Cell-specific induction of distinct oncogenes of the Jun family is responsible for differential regulation of collagenase gene expression by transforming growth factor-beta in fibroblasts and keratinocytes.

Transforming growth factor-beta (TGF-beta) plays a major role in regulating connective tissue deposition by controlling both extracellular matrix production and degradation. In this study, we show that TGF-beta transcriptionally represses both basal and tumor necrosis factor-alpha-induced collagenase (matrix metalloprotease-1) gene expression in dermal fibroblasts in culture, whereas it activates its expression in epidermal keratinocytes. We demonstrate that this differential effect of TGF-beta on collagenase gene expression is due to a cell type-specific induction of distinct oncogenes of the Jun family, which participate in the formation of AP-1 complexes with different trans-activating properties. Specifically, our data indicate that the inhibitory effect of TGF-beta in fibroblasts is likely to be mediated by jun-B, based on the following observations: (a) TGF-beta induces high levels of jun-B expression and (b) over-expression of jun-B mimics TGF-beta effect in inhibiting basal collagenase promoter activity and preventing tumor necrosis factor-alpha-induced trans-activation of the collagenase promoter. In contrast, TGF-beta induction of collagenase gene expression in keratinocytes is preceded by transient elevation of c-jun proto-oncogene expression. Over-expression of c-jun leads to trans-activation of the collagenase promoter in both cell types, suggesting that c-jun is a ubiquitous inducer of collagenase gene expression. Transfection of keratinocytes with an antisense c-jun construct together with a collagenase promoter/reporter gene construct inhibits basal and TGF-beta-induced up-regulation of the collagenase promoter activity, implying that c-jun mediates TGF-beta effect in this cell type. Collectively, our data suggest differential signaling pathways for TGF-beta in dermal fibroblasts and epidermal keratinocytes, leading to cell type-specific induction of two AP-1 components with opposite transcriptional activities.

Transforming growth factor-␤ (TGF-␤) plays a major role in regulating connective tissue deposition by controlling both extracellular matrix production and degradation. In this study, we show that TGF-␤ transcriptionally represses both basal and tumor necrosis factor-␣-induced collagenase (matrix metalloprotease-1) gene expression in dermal fibroblasts in culture, whereas it activates its expression in epidermal keratinocytes. We demonstrate that this differential effect of TGF-␤ on collagenase gene expression is due to a cell type-specific induction of distinct oncogenes of the Jun family, which participate in the formation of AP- 1

complexes with different trans-activating properties. Specifically, our data indicate that the inhibitory effect of TGF-␤ in fibroblasts is likely to be mediated by jun-B, based on the following observations: (a) TGF-␤ induces high levels of jun-B expression and (b) over-expression of jun-B mimics TGF-␤ effect in inhibiting basal collagenase promoter activity and preventing tumor necrosis factor-␣induced trans-activation of the collagenase promoter. In contrast, TGF-␤ induction of collagenase gene expression in keratinocytes is preceded by transient elevation of c-jun proto-oncogene expression.
Over-expression of c-jun leads to trans-activation of the collagenase promoter in both cell types, suggesting that c-jun is a ubiquitous inducer of collagenase gene expression. Transfection of keratinocytes with an antisense c-jun construct together with a collagenase promoter/reporter gene construct inhibits basal and TGF-␤-induced up-regulation of the collagenase promoter activity, implying that c-jun mediates TGF-␤ effect in this cell type. Collectively, our data suggest differential signaling pathways for TGF-␤ in dermal fibroblasts and epidermal keratinocytes, leading to cell type-specific induction of two AP-1 components with opposite transcriptional activities.
Matrix metalloproteases comprise a family of proteolytic enzymes involved in the degradation of the extracellular matrix of connective tissue (for reviews see Refs. 1 and 2). These enzymes play a critical role in a number of physiological and pathological processes involving connective tissue remodeling and/or destruction, as exemplified by embryonic development, wound repair, tumor metastasis, and rheumatoid arthritis. Breakdown of the fibrillar collagen network is initiated by interstitial collagenase (matrix metalloprotease-1), whereas the other components of the matrix are degraded primarily by stromelysins and gelatinases.
The expression of matrix metalloproteases by connective tissue cells is modulated by a variety of cytokines and growth factors (2). In particular, interleukin-1 and tumor necrosis factor-␣ (TNF-␣) 1 are potent activators of fibroblast collagenase gene expression, and their effect is mediated by c-Jun, the product of the c-jun proto-oncogene, which participates in the formation of the transcription factor AP-1 (1,2). In contrast, transforming growth factor-␤ (TGF-␤) has been shown to inhibit fibroblast collagenase gene expression through Jun-B-dependent mechanisms (3), whereas it inhibits the expression of transin, the rat homologue of stromelysin, through Fos-mediated mechanisms (4).
Recent in vivo observations have revealed that during cutaneous wound healing, the expression of collagenase is very low in the dermis, whereas it is markedly elevated in basal keratinocytes at the wound edges (5). In this context, the close topographic proximity of fibroblasts and keratinocytes led us to investigate in vitro the signals that would be responsible for the differential, cell type-specific expression of collagenase during wound healing. We report that TGF-␤, a growth factor with essential wound healing promoting activities (6 -9), is a potent inhibitor of collagenase gene expression in fibroblasts, whereas it strongly up-regulates collagenase expression in keratinocytes. We demonstrate that cell-specific induction of different oncogenes of the Jun family, with opposite trans-activating properties, is responsible for the differential regulation of collagenase gene expression by TGF-␤ in fibroblasts and keratinocytes.

MATERIALS AND METHODS
Cell Cultures-Human dermal fibroblast cultures, established by explanting tissue specimens obtained from neonatal foreskins, were utilized in passages 3-8. The cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum, 2 mM glutamine, and antibiotics (50 g/ml streptomycin, 200 units/ml penicillin-G, 0.25 g/ml Fungizone ). One hour prior to the addition of growth factors, the confluent fibroblast cultures were rinsed with DMEM and placed in DMEM containing 1% fetal calf serum.
Human epidermal keratinocytes obtained by explanting foreskin specimens, were grown in serum-free, low calcium (0.15 mM), keratinocyte growth medium supplemented with epidermal growth factor, hydrocortisone, insulin, and bovine pituitary extract (KGM, Clonetics Corp., San Diego, CA), and utilized in passage 1 to avoid differentiation inherent to subculturing of these cells. One hour prior to the experiments, the confluent keratinocyte cultures were placed in fresh KGM.
Northern Analyses-At the end of incubation with growth factors, cell cultures were subjected to isolation of total RNA as described previously (10). RNA was fractionated in 0.8% agarose gels containing formaldehyde and analyzed by Northern hybridization with 32 P-labeled cDNA probes (11). The [ 32 P]cDNA-mRNA hybrids were visualized by autoradiography, and the steady-state levels of mRNA were quantitated by scanning densitometry using a He-Ne laser scanner at 633 nm (LKB Produkter, Bromma, Sweden).
cDNAs and Plasmid Constructs-The following cDNAs were used for Northern hybridizations to detect specific mRNA transcripts: a 2.0kilobase pair human collagenase cDNA (12), a gift from Dr. Gregory I. Goldberg (Washington University School of Medicine, St. Louis, MO) and a rat glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA was used in control hybridizations to normalize for differences in the loading and transfer of RNA (13). For c-jun, we used a full-length human cDNA in pRSVe expression vector (14); and for jun-B, a fulllength cDNA in pRSVe expression vector (14), both kindly provided by Dr. Michael Karin (UCSD, La Jolla, CA).
To study the transcriptional regulation of collagenase gene expression, the following plasmid construct was used in transient transfection experiments: pCLCAT3, which contains ϳ3.8 kilobase pairs of 5Ј-flanking DNA of human collagenase gene linked to the CAT reporter gene (15), kindly provided by Dr. Steven M. Frisch (La Jolla Cancer Research Foundation, La Jolla, CA). The oncogene expression vectors described above were used in co-transfection experiments. Empty pRSVe was used as filling plasmid in order to transfect the same amount of DNA in every cell plate.
To prepare an antisense c-jun construct, a fragment spanning the region ϩ451 to ϩ617 of the c-jun gene was amplified by polymerase chain reaction (PCR). The PCR amplimers were cloned into a PCRII plasmid vector (InVitrogen Corp., San Diego, CA), and the clones containing the insert were sequenced to ensure the fidelity of the PCR amplification. The PCR products were then inserted in an antisense orientation as XhoI-HindIII fragments into the pRSVe expression vector in order to generate the construct pRSV-ASc-jun.
Transient Transfections of Cultured Cells-Human neonatal foreskin fibroblasts in late logarithmic growth phase were transfected with 1-20 g of plasmid constructs co-transfected with a RSV-promoter/␤-galactosidase construct to allow determination of the transfection efficiency (11). The transfections were performed with the calcium-phosphate/ DNA co-precipitation method (16) followed by a 1.5-min glycerol (15%) shock. Following the glycerol shock, the cells were placed in medium supplemented with 1% fetal calf serum prior to the addition of growth factors. Basal keratinocytes grown in KGM were transfected with a liposome-based method (DOTAP, Boehringer Mannheim), according to the manufacturer's protocol. Five hours after transfections, the medium was replaced with fresh KGM, and growth factors were added 2 h later for a 40-h incubation. At the end of incubation, the cells were harvested and lysed by thee cycles of freeze-thawing in 100 l of 0.25 M Tris-HCl, pH 7.8. Aliquots corresponding to identical ␤-galactosidase activity were used for each CAT assay with [ 14 C]chloramphenicol as substrate (17).
Stable Transfections-To generate stably transfected cultures of NIH3T3 fibroblasts, pRSV-ASc-jun was co-transfected with pRc/CMV (InVitrogen), which expresses the neomycin resistance gene, in a 10:1 ratio. Four days after transfections, G418 (800 g/ml, Life Technologies, Inc.) was added to the culture medium for selection of transfected cells. Fifteen days later, stably transfected cultures of pRSV-ASc-jun NIH3T3 were established by pooling the G418-resistant clones. Control cultures were transfected with pRc/CMV and empty pRSVe to generate pRSVeNIH3T3 cultures.

RESULTS
Cell Type-specific Effect of TGF-␤ on Collagenase Gene Expression-In the first set of experiments, human adult skin fibroblasts were incubated for 24 h without or with TGF-␤ or TNF-␣, both at a concentration of 10 ng/ml. At the end of incubation, RNA was extracted and analyzed for collagenase gene expression by Northern hybridization. Visual observation of the autoradiograms indicated that TGF-␤ markedly downregulated both basal and TNF-␣-induced elevation of collagenase mRNA steady-state levels in parallel fibroblast cultures (Fig. 1A). Quantitation of the autoradiograms by scanning densitometry and correction against GAPDH mRNA levels in the corresponding RNA preparations indicated that the expression of collagenase in TGF-␤-treated cultures was reduced by ϳ80% in comparison with control fibroblast cultures. TNF-␣ enhanced collagenase mRNA levels by ϳ6-fold, but in cultures treated concomitantly with TNF-␣ and TGF-␤, the levels were reduced by ϳ75% in comparison with cultures treated with TNF-␣ alone (Fig. 1B).
In the second set of experiments, confluent keratinocyte cultures were incubated for 24 h without or with TGF-␤ and/or TNF-␣ at concentrations of either 1 or 10 ng/ml. Contrasting with its inhibitory effect observed in dermal fibroblasts, TGF-␤ was found to be a potent enhancer of keratinocyte collagenase gene expression (Fig. 2, second and third lanes). On the other hand, TNF-␣, which is a potent enhancer of collagenase gene expression in fibroblasts (see Fig. 1), had little effect on the basal expression of collagenase in keratinocytes (Fig. 2, fifth FIG. 1. Effect of TGF-␤ on collagenase gene expression in dermal fibroblasts. Confluent fibroblast cultures were incubated in medium containing 1% fetal calf serum without (Ϫ) or with (ϩ) TNF-␣ (10 ng/ml), in the absence (Ϫ) or the presence (ϩ) of TGF-␤ (10 ng/ml). After 24 h, total RNA was extracted and analyzed by Northern hybridization with a collagenase-specific cDNA; a GAPDH cDNA was used as a control. A, autoradiograms. B, densitometric analysis after correction for GAPDH mRNA levels. and sixth lanes) and a minimal effect on the induction exerted by TGF-␤ (Fig. 2, fourth lane). Quantitation of the autoradiograms after normalization of collagenase mRNA levels against those of GAPDH showed a 4.5-5-fold elevation of collagenase mRNA steady-state levels upon TGF-␤ stimulation, whereas TNF-␣ did not stimulate collagenase expression (Fig. 2B).
Transient cell transfections with a collagenase promoter/ CAT reporter gene construct were performed to examine whether TGF-␤ regulates collagenase mRNA steady-state levels through modulation of transcription at the promoter level. Human neonatal fibroblasts were transfected with the collagenase promoter/CAT construct pCLCAT3 and treated with TGF-␤ or TNF-␣ both at 10 ng/ml concentration. Assay of CAT activity after 40 h of incubation indicated that TGF-␤ reduced the promoter activity by 40 -60%, as compared with that of control cultures (Fig. 3). Also TGF-␤ counteracted TNF-␣-induced elevation of the collagenase promoter activity.
In another set of experiments, confluent keratinocyte cul-tures were transfected with the same collagenase promoter/ CAT reporter gene construct and treated with various doses of TGF-␤ (0.1, 1, and 10 ng/ml) for 40 h. Assay of CAT activity revealed a dose-dependent elevation of the promoter activity ( Fig. 4) with a maximal stimulation (ϳ4-fold) observed with 10 ng/ml of TGF-␤. Thus, the enhancement of collagenase gene expression in keratinocytes by TGF-␤ and its inhibition in fibroblasts, as detected at the mRNA level, was mediated, at least in part, by cell was type-specific modulation of the promoter activity.
Cell Type-specific Induction of jun-B and c-jun Proto-oncogene Expression by TGF-␤-It has been previously reported that the expression of c-jun is induced by cytokines such as TNF-␣ and interleukin-1 (18,19) and represents a fundamental intermediate step in cytokine induction of other cellular genes, including those encoding collagenase or stromelysin (reviewed in Refs. 1 and 2).
As shown in Fig. 5, TNF-␣ rapidly (within 1 h of incubation) enhanced the expression of c-jun in dermal fibroblasts (lane 4), and the induction persisted even after 6 h of incubation (lane 5). Also, the steady-state levels of jun-B mRNAs were elevated but to a lesser extent than those for c-jun (lane 4). TGF-␤ alone did not affect either the basal expression of c-jun (lane 7) or the induction of c-jun by TNF-␣ (lanes 9 and 10 versus lanes 4 and  5). In contrast, TGF-␤, either alone or in combination with TNF-␣, strongly elevated the expression of jun-B (lanes 7 and 9), which persisted for at least up to 6 h following the initiation of stimulation (lanes 8 and 10). It appears, therefore, that in fibroblasts, TGF-␤ counteracts TNF-␣-induced collagenase gene expression, but this effect is not due to repression of c-jun transcription. At the same time, TNF-␣ did not alter the expression of jun-B induced by TGF-␤. Accordingly, TGF-␤ stimulation led to a dramatic reduction of the c-jun/jun-B mRNA ratio in fibroblasts, from 1.7 to 0.1 and 0.2 after 1 and 6 h of stimulation, respectively (Fig. 6). In contrast, TNF-␣ elevated the c-jun/jun-B mRNA ratio due to a substantially higher stimulation of c-jun versus jun-B. When TGF-␤ was added simultaneously with TNF-␣, the c-jun/jun-B mRNA ratio remained at levels that were ϳ60 -70% lower than observed with TNF-␣ alone (Fig. 6). Therefore, there is a strong correlation between the c-jun/jun-B mRNA ratio and the level of collagenase expres- sion, suggesting that the modification of the ratio of c-jun/jun-B mRNA is likely to reflect a reduction of the relative amounts of c-Jun versus those of Jun-B when TGF-␤ is present, leading to reduced collagenase gene transcription.
Subsequently, we tested the pattern of expression of oncogenes of the Jun family in the presence of TGF-␤ in epidermal keratinocytes. Contrasting with its lack of effect in fibroblasts, TGF-␤ induced high levels of c-jun expression in keratinocytes, with a maximum enhancement at 1 h following growth factor stimulation (Fig. 7). The high levels of c-jun mRNAs persisted at least 6 h post-stimulation and preceded the induction of collagenase expression. In contrast, the basal expression of jun-B was very low and, although expression was enhanced after 1 h of incubation with TGF-␤, it remained well below the level of expression of c-jun. In fact, using 32 P-labeled cDNA probes with similar specific activities (ϳ2 ϫ 10 7 cpm/g), a 48-h exposure of the autoradiogram was necessary to detect the low jun-B expression, whereas a 14-h exposure was sufficient to easily detect c-jun expression (Fig. 7). Therefore, in keratinocytes, after TGF-␤ stimulation, the ratio c-jun/jun-B is largely in favor of c-jun, ϳ25-fold more c-jun mRNA than jun-B at 1 h after addition of TGF-␤, as estimated by scanning densitometry of the autoradiograms, correction for GAPDH mRNA levels in the RNA preparations, and differences in the exposure times of the autoradiograms. These data contrast the reverse situation observed in dermal fibroblasts in which jun-B expression is boosted by TGF-␤ treatment (see above, Figs. 5 and 6). It should be noted that the extent of stimulation of collagenase gene expression as detected at the mRNA level (ϳ15-fold after 6 h) was more pronounced than the induction of promoter activity observed in transient cell transfection experiments (see Fig. 4). It is conceivable that TGF-␤, in addition to activating the collagenase promoter, may also increase collagenase gene expression by post-transcriptional mechanisms such as stabi- lization of the corresponding mRNA, as described previously for phorbol esters or epidermal growth factor (20,21).
c-jun Trans-activates the Collagenase Promoter in Both Fibroblasts and Keratinocytes-In view of the data described above, it appears that in both cell types studied, enhanced collagenase gene expression correlates with a preceding elevation of c-jun mRNA levels due to TNF-␣ stimulation in fibroblasts and to TGF-␤ stimulation in keratinocytes, respectively. We therefore tested whether over-expression of c-jun could trans-activate the collagenase promoter in both cell types. For this purpose, pCLCAT3 was co-transfected with either an RSV/ c-jun expression vector or an empty RSVe vector as a control. Over-expression of c-jun in both fibroblasts and keratinocytes led to trans-activation of the collagenase promoter, as reflected by measurement of CAT activity in the different cell extracts. The extent of stimulation of the collagenase promoter was similar in both cell types: 5.3 Ϯ 0.7-fold (n ϭ 8) in fibroblasts versus 4.9 Ϯ 0.9-fold (n ϭ 6) in keratinocytes. In contrast, over-expression of jun-B resulted in reduced collagenase promoter activity in both cell types (not shown), attesting to the specificity of the expression vectors, as expected from previous observations by us and others (3,14).
Characterization of the Antisense Activity of pRSV-ASc-jun-Control (pRSVeNIH3T3) and pRSV-ASc-junNIH3T3 fibroblast cultures (see "Materials and Methods") were grown to confluency. Three hours prior to the addition of growth factors, the confluent fibroblast cultures were rinsed with DMEM and placed in DMEM containing 1% fetal calf serum. The cultures were then treated for 6 h with either TNF-␣ or TGF-␤, both at a concentration of 10 ng/ml, prior to Northern analysis. As shown in Fig. 8, significant expression of both c-jun and jun-B was noted in unstimulated pRSVeNIH3T3 cultures (lane 1). TNF-␣ strongly elevated c-jun expression, and jun-B to a lesser extent (lane 2). TGF-␤ slightly elevated jun-B expression but had no effect on c-jun mRNA levels (lane 3). pRSV-ASc-junNIH3T3 cultures showed very little basal expression of c-jun (lanes 4), about 7% of that observed in control cultures, as measured by scanning densitometry and after correction for GAPDH mRNA levels in the same RNA preparations, whereas the basal levels of jun-B mRNA were similar to those of control cultures. TNF-␣ did not elevate c-jun mRNAs in pRSV-AScjun-transfected cultures, but the induction of jun-B by both TNF-␣ and TGF-␤ was similar to that observed in control cultures (Fig. 8, fifth and sixth lanes versus first and second  lanes), indicating that the cells are still responsive to growth factors. Taken together, these results demonstrate that pRSV-ASc-jun prevents both basal and TNF-␣-induced c-jun expression, whereas the construct has no effect on the expression and regulation of jun-B expression. Interestingly, no expression of collagenase was detected in pRSV-ASc-junNIH3T3 cultures, even after TNF-␣ stimulation, indicating that c-jun is fundamental for both basal and cytokine-induced collagenase gene expression. Also, these data demonstrate that the c-jun antisense construct can block c-Jun-mediated transcription.
Antisense c-jun Prevents TGF-␤-induced Elevation of Collagenase Promoter Activity in Epidermal Keratinocytes-We have shown that TGF-␤ stimulates c-jun expression prior to inducing collagenase gene expression in keratinocytes and that overexpression of c-jun by the mean of transfected expression vectors leads to trans-activation of the collagenase promoter in keratinocytes (see above). We therefore wished to examine whether the stimulatory effect of TGF-␤ was directly mediated by c-jun. For this purpose, pRSV-ASc-jun was co-transfected with pCLCAT3 into basal keratinocytes, prior to stimulation with TGF-␤. As shown in Fig. 9, the basal activity of the collagenase promoter was markedly reduced by the antisense c-jun (by about 70%), which implies a role for c-jun in the basal expression of collagenase in keratinocytes. Furthermore, antisense c-jun prevented stimulation of collagenase promoter activity by TGF-␤ (Fig. 9), suggesting that c-jun induction is an essential step for collagenase gene activation by TGF-␤ in keratinocytes.

DISCUSSION
Several studies have shown differences in the pattern of expression and response to extracellular stimuli between c-jun and jun-B in a variety of experimental systems (3,(22)(23)(24). Furthermore, considerable differences in their trans-activation and transforming activities have been reported (3,14,25,26). For example, whereas c-Jun is an efficient activator of the c-jun and collagenase promoters that contain a single TRE, Jun-B is not. In addition, Jun-B counteracts activation of these promoters by c-Jun. However, like c-Jun, Jun-B is a potent activator of FIG. 8. Characterization of pRSV-ASc-jun in stable transfection experiments. In order to verify the activity of the pRSV-ASc-jun antisense construct, pRSVeNIH3T3 and pRSV-ASc-junNIH3T3 fibroblast cultures were grown to confluency. Three hours prior to the addition of growth factors, the confluent fibroblast cultures were rinsed with DMEM and placed in DMEM containing 1% fetal calf serum. The cultures were then treated for 6 h with either TNF-␣ or TGF-␤, each at a concentration of 10 ng/ml. Total RNA was extracted and analyzed by Northern hybridization with 32 P-labeled cDNA probes for collagenase, c-jun and jun-B. A GAPDH cDNA was used as a control. constructs containing multimeric TREs. These differences in the biological actions of the two Jun proteins are due to intrinsic differences in their activation and DNA-binding domains (14,27), allowing fine tuning of the regulation of TRE-driven genes.
The regulatory role of the different Jun proteins is further emphasized by the fact that the corresponding genes are not coordinately expressed in different tissues, as shown in adult mice and during embryogenesis (24,28,29), suggesting a tissue-specific transcriptional regulation of TRE-driven target genes. In this context, it has been recently shown that jun-B and c-jun are selectively up-regulated and functionally implicated in the development of fibrosarcoma (30). It is therefore conceivable that in different inducible systems, increased specificity and precise regulation of TRE-driven transcriptional activation is achieved by interactions between positive and negative transcription factors that belong to the same gene family.
In this study, we have provided the following evidence for a mediation of the inhibitory effect of TGF-␤ on fibroblast collagenase gene expression by Jun-B: (a) TGF-␤ induces high levels of jun-B expression; (b) jun-B expression vectors mimic TGF-␤ action in our experimental system by inhibiting basal collagenase promoter activity and by exerting an antagonistic effect on TNF-␣-induced collagenase gene expression. On the other hand, we have demonstrated that TGF-␤ stimulates collagenase gene expression in keratinocytes and that this effect is mediated by c-jun, as follows: (a) TGF-␤ induces high levels of c-jun expression; (b) c-jun expression vectors trans-activate the collagenase promoter; (c) antisense c-jun expression vectors prevent TGF-␤ activation of collagenase gene expression. A schematic diagram depicting the differential effects of TGF-␤ on collagenase gene expression in fibroblasts and keratinocytes is shown in Fig. 10. These results are the first evidence of differential induction of two oncogenes with opposite trans-activation properties upon stimulation by a single growth factor, TGF-␤, in two different cell types within the same tissue, the skin. TGF-␤ has been shown to reduce collagenase gene expression and activity in cultured fibroblasts. This inhibition results from two distinct mechanisms: (a) TGF-␤ reduces the expression of the collagenase gene and (b) the expression of tissue inhibitor of metalloproteases is elevated by TGF-␤ (31). Our data indicate that the concept of TGF-␤ as a potent inhibitor of matrix remodeling is cell type-specific because this growth factor is a potent activator of collagenase gene expression in epidermal keratinocytes in culture. In that respect, TGF-␤ has been shown previously to up-regulate both 92-and 72-kDa gelatinase activity and gene expression in both fibroblasts and keratinocytes (32,33).
Recently, it has been demonstrated using in situ hybridization in ulcerative skin lesions such as pyogenic granulomas that collagenase is expressed near the advancing edge of the ulceration, within the disrupted epidermis adjacent to an ulcer (5). By contrast, no hybridization signal was detected within the dermis or normal, intact epidermis. Therefore, basal keratinocytes seem to be the primary source of collagenase during wound healing, suggesting that keratinocytes play an essential role in tissue remodeling. It has been suggested that the signals that activate collagenase in keratinocytes are provided by the dermal extracellular matrix. In agreement with this hypothesis is the fact that keratinocytes grown on type I collagen exhibit enhanced collagenase production (34). By contrast, activation of fibroblast collagenase expression may be mediated by soluble factors such as interleukin-1 or TNF-␣, rather than by the extracellular matrix. Skin injury is accompanied by release of interleukin-1, which in turn may activate fibroblasts but not keratinocytes to produce collagenase (34). Our data provide an alternative model for the cell-specific activation of collagenase gene expression during wound healing in which TGF-␤, which is present in abundant amounts in the healing wound bed, could simultaneously turn off the expression of collagenase in fibroblasts while activating that of keratinocytes directly in contact with the dermis. We hypothesize that collagenase-secreting keratinocytes, possibly in response to TGF-␤, may be able to migrate to close the wound. This hypothesis is supported indirectly by a previous study indicating that TGF-␤ stimulates the outgrowth of epidermal cells from skin explant cultures (35).
In conclusion, this study has provided the first evidence for cell type-specific, differential induction of two transcription factors of the same family with antagonistic trans-activation properties, leading to opposite regulation of collagenase gene expression in fibroblasts and keratinocytes by TGF-␤. FIG. 10. Schematic representation of the putative mechanisms for differential regulation of collagenase gene expression by TGF-␤ in fibroblasts and keratinocytes. Left, TGF-␤, through interactions with specific receptors on the keratinocyte surface, induces high levels of expression of c-jun, which, after dimerization, is responsible for the trans-activation of the collagenase promoter. Right, TGF-␤ induces high levels of expression of jun-B, and its product participates in the formation of AP-1 complexes with inhibitory activity on collagenase gene expression. Transduction mechanisms (1; to be identified) induce transcription of c-jun in keratinocytes and of jun-B in fibroblasts (2). Jun products are translated in the cytoplasm (3) and translocate into the nucleus (4) to form AP-1 complexes that modulate collagenase gene transcription (5). TGF-␤-induced c-jun expression in keratinocytes leads to increased collagenase gene transcription (ϩ) and increased collagenase production (ÀÀ). In fibroblasts, increased jun-B expression represses collagenase transcription (Ϫ) with subsequent decrease in collagenase production (ÄÄ).