Cysteine Proteinases and Matrix Metalloproteinases Play Distinct Roles in the Subosteoclastic Resorption Zone

Authors

  • Vincent Everts,

    Corresponding author
    1. Experimental Oral Biology Group, Department of Cell Biology and Histology, Amsterdam, The Netherlands
    2. Experimental Oral Biology Group, Department of Periodontology, Academic Centre for Dentistry (ACTA), Amsterdam, The Netherlands
    • Vincent Everts, Academic Medical Center, Department of Cell Biology and Histology, P.O. Box 22700, 1100 DE Amsterdam, The Netherlands
    Search for more papers by this author
  • Jean-Marie Delaissé,

    1. Department of Basic Research, Center for Clinical and Basic Research, Ballerup, Denmark
    Search for more papers by this author
  • Wolf Korper,

    1. Experimental Oral Biology Group, Department of Cell Biology and Histology, Amsterdam, The Netherlands
    Search for more papers by this author
  • Wouter Beertsen

    1. Experimental Oral Biology Group, Department of Periodontology, Academic Centre for Dentistry (ACTA), Amsterdam, The Netherlands
    Search for more papers by this author

  • Part of this work has been published in the proceedings of the conference: The Biological Mechanisms of Tooth Eruption, Resorption and Replacement by Implants, EBSCO Media, p. 85–97, 1994 and at the IADR Conference in Singapore, 1995 (Everts et al., J Dent Res 74:466, abstract 527).

Abstract

Digestion of calvarial bone by osteoclasts depends on the activity of cysteine proteinases and matrix metalloproteinases (MMPs). It is unknown, however, whether these enzymes act simultaneously or in a certain (time) sequence. In the present study, this was investigated by culturing mouse calvarial bone explants for various time intervals in the presence or absence of selective low molecular weight inhibitors of cysteine proteinases (E-64, Z-Phe-Tyr(O-t-Bu)CHN2 or CA074[Me]) and MMPs (CI-1, CT1166, or RP59794). The explants were morphometrically analyzed at the electron microscopic level. All proteinase inhibitors induced large areas of nondigested demineralized bone matrix adjacent to the ruffled border of actively resorbing osteoclasts. The appearance of these areas proved to be time dependent. In the presence of the cysteine proteinase inhibitors, a maximal surface area of demineralized bone was seen between 4 and 8 h of culturing, whereas the metalloproteinase inhibitors had their maximal effect at a later time interval (between 16 and 24 h). Because different inhibitors of each of the two classes of proteolytic enzymes had the same effects, our data strongly suggest that cysteine proteinases attack the bone matrix prior to digestion by MMPs. In line with the view that a sequence may exist were differences in the amount of proteoglycans (shown with the selective dye cuprolinic blue) in the subosteoclastic demineralized areas induced by the inhibitors. In the presence of the cysteine proteinase inhibitor, relatively high levels of cuprolinic blue precipitates were found, whereas this was less following inhibition of metalloproteinases. These data suggested that cysteine proteinases are important for digestion of noncollagenous proteins. We propose the following sequence in the digestion of calvarial bone by osteoclasts: after attachment of the cell to the mineralized surface an area with a low pH is created which results in dissolution of the mineral, then cysteine proteinases, active at such a low pH, digest part of the bone matrix, and finally, when the pH has increased somewhat, MMPs exert their activity.

INTRODUCTION

RESORPTION OF BONE IS accomplished by the multinucleated osteoclast. This cell attaches to mineralized surfaces (e.g., bone) and forms a unique resorption area, the ruffled border, which is sealed from the cell-surrounding environment by the clear zone. One of the first steps in the resorption sequence is acidification of the ruffled border area.1 The low pH results in dissolution of the mineral, thus exposing the collagenous and noncollagenous bone matrix. A subsequent enzymatic attack of the matrix constituents leads to the final degradation of the bone (reviewed in Ref. 2). Although most data in the literature coincide with this sequence, it is still not entirely clear which proteolytic enzymes are responsible for the digestion of the various organic components of the bone matrix.

One of the first indications for the involvement of proteolytic enzymes in bone digestion came from studies of Delaissé and coworkers.3 These authors demonstrated that inhibition of the activity of certain classes of proteinases, in particular the cysteine proteinases, resulted in a strongly decreased bone degradation. Since then numerous studies have provided convincing evidence that this class of enzymes is indeed crucial for the resorptive process.4–15

It is not clear, however, whether these enzymes digest all constituents of the organic matrix. In this respect it is of interest to note that participation of another class of proteolytic enzymes, the matrix metalloproteinases (MMPs, e.g., collagenases and gelatinases) has been implicated.

By using an inhibitor of MMPs, bone digestion was strongly inhibited,16 thus indicating that these enzymes are somehow involved. Their actual participation in the osteoclast-mediated digestion was shown in 1992.7 In the presence of a MMP inhibitor, large areas of demineralized nondigested bone matrix were found adjacent to the ruffled border of actively resorbing osteoclasts in calvarial explants. These observations indicated that demineralization continued, whereas digestion of the matrix was inhibited thus providing convincing evidence for the involvement of MMPs in the subosteoclastic degradation of bone matrix. Because an inhibitor of cysteine proteinases (E-64) had similar effects,6,7 it was concluded that members of both classes of enzymes participated in matrix degradation.

Further support for MMP participation was provided by immunolocalization,17,18 in situ hybridization studies,19–22 and studies in which inhibitors (among which the naturally occurring tissue inhibitor of metalloproteinases [TIMP]) were used.18,23,24 Not all data, however, coincide with the view that MMPs are involved. For as yet unknown reasons, resorption induced by isolated osteoclasts seeded on mineralized substrates (such as bone and dentin slices) does not seem to depend on the activity of this class of enzymes.5,14 Under these conditions, participation of cysteine proteinases was confirmed, but blocking the activity of MMPs did not affect the resorption by osteoclasts. These findings strongly suggest that MMPs are not always a prerequisite for osteoclastic digestion of mineralized tissues. Although it is not clear yet whether this MMP independent degradation is due to differences in the model systems used (isolated osteoclasts versus tissue culture systems), participation of these enzymes in digestion of calvarial bone in explant culture was demonstrated convincingly.7,23,25

The involvement of MMPs in bone resorption would seem to be complicated by the fact that most, if not all, MMPs have a neutral pH optimum, whereas the resorption lacuna is characterized by a much lower pH (4–6).1,26 This acidified environment is optimal for the activity of cysteine proteinases but probably not for metalloproteinases. Since members of both classes of enzymes seem to participate, the intriguing question arises, How do they accomplish that? Do they exert their activity simultaneously or in a certain time sequence? In the present study, we have attempted to answer this by incubating calvarial bone explants for different time intervals with a variety of inhibitors of cysteine proteinases and MMPs. Resorption was quantitatively analyzed at the electron microsocopic level.7 In addition, explants cultured with or without inhibitors were stained with cuprolinic blue, a dye selective for proteoglycans,27 to analyze the presence of these noncollagenous proteins in the resorption lacunae.

MATERIALS AND METHODS

Materials

Parathyroid hormone (PTH; 69.6 μg of PTH/mg of solid), indomethacin, acetazolamide, and E-64 were obtained from Sigma Chemical (St. Louis, MO, U.S.A.). Medium 199 and fetal calf serum were from GIBCO (GIBCO Laboratories, Grand Island, NY, U.S.A.). The MMP inhibitor CI-116 was a gift from Dr. C. Sear (Searle Research and Development, High Wycombe, U.K.), the MMP inhibitor RP5979428 was a gift from Dr. B. Terlain (Rhone-Poulenc Santé, Vitry sur Seine, France), and the MMP inhibitor CT116625,29 was a gift from Dr. A.J.P. Docherty (Celltech, Slough, U.K.). The cysteine proteinase inhibitor Z-Phe-Tyr(O-t-Bu)CHN230 was a gift from Dr. E. Shaw (Friedrich Miesher Institut, Basel, Switzerland), the cysteine proteinase inhibitor CA074 was a gift from Dr. M. Tamai (Taisho Pharmaceutical Co. Ltd., Saitama, Japan), and its methylated form (CA074-Me) was a gift from Dr. D.J. Buttle (Institute of Bone and Joint Medicine, Sheffield, U.K.). The inhibitors used were specific for either MMPs or cysteine proteinases.16,25,28,30,31

Tissue culture

Calvariae (frontal and parietal bones) of 5-day-old NMRI mice were aseptically removed and cultured in medium 199, 2.5% fetal calf serum, and 0.1 μM PTH.3 The inhibitors were used in concentrations that have a maximal effect on resorption: E-64, 42 μM4; CI-1, 40 μM16; RP59794, 25 μM5; CT1166, 10 nM18; Z-Phe-Tyr(O-t-Bu)-CHN2, 0.45 μM32; and CA074/CA074(Me), 10 μM.12 The explants were cultured for different time intervals up to 24 h. Following culturing, the explants were processed for microscopic examination (see below).

Microscopy

Calvarial explants were processed for light and electron microscopic examination as described previously.6 The tissue was fixed in 4% paraformaldehyde and 1% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) for 2 days at room temperature. After fixation, each explant was cut into four parts and postfixed for 60 minutes in 1% OsO4 in the same buffer. The explants were washed in buffer, dehydrated through a series of ethanol, and embedded in epoxy resin (LX-112).

Semithin transverse sections (1–2 μm) were cut and stained with methylene blue. Ultrathin sections were cut with a diamond knife, stained with uranyl and lead, and examined in a Zeiss EM-10 electron microscope (Zeiss, Jena, Germany).

Cuprolinic blue

Calvarial explants were cultured for 24 h in the presence or absence of the cysteine proteinase inhibitor E-64 or the MMP inhibitor CI-1. The explants were fixed for 16 h in 6% paraformaldehyde and immersed in a cuprolinic blue solution (0.2% cuprolinic blue, 2.5% glutaraldehyde, 0.2 M magnesium chloride in 0.025 mM sodium acetate buffer, pH 5.627). The tissue was washed three times for 10 minutes in the same solution without cuprolinic blue, washed in 1% sodium tungstate (3 × 10 minutes), and dehydrated in ascending concentrations of ethanol, the 30% and 50% concentrations containing 1% sodium tungstate. The tissue was embedded in LX-112.

All resorption pits covered by osteoclasts were micrographed at a magnification of ×25,000. The micrographs were coded, randomized, and the resorption areas were evaluated and scored according to the number of cuprolinic blue stain precipitates into one of the three following categories: many precipitates (+), no precipitates (−), and a few precipitates (±). Evaluation of randomized micrographs was performed independently by two persons, who were not aware of the treatment of the explants.

Morphometric analysis of bone resorption

From each fragment of each calvarial explant, we studied one ultrathin section. Each osteoclast that was attached to the bone surface together with adjacent bone was micrographed at a low maginification (×950). In addition, all areas of demineralized bone not occupied by an osteoclast were micrographed. The micrographs were printed at a final magnification of ×5700, coded, randomized, and the surface area of demineralized bone matrix (DA) was quantitatively analyzed using a computerized X-Y tablet (MOP-Videoplan, Kontron, Munchen, Germany). The data were expressed as mean of square micrometer of demineralized area per osteoclast ± SEM of five or six explants.

Statistical analysis

The data were statistically analyzed using Kruskal-Wallis nonparametric analysis of variance test followed by Tukey-Kramer's multiple comparison test. Effects were considered statistically significant when p < 0.05 (two-tailed).

RESULTS

Time-dependent effect of proteinase inhibitors on osteoclastic bone degradation

In line with data presented previously,7 in calvarial explants incubated for 24 h in the presence of the cysteine proteinase inhibitor E-64 or the MMP inhibitor CI-1, large areas of demineralized bone (Fig. 1) were noted adjacent to the ruffled border of almost all osteoclasts. At this time interval, both types of inhibitors had a similar effect on the occurrence of these mineral-free areas (Fig. 2).7 In addition, at the 24 h time point, inhibitor-treated explants also exhibited demineralized areas without osteoclasts attached to them (Fig. 3). This phenomenon was seen with both types of inhibitors and was only apparent at the later time intervals (>16 h). In control explants, such areas were not found at any time point studied.

Figure FIG. 1.

(A) Low-power electron micrograph of an osteoclast (OC) attached to the bone surface (B) of a control explant. (B) Higher magnification of the resorption area; note the presence of mineralized bone in direct contact with the ruffled border (RB) membrane. (C) Osteoclast of an explant incubated for 24 h with the MMP inhibitor CI-1. A large area of bone matrix free of mineral (DA) is seen adjacent to the ruffled border (RB) of the osteoclast. OC, osteoclast; RB, ruffled border; B, bone; DA, demineralized matrix. Magnifications: (A) ×7300; (B) ×12,500; (C) ×7300.

Figure FIG. 2.

Effect of the cysteine proteinase inhibitor E-64 (42 μM) and the MMP inhibitor CI-1 (40 μM) on the occurrence of demineralized areas in calvarial bone explants following a 24 h culture period. Data are expressed as square mictrometers of demineralized area per osteoclast (μm2 of DA/OC); each value represents the mean ± SEM of five explants. *p < 0.01 compared with controls.

Figure FIG. 3.

Demineralized areas (DA) without osteoclasts attached to them. Explants were cultured for 24 h in the presence of the metalloproteinase inhibitor CI-1. Note the gradual decrease in density of mineral (asterisk), which is similar to the decrease seen in the subosteoclastic area (compare with Fig. 1C). Such a gradual decrease in mineral density is not seen at the edges of the resorption pits. At these sites, a sharp demarcation is apparent between mineralized bone and the nonmineralized collagen fringe (arrowheads). OC indicates part of an osteoclast at some distance from the demineralized area (DA). Magnifications: (A) ×7300; (B) ×7300.

The occurrence of subosteoclastic demineralized areas proved to be time-dependent as well as to depend on the type of inhibitor used. Already within a culture period of 4–8 h the effect of the cysteine proteinase inhibitor (E-64) was maximal (Figs. 4, 5, and 6). The MMP inhibitor, however, reached its maximal effect at a later time interval, between 16 and 24 h (Fig. 4).

Figure FIG. 4.

Time response of the cysteine proteinase inhibitor E-64 (42 μM) and the MMP inhibitor CI-1 (40 μM) on the surface area of demineralized bone matrix. The effect of E-64 is maximal at the 8 h time point, whereas the effect of CI-1 is reached later. Data are expressed as square micrometers demineralized area per osteoclast (μm2 DA/OC). Values represent the mean ± SEM of five or six explants. *p < 0.05; **p < 0.001 compared with controls of the different time intervals.

A number of explants was also incubated with a combination of the two inhibitors and cultured for 8 or 24 h. The combined presence of the inhibitors resulted at both time intervals in a maximal surface area of demineralized matrix being similar to the one found with E-64 alone; the presence of the MMP inhibitor did not result in an additional effect (data not shown; see Ref. 7).

Effect of different MMP inhibitors

Although others have shown that synthetic metalloproteinase inhibitors are fast acting in cell culture33 and tissue explant34 systems, we investigated whether the relatively late effect of CI-1 was due to the inhibitor used. Therefore, two other nonrelated inhibitors (RP59794 and CT1166) were tested. Also in the presence of these MMP inhibitors the occurrence of demineralized areas was not found at the early time points (Fig. 5). However, after a culture period of 24 h, the surface area of demineralized matrix was similar to the one found with CI-1 or E-64 (data not shown).

Figure FIG. 5.

Effect of different MMP inhibitors and the cysteine proteinase inhibitor E-64 on the surface area of demineralized bone matrix. (A) Effect of E-64 (42 μM), CI-1 (40 μM), and RP 59794 (25 μM) on the subosteoclastic surface area of demineralized bone after 4 h and 8 h of culturing. Data are expressed as square micrometers of demineralized area per osteoclast (μm2 of DA/OC). Values represent the mean ± SEM of five explants. *p < 0.001 compared with controls and metalloproteinase inhibitor-incubated explants; **p < 0.001 compared with metalloproteinase inhibitor-incubated explants. (B) Effect of E-64, CI-1, and CT1166 (10 nM) on the subosteoclastic area of demineralized bone (4 h culture period). Data are expressed as square micrometers of demineralized area per osteoclast (μm2 of DA/OC). Values represent the mean ± SEM of five explants. *p < 0.05 compared with controls and metalloproteinase-incubated explants.

Apart from the demineralized area expressed as mean values (see above), we also established the number of osteoclasts with or without such an area and categorized these according to the size of the mineral-free resorption zone (Table 1). The data demonstrated that in calvarial explants treated with either inhibitor osteoclasts were found in close contact with a demineralized area, whereas such a phenomenon was hardly noted in the control explants. The highest number of osteoclasts adjacent to a demineralized area was seen in the E-64–incubated calvariae (86% and 85% at the 4 and the 8 h time interval, respectively; Table 1).

Table TABLE 1. OSTEOCLASTS ATTACHED TO DEMINERALIZED BONE MATRIX
original image

Effect of different cysteine proteinase inhibitors

In a series of experiments, we compared the effect of two other more selective cysteine proteinase inhibitors with the effect of the general inhibitor, E-64. The other inhibitors were Z-Phe-Tyr(O-t-Bu)-CHN2, which is a relatively selective cathepsin L inhibitor,30 and CA074, a cathepsin B inhibitor.31 The latter inhibitor was used also in a methylated form. The methylated inhibitor is inactive in the extracellular space, but following its penetration of the plasma membrane, it is activated by cytoplasmic esterases.31

Calvarial explants were incubated for 4 h in the presence of these inhibitors and analyzed. The morphometric data demonstrated that the general inhibitor (E-64) induced the highest level of subosteoclastic demineralized matrix, followed by the cathepsin L inhibitor and the methylated cathepsin B inhibitor (Fig. 6). The nonmethylated CA074 did not induce demineralized areas. Also in this experiment, the effect of the metalloproteinase inhibitor (CI-1) was less than the effect of either of the three effective cysteine proteinase inhibitors.

Figure FIG. 6.

Effect of different cysteine proteinase inihibitors on the surface area of subosteoclastic demineralized bone matrix. Calvarial explants were cultured for 4 h in the absence or presence of the cysteine proteinase inhibitors, E-64 (42 μM), CA074 (CA, 10 μM), CA074-Me (CA-Me, 10 μM), and Z-Phe-Tyr(O-t-Bu)-CHN2 (Z-Phe, 0.45 μM) or the MMP inhibitor CI-1 (40 μM). Data are expressed as square micrometers of demineralized area per osteoclast (μm2 of DA/OC). Each value respresents the mean ± SEM of six explants. *p < 0.05, **p < 0.001 compared with controls.

Preculture

A possible explanation for the relatively late effect of the MMP inhibitors is that, due to the isolation of the calvariae, production of these enzymes was low. Since during the initial phase of culturing, PTH up-regulates the synthesis of metalloproteinases,35 we precultured the explants for 4 h without inhibitors but with PTH. This was followed by a subsequent 4 h incubation in the presence of both types of inhibitors (and PTH). Also under these conditions, the effect of the cysteine proteinase inhibitor was much stronger than the effect of the metalloproteinase inhibitor and did not differ from explants which were not precultured (Fig. 7).

Figure FIG. 7.

Effect of preculturing of calvarial explants on the appearance of demineralized areas. Calvarial explants were cultured for 4 h in the presence of the cysteine proteinase inhibitor E-64 (42 μM) or the MMP inhibitor CI-1 (40 μM). One series of explants (4 + 4 h) was, prior to incubation with the inhibitors, precultured for 4 h with PTH. Data are expressed as square micrometers of demineralized area per osteoclast (μm2 of DA/OC). Values represent the mean ± SEM of five explants. *p < 0.05 compared with metalloproteinase inhibitor-incubated explants.

Effect of azetazolamide and indomethacin

To investigate whether interfering with acidification influenced the response to the inhibitors, explants were incubated with the carbonic anhydrase inhibitor azetazolamide (75 μM), a compound shown to strongly inhibit osteoclastic bone resorption.36 In calvarial explants cultured for 24 h in the presence of this compound as well as inhibitors, demineralized areas were absent. Approximately 90% of the bone-lining osteoclasts proved to be attached to mineralized bone, whereas in control explants incubated with inhibitors but without azetazolamide, up to 80% of the osteoclasts was associated with demineralized areas.

It is likely that isolation and subsequent culturing of calvariae results in an increased synthesis of prostaglandins. Since these compounds modulate osteoclastic bone resorption,37,38 probably by inducing a higher level of MMPs,35 we analyzed osteoclastic activity in the presence of indomethacin, a potent inhibitor of prostaglandin synthesis. Explants were cultured for 24 h with or without indomethacin and with or without the proteinase inhibitors E-64 and CI-1. In the presence of indomethacin, the subosteoclastic surface area of demineralized bone proved to be inhibited by ∼50% (Fig. 8). This effect was seen with both types of inhibitors, suggesting that osteoclastic bone digestion is partially mediated by prostaglandins.

Figure FIG. 8.

The effect of indomethacin on the appearance of subosteoclastic demineralized areas. Calvarial explants were incubated for 24 h in the absence or presence of the cysteine proteinase inhibitor E-64 (42 μM) or the MMP inhibitor CI-1 (40 μM). To one series of explants the prostaglandin synthesis inhibitor indomethacin (14 μM) was added. Data are expressed as square micrometers of demineralized area per osteoclast (μm2 of DA/OC). Each bar represents the mean ± SEM of five explants. *p < 0.05; **p < 0.01 compared with controls.

Cuprolinic blue

In resorption lacunae, cuprolinic blue precipitates were abundantly present. The precipitates were seen in demineralized areas created in the presence of both types of inhibitors. Differences in the amount of precipitate in the resorption areas, however, were noted among the two inhibiting compounds (Fig. 9). In explants incubated with E-64, more resorption lacunae contained precipitates than in those incubated with the MMP inhibitor (Fig. 10).

Figure FIG. 9.

(A) Cuprolinic blue precipitates (arrows) in a resorption lacuna adjacent to an osteoclast of an explant cultured for 24 h in the presence of the cysteine proteinase inhibitor, E-64. (B) Resorption lacuna of an osteoclast cultured with the MMP inhbitor, CI-1. Note the almost complete absence of stain precipitates. Magnifications: (A) ×58,000; (B) ×58,000.

Figure FIG. 10.

Mean percentage of resorption lacunae (%DA ± SEM) with many (+), without (−), or with a few (±) cuprolinic blue stain percipitates. Calvarial explants were cultured for 24 h with inhibitors (E-64 and CI-1) and stained with cuprolinic blue to detect proteoglycans in the demineralized resorption lacunae.

DISCUSSION

The data presented in this study show that under the influence of different inhibitors, which selectively block the activity of cysteine proteinases, or MMPs, demineralized areas were formed in the subosteoclastic region of actively resorbing osteoclasts. This finding is in line with previous observations7 and indicates that digestion of calvarial bone matrix was disturbed whereas demineralization proceeded. Osteoclastic degradation of calvarial bone matrix thus depends on the activity of both classes of enzymes. An intriguing finding of the present study was the time-dependent effects of the proteinase inhibitors used. Blocking activity of cysteine proteinases resulted in a maximal effect already after a relatively short incubation period, between 4 and 8 h. The maximal effect of the MMP inhibitors, however, was found at a later time point (16–24 h). These data strongly suggest that cysteine proteinases and MMPs do not play the same role in subosteoclastic digestion of bone matrix.

Different roles of enzymes

What different roles do these enzyme classes play? It could be assumed that both classes of proteinases act simultaneously in the resorption zone, though at different rates, with the cysteine proteinases being more active than the MMPs. This view is supported by the finding that both types of inhibitors induce demineralized areas as soon as they are added (see Table 1) but that these areas grow faster in the presence of the cysteine proteinase inhibitors. However, if this interpretation were true, then the simultaneous presence of both types of inhibitors would result in the formation of larger demineralized areas than when added separately, and the rate of increase in demineralized fringes would remain higher in the presence of cysteine proteinase inhibitors than with MMP inhibitors. Our observations show that a higher activity of cysteine proteinases is very unlikely. First, the presence of both inhibitors does not induce a stronger effect (see also Ref. 7) and second, as shown in Fig. 4, the inhibitory effect of the metalloproteinase inhibitors increases with time and reaches the same level as with the cysteine proteinase inhibitors. This increase in time could mean, however, that culturing of the calvariae with PTH up-regulates MMPs, so that the participation of MMPs would become more important as compared with that of the cysteine proteinases. Our preculture experiments do not support this option since the addition of inhibitors following an initial culture period of 4 h with PTH did not result in an increased effect of the MMP inhibitor (see Fig. 7). Similarly, it could be speculated that endogenous prostaglandin E2 might progressively up-regulate metalloproteinase synthesis during the culture period.35 If this was indeed the case, inhibition of prostaglandin E2 production with indomethacin would keep the participation of MMPs low, and after a 24 h culture period the metalloproteinase inhibitor-induced demineralized fringes would not develop to the same extent as those induced by the cysteine proteinase inhibitor. Our data demonstrate that this is not true: the presence of indomethacin does not affect the relative inhibitory effects of the two types of inhibitors (Fig. 8).

According to the above results, the differences in the roles of MMPs and cysteine proteinases are not merely quantitative. It may be assumed that qualitative differences exist. One possibility is that MMPs participate in the migration of osteoclasts.39,40 According to the latter authors, blocking the activity of MMP inhibits the movement of osteoclasts. If this is indeed the case and the cells still secrete H+ at a later stage, but produce less cysteine proteinases, then demineralized fringes would be generated because of the lack of cysteine proteinases and not because of an inhibited MMP activity in the resorption zone. This is not very likely because we have found that resorption pits (demineralized areas), which were formed during the culture period, were left by the osteoclast. This phenomenon was seen in the presence of all inhibitors and thus indicated migration of osteoclasts, also after blocking the activity of metalloproteinases (see Fig. 3).

A feasible explanation for our findings is that the two enzyme classes act in sequence and may participate in the digestion of different matrix components. A series of independent observations do indeed suggest a sequential action of the enzymes. First, the simultaneous addition of a MMP as well as a cysteine proteinase inhibitor does not lead to a stronger effect than a separate addition of the compounds, suggesting that cysteine proteinases and MMPs act in series rather than in parallel. Since during the initial phase cysteine proteinase inhibitors have a stronger effect than MMP inhibitors, cysteine proteinases are likely to act first. Second, cysteine proteinases appear more active in proteoglycan digestion as compared with MMPs (Fig. 10). Because digestion of proteoglycans has been shown to precede degradation of collagenous components,41 and cysteine proteinases are known to digest proteoglycans quite effectively,42 it may be assumed that cysteine proteinases are required for the initial attack of the matrix and that the MMPs start later. Third, a sequence of enzyme activities seems to coincide with the pH profile of the resorption lacuna. After attachment of the osteoclast to the mineralized surface, a low pH is created in the resorption lacuna.1,26 This acidified environment leads to dissolution of the mineral crystallites and exposure of the organic matrix. We suggest that during this initial phase with the low pH the cysteine proteinases are active. Later, when the pH has increased due to a higher level of phosphate and/or Ca2+,43,44 the other class of enzymes, the MMPs, digest the rest of the matrix.

Which enzymes participate in the digestion?

Cysteine proteinases:

Because some of the inhibitors used in the present study are relatively selective for certain members of the two proteinase classes studied, we are also able to give some clues as to the nature of the enzymes involved in the digestion. CA074 is quite selective for cathepsin B,31 whereas Z-Phe-Tyr(O-t-Bu)-CHN2 is more selective for cathepsin L.30 Both inhibitors induced comparable effects with respect to the subosteoclastic demineralized areas, suggesting that both members of the cysteine proteinases participated in the digestion.12 According to the latter authors, cathepsin B is more important for processes taking place intracellularly, whereas cathepsin L is considered to exert its activity primarily in the (extracellular) ruffled border area. Since we also found the demineralization effect using the cathepsin B inhibitor, our data seem to suggest that this enzyme has at least some activity in the extracellular space. What we do not know though, is whether the inhibitors used interfere with the recently described osteoclast-cysteine proteinase, cathepsin K.45,46 Deficiency of this enzyme has been shown to occur in pycnodysostosis,47 a rare osteopetrosis-like disease characterized by the presence of large mineral-free areas of bone matrix adjacent to the osteoclasts.48 It is likely that the occurrence of these areas is indeed due to a decreased activity of cysteine proteinases,6 probably cathepsin K. So far it cannot be ruled out, however, that the activity of metalloproteinases is unaffected in pycnodysostosis. After all, our in vitro data demonstrate also that inhibition of these enzymes causes similar demineralization effects.

Matrix metalloproteinases:

With regard to the nature of the MMPs involved in matrix digestion, only the data obtained from one of the three inhibitors may give us some information, since the Ki values of this inhibitor (CT1166) have been extensively documented.25 This inhibitor blocks primarily the activity of gelatinases A and B (MMP-2 and MMP-9, respectively) at the concentration used. Interstitial collagenase (MMP-1) is not inhibited under these conditions. The effect of CT1166 on the amount of demineralized matrix was comparable with the one found with the general inhibitors, suggesting that gelatinases play an important role in digestion. Of the two known gelatinases, gelatinase B (MMP-9) is a likely candidate because its presence in osteoclasts has been shown by immunocytochemistry18 and in situ hybridization.20,21,49

CONCLUSION

In conclusion, the data presented in this study strongly suggest a sequential action of enzymes participating in osteoclastic degradation of calvarial bone. We propose that first mineral crystallites are dissolved due to the low pH created by the osteoclast in the resorption site. Subsequently, during this low pH phase, cysteine proteinases are secreted in this area and exert their activity. Finally, the pH increases somewhat and MMPs digest the rest of the bone matrix.

Acknowledgements

The authors thank Mr. C.E. Gravemeijer for careful handling of the photographic material.

Ancillary