. 3A). In addition to and consistent with all of these findings, protein expression of BMP-1 and PCPE was not altered by TNF-␣ treatment as revealed by Western blot analyses (Fig. 4). The results together indicate that the diminished collagen deposition in TNF-␣-treated MC3T3-E1 cells is related to down-regulation of lysyl oxidase but not BMP-1, mTLD, or PCPE. DISCUSSION The present studies demonstrate that, in the presence of TNF-␣, less collagen accumulates in the extracellular matrix of osteoblast cultures and that the deposited collagen contains diminished levels of mature collagen cross-links, Pyd and Dpd. These data support the proposition that these effects are caused in part by a significant down-regulation of lysyl oxidase expression and activity by TNF-␣.
MC3T3-E1 cells derived from newborn mouse calvaria are capable of fully differentiating into osteoblast-like cells during culture, forming an extensive collagenous extracellular matrix that mineralizes in later stages through a process similar to that in bone formation in vivo (39). This study shows that TNF-␣ decreases collagenous protein accumulation in the extracellular matrix and that type I collagen mRNA expression and collagen synthesis are not affected significantly in agreement with other studies (6,40). Most important, these findings with or without 1, 10, or 100 ng/ml TNF-␣ for 24 h. In A, 10 g of total RNA was subjected to Northern blot analyses and hybridized with probes for lysyl oxidase, mTLD/BMP-1, PCPE, and 18 S rRNA. In B, autoradiograms exposed for varying lengths of time were quantitated by scanning densitometry for lysyl oxidase normalized to 18 S rRNA. Values represent the mean Ϯ S.D. obtained from three scanning densitometry determinations. Experiments were performed three times with the same results. suggest that TNF-␣ regulates extracellular rather than intracellular events of collagen biosynthesis. In fact, we show that TNF-␣ affects the posttranslational extracellular cross-linking of collagen by down-regulating lysyl oxidase. Insufficiencies in the post-translational modification of type I collagen in bone in vivo affect the mineralization density and crystal structure (41), and TNF-␣ negatively affects bone formation (1,4).
The data now presented show that TNF-␣ reduced lysyl oxidase steady-state mRNA expression, and this regulation is concentration-dependent as 10 -100 ng/ml TNF-␣ down-regulated lysyl oxidase mRNA levels by up to 50%. The TNF-␣ concentrations used are in a physiologically important range. For example, in vivo, the TNF-␣ concentrations of 10 ng/ml TNF-␣ were found in chronic inflammatory lesions (9). Decreased lysyl oxidase gene expression was detected after 16 h of treatment, and decreased lysyl oxidase steady-state mRNA levels were accompanied by a reduction in the amount of a 32-kDa lysyl oxidase protein. Regulation of lysyl oxidase enzyme activity can occur in parallel with mRNA changes (42,43), and a decrease in lysyl oxidase enzyme activity was found following TNF-␣ treatment.
Lysyl oxidase catalyzes the oxidative deamination of lysine and hydroxylysine residues in tropocollagen to generate reactive peptidyl aldehydes that undergo condensation reactions to ultimately form lysine-derived cross-links (12,13). Proper functioning of lysyl oxidase is crucial for collagen cross-linking and subsequent accumulation of insoluble collagen. Disruption of the cross-linking process can result in severe structural collagen changes and dysfunction of the tissue. For example, inhibition of lysyl oxidase activity leads to osteolathyrism where bones are thickened, extremely fragile, and soft leaving them with an increased risk for deformities and fractures (44,45). Inhibition of lysyl oxidase enzyme activity negatively affects bone and cartilage formation and function in vivo and in vitro (44,46,47). As we have published earlier (11), experimental inhibition of lysyl oxidase enzyme activity with the specific inhibitor ␤-aminoproprionitrile fumarate increased the accumulation of abnormal collagen fibrils in MC3T3-E1 cells. ␤-Aminoproprionitrile fumarate treatment abnormally increased fiber diameters, and the deposited collagen was characterized by enhanced solubility.
Pyridinium cross-links, Pyd and Dpd, are the primary crosslinks of mature type I collagen in bone (48,49). Their formation depends directly on lysyl oxidase activity. As noted, we found that significantly less Pyd and Dpd collagen cross-links were formed in the matrix of TNF-␣-treated MC3T3-E1 cell cultures. A decrease of mature collagen cross-links affects the biomechanical integrity, rigidity, and strength of bone (41), and TNF-␣ disrupts the extracellular matrix structure of osteoblasts (7). Electron microscopic studies of limb bud cultures reveal that sustained TNF-␣ treatment for 12 days leads to loosely packed collagen fibrils (8), and less cross-striated collagen fibrils were observed. In chronic inflammatory lesions where significant levels of TNF-␣ are found, accumulated collagen is disorganized (9). This study demonstrates for the first time that diminished lysyl oxidase and diminished collagen cross-linking may be important contributors to the effects of TNF-␣ on osteoblast extracellular matrix alterations.
Our data suggest that down-regulation of lysyl oxidase by TNF-␣ contributes to diminished deposition of mature collagen. It is known that TNF-␣ induces the synthesis of gelatinases in MC3T3-E1 cells, which preferentially cleave denatured collagen chains (50 -54). In addition, it has been shown that MMP-13 is expressed by MC3T3-E1 cells and is stimulated by TNF-␣ (53, 55). MMP-13 is the major interstitial collagenase produced by osteoblasts (56). The lack of proper collagen maturation because of inhibited lysyl oxidase expression most probably results in a higher capability of collagenases and gelatinases to degrade collagen molecules and promote matrix degradation in the course of inflammation (12). Indeed, we suspect that diminished collagen accumulation in response to TNF-␣ reported here depends partly on stimulation of proteolytic activity. This notion is based on our findings that inhibition of lysyl oxidase in MC3T3-E1 cultures with ␤-aminoproprionitrile fumarate that does not affect collagenase or gelatinase activity resulted in increased total collagen deposition, but the collagen was abnormal in structure (11). The ability of TNF-␣ to simultaneously inhibit lysyl oxidase biosynthesis and stimulate proteolytic activity seems likely to account for the decreased total amount of collagen deposited reported here.
Currently, little is known regarding procollagen C-proteinases and PCPE regulation, in general, and in osteoblast-like cells, in particular. As we have recently shown, BMP-1 and mTLD are expressed constitutively during the differentiation of phenotypically normal murine osteoblasts (11). It is unknown whether the same factors that regulate collagen biosynthesis also control the expression of procollagen C-proteinases and PCPE. As noted, we wished to investigate whether TNF-␣, which regulates collagen accumulation, also controls the expression of procollagen C-proteinases and PCPE. It has been shown previously that transforming growth factor-␤ increases the levels of BMP-1 and mTLD in fibrogenic cells and keratinocytes and that PCPE remained unchanged (57,58). In a study by Ogata et al. (34) conducted in lipocyte-like liver stellate cells, TNF-␣ down-regulated PCPE mRNA after 24 h. The data now presented show that in phenotypically normal MC3T3-E1 cells, TNF-␣ did not alter steady-state mRNA levels of BMP-1, mTLD, or PCPE. The cognate protein expression was not affected by TNF-␣ in these cells. Correspondingly, the processing of procollagen type I C-propeptides remained unaltered in our cell cultures (data not shown). As procollagen C-proteinases, especially BMP-1, are involved in the processing of numerous extracellular matrix components, stable BMP-1 and mTLD expression seems to be essential and is not affected by TNF-␣. This stable expression seems to assure the precise regulation of the deposited collagenous matrix and maintain bone tissue homeostasis.
In conclusion, we demonstrated that TNF-␣ regulates lysyl oxidase expression and activity in MC3T3-E1 cells. Lysyl oxidase is the key enzyme required for collagen cross-linking in the extracellular matrix and is essential for the accumulation of a functional collagen matrix. Less lysyl oxidase activity leads to perturbed collagen deposition that is prone to degradation by proteinases. As it is probable that these effects of TNF-␣ occur in vivo, this inflammatory cytokine secreted in excess by activated monocytes and macrophages could contribute to net bone resorption by these mechanisms in inflamed mineralized tissues. In fracture healing where TNF-␣ is secreted during the initiation of the repair process as well as in later stages of bone formation (59), structural alterations of the collagen molecule might promote tissue remodeling during different phases of the healing process. Therefore, elucidation of how TNF-␣ inhibits the posttranslational extracellular processing of the fibrillar collagens provides new insights into the mechanisms that could contribute to diminished osteoblast function, bone remodeling, and bone resorption.