Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Emphysema, a disease characterized by destruction of alveolar walls, has generally been accepted as an irreversible lesion (1, 2). The observation that retinoic acid can partially reverse emphysema in an animal model (3), however, has challenged this concept and engendered considerable excitement for the potential development of novel therapeutic approaches in emphysema. Retinoids are believed to play an important role in regulating the alveolarization process during lung development (4, 5). This has led to the concept that neoalveolarization induced by retinoic acid following the development of emphysema may be, in part, a recapitulation of the alveolarization process. Other, non-exclusive mechanisms for retinoid activity in emphysema may also be possible.

The observation that individuals with homozygous 1 protease inhibitor deficiency are unusually susceptible to the development of emphysema led to the "protease-anti-protease" hypothesis of emphysema (1, 6). In this concept, proteolytic activity in excess of anti-proteolytic defenses can lead to the development of emphysema (1). It is now believed that other destructive processes, including oxidants (9, 10) and potentially toxic peptides such as defensins, may also play a role in tissue destruction leading to emphysema. In its expanded form, the protease-anti-protease hypothesis remains the most widely accepted concept for the development of emphysema (11). Agents that inhibit these destructive processes have the potential for altering the development of this disease.

Retinoids are capable of regulating a large number of metabolic processes (5, 12), including matrix metalloproteinases (MMPs). It is reasonable to hypothesize, therefore, that retinoids might be able to modulate extracellular matrix turnover and, by such a mechanism, could regulate the tissue remodeling believed to play an important role in emphysema. The current study, therefore, was designed to evaluate the ability of retinoids to modulate fibroblast-driven extracellular matrix degradation. The in vitro system in which fibroblasts are cultured in a three-dimensional matrix composed of type I collagen was used as a model system to assess this hypothesis.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Type I collagen was extracted from rat-tail tendons (RTTC) by a previously published method (16, 17). Briefly, tendons were excised from rat tails, and the tendon sheath and other connective tissues were removed carefully. After repeated washing with tris-buffered saline (TBS, 0.9% NaCl, 10 mM Tris, pH 7.5) and 95% ethanol, type I collagen was extracted in 6 mM hydrochloric acid at 4°C for 24 h. Protein concentration was determined by weighing a lyophilized aliquot from each lot of collagen solution. Sodium dodecyl sulfate polyacrylamide gel electrophoresis routinely demonstrated no detectable proteins other than type I collagen. Human neutrophil elastase (NE) was purchased from Elastin Products Company (Owensville, MO). All-trans retinoic acid (RA) and 1-antitrypsin (1-PI) was purchased from Sigma (St. Louis, MO). Human recombinant tumor necrosis factor (TNF)-, human recombinant interleukin (IL)-1, and human recombinant interferon (IFN)- were purchased from R&D Systems (Minneapolis, MN). Tissue culture supplements and media were purchased from GIBCO (Life Technologies, Grand Island, NY). Fetal calf serum (FCS) was purchased from Biofluid (Rockville, MD).

Cell Cultures

Human fetal lung fibroblasts (HFL-1) were obtained from the American Type Culture Collection (Rockville, MD). The cells were cultured in 100-mm tissue culture dishes (Falcon, Becton Dickinson Labware, Lincoln Park, NJ) with Dulbecco's modified Eagle's medium (DMEM), supplemented with 10% FCS, 50 U/ml penicillin, 50 µg/ml streptomycin, and 0.25 µg/ml fungizone. The fibroblasts were passaged every 3-5 d. Subconfluent fibroblasts were trypsinized (trypsin-EDTA; 0.05% trypsin, 0.53 mM EDTA-4 Na) and used for collagen gel culture. Fibroblasts used in these experiments were between cell passages 14 and 19.

NE was dissolved in phosphate-buffered saline (PBS) to a stock solution of 1 mg/ml. TNF-, IL-1, and IFN- were dissolved in PBS to a stock solution of 1 µg/ml. RA was dissolved in dimethyl sulfoxide (DMSO) to a stock solution of 1 mM.

Preparation of Collagen Gels

Collagen gels were prepared as described previously (17). Briefly, the appropriate amount of RTTC was mixed with distilled water, 4× concentrated DMEM, and cell suspension so that the final mixture resulted in 0.75 mg/ml of collagen, 4.5 × 10⁵ cells/ml, and a physiologic ionic strength. Fibroblasts were routinely added last to minimize damage during the preparation of collagen gels. One-half milliliter of the mixture was cast into each well of 24-well tissue culture plates (Falcon, Franklin Lakes, NJ). Gelation occurred in about 20 min at room temperature, after which the gels were released and transferred to 60-mm tissue culture dishes containing 5 ml of serum-free DMEM and stored at 37°C with 5% CO₂ for 4-5 d. To demonstrate the effects of cytokines and elastase on collagen gel contraction and collagen degradation, cytomix (18) (TNF- 10 ng/ml, IL- 1 5 ng/ml, and IFN- 10 ng/ml), NE, or the combination of NE with cytokines was added to culture media. To investigate the effect of RA on gel contraction and collagen degradation, RA (1 µM) was added into the media containing NE, cytomix, or the combination. Gel area was measured daily using an image analysis system (Optomax, Hollis, NH).

Hydroxyproline Assay

The amount of hydroxyproline, which is directly proportional to type I collagen content in the gels, was measured by spectrophotometric determination (19, 20). Briefly, the media surrounding gels was completely removed, and the gels were transferred to glass tubes (KIMAX, Fisher Scientific, St. Louis, MO) with 2 ml of 6N HCl. Oxygen was removed by ventilating with pure nitrogen for 30 s. The gels were then hydrolyzed at 110°C for 12 h. The samples were dried with a vacuum centrifuge, then dissolved in dH₂O. Hydroxyproline in the samples reacted with oxidant (1.4% Chloramine T in acetate/citric acid buffer, Sigma) and Ehrlich's reagent (0.4% p-Dimenthylamino-benzaldehyde [Sigma], in 60% perchloric acid, Fisher Chemical) at 65°C for 25 min, and absorbance was measured at 550 nm (Ultrospec 2000, Pharmacia Biotech, Piscataway, NH).

Gelatinase Activity Assay

To determine the presence of active gelatinase, gelatin zymography was prepared. Conditioned media were concentrated 10-fold by lyophilization, followed by solution in distilled water. Gelatin zymography was performed by a modification of previously published procedures (21, 22). The samples were dissolved in 2-fold electrophoresis sample buffer, and heated for 5 min at 95°C. Thirty microliters of each sample were then loaded in each lane, and electrophoresis was performed at 45 mV per gel. After this, the gels were soaked with 2.5% (v/v) Triton-X 100 gently shaking at 20°C for 30 min, then incubated in the metalloproteinase buffer (0.06 M Tris-HCL, pH 7.5, containing 5 mM CaCL₂ and 1 µM ZnCL) for 18 h at 37°C. The gels were stained with 0.4% (w/v) Coomassie blue and rapidly destained with 30% (v/v) methanol, 10% (v/v) acetic acid. The gels were dried between dialysis membranes (cellophane sheets; Pharmacia Biotech, San Francisco, CA).

Immunoblot Analysis of Metalloproteinases

To further identify the MMPs produced, immunoblots were performed. Supernatant media from three-dimensional cultures were precipitated with ethanol, resuspended in an equal volume dH₂O and 2× sample buffer (0.5 M Tris-HCl, pH 6.8, 10% sodium dodecyl sulfate [SDS], 0.1% bromphenol blue, 20% glycerol). After heating for 3 min at 95°C, 30 µl of each sample was loaded into each well, and electrophoresis was performed with mini-protein 3 cells (Bio-Rad, Hercules, CA). The proteins were transferred to polyvinyl difluoride (PVDF) membranes (Bio-Rad) in electroblotting buffer (20 mM tris, pH 8.0, 150 mM glycine, 20% MeOH) at 20 V for 35 min with semi-dry electrophoretic transfer cell (Bio-Rad). The blots were blocked in 5% nonfat milk in PBS-Tween at room temperature for 1 h and then exposed to primary antibodies (mouse anti-human MMP-1, MMP-3, TIMP-1, and TIMP-2; Calbiochem, Cambridge, MA) for 1 h. Target protein was subsequently detected using rabbit anti-mouse IgG horseradish peroxidase (Rockland, Gilbertsville, PA) in conjunction with an enhanced chemiluminescence detection system (ECL; Amersham Pharmacia Biotech, Little Chalfont, Buckinghamshire, England).

NE Activity Assay

To investigate the effect of RA on functional NE, NE activity was quantified (23). NE (40 nM), RA (1 µM), or cytomix alone or in various combinations were added to the media in which collagen gels with or without fibroblasts were floated. 1-Protease inhibitor (1-PI) was used as a control elastase inhibitor (100 nM). The conditioned media were harvested at 2, 12, 24, and 48 h of incubation. Elastase activity was measured using a synthetic substrate, methoxy-succinyl-alanyl-prolyl-valyl-p-nitroanilide (Calbiochem-Novabiochem Co., La Jolla, CA). The supernatants (100 µl) were incubated with 200 µl of 0.2 M substrate in 0.1 M HEPES, 0.5 M NaCl, and 10% DMSO at pH 7.5. After incubation for 24 h, absorbance of the product, p-nitroanilide, was measured at 414 nm. Purified human NE was used as a standard, and the elastase activity was expressed as ng/ml.

Statistical Evaluation

Results are expressed as the means of three determinations ± SEM except as described otherwise. The gel contraction assay, hydroxyproline measurement, and zymograms were confirmed by repeating experiments on separate occasions at least three times. The data shown in each figure were taken from single, representative experiments except as described. Grouped data were evaluated by analysis of variance (ANOVA), followed by Bonferroni correction. Differences between two groups were analyzed by unpaired Student's t test and comparisons were considered statistically significant if P < 0.05.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Effect of RA on Fibroblast-Mediated Collagen Gel Contraction

By Day 5, NE augmented and cytomix inhibited fibroblast-mediated collagen gel contraction, and NE plus cytomix together further enhanced the gel contraction compared with elastase alone. RA resulted in a significant inhibition of the synergistic augmentation of contraction observed when elastase and cytomix were added together. In the absence of RA, on average, the two agents together resulted in gel contraction to 2.5 ± 0.3% of original size, while in the presence of RA, gels contracted to 15.1 ± 2.1% of original size (P < 0.01, Figure 1A).

View larger version (19K): [in this window] [in a new window]   Figure 1.   Effect of RA on fibroblast-mediated collagen gel contraction in the presence of NE and cytomix. Fibroblasts (4.5 × 105/ml) were cast into collagen gels and floated in serum-free DMEM alone or supplemented with RA (1 µM) either alone or with NE (15 nM), cytomix (TNF-, 10 ng/ml; IL-1, 5 ng/ml; and IFN-, 10 ng/ml), or both. On Day 5, gel size was measured with an image analyzer and hydroxyproline amount was measured, as described in MATERIALS AND METHODS. Open bar: control; closed bar: +RA. The data presented are means ± SEM for triplicate gels in a single experiment. Error bars not shown are contained with plot symbols. NE: neutrophil elastase; RA: retinoic acid. (A) Gel size. Vertical axis: gel area (expressed as percentage of original size); horizontal axis: additions. (B) Hydroxyproline measurement. Vertical axis: hydroxyproline amount in each gel (µg/ gel); horizontal axis: additions.

The Effect of RA on Collagen Degradation

To determine if RA altered collagen degradation, hydroxyproline content within the gels was determined. Over the five-day culture period, NE and cytomix added together resulted in nearly complete degradation of the collagen within the gel (Figure 1B). RA had no effect on collagen degradation when added either alone or in the presence of cytomix or NE. However, RA dramatically reduced the synergistic degradation that occurred in the presence of NE and cytomix together (7.3 ± 0.4 versus 17.9 ± 1.4 µg/gel).

To determine if RA inhibited collagen degradation in a concentration-dependent manner, various concentrations of RA were added to fibroblast cultures containing both NE and cytomix. In the presence of this combination without RA, 98% of the collagen was degraded. Over the concentration range 0.01 µM to 1 µM, RA demonstrated a concentration-dependent inhibition of collagen degradation (Figure 2).

View larger version (15K): [in this window] [in a new window]   Figure 2.   Effect of RA on collagen degradation induced by cytomix and elastase. Fibroblasts were cast into collagen gels and floated in media containing cytomix along or combined with NE, with or without various concentrations of RA. After 5 d, hydroxyproline content in the gels was determined. Vertical axis: hydroxyproline content; horizontal axis: culture conditions. NE: neutrophil elastase; RA: retinoic acid. *P < 0.0083 by Bonferroni test.

Effect of RA Inhibition of Collagen Degradation Induced by NE in Combination with TNF-, IL-1, or IFN-

To determine if RA was affecting an interaction of any one component of cytomix with NE, each component was added to fibroblast gel cultures alone and in combination with NE in the presence and absence of RA. As shown in Table 1, individual components resulted in minimal degradation of collagen contained in the floating gels. NE resulted in increased degradation of collagen in the presence of TNF- (P < 0.01) and in a smaller but significant augmentation of degradation in the presence of IL-1 (P < 0.05). NE added with IFN- was not significantly different with regard to collagen degradation than NE alone. NE added with the individual cytokines, however, was not as effective in leading to collagen degradation as was the combination with cytomix (Figure 1B and Table 1). The addition of RA resulted in an attenuation of collagen degradation which was most marked for the combination of RA and TNF- (Table 1).

Effect of RA on NE Activity

To determine if RA resulted in direct inhibition of NE activity, cultures were prepared with and without fibroblasts and incubated with NE, cytomix, or their combination in the presence and absence of RA. As a control, the NE inhibitor 1 protease inhibitor (1-PI) was applied. In the presence of 1-PI, no NE activity was detected at any time (data not shown). At 2 h, RA had no effect on NE activity in gels containing fibroblasts or in the absence of cells (Figures 3A and 3B). Interestingly, a dramatic decrease occurred in NE activity over 12 h of culture, perhaps due to autodegradation of the NE. In the absence of fibroblasts, this resulted in complete loss of NE activity over 12 h (Figure 3B). In the presence of fibroblasts, however, residual elastase activity was detectable throughout the culture (Figure 3A). At 12 h, only 3.9 ± 4.4% of the initial elastase activity was observed, after which elastase-like activity gradually increased with culture time. By 48 h, it increased to 28.7 ± 1.5% of initial activity. RA inhibited elastase-like activity significantly only at 48 h (15.9 ± 2.6% of the initial activity, P < 0.05). Cytomix added with elastase (Figure 3C) resulted in more elastase-like activity than elastase alone (Figure 3A). This was significantly inhibited by RA (Figure 3C, 25.8 ± 1.2% versus 18.8 ± 1.4% of initial elastase activity at 48 h, P < 0.05).

View larger version (22K): [in this window] [in a new window]   Figure 3.   Effect of RA on NE activity. Collagen gels were prepared with or without fibroblasts. The gels were floated in media supplemented with or without cytomix, NE, and RA. Gels were incubated at 37°C for 48 h. Media were harvested at various times, and elastase activity was assayed. Vertical axis: neutrophil elastase activity; horizontal axis: culture conditions. The data presented are means ± SEM from three separate experiments. (A) with HFL-1; (B) without HFL-1; (C ) with HFL-1. RA: retinoic acid; NE neutrophil elastase; CM: cytomix.

Effect of RA on MMP-2 and MMP-9 Activity: Gelatin Zymography

Gelatinases released by fibroblasts cultured in three-dimensional collagen gels were assessed by gelatin zymography (Figure 4). Under control conditions, fibroblasts released primarily MMP-2 (gelatinase A) into surrounding medium as identified by its characteristic molecular weight of 72 kD (latent form) and 66 kD (active form). This identification was confirmed by Western blotting (data not shown). Exposure to NE appeared to increase the amount of MMP-2 and to increase the percentage present in the active form (Figure 5A). RA did not markedly affect MMP-2 under either control conditions or in the presence of NE (Figure 4A).

View larger version (68K): [in this window] [in a new window]   Figure 4.   Gelatin zymography. Cultures incubated with various additions were incubated for 5 d, after which media were harvested for gelatin zymography. (A) No cytokines. (B) Cytomix. (C ) Individual cytokines. Other additions and molecular weight markers are indicated. NE: neutrophil elastase; RA: retinoic acid; CM: cytomix. The conditioned media from HT1080 was used as a positive control (P).

View larger version (80K): [in this window] [in a new window]   Figure 5.   Immunoblot for MMP-1. Cultures incubated with various additions were incubated for 5 d, after which media were harvested for SDS-PAGE, followed by Western blotting with antibodies for MMP-1. (A) No cytokines. (B) Cytomix. (C ) Individual cytokines. Other additions and molecular weight markers are indicated. The conditioned medium from HT1080 was added as a positive control. P: positive control; RA: retinoic acid; NE: neutrophil elastase; CM: cytomix.

In the presence of cytomix, a marked induction MMP-9 (gelatinase B) was identified by its characteristic molecular weight of 92 kD (Figure 4B) and confirmed by Western blotting (data not shown). Retinoic acid increased the amount of detectable latent MMP-9 (Figure 4B). Within the detectable limits of zymography, exposure to NE resulted in complete conversion of latent MMP-9 to the active 83 kD form in the presence of cytomix alone. When RA and NE were present together, however, there was only partial conversion of MMP-9 from the 92 kD to the 83 kD form (Figure 4B).

The individual cytokines contained in the cytomix were also evaluated. Both TNF- and IL-1 increased the amount of latent MMP-9 release, while IFN- had no detectable effect (Figure 4C). RA increased the apparent amount of MMP-9 induced in the presence of cytokines, and exposure to NE resulted in partial conversion to lower molecular weight forms corresponding to active MMP-9. This conversion was only partial and, in this respect, differed from the complete conversion that occurred with NE in the presence of cytomix. RA reduced the amount of detectable MMP-9 when NE was added to individual cytokines (Figure 4C).

Effect of RA on Induction of MMP-1 Expression and Activation

Immunoblotting was performed to determine the effect of RA on MMP-1 (Figure 5). Under control conditions, a small amount of MMP-1 was detectable with an apparent molecular size of 52 kD, corresponding to the latent form of MMP-1 (Figure 5A). Minimal MMP-1 was detectable under control conditions or with RA. Neutrophil elastase alone induced MMP-1. RA reduced the detectable MMP-1 in the presence of NE (Figure 5A).

In the presence of cytomix, a marked induction of MMP-1 release was observed (Figure 5B). RA reduced the MMP-1 production induced by cytomix. Co-exposure to NE, however, subsequently converted some of the MMP-1 to lower molecular weight forms corresponding to 42 kD and 20 kD, consistent with the size expected for active MMP-1. RA added with NE reduced the amount of MMP-1 detectable in both higher and lower molecular weight forms (Figure 5B).

The individual cytokines were less potent than cytomix in induction of MMP-1. TNF- was more effective than IL-1, while IFN- appeared to have a minor effect on MMP-1 induction. RA appeared to attenuate the TNF- and IL-1 induction of MMP-1. Exposure to NE resulted in the conversion of MMP-1 to lower molecular weight forms corresponding to active MMP-1. RA added with NE appeared to reduce the amount of active MMP-1 in the presence of TNF- or IL-1 (Figure 5C).

Effect of RA on MMP-3

Immunoblotting was performed to determine the effect of RA on MMP-3 (Figure 6). Fibroblasts incubated without cytokines produced no detectable MMP-3, and neither the presence of NE nor RA altered this (Figure 6A). In the presence of cytomix, however, detectable MMP-3 was observed with an apparent molecular size of both 57 kD, and smaller amounts were detectable in lower molecular weight forms (Figure 6B). Co-exposure to NE resulted in increased amount of MMP-3 detectable in lower molecular weight forms (Figure 6B). RA reduced the cytomix induction of both latent and active forms of MMP-3.

View larger version (62K): [in this window] [in a new window]   Figure 6.   Immunoblot for MMP-3. Cultures incubated with various additions were incubated for 5 d, after which media were harvested for SDS-PAGE, followed by Western blotting with antibodies for MMP-3. (A) No cytokines. (B) Cytomix. (C ) Individual cytokines. Other additions and molecular weight markers are indicated. The conditioned medium from HT1080 was added as a positive control. P: positive control; RA: retinoic acid; NE: neutrophil elastase.

Both IL-1 and TNF- were able to induce the release of detectable MMP-3 with IL-1 being considerably more effective than TNF- (Figure 6C). Interferon- alone did not induce the release of MMP-3. Increased amount in MMP-3 to lower molecular weight forms was observed when NE was added in the presence of IL-1, and this was reduced in the presence of RA. RA also appeared to reduce the amount of MMP-3 detectable in the latent form when added to TNF- or IL-1 in the presence of NE, an effect that was also observed in the presence of cytomix.

Effect of RA on TIMP-1 and TIMP-2

As MMP activity can be regulated by the presence of inhibitors, prominently TIMPs, the effect of RA on release of TIMP-1 and TIMP-2 was investigated by immunoblotting (Figure 7). Detectable TIMP-1 was released from fibroblasts cultured in three-dimensional collagen gels under control conditions. Bands were detectable both at 30 kD corresponding to free TIMP-1 and to higher molecular weight forms consistent with TIMP-1 complexed with MMPs. RA resulted in a reduction in the amount of TIMP-1 detectable in higher molecular weight forms. NE reduced the amount of free TIMP-1 while RA added together with NE appeared to increase free TIMP-1 as well as increase the amount of TIMP-1 present in higher molecular weight forms (Figure 7A).

View larger version (83K): [in this window] [in a new window]   Figure 7.   Immunoblot for TIMP-1 and TIMP-2. Fibroblasts were cast into collagen gels and floated in medium with or without cytomix and elastase in the presence or absence of RA. After 5 d, media were harvested and subjected to SDS-PAGE, followed by Western blotting with antibodies for TIMP-1 (A and B) or TIMP-2 (C and D). Additions and molecular weight standards are indicated. The conditioned medium from HT1080 was added as a positive control. RA: retinoic acid; NE: neutrophil elastase; CM: cytomix; P: positive control.

Cytomix resulted in significant induction of TIMP-1 release. The amount of detectable TIMP-1 was reduced in the presence of NE while RA increased it (Figure 7B).

In contrast to TIMP-1, no detectable TIMP-2 was recovered under control conditions (Figure 7C). In the presence of cytomix, however, TIMP-2 could be readily detected migrating with a molecular size corresponding to 72 kD (Figure 7D). RA reduced this band, and exposure to NE nearly completely eliminated it.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

The current study demonstrates that RA can modulate the degradation of type I collagen by fibroblasts cultured in three-dimensional gels. In the presence of cytokines, which induce matrix metalloproteinase production, and NE, which results in activation of latent MMPs, nearly complete degradation of extracellular type I collagen can occur. RA can inhibit this degradation in a concentration-dependent manner. The effect of RA was observed with TNF- and IL-1 separately and when combined with IFN- together as cytomix (18). The degradation of collagen was associated with augmented fibroblast-mediated contraction of the three-dimensional collagen gels, and this was also inhibited by RA. RA did not appear to inhibit NE directly but did result in inhibition of MMP activation. Taken together, the results of the current study support the concept that RA can modulate tissue remodeling by altering cytokine-driven tissue degradation.

RA is a multi-functional hormone capable of modulating a variety of developmental and inflammatory processes (5, 12, 13). In the lung, RA is believed to play a pivotal role in the formation of alveolar structures (4). At the time of alveolarization, lung fibroblasts contain large quantities of RA (24), and retinoid receptors are expressed within the lung at high levels (25). Exogenous administration of retinoids can accelerate alveolarization in neonatal animals (4).

Destruction of alveoli is the characteristic and defining feature of pulmonary emphysema (2). The observation that individuals with a homozygous deficiency of 1 protease inhibitor are at increased risk for the development of emphysema led to the "protease anti-protease" hypothesis of emphysema (2). This model, now greatly expanded, suggests that the unopposed action of proteases, oxidants and possibly other destructive moieties cause tissue destruction in emphysema (9, 10). Cigarette smoke may lead to emphysema by initiating inflammatory responses thus inducing the production of proteases and other destructive factors (26). In addition, cigarette smoke may deplete anti-protease and anti-oxidant defenses (27, 28).

The specific proteolytic enzymes responsible for the development of emphysema are not entirely defined. Animal models with exogenous protease administration suggests that enzymes with elastolytic activity are required (7). NE and other serine proteases that can be inhibited by 1 protease inhibitor, have been suggested to play an important role (2). Studies with genetically manipulated mice, however, have established a role for matrix metalloproteinases, in particular MMP-12, an elastolytic enzyme derived from macrophages (29). Evidence has also developed that collagenase may contribute to the development of emphysema (30). Roles for these various proteases are not exclusive. Not only can they synergize with regard to degradation of extracellular matrix, but various proteases can augment each other's activities by several mechanisms. In this context, some proteases are released as latent precursors that can be activated through the proteolytic action of other proteases. In addition, proteolytic cleavage of anti-proteases can remove inhibitory mechanisms. The current study is consistent with a collaborative interaction among proteases leading to tissue destruction (11, 31).

The "classical" concept that once developedthat emphysema is irreversiblehas been challenged by recent animal studies (3). Using the well-described model of porcine pancreatic elastase-induced emphysema in the rat, Massaro and Massaro demonstrated that the administration of retinoids can induce neoalveolarization with partial reversal of the emphysema. A recapitulation of the developmental processes regulated by retinoids has been suggested as a mechanism for this effect. The current study suggests that retinoids may be able to alter the development of emphysema by altering tissue destruction as well. As noted above, transgenic mice that over-express collagenase developed emphysema, osteoarthritis and delay of epidermal wound healing (33). In this context, macrophages, as well as mesenchymal cell production of MMP-1, are known to contribute to tissue destruction in a variety of settings, including emphysema, osteoarthritis, osteoporosis, ulcers and tumor invasion (33, 35). It seems likely that normal lung structure maintenance requires ongoing tissue deposition and removal. That retinoids may be able to modulate this process, both by regulating alveolarization and by regulating tissue destruction, is consistent with coordinated regulation of these processes.

The current system utilized the in vitro model system of fibroblasts cultured in three-dimensional collagen gels. This culture system has been suggested to more closely model in vivo tissues than routine culture on plastic dishes (39). Fibroblasts cultured in the three-dimensional matrix align themselves on the collagen fibers and have altered synthetic and functional activities. The ability of fibroblasts to contract extracellular matrix has been used as a model for both tissue repair and the development of contracted fibrotic scars (39, 40). Cytomix is able to inhibit fibroblast-mediated collagen gel contraction (41). This process is mediated by induction of fibroblast production of PGE₂ and nitric oxide, which in turn inhibit gel contraction in a paracrine manner. In the absence of other stimuli, however, cytomix alone does not result in degradation of extracellular matrix.

When cultured in three-dimensional gels, fibroblasts are also able to secrete metalloproteinases. The production of matrix-degrading enzymes, including MMPs by fibroblasts, is under cytokine control. In the current study, IL-1 and TNF- administered individually and in combination with IFN- were able to augment the production of several matrix-degrading enzymes. Degradation of the extracellular collagenous matrix, however, was minimal, presumably because the matrix metalloproteinases were largely secreted in their latent form. In the presence of NE, however, conversion of the MMPs to active forms was documented although it is possible that the effect of elastase is indirect. Concurrently, degradation of the extracellular matrix took place. These results are consistent with a collaborative interaction among proteases leading to tissue destruction (11, 31).

RA was able to block the degradation of extracellular collagen caused by fibroblasts exposed to cytokines in the presence of NE. This inhibitory activity of RA was associated with a decrease in the activation of MMPs. These results suggest that RA, in addition to inducing anabolic effects, can also modulate the proteolytic activities likely present in an inflammatory milieu. This inhibitory activity of RA was not due to a direct effect on NE. Inhibition of the production of MMP-1 and/or -3 may account for the action of RA. The initial decline in elastase activity observed in fibroblast cultures and in cell-free gels is consistent with auto-degradation of the elastase. The subsequent increase in elastase activity in fibroblast cultures is consistent with production of amidolytic activity not directly identified in the current study. RA appeared to decrease the production of this activity as well. As RA is able to affect multiple MMPs, however, it seems likely that retinoids may be acting at several steps in a proteolytic cascade, which ultimately leads to degradation of collagen.

The model system used in the current assay involves embedding fibroblasts in fibers composed of reconstituted type I collagen in a native confirmation. In vivo extracellular matrices are more complex and contain a variety of matrix components in addition to collagen. In addition to modulating the degradation of collagen, the current study suggests that retinoids, by altering proteolytic activities released by fibroblasts, are likely able also to modulate the degradation and turnover of connective tissue matrix components. As tissue development likely involves turnover and remodeling of extracellular matrix, the ability of retinoids to modulate matrix degradation may be closely connected to their effects on tissue development and repair.

The observation that retinoids have the potential for reversing emphysema raises exciting new possibilities for the treatment of a devastating disease previously thought to be irreversible. The mechanisms by which retinoids may alter emphysematous tissues remain to be fully defined. The current study suggests that in addition to stimulating anabolic effects, retinoids may modulate proteolytic processes including those thought to contribute to tissue destruction in emphysema.