Pre- and Post-translational Regulation of Lysyl Oxidase by Transforming Growth Factor- (cid:98) 1 in Osteoblastic MC3T3-E1 Cells*

The final enzymatic step required for collagen cross- linking is the extracellular oxidative deamination of peptidyl-lysine and -hydroxylysine residues by lysyl ox- idase. A cross-linked collagenous extracellular matrix is required for bone formation. The goals of this study were to compare the transforming growth factor (TGF)- (cid:98) 1 regulation of lysyl oxidase enzyme activity and steady state mRNA levels to changes in COL1A1 mRNA levels in MC3T3-E1 osteoblastic cells. TGF- (cid:98) 1 increased steady state lysyl oxidase and COL1A1 mRNA levels in a dose- and time-dependent manner. The increase in lysyl oxidase mRNA levels was transient, peaking at 12 h and 8.8 times controls in cells treated with 400 p M TGF- (cid:98) 1. COL1A1 steady state mRNA levels increased maximally to 3.5-fold of controls. Development of increased lysyl oxidase enzyme activity was delayed and was of slightly lower magnitude than the increase in its mRNA levels. This suggested limiting post-translational processing of lysyl oxidase proenzyme. Pulse-labeling/immunopre-cipitation studies demonstrated slow proenzyme secre- tion and proteolytic processing. Development and appli-cation of an independent assay for lysyl oxidase

Lysyl oxidase is the extracellular enzyme that catalyzes the oxidative deamination of lysine and hydroxylysine residues in collagen and elastin precursors to form peptidyl-␣-aminoadipic-␦-semialdehyde, or peptidyl-␦-hydroxy-␣-aminoadipic-␦semialdehyde, respectively. Spontaneous condensation reactions of these resultant aldehydes leads to the formation of lysine-derived cross-links found in mature collagens and elastin (1).
A cross-linked collagen matrix is required for differentiation of osteoblastic cells and for bone mineralization (2). Type I collagen synthesis and accumulation have been shown to be uncoupled processes in developing MC3T3-E1 osteoblastic cells (3). These cells accumulated collagen in the extracellular matrix when the rate of collagen synthesis was decreasing. This suggested post-translational control of collagen fibril accumulation. Similarly, in primary rat calvaria osteoblast cell cultures, TGF-␤1 1 -stimulated extracellular collagen accumulation was controlled by unidentified post-translational steps (4). Lysyl oxidase-mediated control of collagen accumulation in bone cell cultures has been described (5). In this study, treatment of chick osteoblast cultures with the lysyl oxidase inhibitor, ␤-aminopropionitrile (BAPN), reduced the accumulation of collagen in the cell layer by 50%. Turnover of collagen deposited into the extracellular matrix was also markedly increased in BAPN-treated cultures. Further evidence for the importance of lysyl oxidase in bone formation was illustrated by in vivo studies, where growing chicks and rats fed with BAPN developed skeletal deformities known as osteolathyrism (6,7).
TGF-␤ increased accumulation of extracellular matrix proteins in different cell and organ culture models (4, 8 -10). In vivo, the highest levels of TGF-␤ were found in platelets and bone (11,12). TGF-␤-mediated inhibition of osteoblastic cell growth was accompanied by increased synthesis of type I collagen, fibronectin, and proteoglycans (13)(14)(15). TGF-␤ also decreased the expression of extracellular matrix degrading enzymes, and induced genes for protease inhibitors in different cell types (16,17). Among the genes that were induced by TGF-␤ in osteoblastic MC3T3-E1 cells, steady-state lysyl oxidase mRNA levels increased almost 10-fold (18).
Although lysyl oxidase enzyme activity was not measured in the studies cited above, taken together they raise the possibility that like collagens, lysyl oxidase is a major target for TGF-␤ in osteoblasts. If true, then lysyl oxidase regulation by TGF-␤ could contribute to increased collagen accumulation by osteoblastic cells stimulated by this growth factor.
The biosynthesis of lysyl oxidase is complex and requires numerous post-translational modifications, each of which could potentially control activation of lysyl oxidase enzyme activity. Lysyl oxidase is synthesized as a 50-kDa glycoprotein precursor, and is then secreted and proteolytically processed to the mature 32-kDa enzyme (19). Recently, the cleavage site of the NH 2 -terminal propeptide region of the lysyl oxidase precursor was reported (20). Lysyl oxidase is a copper metalloenzyme, and in addition contains an organic cofactor believed to be a quinone derived from post-translational modification of a tyro-sine residue (1,21). The mechanisms of post-translational constitution of the metal and organic cofactors with lysyl oxidase proenzyme have not yet been defined.
In the present study, we report that TGF-␤1 dramatically increased the levels of lysyl oxidase mRNA and enzyme activity in MC3T3-E1 osteoblastic cultures. The effects on steady-state lysyl oxidase and COL1A1 mRNA levels were compared. Furthermore, we found that post-translational mechanisms may control the expression of fully active lysyl oxidase.

EXPERIMENTAL PROCEDURES
Materials-Recombinant human transforming growth factor ␤-1 (TGF-␤1) was purchased from Austral Biologicals, San Ramon, CA. TRI Reagent LS was obtained from Molecular Research Center, Inc., Cincinnati, OH. Dulbecco's modified Eagle's medium, newborn bovine serum, trypsin-EDTA solution, penicillin-streptomycin solution, Dulbecco's phosphate-buffered saline, non-essential amino acids solution, bovine albumin fraction V, ascorbic acid, and ␤-glycerophosphate were purchased from Sigma. The Protein Fusion and Purification System was purchased from New England Biolabs, Beverly, MA. All other chemicals were of reagent grade.
Cell Culture-Murine osteoblast-like MC3T3-E1 cells were provided by Dr. Louis Gerstenfeld, Children's Hospital, Boston, MA, and Dr. Renny Franceschi, University of Michigan, Ann Arbor. Cells were plated onto 100-mm tissue culture dishes in Dulbecco's modified Eagle's medium, containing 10% heat-treated (56°C, 30 min) newborn calf serum plus 1% non-essential amino acids and 100 g/ml of each penicillin and streptomycin. Cultures were maintained at 37°C in a fully humidified atmosphere of 5% CO 2 in air. Media were changed every 3 days. Cells in the logarithmic growth phase were dissociated with trypsin/EDTA, and inoculated at 200,000 cells/plate. After 2.5 days the subconfluent cells were then fed with serum-free medium, containing 0.1% bovine serum albumin, 50 g/ml ascorbate, and 10 mM ␤-glycerophosphate and cultured for an additional 24 h. Cells were then refed with fresh media plus or minus TGF-␤1, for the appropriate period of time. Experiments were performed after no more than 3 passages from one set of frozen cell stocks.
Assay of Lysyl Oxidase Activity-Lysyl oxidase enzyme activity was measured in the conditioned media and cell layers, using recombinant human [ 3 H]tropoelastin substrate (22). Media samples (0.5 ml) were assayed in quadruplicate in a final volume of 1 ml, containing 0.1 M borate, 0.15 M NaCl, pH 8.0, and 160,000 cpm of [ 3 H]tropoelastin in the presence and absence of 5 ϫ 10 Ϫ4 M BAPN. Reactions were incubated for 90 min at 37°C, followed by distillation under vacuum. Radioactivity in 0.5-ml aliquots of distillate was determined by liquid scintillation spectrometry. Units of enzyme activity were defined as dpm released above the BAPN control. Enzyme activity was normalized to cell number, after dissociating cells from the same plate with trypsin/EDTA, and counted using a hemocytometer.
RNA Isolation and Northern Analysis-Total RNA was prepared using Tri-Reagent LS (23). Ten micrograms of denatured RNA was applied per lane and separated by electrophoresis on 1% agarose gels, containing 18% formaldehyde, and transferred to GeneScreen (DuPont) nylon membranes using 10-fold SSC (1 ϫ SSC, 0.15 M NaCl, 0.015 M sodium citrate, pH 7.0). Membranes were baked at 80°C for 1 h and prehybridized at 42°C for 4 h in hybridizing solution (50% formamide, 1-fold SSPE (1 ϫ SSPE, 0.15 M NaCl, 0.01 M NaH 2 PO 4 , 0.02 M EDTA, pH 7.4), 10% 5-fold Denhardt's solution, 0.2% SDS, 0.05% Na 2 PP i and 1% salmon sperm DNA). Membranes were then hybridized for 18 h at 42°C with a 32 P-labeled COL1A1 cDNA (24) or 32 P-labeled mouse lysyl oxidase cDNA probe (25). For normalization, blots were stripped and rehybridized with a radiolabeled 18 S ribosomal probe (26). 1 ϫ 10 6 cpm of probe was used per ml of hybridizing solution. Probes were labeled using the random primer method (27). The membranes were washed as described and subjected to autoradiography (28). Autoradiograms were assessed and normalized by densitometric scanning. Values for standard error were derived from triplicate scans of films. Experiments were performed at least twice.
Pulse Labeling and Immunoprecipitation-Cells were grown as described above and then fed with serum free medium Ϯ TGF-␤1 (400 pM) for the appropriate period of time. Cultures were then refed and incubated for 20 min with 7 ml/plate of serum-free and methionine-free Dulbecco's modified Eagle's medium Ϯ TGF-␤1 (400 pM). Cultures were then placed in fresh media Ϯ TGF-␤1 (400 pM) supplemented with 50 Ci/ml [ 35 S]methionine (1175 Ci/mmol; DuPont NEN, catalog number NEG-072). Following incubation, media and cell layers were harvested and prepared for immunoprecipitation (19). Constant counts/min from the cell layer, and medium were used for immunoprecipitation with rabbit anti-bovine aortic lysyl oxidase (19). Samples were subjected to SDS-PAGE, and the gels were treated with Resolution (EM Corp., Chestnut Hill, MA) and dried under vacuum. Gels were placed on Kodak XAR-5 film with an intensifying screen at Ϫ80°C for 5-30 days.
Rat Lysyl Oxidase Fusion Protein Production and Purification from Escherichia coli-A bacterial expression vector was constructed utilizing rat lysyl oxidase cDNA (28,29) cloned downstream and in-frame with the malE gene of E. coli in vector pMAL-c2 (30,31). The construct encodes the lysyl oxidase proenzyme beginning at amino acid residue 25 cloned into the XmnI site of pMAL-c2. The cloning strategy resulted in introduction of no non-lysyl oxidase residues into the vector. Although not relevant to the present study, if the fusion protein were cleaved by Factor Xa as designed, the resultant 43,748-Da lysyl oxidase proenzyme would begin four amino acid residues to the carboxyl side of the predicted signal peptide cleavage site (28).
The lysyl oxidase construct was cloned in three steps. Fragment I was lysyl oxidase coding region from bp 42 to bp 233 derived from a PstI fragment cloned into pUC 18 vector as described below (28,29). Fragment II was lysyl oxidase coding region from bp 234 to bp 939 (PstI/ HindIII fragment). Fragment III was a HindIII fragment containing lysyl oxidase coding region from bp 940 through the end of the coding region, and extending to the first EcoRI site in the 3Ј-untranslated region. Fragments II and III were derived from the lysyl oxidase transcription vector pBSCOD (19). pUC 18 containing fragment I was digested with Eco0109I and recessed 3Ј termini were filled in using the Klenow fragment of E. coli DNA polymerase. Then the fragment was digested with PstI, yielding the 230-bp blunt/PstI fragment I. pBSCOD containing Fragments II and III was digested with PstI/HindIII, yielding Fragment II. This DNA was cloned into a PstI/HindIII-digested pMAL-p2 vector. pMAL-p2 DNA containing Fragment II was isolated from positive clones, digested with Xmn I/PstI, and ligated with blunt/PstI Fragment I, and correct clones were isolated. This pMAL-p2 plasmid containing Fragments I and II was digested with HindIII and 600-bp HindIII/HindIII Fragment III was cloned into this site. The resulting construct was lysyl oxidase cloned into pMAL-p2 (pLO/P2). As fusion protein expression from pLO/P2 was unacceptably low, we then recloned the NdeI/NcoI fragment of LO/P2 into the NdeI/NcoI site of pMAL-c2 resulting in construct pLO/C2. Limited DNA sequence analyses of these constructs revealed the correct nucleotide sequences at the junctions of lysyl oxidase Fragments I, II, and III. Detailed restriction mapping also confirmed the structure of these constructs.
pLO/C2 was used to transform E. coli strain BL21 (DE3)pLys(S) (Novagen, Inc). Two liters of logarithmic phase cells were induced with 0.3 mM 5-bromo-4-chloro-3-indoyl-␤-D-glucopyranoside for 4.5 h, collected by centrifugation at 5000 ϫ g, and stored at Ϫ80°C. The pellet was thawed and resuspended in 200 ml of 50 mM Tris, 2 mM EDTA, pH 8.0. At room temperature, 1.2 ml of 0.2 M phenylmethylsulfonyl fluoride, 25 ml of 10 mM MgCl 2 , and 10 mg of protease-free DNase I dissolved in 8 ml of 0.15 M NaCl were added to the lysed cell pellet and stirred for 10 min. The inclusion bodies were collected by centrifugation, and then solubilized in 280 ml of 8 M urea denaturing buffer, pH 7.8 (8 M urea, 20 mM sodium phosphate, 500 mM sodium chloride, pH 7.8), and dialyzed against 10 liters of 1 M urea denaturing buffer, pH 7.8, at 4°C overnight, followed by several changes of amylose column buffer (20 mM Tris-Cl, 200 mM NaCl, 1 mM EDTA, 10 mM ␤-mercaptoethanol) over 24 h. The soluble proteins were then applied to an amylose column (New England Biolabs) as described (31). The fusion protein was eluted with column buffer containing 10 mM maltose, with a typical yield of 6.4 mg from 2 liters of culture. The eluted fusion protein was further purified by chromatography on hydroxylapatite as follows. Nine mg of fusion protein dialyzed against 10 mM KH 2 PO 4 , pH 7.2, was applied to a 10-ml hydroxylapatite column (Sigma) equilibrated in the same buffer. The column was successively washed with 40 ml of 10 mM KH 2 PO 4 , pH 7.2, 40 ml of 500 mM KH 2 PO 4 , pH 7.2. The column was then eluted with 6 M urea, 20 mM KH 2 PO 4 , pH 7.2. The yield was 3.21 mg or 35%. The purified fusion protein was exhaustively dialyzed into 50 mM Tris, pH 7.8, prior to use. Purity was assessed by SDS-PAGE.
NH 2 -terminal Analysis of Cleaved Lysyl Oxidase Fusion Protein-MC3T3-E1 cells were treated with 400 pM TGF-␤1 in serum-free media containing 0.01% bovine serum albumin, ascorbate, and ␤-glycerol phosphate for 24 h. Three ml of medium was then mixed with 180 g of fusion protein (3.5 ml in 50 mM Tris, pH 7.8) and incubated at 37°C for 3 h (final volume of 6.5 ml). Proteins were precipitated overnight by adding trichloroacetic acid to 10% after cooling sample to 4°C, collected by centrifuging at 10,000 ϫ g, washed twice with cold acetone, and then subjected to 10% SDS-PAGE. Sample was electroblotted onto polyvinylidene difluoride membranes (Bio-Rad) and stained with Ponceau S (Sigma). The 32-kDa band was excised and submitted to the Harvard Microchemistry Facility, Cambridge, MA, for NH 2 -terminal sequence analysis.

Dose-dependent Up-regulation of Lysyl Oxidase by TGF-␤1-
Preconfluent MC3T3-E1 cells were cultured in serum-free media supplemented with 0 -400 pM TGF-␤1 for 24 h. A dose-dependent up-regulation in the steady-state lysyl oxidase and COL1A1 mRNA levels was found (Fig. 1). Lysyl oxidase mRNA increased 8.8-fold in the presence of 400 pM TGF-␤1. COL1A1 steady-state mRNA levels also increased but reached a modest level of 3.5-fold of control at 400 pM TGF-␤1. Lysyl oxidase enzyme activity in the cell culture media was stimulated by 4 and 40 pM TGF-␤1 comparable to increases observed in the mRNA levels (Table I). In contrast, 400 pM TGF-␤1 caused a smaller increase in lysyl oxidase enzyme activity (5.4-fold) compared to the 8.8-fold increase in lysyl oxidase steady-state mRNA levels. No enzyme activity was detected in the cell layers (data not shown). These experiments were performed twice yielding similar results.
Time-dependent Regulation of Lysyl Oxidase-The time-dependent regulation of steady-state lysyl oxidase mRNA levels and enzyme activity by 400 pM TGF-␤1 was determined. TGF-␤1 increased steady-state lysyl oxidase mRNA levels after 3 h of treatment, reaching a maximum at 12 h (Fig. 2 B, lanes  5 and 6). Lysyl oxidase enzyme activity in the cell culture media was stimulated to a lesser degree, and peaked later, at 18 h ( Fig. 2A). The delay between steady-state lysyl oxidase mRNA levels and enzyme activity suggests that post-translational mechanisms may limit the formation of fully active lysyl oxidase. The following experiments were performed to investigate potential limiting steps in the biosynthesis of fully active lysyl oxidase.
Synthesis and Secretion of the Lysyl Oxidase Precursor-We tested whether TGF-␤1 increased the synthesis of lysyl oxidase precursor protein parallel to changes in steady-state lysyl oxidase mRNA. Cells were pretreated with or without 400 pM TGF-␤1 for 4 or 15 h, and then pulse labeled with [ 35 S]methionine for 3 h. Radioactive media and extracts of cell layers were then immunoprecipitated with anti-bovine lysyl oxidase as described previously (19). Fig. 3 shows autoradiograms of lysyl oxidase immunoprecipitates from the media (lanes 1-5) and cell layers (lanes 6 -9) of these cultures. TGF-␤1 increased the labeled pool of 50 Ϯ 5 kDa lysyl oxidase precursor in the cell layer after 4 and 15 h of growth factor pretreatment compared to control cultures (Fig. 3, lanes 7 and 9). A 5.5-fold increase in the amount of labeled proenzyme was detected after 15 h of growth factor pretreatment compared to 4 h of pretreatment, revealed by scanning laser densitometry of autoradiograms. This change was similar to that seen at the mRNA level (Fig.  2B), where a 6.9-fold increase in lysyl oxidase steady state   mRNA levels occurred between 3 and 12 h of treatment with TGF-␤1. Thus, as expected, increases in cell layer proenzyme labeling closely paralleled changes in lysyl oxidase steady-state mRNA. Under these conditions of pulse-labeling, no lysyl oxidase proenzyme was detected in unstimulated control cultures (Fig. 3, lanes 6 and 8).
Levels of radioactive secreted proenzyme from growth factor pretreated cells were slightly lower after 15 h compared to 4 h under these conditions of pulse labeling (Fig. 3, lanes 2 and 4). Control media had no detectable radiolabeled proenzyme (Fig.  3, lanes 1 and 3). No fully processed labeled 32-kDa lysyl oxidase was detected under these conditions in control or growth factor pretreated cultures (Fig. 3, lanes 1-4). Consistent with previous studies (19), the immunoprecipitations were shown to be specific (Fig. 3, lane 5). These findings suggest that secretion and extracellular proteolytic processing of lysyl oxidase requires more than 3 h in MC3T3-E1 cells. Thus, due to a large unlabeled intracellular lysyl oxidase pool after 15 h of TGF-␤1 pretreatment, secreted proenzyme was likely of lower specific radioactivity than that secreted from cells pretreated for only 4 h. This would account for the lower labeling of secreted proenzyme at 15 h compared to 4 h of growth factor pretreatment.
Accumulation of Media Lysyl Oxidase Proenzyme in TGF-␤1stimulated Cultures-We next tested whether proteolytic processing may be insufficient to fully convert the lysyl oxidase proenzyme secreted from TGF-␤1-stimulated cells. Cells were treated with or without 400 pM TGF-␤1 for 12 and 18 h in the continuous presence of radioactive methionine. Media and cell layer extracts were then immunoprecipitated with lysyl oxidase antibodies. This experiment revealed the molecular forms of lysyl oxidase that accumulate in control and growth factortreated cell cultures over relatively long periods of pulse labeling.
As shown in Fig. 4 (lanes 1-4), TGF-␤1 increased the pool of lysyl oxidase precursor in the conditioned media in a time-dependent manner. Similarly, a greater amount of labeled 32-kDa lysyl oxidase was found at 18 h compared to 12 h. This is consistent with increased enzyme activity observed after 18 h of growth factor treatment (Fig. 2). Densitometric scanning of autoradiograms from 12-and 18-h immunoprecipitates showed 3-fold increases in the 50 Ϯ 5 kDa precursor (12 h, 689 Ϯ 5.6; 18 h, 1, 964 Ϯ 25.5) and in the 32 Ϯ 3 kDa form of lysyl oxidase (12 h, 378 Ϯ 33.2; 18 h, 1, 121 Ϯ 29.7) in TGF-␤1-treated cells. The accumulation of 50 Ϯ 5-kDa lysyl oxidase only in the media of TGF-␤1-treated cultures suggests that the conversion to the 32 Ϯ 3-kDa species was limited by extracellular proteolytic processing activity.
Scanning laser densitometry of Fig. 4, lanes 1-4, revealed that the level of labeled 32 Ϯ 3-kDa lysyl oxidase increased by 13-and 20-fold in TGF-␤1-treated cells compared to unstimulated cells at 12 and 18 h, respectively. Assuming that the 32 Ϯ 3-kDa molecular form of lysyl oxidase was fully active and that the specific radioactivity of methionine pools was comparable in these continuously labeled cultures, corresponding increases in lysyl oxidase enzyme activity by TGF-␤1 of 10 -20-fold would be expected. Interestingly, we observed at most only a 6-fold stimulation of enzyme activity at 18 or 24 h (Table I and Fig. 2). We conclude that lysyl oxidase is unlikely to be fully active in TGF-␤1-stimulated cultures (see "Discussion").
Results obtained from the cell layer suggest that relatively low levels of lysyl oxidase proenzyme accumulation occurred in this fraction compared to the media (Fig. 4). Moreover, absence of the 32 Ϯ 3-kDa molecular form in the cell layer correlated well with the absence of enzyme activity. Thus, in these MC3T3-E1 osteoblastic cell cultures, lysyl oxidase was synthe-sized as a 50 Ϯ 5-kDa precursor (Fig. 3), secreted and processed proteolytically to a 32 Ϯ 3-kDa molecular species that remained soluble (Fig. 4). This is distinct from studies in neonatal rat aorta smooth muscle cells, and rat lung fibroblasts, where lysyl oxidase enzyme activity and the 32 Ϯ 3-kDa molecular form were shown to accumulate in the cell layer after extracellular proteolytic processing (19,32).

Extracellular Proteolytic Conversion of Lysyl Oxidase Precursor to the 32-kDa Molecular
Form-An independent assay for lysyl oxidase processing proteinases was developed in order to verify results found in pulse-labeling studies suggesting limiting levels in TGF-␤1-stimulated cultures. We produced the lysyl oxidase proenzyme as a fusion protein with the maltosebinding protein in E. coli (see "Materials and Methods" and Fig.  5). This 86-kDa fusion protein was used as the substrate to assay for production of a 32 Ϯ 3-kDa protein immunoreactive with lysyl oxidase antibodies. To validate this assay, we also established the NH 2 -terminal amino acid sequence of the immunoreactive cleavage product generated from the fusion protein by MC3T3-E1 conditioned media (see below).
Media obtained from MC3T3-E1 cells treated for 24 h with 400 pM TGF-␤1 was incubated with fusion protein in the presence and absence of 20 mM EDTA for up to 3 h, and then analyzed by Western blotting. As shown (Fig. 6) MC3T3-E1 media converted the fusion protein to a discrete 32 kDa Ϯ 3-kDa molecular species clearly immunoreactive with lysyl oxidase antibody in a time-dependent manner. This activity was fully inhibited by 20 mM EDTA, suggesting that this activity is a metalloenzyme (Fig. 6, lane 8). Conversion to 32 kDa Ϯ 3-kDa product was linear with respect to time up to 50% conversion (Fig. 6). If reactions were carried out beyond 50% conversion, production of 32-kDa product as a function of time was no longer linear (Fig. 6, lanes 6 and 7 in inset were not plotted). Production of 32 Ϯ 3-kDa product was also linear with respect to volume of culture medium up to 50% conversion (not shown). Thus, linearity was maintained up to 50% conversion with respect to time of incubation, and volume of MC3T3-E1 culture media. It is likely that loss of linearity beyond 50% conversion was due to substrate depletion or product inhibition. 1,10-Phenanthroline (1 mM) inhibited production of 32 Ϯ 3-kDa product by 87% (Fig. 6, lane 9). This further supports that lysyl oxidase processing activity is a metalloenzyme(s). NH 2 -terminal analysis was performed on the 32 Ϯ 3-kDa product generated by culture media from TGF-␤1-stimulated cells (see "Methods and Materials" for details). As shown in Table II, the analysis is consistent with the sequence Asp-Asp- Pro-Tyr-Ser. This is the same NH 2 -terminal sequence identified for active lysyl oxidase (20). This sequence does not occur in the maltose-binding protein. Thus, the Western blot assay detects activities that result in cleavage of the lysyl oxidase fusion protein at the correct site.
Processing protease activity as a function of time of TGF-␤1 treatment of MC3T3-E1 cells was determined under valid assay conditions established above. These results were compared to changes in lysyl oxidase enzyme activity in the same media. Lysyl oxidase enzyme activity and processing protease activity were not changed by 400 pM TGF-␤1 after 3-6 h of treatment (Fig. 7). In contrast, from 12 h, both activities were higher in media from growth factor-treated cultures. Interestingly, the maximum increase in lysyl oxidase enzyme activity was greater than the processing protease. Thus, at 15 h, lysyl oxi-dase activity was 3.4-fold of control levels; whereas processing activity was increased 1.7-2-fold. These results demonstrate that increased lysyl oxidase enzyme activity induced by TGF-␤1 was accompanied by increased proteolytic processing activity. The degree of stimulation of proteolytic processing activity by TGF-␤1 was less than the degree of stimulation of lysyl oxidase enzyme activity. These results support the concept that proteolytic processing of lysyl oxidase may partially limit the production of fully active 32 Ϯ 3-kDa lysyl oxidase in these osteoblastic cell cultures. DISCUSSION TGF-␤1 is present in high amounts in bone (12), and has been shown to stimulate the synthesis of type I collagen, fibronectin, and proteoglycans by osteoblastic cells (14,15). Type I collagen synthesis and accumulation have been shown to be  (40). Replicate scans did not vary by more than 5% conversion. Markers were Bio-Rad prestained low molecular weight standards.  (20) and to the corresponding sequence predicted from rat lysyl oxidase cDNA (28,29)   MC3T3-E1 cells were treated with 0 or 400 pM TGF-␤1 as described under "Methods and Materials." Plates were harvested at intervals, and proteinase (see Fig. 6) and lysyl oxidase enzyme activities were determined from the conditioned media, normalized to cell number determined from the same plate. Values for lysyl oxidase activity are means Ϯ S.E. of quadruplicate assays. Values for processing proteinase activity are means of duplicate determinations Ϯ S.E. The volume of media assayed for processing proteinase activity was 125 l. This enabled all assays to be in the linear range. uncoupled processes in developing MC3T3-E1 cells (3). Moreover, TGF-␤1-stimulated collagen accumulation in cell layers of rat calvaria osteoblastic cultures was found to be partially controlled by unidentified post-translational mechanisms (4). In a separate study, lysyl oxidase mRNA was found to increase 8-fold in MC3T3-E1 cells treated with TGF-␤1 (18). Recently, lysyl oxidase was implicated in regulating collagen accumulation, and in decreasing collagen turnover in mineralizing chick osteoblastic cultures (5). It was suggested that collagen crosslinking increases irreversible retention of collagen fibrils in the extracellular matrix. This would amplify the efficiency of collagen insolubilization, reduce the pool of soluble collagen, and thereby decrease collagen susceptibility to proteolytic degradation.
In the present study, we investigated whether lysyl oxidase, like COL1A1 is a major target for TGF-␤1 in MC3T3-E1 osteoblastic cultures. We now report that TGF-␤1 significantly increased steady state lysyl oxidase mRNA levels, lysyl oxidase proenzyme, and its extracellular proteolytic processing, in a dose-and time-dependent manner. Increases in lysyl oxidase mRNA by 400 pM TGF-␤1 were followed to a lesser degree by increases in enzyme activity. Moreover, we found a 6-h delay between the growth factor-stimulated peak in lysyl oxidase steady state mRNA levels and enzyme activity. This suggests that lysyl oxidase biosynthesis is slow, and may be limited by post-translational steps in MC3T3-E1 osteoblastic cells.
Post-translational control of lysyl oxidase was supported by short-term pulse-labeling/immunoprecipitation experiments following TGF-␤1 treatment. Due to the short-term pulse labeling and low level of processing protease(s), the resulting media (extracellular) labeled lysyl oxidase were not processed to the 32-kDa molecular species. Cell-associated radiolabeled lysyl oxidase proenzyme was increased in proportion to TGF-␤-dependent changes in steady state lysyl oxidase mRNA levels, indicating that there was no block of translation of lysyl oxidase pro-protein in TGF-␤1-stimulated cells. Experiments with long periods of continuous pulse labeling resulted in increased accumulation of both 50-kDa precursor and 32-kDa mature enzyme in the media. This supported the concept that lysyl oxidase processing protease(s) partially limited production of the 32-kDa molecular form of lysyl oxidase.
We produced the lysyl oxidase proenzyme as a fusion protein with the maltose-binding protein in E. coli in order to develop an independent assay for lysyl oxidase processing protease(s). Interestingly, the maximum increase in lysyl oxidase enzyme activity (3.4-fold of control) at 15 h was greater than the processing protease activity (1.7-2-fold of control). This finding supported the pulse-labeling studies suggesting that proteolytic processing of lysyl oxidase proenzyme may partially limit the production of the 32-kDa molecular form of lysyl oxidase. Moreover, the inhibition by EDTA and 1,10-phenanthroline, confirmed the notion that this activity is a metalloproteinase (19,20). 400 pM TGF-␤1 caused an increase in steady state lysyl oxidase mRNA and in both the 32 kDa Ϯ 3-kDa and 50 Ϯ 5-kDa kDa lysyl oxidase protein molecules, whereas the increase in enzyme activity occurred to a lower extent. These findings may be related to previous studies where a catalytically inactive 32-kDa form of lysyl oxidase was isolated from bovine aorta (33), and to observations indicating that purified lysyl oxidase is not fully active (34). Furthermore, TGF-␤1 treatment might not have increased the levels of proteins involved in copper transport and metabolism, which could also have impaired development of full lysyl oxidase enzyme activity (35)(36)(37). It is also currently unknown whether the 50-kDa molecule secreted by cultured cells is catalytically active, although indirect evi-dence suggests that this molecular form is inactive. Thus, cell cultures exhibiting lysyl oxidase activity in the cell layer such as neonatal rat aorta smooth muscle cells and rat lung fibroblasts have been shown to accumulate an insoluble 32-kDa form of lysyl oxidase (19,32). In contrast, cultures such as the MC3T3-E1 cells grown as described in the present study accumulate no 32-kDa lysyl oxidase in the cell layer, and exhibit no detectable activity in this fraction. All activity was found in the medium. Taken together these studies show that lysyl oxidase activity correlates well with the presence of the 32-kDa molecular form of lysyl oxidase, and not with the 50-kDa molecular form. It is of interest, however, that an active 45-kDa lysyl oxidase has been demonstrated in rat skin tissues (38).
As MC3T3-E1 cells produce relatively high levels of soluble lysyl oxidase enzyme activity, and 32 kDa-and 50 kDa-immunoreactive proteins, they may provide a source of lysyl oxidase that will allow direct analysis of the activity of these different molecular forms. Elucidation of the mechanisms of activation of lysyl oxidase activity in osteoblasts is required in order to evaluate the ultimate importance of these pathways in bone formation and pathology. It is of interest that lysyl oxidase dependent cross-linking has been shown to be decreased in human osteoporotic bone, and is likely related to the known bone weakness associated with this disease (41).
MC3T3-E1 cells have been shown to undergo phenotypic changes characteristic of the three proposed stages of osteoblast differentiation, resulting in the production of a mineralized extracellular matrix (3,39). We have now shown that TGF-␤1 up-regulated lysyl oxidase activity and mRNA levels consistent with COL1A1 mRNA regulation in preconfluent MC3T3 E1 cells. It will be of interest to further these studies in long-term mineralizing organ and cell cultures, analyzing for phenotype-and growth factor-dependent changes in lysyl oxidase regulation and collagen cross-linking and accumulation. Such studies may reveal minimum and optimum levels of lysyl oxidase activity required for collagenous matrix accumulation. These experiments would further define possible connections between the regulation of lysyl oxidase and TGF-␤1-mediated increases in bone extracellular matrix.