The Expression of Metalloproteinase-2, −9, and −14 and of Tissue Inhibitors-1 and −2 Is Developmentally Modulated During Osteogenesis In Vitro, the Mature Osteoblastic Phenotype Expressing Metalloproteinase-14



During osteogenesis, in vitro, of tibial-derived rat osteoblasts (ROB) and derived clones, changes occur in the interactions of mature osteoblasts with the endogenous extracellular matrix (ECM) and these culminate in the formation of tridimensional nodules, which become sites of mineral deposition. We investigated if these changes might be mediated by remodeling of ECM, and we focused our study on the neutral metalloproteinases (MMPs), known agents of matrix remodeling, and on their tissue inhibitors (TIMPs). We report that during in vitro differentiation, osteoblasts express the secreted MMP-2 and −9 and the membrane gelatinase MMP-14. These, along with the tissue inhibitors TIMP-1 and −2, are developmentally regulated according to the maturation stage of osteoblasts. Their levels change in a similar association with osteoblast phenotypic maturation in different populations of ROB, which take different times to complete osteogenesis in vitro. MMP-14 expression coincides in both cell populations with the mature osteoblastic phenotype and is localized in the cells forming nodules. MMP-2 and −9 are expressed diffusely in the osteoblast population. Developmentally associated changes in the activation of MMP-2 are detected, associated in their timing with the expression of MMP-14 in both populations of ROB, and MMP-14 activates pro-MMP-2 in vitro. Expression of messenger RNAs (mRNAs) for the three MMPs increases up to the time of nodule formation. At this stage, TIMP-1 mRNA levels are lowest. TIMP-2 mRNA decreases throughout osteogenesis. In situ hybridization in 7-day-old rat tibias shows the strongest expression of MMP-14 among osteogenic cells, in lining osteoblasts on the newly formed trabeculae under the growth plate, and on the endosteal surface of cortical bone. Our data support the concept that the developmentally regulated expression of MMP-14 triggers localized proteolysis within the osteogenic population, concomitant in vitro to nodule formation.


OSTEOGENESIS IN VITRO is accompanied by the expression of specific and sequentially activated osteoblast phenotypes, by dynamic modeling of the extracellular matrix (ECM) and morphogenesis of nodules, in which mineral is deposited and osteocalcin and bone sialoprotein are preferentially localized.(1–5) Nodule formation involves changes in cell adhesion and may require reorganization of the ECM. Neutral metalloproteinases (MMPs) are a family of proteolytic enzymes involved in remodeling of ECMs and play a role in normal morphogenesis.(6,7) They include collagenases (MMP-1, −7, −8, −12, and −13), stromelysins (MMP-3 and −10), gelatinases (MMP-2 and −9), and transmembrane gelatinases or MT-MMP (MMP-14-18),(8–11) which degrade, with different specificity, proteoglycans, and glycoproteins in the ECM, native collagens, and denatured interstitial collagens.(6,7,12) They also proteolytically activate other MMPs(13–16) and release and activate growth factors.(6,17) Secreted MMPs are proenzymes, catalytically latent, activated by proteolysis in the pericellular and extracellular environment, subject to regulation by binding with tissue inhibitors (TIMPs).(18) Membrane enzymes also are synthesized as latent forms, and require activation by intracellular or pericellular convertases.(11) MMP expression is affected by hormones, cytokines, and growth factors in many cell types and in cells of the osteoblastic lineage.(19–29)

The data presently available on MMP expression during osteogenesis suggest that collagenases and stromelysins be involved in the early events of mesenchymal determination to the osteoblastic lineage while gelatinases are associated also with osteoblastic differentiation. The expression of MMP-13 is under the control of the bone-specific transcriptional regulator core binding factor alpha-1 (CBFA-1),(30) and it is restricted to preosteoblasts during bone morphogenesis.(31–34) The stromelysins (MMP-3 and −10) are detected in developing human bone,(35) MMP-3, and collagenase (MMP-13) in mouse preosteoblasts.(28) The gelatinase MMP-2 is expressed in MC3T3-E1 mouse preosteoblasts,(36) in human osteoblasts,(21,37–39) and in cultured mouse calvaria.(28) The gelatinase MMP-9 is expressed in cultured mouse calvaria,(27) in rat long bone,(29) and in developing human osteophytes(40) and is required for growth plate ossification, induced by parathyroid hormone (PTH) during fetal bone formation.(26,41) MMP-9 is not expressed in human osteoblasts in vitro.(39) The expression of the transmembrane gelatinase MMP-14 during mouse embryogenesis is associated with expression of TIMP-2(8) and activation of MMP-2.(42) It also is detected in enamel and pulp organs increasingly with dentine maturation.(43)

Distinct from expression, the activation of proenzymes is critical for MMP function. Localized activation is responsible for focalized proteolysis by MMPs and results in directional cell migration; moreover it causes proteolytic events during differentiation and morphogenesis.(7,42–46) Stoichiometric binding with the tissue inhibitors TIMP-1 and TIMP-2(47–49) regulates activation. The membrane gelatinase MMP-14 activates pro-MMP-2, mediated by binding with TIMP-2(50–53) producing MMP-2 derivatives of decreasing molecular size, with increasing molar enzymatic activity(50) and reduced efficiency of binding to tissue inhibitors.(51,54) Activated forms of MMP-2 were detected in tumor cells, in normal fibroblasts stimulated with concanavalin A, or 12–0-tetradedecanolyphorbol-13-acetate (TPA),(55) in normal osteoblasts proliferating on collagen gel or stimulated with monocyte conditioned medium (CM).(21) In many normal and tumor-derived cell types, the activation of pro-MMP-2 is regulated by the interaction of cells with the substrate and by the binding of the enzyme to integrins.(56–64) Binding by integrins of MMP-2 causes focalized proteolysis required for angiogenesis and tumor invasion in vivo.(44) Activated forms of MMP-2 can activate pro-MMP-9,(65) which also is activated by stromelysin(7) and thrombin.(66) Activated MMP-2 can degrade fibronectin,(67) BM-40,(68) insulin-like growth factor-binding protein 3 (IGFBP-3),(17) and IGFBP-5(36) in fragments with new biological activities, affecting cell adhesion, migration, and differentiation.(69)

Changes ensuing from ECM proteolysis elicit cellular responses through integrin and nonintegrin receptors(56,61,70–72) and, in turn, integrin occupancy can affect differentiation and MMP expression.(59,62,63) During osteogenesis, the modulation of synthesis and deposition of the ECM and of the associated endogenous growth factors and cytokines change the osteoblast microenvironment. The topologically restricted organization of some matrix proteins and the turnover of the ECM are concomitant to cell multilayering to form nodules in the osteoblast cultures in vitro.(5,73,74) All these events might be involved in creating a microenvironment suitable for the deposition of mineral in the nodules.

Because MMPs and TIMPs might play a role during these processes, and the timing of expression and activation of MMPs and of expression of tissue inhibitors in osteoblasts during the osteogenic progression in vitro has not been described, we investigated these issues and also studied the cellular expression of MMPs. We found that differentiating tibial-derived rat osteoblasts (ROBs) express, in a developmentally regulated fashion, MMP-2, −9, and −14 and TIMP-1 and −2. Expression of MMP-14 is restricted to mature osteoblasts and determines activation of pro-MMP-2. TIMP-1 messenger RNA (mRNA) is down-regulated at the time of nodule formation and TIMP-2 mRNA decreases throughout osteogenesis. We found a similar pattern of expression of MMPs and TIMPs, related to the progression of the osteoblastic phenotype during differentiation in two populations of osteoblasts,(2,3) which take different time courses to traverse osteogenesis and to accomplish mineralization of nodules. The cellular localization for each MMP during differentiation showed that mRNA for MMP-14 is preferentially localized in mature osteoblasts forming nodules, whereas MMP-2 and −9 mRNA expression is detected earlier in the culture. In vivo, in sections of tibia from 7-day-old rats, the strongest expression of the mRNA for MMP-14 was detected, among cells of the osteoblastic lineage, specifically in the mature osteoblasts.


ROB cultures

Primary cultures were obtained from cells migrating spontaneously out from enzymatically cleaned 7-day-old rat tibial diaphyseal fragments, in Ca-free Coon's modified F12, 10% fetal calf serum (FCS; Seromed, Berlin, Germany) as already described.(2,3,5) After the first trypsinization, cultures were grown in Coon's modified F12, 10% FCS. Cells were passaged when they reached confluence at 1 × 104 cells/cm2. We will refer to ROB when early passage cultures were used (second-fourth passage of primary) and to propagated ROB when cells were utilized between passage 46 and 60. On experimental day 0, coincident with confluence of the cultures, the medium was changed to differentiation medium (Coon's F12 supplemented with 100 μg/ml of ascorbic acid and 10 mM β-glycerophosphate) and afterward was changed every second day. These culture conditions are permissive for osteogenic differentiation. The sequence of phenotypic expression during differentiation was analyzed by monitoring the level of alkaline phosphatase, nodule formation, mineralization, and quantitatively by45Ca+2 incorporation, as previously reported.(2,3) The pattern of changes of these phenotypes followed a similar sequence in early passage and propagated cultures, while the osteogenic differentiation required about 1/3 of the time in propagated populations of ROBs after 46 passages than in early passages of ROB.(3) DNA determinations to establish cell equivalents were done by fluorimetry, as previously described,(2,3) and DNA content of the cultures did not change appreciably during differentiation in vitro.(2,3)

Scanning electron microscopy

Cells cultured on slides were fixed at different times during osteogenesis in 2% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.4, at 4°C for 24 h. They were rinsed twice in the same buffer, postfixed in 1% OsO4 in cacodylate buffer for 30 minutes, rinsed in the same buffer, dehydrated in increasing ethanol concentrations, and critical point dried from liquid CO2. Samples were coated with a 20-nm layer of gold in an argon atmosphere flow discharge sputter coating unit (Polaron E 5100) and examined in an ISI SS40 scanning electron microscope (International Scientific Instruments, Santa Clara, CA, U.S.A.) operated at an accelerating voltage of 10–20 kV.


CMs were obtained by 5 h incubation in serum-free medium of cultures at different times during osteogenesis. The CMs were centrifuged immediately to remove cell debris and the supernatant was stored frozen. Cell monolayers (i.e., cell-matrix compartments) also were collected, after three rinses with cold phosphate-buffered saline (PBS), in 25 mM Na citrate, 0.5% Sarcosyl, pH 7, and stored frozen. Samples were stored with 0.02% NaN3 and maintained at −20°C. Immediately after thawing the samples were concentrated by cold ethanol precipitation (46/100, vol/vol) for 1.5 h on ice followed by centrifugation and the pellet was resuspended in running buffer for electrophoresis. Electrophoresis was on acrylamide-sodium dodecyl sulfate (SDS) gels, containing 0.28% wt/vol gelatin (type A) or 1 mg/ml casein and run at 6–8°C in a water-cooled box. After migration, the gel was rinsed twice for 30 minutes in 2.5% Triton X-100 and incubated 16–18 h at 37°C in 40 mM Tris-HCl, 0.2 M NaCl, and 10 mM CaCl2. Addition of 1 μM to 1 mM(39) of Zn+2 to the incubation buffer did not change the pattern of MMPs obtained. When indicated p-amino-phenylmercuric acetate (APMA; at final concentration of 1 mM) was added to CM for 2 h at 37°C to activate the MMPs. To inhibit cation-dependent enzymatic activities, 10 mM EDTA was added to the incubation buffer. The gels were stained with 0.2% Coomassie blue in 50% methanol and 10% acetic acid, usually overnight, and destained in 50% methanol and 10% acetic acid. In each lane of the gels, unless otherwise indicated, we loaded equivalent per-DNA amounts of samples (CM) or cell-matrix (c-m) lysates collected from each time point (days) during differentiation. This corresponded, within 5% variations, with equivolumes of CM or c-m.

Western blotting

Equivalent-DNA amounts of CM, without or after reduction with final 0.1 mM DTT, were run on 10% acrylamide-SDS gels. The samples were electrotransferred to nitrocellulose and the membranes were incubated at 4°C for 5 h in Tween 20 Tris-buffered saline (TTBS), 20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 0.05% Tween 20 containing 5% powdered cow's milk (food item, local supermarket), and then 3 h in TTBS-5% milk, added with the specific antibody, 1 h in TTBS-5% milk added with a biotinylated antiserum against rabbit immunoglobulin G (IgG; Jackson Immunoresearch, U.S.A.), and finally 1 h with Avidin peroxidase (Jackson Immunoresearch, West Grove, PA, U.S.A.). The antisera for MMP-2 (Ab 45) and MMP-9 (Ab 110) were a kind gift from Dr. Stetler-Stevenson, National Institutes of Health (NIH), and were described by Stetler-Stevenson.(75)

Cell membrane preparation and MMPs activation in vitro

CMs were collected as usual after 5 h. Cell layers from a 100-mm-diameter Petri dish were washed three times with cold TBS (50 mM Tris-HCl, pH 7.4, and 150 mM NaCl). Eight hundred fifty microliters of 1.5% Triton X-114, 2 mM CaCl2, and 2 mM MgCl2 (TBS2+) were then added and the dishes were set on ice, on a basculating tray, for 20 minutes. The cell layer was scraped from the dish, homogenized, and centrifuged at 8000g for 2 minutes at 4°C. The supernatant was kept and partitioned into detergent and aqueous phases by incubation at 37°C for 5 minutes and centrifuged at 5000g for 2 minutes. The upper aqueous phase was collected and frozen. The lower detergent phase was washed with 1 ml of cold TBS2+ and repartitioned three times. The detergent phase was mixed with 600 μl of ice-cold CM collected in 0.02% NaN3 at day 0 or day 4 of differentiation and the mixture was incubated at 37°C for 72 h. Separate aliquots of the membrane preparations and of the CMs were added with fresh medium, incubated, and run in the zymograms as controls. After incubations, the samples were concentrated by cold ethanol precipitation and analyzed by gelatin zymography.

Total RNA preparation and Northern blotting

Total RNA was extracted according to Chomczynski and Sacchi.(76) RNA (20 μg) was separated on a denaturing gel of 1.2% agarose, transferred onto nylon membranes (Hybond-N+; Amersham, Buckinghamshire, U.K.), and cross-linked by UV exposure. In each experiment, five membranes were prepared, loaded with the same amounts of total RNA for each time point. Probes were a 1.1-kilobase (kb) complementary DNA (cDNA) coding for MMP-2, a 1.0-kb cDNA coding for MMP-9, a 0.6-kb cDNA coding for TIMP-1 and a 1.0-kb cDNA coding for TIMP-2 (all obtained from Dr. W.G. Stetler-Stevenson, NIH(75)), and a 1.8-kb cDNA coding for MMP-14 (obtained from Dr. J.M. Lewalle, University of Liège, Belgium(16)). The 1 × 106 counts per minute (cpm)/ml of [3H]deoxycytosine triphosphate (dCTP) randomly primed cDNA were used and hybridization was performed in 50% formamide, 5× standard saline phosphate-EDTA buffer (SSPE) (1× SSPE is NaCl, 0.18 M; Na2HPO4, 10 mM; and EDTA, 1 mM, pH 7.4), 5× Denhardt's solution, and 1% SDS, added with 0.1 mg/ml denatured salmon sperm DNA at 42°C overnight. After hybridization, the blots were washed, 20 minutes each, at room temperature in 2× SSC (SSC)/0.1% SDS; at 42°C in 2× SSC/0.1% SDS, 1× SSC/0.1% SDS, and 0.2× SSC/0.1% SDS; and finally at 55°C in 0.2× SSC/0.1% SDS. Each membrane, after hybridization with the specific probe, was stripped and utilized for subsequent hybridization with a probe for detection of 18s and 28s ribosomal RNA (rRNA; Xenopus genomic 45s fragment for rRNA, cloned in the HindIII site of pBR322; obtained from Dr. Castagnola, CBA, Genova, Italy). The dry membranes were exposed to Hyperfilm (Amersham) at −80°C, utilizing intensifying screens. Quantitation was obtained using a Macintosh scanner with an Image 1.54 application and ratios were calculated for each membrane between the quantity of specific mRNA and that of 28s rRNA. The data are expressed in arbitrary units obtained by setting equal to 1 the higher value for the ratio between mRNA and 28s RNA in each membrane. Experiments were repeated with two different total RNA preparations and the data presented are the average of very similar results (SD less than 5% for all points).

Biosynthetic labeling and immunoprecipitation

At each stated time during differentiation, the cultures were rinsed extensively with warm PBS and preincubated for 1 h in serum, methionine, and cysteine-free differentiation medium. The medium was then substituted with similar medium containing 150 μC/ml of trans-labeled35S methionine (Amersham) for 5 h. The CM and c-m compartments were collected as detailed previously and stored frozen. Equicounts of aliquots of c-m and CM were immunoprecipitated overnight at 4 C°, with the selected specific antiserum (described previously). Pansorbin (Calbiochem, La Jolla, CA, U.S.A.) was then added, before further incubation for 2 h at 4°C. The pellet obtained by centrifugation was rinsed extensively with a buffer containing 5 mM EDTA, 0.1% Triton X-100, 0.1% SDS, and 10 mM NaCl. The pellet was resuspended in running buffer and boiled and the supernatant from centrifugation was loaded on the gels.


Cells were fixed and permeabilized as reported previously.(3) Briefly, all the procedure was carried on at 4°C in PBS added with 0.1% heat-denatured bovine serum albumin (BSA). The incubation with the primary antibody was overnight and that with the secondary antibody was for 3 h. Antisera in rabbit for MMP-2 and −9 were the same as indicated previously, utilized at 1/50 dilution. Antiserum in rabbit for MMP-14 was from Biotrend (Colonia, Germany) and was utilized at 1/100 dilution. The dilutions of antisera were chosen, after testing various concentrations, to achieve a reasonable balance between the strong signal from nodules and the signal from cells in the monolayer for each immunostain. This was done in order to achieve the possibility of photographic illustration of expression in cultures at day 7. No comparative or quantitative indication on the actual amounts of each protein can be deduced from these illustrations. The second antibody was Cy3-conjugated anti-rabbit donkey IgG (Jackson Immunoresearch) and was used at 1/100 dilution. Controls were run without the first antibody (not shown).

In situ hybridization

The sequences for MMP probes utilized for Northern hybridization were subcloned into the EcoRI site at 701 base pairs (bp) of the pBluescript SK plasmid, and, after appropriate cutting to linearize the plasmid, they were transcribed from T3 or T7 promoters in order to obtain sense and antisense riboprobes for each sequence. Templates for each probe varied in size from 750 to 1400 bp. Probes were purified and labeled with digoxigenin (Boehringer-Mannheim Biochemicals, Mannheim, Germany) and hybridization was revealed by alkaline phosphatase-conjugated antidigoxigenin antibodies, according to protocols supplied by the producer. Safranine O was used as counterstain (10 s). Tibias were fixed in RNase-free 4% paraformaldehyde overnight and processed into paraffin wax. Sections (5 μm thick) were treated with 0.2N HCl for 20 minutes, with 10 μg/ml proteinase K in PBS for 30 minutes at 37°C, and with 0.2% cold glycine in PBS for 2 minutes. They were then postfixed in 4% paraformaldehyde for 15 minutes and treated with 0.25% acetic anhydride in 0.1 M triethenolamine for 10 minutes. Hybridization was in 4× SSC and 50% formamide at 42°C overnight. After hybridization and RNase treatment, successive washes were done at progressively higher stringency, the final one in 0.2× SSC at 45°C.

Cells in culture were fixed in 3.7% paraformaldehyde and 2% sucrose for 15 minutes. They were treated with 60 μg/ml collagenase and 20 μg/ml hyaluronidase for 5 minutes, before digestion with 10 μg/ml proteinase K for 30 minutes, in order to partially remove excess of the ECM and grant access of the riboprobe into the cells. An equal concentration (10 ng/μl) for each probe was utilized in all the hybridizations.

All chemicals, unless otherwise stated, were from Sigma, St. Louis, MO, U.S.A.


Operationally, we divide osteogenesis in vitro into three phases: in the first phase collagen and alkaline phosphatase expression increase and matrix is accumulated, in the second phase expression of alkaline phosphatase decreases and cell multilayering occurs in limited zones of the cultures determining the formation of discrete nodules. In the third phase of osteogenesis mRNA for osteocalcin and incorporation of45Ca+2 in the ECM increase and osteocalcin, bone sialoprotein, and mineral are detected preferentially in the nodules.

Identity of the MMPs expressed by differentiating osteoblasts

Analysis of the MMPs secreted in CMs collected at 7 days and 14 days during differentiation of ROB by gelatin zymography shows a thin band of the approximate molecular size of pro-MMP-9 (92 kDa) and larger bands of the size expected for pro-MMP-2 (72 kDa) and for its major activated forms (69 kDa and 64–59 kDa). Both proenzymes are activated by APMA, which also reveals MMP-9 at 82 kDa, and all the activities are inhibited by treatment with EDTA. Zymograms on casein of the same CMs, containing 4-fold the amount of the same samples, do not reveal digestion of the substratum after treatment with APMA (Fig. 1A). Treatment of the cultures for 6 h or 24 h with 20 nM PTH before collection of the media does not change the pattern of MMP expressed at 0 days and at 7 days of differentiation (not shown). In Western blotting of reduced samples of CM, specific antibodies for MMP-2 and −9 identify, respectively, a doublet of bands (pro-MMP-2 at 69 kDa and the activated form at 66 kDa) and a single band (pro-MMP-9 at 85 kDa; Fig. 1B).

Figure Figure 1.

Identification of secreted MMPs expressed by ROBs. (A) Zymography on gelatin (a-c) and casein (d and e) of media collected from early passage ROB at day 7 and day 14 during differentiation (indicated above). Samples were with or without pretreatment with APMA and ±EDTA, as indicated below. Dots are the positions of the mw standards, in decreasing order of 97.5, 69, and 46 kDa. Numbers at left indicate estimated molecular sizes. (B) Western blotting of CM at day 3, with antiserum against MMP-2 (a), and against MMP-9 (b). Samples were reduced before electrophoresis. Molecular sizes indicated at the sides.

MMP expression and activation varies during osteogenesis in vitro

Figure 2A shows the pattern of alkaline phosphatase expression and of incorporation of45Ca+2 in early passage differentiating ROB, and Fig. 2B shows the morphological changes detected by scanning electron microscopy, which occur during the process of osteogenesis in vitro. These serve as reference for the timing of the expression of MMPs (Figs. 2C and 2D). CMs collected at different times during osteogenesis (Fig. 2C) contain pro-MMP-9 (92 kDa), activated MMP-9 (82 kDa), and increasing amounts of pro-MMP-2 (72 kDa) and MMP-2 at 69 kDa, and, gradually becoming detectable during differentiation, activated derivatives of MMP-2 of smaller molecular sizes. These are the 64- to 59-kDa forms, very evident already after 4 days in differentiating conditions, and multiple bands at 43–37 kDa, resulting from further proteolysis of the 69-kDa/MMP-2 with loss of the COOH terminal fragment, present by 7 days. Pro-MMP-2 and 69 kDa/MMP-2 also are detected in the c-m compartments prepared at all times of osteogenesis. After a lag in differentiating conditions, a band at 66 kDa also is detected in c-ms, the molecular size of MMP-14, most evident in the second phase of osteogenesis and decreasing again at mineralization. MMP-9 activity is undetectable throughout. Analysis of the c-ms and CMs on casein gels, even when using four times the amount of material as utilized in zymograms on gelatin, does not reveal digestion bands. This suggests that no casein digesting MMPs are produced at any time during osteogenesis (as already shown for CMs collected at 7 days and 14 days). We attempted, but failed to detect TIMPs in CMs by reverse zymography on gelatin (not shown).

Figure Figure 2.

Correspondence between the osteogenic phenotypes and the expression of secreted and cell-associated MMPs. (A) Alkaline phosphatase and incorporation of45Ca were determined during differentiation of early passage ROB. The arrow indicates the time when nodules were first detected by optical microscopy in the cultures. SEM is less than 5% at all points. (B) Scanning electron microscopy of a differentiating culture of primary ROB. Magnification is from left to right: ×1280, ×430, and ×370. (C) CMs from early passage ROB were loaded in equal amounts on acrylamide (10%) gel containing gelatin. (D) The c-m compartments corresponding to the foregoing CMs were loaded in twice the per-DNA amounts than the CMs. Gels were acrylamide (10%) containing gelatin. Molecular sizes are indicated on the sides. (A-D) Days in differentiation culture condition are indicated below each panel.

Identification of MMP-2 secreted during osteogenesis by Western blots

Western blots with a polyclonal antiserum against MMP-2 of CMs collected during osteogenesis shows pro-MMP-2, 69 kDa MMP-2, identifies both the 64- and 59-kDa forms as derivatives of MMP-2, and confirms their production only in CM collected in the second and third phases of osteogenesis (Fig. 3). Lower molecular-size bands (43-37 kDa) are faintly detectable in these same CMs, only by overloading the gels (not shown). The relative amounts detected in each lane of the Western blots for proenzyme, 69-kDa form and activated forms of lower molecular mass of MMP-2 differ from the observations by zymography, as expected from the fact that activated forms of 64–59 kDa have higher specific activity than the 69-kDa form.

Figure Figure 3.

Identification of MMP-2 secreted during osteogenesis. CMs were collected from early passage ROB at different times in differentiating culture conditions and equal amounts were processed by Western blotting for immunodetection with antiserum against MMP-2. Samples were unreduced. The numbers to the right indicate molecular sizes calculated on the basis of mw standards.

We also attempted to detect TIMP-1 and TIMP-2 in c-m's by Western blotting with the specific antisera and by utilizing similar amounts of samples as utilized for detection of MMPs, without success.

In vitro activation of pro-MMP-2 by the 66-kDa membrane-associated gelatinase

The gelatinolytic activity at 66 kDa, detected in the c-m compartment, having highest expression in the second phase of osteogenesis, has been tentatively identified with MMP-14 on the basis of the molecular size. This gelatinase is associated with cell membranes and its absence or presence in purified membrane preparation correlates with the capability to process in vitro pro-MMP-2 and 69 kDa/MMP-2 to gelatinases of smaller molecular size. In Fig. 4, purified membrane preparations from ROB at 4 days of osteogenic differentiation contain the 66-kDa gelatinase and promote the formation of two activated gelatinolytic bands at about 64 kDa and 59 kDa from pro-MMP-2 and 69 kDa MMP-2. Membrane preparations obtained from ROB at day 0 of osteogenesis and not expressing detectable amounts of the 66-kDa gelatinase are unable to process MMP-2 in vitro. Therefore, the 66-kDa gelatinase is a membrane gelatinase functionally akin to MMP-14.

Figure Figure 4.

The 66-kDa membrane-associated gelatinase processes pro-MMP-2 and 69 kDa MMP-2 in vitro. Membranes were prepared from the same number of cells (equi-DNA amounts) at the beginning of culture in differentiating conditions and 4 days later. Membrane preparations were incubated with CM containing the pro and the 69-kDa forms of MMP-2 (CM day 0, lane 1). After incubation, the samples were examined by zymography on gelatin. Samples were (1) CM day 0, incubated alone; (2) CM day 0, incubated with membranes from cells at day 0; (3) CM day 0, incubated with membranes from cells at day 4; (4) membranes from cells at day 4; and (5) membranes from cells at day 0, lacking the 66-kDa gelatinase. The numbers on the sides are molecular weights calculated on the basis of mw standard.

Developmentally modulated expression of mRNAs for MMPs and for tissue inhibitors

Total RNAs are prepared at the beginning of culture in differentiating conditions (day 0), at the formation of nodules (day 15), and at mineralization (day 30). Expression of all the genes is modulated during differentiation (Fig. 5). Quantitation of the amounts of mRNAs for MMP-2, MMP-9, MMP-14, TIMP-1, and TIMP-2 in total RNA extracted during osteogenesis has been obtained by normalization against 28s rRNA of Northern blots of total RNAs with cDNA probes specific for each sequence listed previously. MMP-2 and MMP-14 mRNA levels increase during osteogenesis; the increment of MMP-14 mRNA between days 0 and 15 is more than 50%, whereas that of MMP-2 mRNA is about 20%. MMP-9 mRNA more than doubles during the first phase of osteogenesis and decreases slightly at mineralization. TIMP-1 mRNA has an inverse pattern of expression to the one of MMP-9, and at nodule formation it is reduced at about 60% in comparison with day 0, to increase again in mineralized cultures. TIMP-2 mRNA is detected as two bands (3.5 kb and 1.0 kb), which are expressed differentially, with the larger transcript decreasing slightly coincident with nodule formation and the smaller transcript decreasing continuously during osteogenesis. As a whole, TIMP-2 transcripts decrease by about 20% by the second phase of osteogenesis, remaining steady at mineralization.

Figure Figure 5.

The expression of mRNAs for MMPs −2, −9, and −14 and of tissue inhibitors 1 and 2 is modulated during osteogenesis. Total RNA isolated from early passage ROB at successive times during culture in differentiating conditions was hybridized with the specific probes indicated on the left side of each Northern blot. In the right column are the densitometric scans, normalized to that 28s rRNA from the corresponding Northern probe with a ribosomal cDNA (not shown). The amounts of each mRNA are indicated on an arbitrary scale (for TIMP-2, the two mRNAs are shown separately, the right column being the 1-kb transcript). The results of two experiments were averaged. SD was less than 5%. Numbers at the bottom indicate days in differentiating culture conditions.

MMP expression during osteogenesis in populations of propagated ROB has a similar schedule of developmental modulation as in early passage osteoblast population

Propagated ROB populations express a similar sequence of changes in the osteogenic phenotypes as early passage populations, although they traverse osteogenesis in vitro with increasing synchronism and within a shorter time (Fig. 6A). All developmentally regulated events of osteogenesis studied are accelerated in these cultures. We show here that the pattern of modulation in the expression of MMP-2 and of its activation is regulated temporarily in a fashion coordinated to the differentiation sequence. Activated forms of MMP-2 are already detected in CM at 2 days of differentiation by zymography (Fig. 6B), and by immunoprecipitation (Fig. 6E). Immunoprecipitation of the radiolabeled CMs also shows that 69, 64, and 59 kDa MMP-2 immunoreactive bands are processed from the secreted pro-MMP-2 in 5 h. Expression of MMP-14, between days 2 and 8, also is observed in zymograms on gelatin of c-m of propagated ROB (not shown). Metabolic radiolabeling followed by immunoprecipitation of the c-m compartments with the specific antisera shows newly synthesized pro-MMP-2 (Fig. 6C) and pro-MMP-9 (Fig. 6D), accompanied by two bands at circa 30 kDa and 21 kDa, the molecular sizes, respectively, of TIMP-1 and TIPM-2. Proteins of these sizes were co-immunoprecipitated with antisera against MMP-2 also from CM, faintly visible, and therefore in lesser amounts, relative to newly synthesized MMPs, than from the corresponding c-m compartment (Fig. 6E). TIMPs were not detectable in these CM by reverse zymography (not shown). Figures 6F and 6G show the identification, in CM collected during osteogenesis, of the secreted MMP-2 and MMP-9 by Western blotting with the specific antibodies.

Figure Figure 6.

Developmentally regulated expression of MMPs in populations of propagated ROB. (A) Alkaline phosphatase and incorporation of45Ca were determined during differentiation of ROB at the 46 passage. The arrow indicates the time when nodules were first detected by optical microscopy in the cultures. SEM is less than 5% at all points. (B) CM from propagated ROB was loaded in equal amounts on acrylamide (10%) gel containing gelatin. (C) Equicounts of metabolically radiolabeled c-m's collected at different times during osteogenesis from propagated ROB were immunoprecipitated with antiserum against MMP-2 or (D) against MMP-9. (E) Equicounts of the corresponding metabolically radiolabeled CMs were immunoprecipitated with antiserum against MMP-2. (F) Equiamounts of CM were processed for Western blotting with antiserum against MMP-2. (G) MMP-9. Gels were 10% acrylamide and samples were unreduced. In A-G, numbers below each panel are days in differentiating culture conditions. On the sides of each panel are the molecular sizes calculated on the basis of mw standards.

Northern blots of total RNAs were prepared from propagated ROB, at the phases of osteogenesis phenotypically corresponding to those described previously for cultures at early passage. These, hybridized with the probes specific for MMPs and TIMPs, reproduce the pattern for the modulation of the expression of all five genes obtained for early passage ROB, in a shorter time lapse and with changes of similar entities (not shown).

Identification of cells expressing MMPs by immunofluorescence and in situ hybridization in vitro and ex vivo in tibias

Both immunofluorescence (not shown) and in situ hybridization at time 0 of propagated ROB cells indicated the constitutive expression of MMP-2, diffusely in all the cells. MMP-9 was only faintly detectable and MMP-14 was not detected with either procedure at this time. In the second phase of osteogenesis, both immunofluorescence and in situ hybridization showed positivity for MMP-9 and MMP-14 (Fig. 7 and Fig. 8B show cultures at day 7). MMP-9 was detected since day 3, and MMP-14 at day 5, in coincidence with the formation of nodules. The association of the expression MMP-14 mRNA with that of pro-MMP-2 and pro-MMP-9 therefore is restricted during osteogenesis to cells in the nodules.

Figure Figure 7.

Localization by immunofluorescence of MMP-2, −9, and −14 in mature osteoblasts. Propagated ROB (60 passages) at day 7 in differentiation conditions were probed with antiserum against MMPs, as detailed. Enlargement is ×200.

Figure Figure 8.

In situ hybridization for MMP-2, −9, and −14 in differentiating osteoblasts. Propagated ROB (60 passages) at day 0 and day 7 in differentiation conditions. On the side of each panel the antisense probe utilized is indicated. Enlargements are ×400. Controls (C) are hybridized with the sense probe for MMP-2, and the absence of stain in control at 7 days also confirms lack of a specific retention of the probe within the multilayered nodule.

In 7-day-old rat tibias, the expression of MMP-14 mRNA (Fig. 9) was strongest in lining osteoblasts of newly formed trabeculae adjacent to the growth plate and at the endosteal surface of the bony shaft. Lower positivity was associated with osteocytes and periosteal cells, bone marrow cells being negative. These results indicate that in vivo the expression of MMP-14 is developmentally regulated in cells of the osteoblastic lineage.

Figure Figure 9.

Localization by in situ hybridization of MMP-14 in bone cells of 7-day-old rat tibia. (a-d and f, h, and i) antisense probe for MMP-14; (e and g) sense probe, as negative control. All but c and g were counterstained with safranine. (a, b, f, and g) Cortical bone; (c, d, e, h, and i) Trabecular bone. Long arrows point to osteoblasts and short arrows point to osteocytes. P, periosteum; B, bone; M, bone marrow. Enlargements are (a) ×400; (b and h) ×630; (c-g) ×200; (i) ×1200.


Our primary aim in this study was to identify the neutral MMPs produced by rat tibial-derived osteoblasts and to establish if their expression and activity were related to the events of osteogenic differentiation in vitro and to nodule morphogenesis. For this second purpose, we also have compared the expression of MMPs in two osteoblastic populations, differing in the timing taken for accomplishing osteogenesis in vitro.

In both populations, we identify the MMP activities primarily expressed as MMP-2 and MMP-9 by molecular size and immunodetection, and MMP-14 by molecular size, localization in the cell membrane, and function.

MMP-2, MMP-9, and MMP-14 and also TIMP-1 and TIMP-2 are expressed in a developmentally modulated fashion during osteogenic differentiation in vitro. A sequential pattern of changes is detected in enzyme activity for MMPs and in mRNA levels for the three MMPs and for the two TIMPs, which is similarly associated in the early and late passage populations to the other osteoblast phenotypes and equally developmentally regulated. This finding strongly confirms that the observed regulation of the expression of MMPs and TIMPs is an intrinsic feature of the osteogenic progression.

The activation of pro-MMP-2 to higher specific activity and lower molecular-size forms also is modulated according with the stage of osteogenic differentiation and follows the expression of MMP-14 in osteoblast membranes. The relative amounts of activated forms of MMP-2 in CM increase, concomitant with the up-regulation of MMP-14 in the corresponding c-m compartments. This is consistent with the processing of pro-MMP-2/69 kDa MMP-2 observed by us in vitro by membrane preparations and, in general, with the function described for MMP-14 in mesenchymal cells.(8,42)

MMP-9 is detected in c-m in radiolabeled samples and in CMs by zymography. The 82-kDa active form of MMP-9 also is faintly represented in the CMs from differentiating cells, suggesting that modulation of the activation of MMP-9 during osteogenesis might occur also. Nonetheless, given the relatively low amounts of MMP-9 at all times, it was not possible to follow the timing of its activation by zymography on gelatin. In osteoblast CM, we have not detected other MMPs known to activate pro-MMP-9 besides MMP-2, suggesting that activation of pro-MMP-9 in this cell system may depend on MMP-2 activity.

Proteins of the molecular size of TIMP-1 and TIMP-2 were coimmunoprecipitated with antisera against MMP-2 and MMP-9 from metabolically labeled cells and in lesser relative amounts from corresponding CMs, suggesting that association of newly synthesized TIMPs with MMPs occurs prevalently in a cellular and/or pericellular location. Retention of TIMP-2 in the c-m compartment is consistent with its role in mediating the activation of MMP-2 by MMP-14 at low level.(16) We have failed to detect TIMP-1 and −2 by Western blots of CM compartments, by immunofluorescence with the specific antibodies, and by reverse zymography. Others have reported variable amounts of TIMP-1 and −2 in human osteoblasts.(37–39)

The increase observed at the beginning of differentiation in the expression of MMPs mRNAs is associated to the deposition of the ECM by osteoblasts, suggesting that endogenous ECM and/or integrin engagement may transcriptionally regulate MMPs expression, in analogy with what was described in other cell types.(58–63)

Although after the initiation of osteogenic differentiation and up to the stage of nodule formation there is an increase in the three MMPs and in their mRNA levels, at the stage of nodule formation, TIMP-1 mRNA level is at its minimum and TIMP-2 mRNA is down-regulated, relative to day 0. The highest concentration of proteolytically active MMPs since the onset of osteogenesis therefore might be achieved in osteoblast cultures at this stage. The coexpression of MMP-2, −9, and −14 has the potential to induce a proteolytic cascade, MMP-14 determining activation of pro-MMP-2, followed by activation of MMP-9 by activated MMP-2. These can degrade efficiently the ECM in an environment in which there is a relative scarcity of TIMPs. The establishment of a balance favoring MMPs over their inhibitors therefore might be relevant in this phase of osteogenesis and is reflected in high turnover of proline radiolabeled proteins at this stage (P. Manduca, unpublished data, May, 2000).

Later, in mineralized cultures, the correlation between mRNA and enzyme activity for MMP-14 is lost because the mRNA level increases while enzyme activity decreases.

Still unexplained features of osteogenesis in vitro are the mechanism and the requirements for the formation of discrete nodules. Nodules involve the specification of new topographic relationships within a population of cells progressing into osteogenesis, and this occurs even when the population is homogenous, as in cloned osteoblasts.(3) In the nodules, the cells undergo morphogenetic events possibly involving adhesive and migratory changes, and the nodules become privileged sites for mineral deposition.

We speculate that the stochastic occurrence of a pericellular or cell membrane-localized event(s) in the second phase of osteogenesis, here identified in the onset of expression of MMP-14, is responsible of the creation of topologically restricted differences within the cell population and of the formation of nodules, by promoting proteolytic remodeling of the ECM and/or of the adhesion behavior of a subpopulation of osteoblasts.

Consistent with this hypothesis, the expression of the membrane gelatinase is detected after a lag since the beginning of osteogenesis and in cells located in nodules, distinct from the widespread localization occurring already at earlier time points during osteogenesis in monolayer and in nodules, of MMP-2 and MMP-9. The expression of the membrane gelatinase activity and of MMP-14 mRNA therefore are restricted in time and sites in the osteogenic populations and are characteristic for cells involved in the morphogenesis of nodules.

In agreement with the developmental regulation of the expression of MMP-14 during osteogenesis and with its preferential expression in mature osteoblasts in vitro, in 7-day-old rat tibias the cells in the osteogenic lineage also expressed differentially MMP-14 mRNA. The level of expression changes according to their maturation stage, as shown by in situ hybridization, with a strong signal in osteoblasts, low in periosteal cells (preosteoblasts) and in osteocytes and absent in bone marrow cells. This suggests that among osteogenic cells, the expression of MMP-14 is a characteristic, also in vivo, of mature osteoblasts. MMP-14 expression also was detected in osteoclasts and chondroblasts (P. Manduca, unpublished data, May, 2000).

Distinct from the relevance of coexpression of MMP-2, −9, and −14 for localized proteolysis and nodule morphogenesis, the activation of secreted MMPs in the middle phase of osteogenesis in vitro also might play a more general role affecting the entire osteoblast population through the production of diffusible and biologically active fragments derived from proteolysis of the ECM and/or through the release and/or the activation of new functions for proteins, proteoglycans, glycoproteins, hormones, and cytokines stored within the matrix. The resulting changes might control osteogenic progression via feedback effects on proteolysis or on other functions and could be involved in vivo in signaling by paracrine mechanisms to other cell types (P. Manduca, unpublished data, May, 2000). Recently, the relevance of MMP-14 in osteogenesis was illustrated by the generation of double-deficient mice and by the demonstration of severely deficient osteogenic activity in vitro of marrow osteogenic cells derived from these mice.(77)

Recently, it was shown that other proteases are regulated developmentally in their expression during osteogenesis.(78) The relevance for the osteogenic process of the proteolytic events determined by different proteinases deserves further investigation, as does the issue of substrate specificities and of complementarity in function for these enzymes.

In summary, we have shown that in two distinct osteogenic populations of tibial-derived ROBs, the expression of MMP-2, MMP-9, and MMP-14 and TIMP-1 and TIMP-2 and the activation of MMP-2 are regulated and can be considered as part of the developmental sequence of osteogenesis. We show coexpression of MMP −2, −9, and −14 only in mature osteoblasts involved in nodule formation in vitro and progressive activation of MMP-2, up to the second phase of osteogenesis. We also show that MMP-14 expression is specific for mature osteoblasts in vivo.


We acknowledge the courtesy of Dr. W.G. Stetler-Stevenson (NIH) and Dr. J.M. Lewalle (Liegé, Belgium), who made antisera and DNA probes available to us, and of Profs. A. Abbondandolo (IST, Genova, Italy) and D. Zacheo and A. Zicca (Dip. Medicina Sperimentale, Genova) in offering access to instruments not otherwise available to us. This work was supported by funds from Italian Ministero della Università e della Ricerca Scientifica e Tecnologica (Progetti Nazionali) and from Progetto Biotecnologie, Consiglio Nazionale della Ricerca (P.M.).