Transcription from the Stromelysin Promoter Is Induced by Interleukin-1 and Repressed by Dexamethasone *

The stromelysin gene encodes a potent tissue-degrading proteinase whose activity is important in tissueremodeling processes such as wound healing, the inflammatory reaction, rheumatoid arthritis, tumor invasion, and possibly embryonic development. In light of the ability of interleukin-1 to amplify, and ability of glucocorticoids to attenuate the inflammatory response, we tested interleukin1 and dexamethasone for regulatory effects on stromelysin gene expression. We report that interleukin1 induces the stromelysin gene, and dexamethasone diminishes the level of induction by interleukin-1, epidermal growth factor, phorbol ester, and cAMP elevation (elicited by cholera toxin). Similar responses are conferred upon a chloramphenicol acetyltransferase coding sequence by a 700-base pair stromelysin 5”flanking fragment, implying transcription regulation by sequence elements in this region.


Transcription from the Stromelysin Promoter Is Induced by
Interleukin-1 and Repressed by Dexamethasone* (Received for publication, July 16, 1987)  The stromelysin gene encodes a potent tissue-degrading proteinase whose activity is important in tissueremodeling processes such as wound healing, the inflammatory reaction, rheumatoid arthritis, tumor invasion, and possibly embryonic development. In light of the ability of interleukin-1 to amplify, and ability of glucocorticoids to attenuate the inflammatory response, we tested interleukin-1 and dexamethasone for regulatory effects on stromelysin gene expression. We report that interleukin-1 induces the stromelysin gene, and dexamethasone diminishes the level of induction by interleukin-1, epidermal growth factor, phorbol ester, and cAMP elevation (elicited by cholera toxin). Similar responses are conferred upon a chloramphenicol acetyltransferase coding sequence by a 700-base pair stromelysin 5"flanking fragment, implying transcription regulation by sequence elements in this region.

Steven M. FrischS and H. Earl Ruleyg
Extracellular matrix degradation is a key element in such physiologic processes as wound healing (l), inflammation (a), differentiation, and development (3,4) and in the pathology of rheumatoid arthritis (5) and tumor invasion (6). Degradation of extracellular matrix/connective tissue absolutely requires the cooperation of multiple proteinases. Collagenase alone is capable only of introducing a single cleavage in the collagen triple helix, leaving further collagen degradation and degradation of other matrix components to be dealt with by different proteinases.
The control of genes encoding the relevant proteinases is only beginning to be elucidated. The large number of reports of elevated extracellular proteinase activity in tumor cells (6) suggests that activated oncogenes, possibly by influencing normal growth factor or lymphokine signal transduction pathways, play a regulatory role with regard to these proteinase genes. Recently, we reported the molecular cloning of a cDNA encoding rabbit stromelysin, a phorbol ester-induced, secreted metalloproteinase which has potent, and fairly promiscuous, matrix-degrading activity (7); Matrisian et al. (8) reported the cloning of transin, an epidermal growth factor-induced gene of unknown function. We (7) and they (9) subsequently found transin and stromelysin to be very homologous or identical except for species differences. The rat transin gene was found also to be inducible by transformation mediated by polyoma, Rous sarcoma virus, or activated Ha-ras (lo), consistent with the SV40 transformation inducibility of human stromelysin observed previously (11). As in the rabbit synovial fibroblast system, the transin gene was inducible by cytochalasin B (lo), but the tumor promoter, TPA,' induced transin only in the presence of serum, while TPA induced the rabbit stromelysin gene only in the absence of serum (12).
While these second messenger modulators are useful for revealing the signal transduction pathways that have the potential of inducing the stromelysin gene, it is also important to identify the possible physiologic inducers and repressors of this highly tissue-destructive enzyme. Interleukin-1 (11-1) is an attractive candidate, considering its central role in promoting the inflammatory response (13) and, in particular, its stimulation of connective tissue-degrading activity (e.g. Ref. 14) partly attibutable to collagenase gene induction (15,16).
We also chose to examine the effects of the synthetic glucocorticoid dexamethasone on stromelysin gene expression in light of the potent anti-inflammatory (17) and anti-arthritic (18) properties of the glucocorticoids and, in particular, their repression of collagenase gene expression (19). Also, we reasoned that, because fibronectin is readily digested by stromelysin (20), repression of stromelysin synthesis by glucocorticoids would be consistent with (or possibly responsible for) the phenomenon that dexamethasone causes restoration of normal levels of fibronectin accumulation in certain transformed cell lines (21).
Induction of stromelysin gene expression by interleukin-1 and repression by dexamethasone are reported in this paper. DNA elements located in the region -700 to -20 were found to be sufficient to confer both of these regulatory responses upon a heterologous coding sequence, chloramphenicol acetyltransferase, as well as the regulatory responses reported previously (induction by EGF, cAMP elevation, and TPA).
Sequence analysis of the stromelysin 5'-flanking region revealed some interesting homologies with the corresponding regions of other, similarly regulated genes.
Filter Blot Hybridization-Total RNAs were prepared, Northern blotted, and hybridized as described previously (22); oligonucleotide primed restriction fragments were isolated from low-melt agarose gels and labeled by the method of Feinberg and Vogelstein (23). A rabbit genomic DNA library in X Charon 4A was screened by standard methods (24). Southern blots were performed using the alkaline transfer method (25).
Other Nucleic Acid Methods-DNA sequencing was performed by double stranded plasmid sequencing (26) of exonuclease 111-mung bean nuclease deletion mutants (27). mids pSLCATMPlO and pSLCATMP16 (having -700 bp and -6.1 The stromelysin promoter chloramphenicol acetyltransferase plaskbp of 5"flanking region, respectively) were constructed by digesting a bluescript plasmid containing a 14.3-kbp EcoRI-EcoRI insert of stromelysin DNA with Sac1 and either ScaI (MP10) or SmaI (MP16). Fragments were made blunt-ended with T4 DNA polymerase, HindIII linkers were ligated with T 4 DNA ligase, and, following HindIII digestion, fragments were isolated on low-melt agarose gels; fragments were ligated in the gel with HindIII-cut pUCCAT (a derivative of pSV2CAT in which the SV40 promoter/enhancer has been replaced by a pUC13 polylinker; provided by M. Gilman, M.I.T.).
Primer extension analysis was performed by hybridizing a 433-bp SinI-Sin1 fragment of stromelysin cDNA (3"end-labeled with Klenow DNA polymerase) with 10 pg of total RNA from TPA-induced rabbit synovial fibroblasts in hybridization buffer containing 80% formamide, 0.4 M NaC1.40 mM PIPES, pH 6.4, 1 mM EDTA for 5 h a t 50 "C. After ethanol precipitation, primers were extended by reaction with reverse transcriptase (Life Sciences, Inc.) in a 60-pl reaction containing 0.15 mM unlabeled deoxynucleoside triphosphates under standard conditions (28) for the enzyme; products were analyzed on a 6% sequencing gel.
Ribonuclease protection analysis was performed by transcribing an XbaI-cut exoIII deletion mutant in blue scribe (Stratagene) containing -650 bp of stromelysin sequence, from +115 to -500 (relative to the presumed start site deduced by primer extension) with T3 RNA polymerase under conditions recommended by the supplier (Stratagene). Bluescript plasmid linearized with various restriction enzymes and transcribed with T3 RNA polymerase served as RNA molecular weight markers. T 3 transcripts were hybridized with 10 pg of total TPA-induced rabbit synovial cell RNA under the primer annealing conditions above, except at 52 "C, diluted 10-fold, and digested with 40 pg/ml RNase A and 2 pg/ml RNase T1 for 45 min a t 30 "C followed by proteinase K treatment and phenol extraction; the protected RNA products were then analyzed on a 6% sequencing gel.
Chloramphenicol acetyltransferase assays were performed by calcium phosphate transfecting 10 pg of plasmid DNA per 100-mm dish of subconfluent rabbit synovial fibroblasts for 6 h, using the method of Spandidos and Wilkie (29), washing with serum free medium, and incubating for 16 h in serum free medium prior to the addition of stromelysin gene regulators. After 24 h of induction, cells were harvested and protein extracts prepared by lysis in 0.25 M Tris, pH 7.5, 0.5% Triton X-100; extracts were assayed according to Gorman (30). Protein concentrations were checked by using the Bio-Rad protein assay reagent. &Galactosidase assays were performed on extracts from cells that had been co-transfected with RSVBgaI as described by Gorman (30) except that SDS (0.2%) was added at the end of the reaction period to prevent precipitation of Triton X-100 caused by the addition of sodium carbonate.

Regulatory Behavior of the Endogenous Stromelysin Gene-
Total RNA was isolated from rabbit synovial fibroblasts and analyzed on Northern blots using a stromelysin cDNA as probe. 11-1 elicited a dose-dependent ( Fig. la) induction (maximum 63-fold) of stromelysin mRNA levels in rabbit synovial fibroblasts, with a substantial effect observed even at 4 ng/ ml (100 p~) .
By contrast, the inflammatory mediator tumor necrosis factor, whose biologic activities are often similar to those of 11-1, had no effect on stromelysin mRNA levels even a t 2 pg/ml (data not shown). Rabbit synovial fibroblasts were serum-starved for 16 hand incubated in serum free medium without additions ( N ) or with EGF ( E ) , TPA (T),  ( I ) at the doses indicated in ng/ml, or cholera toxin + IBMX (C). + orrefers to the presence or absence of dexamethasone 2 h prior to, and throughout, the induction period of 24 h. Six pg of each RNA were Northern blotted and hybridized with a '"P-labeled fragment of a stromelysin cDNA clone as described under "Materials and Methods." 6, confirmation that the lanes in a had equal amounts of intact mRNA by in vitro translation. Total RNAs (0.5 pg, which was found to be subsaturating) were translated in rabbit reticulocyte lysate reactions and analyzed by SDS-polyacrylamide gels. The stromelysin gene was induced by several other factors as well. TPA elicited a large (120-fold) induction of stromelysin mRNA levels ( Fig. la), consistent with our previous study (7) showing that stromelysin mRNA constitutes 1-2% of the total mRNA of TPA-treated rabbit synovial fibroblasts. EGF had a weaker effect, inducing stromelysin mRNA levels by &fold (Fig. la). These results were in contrast to the rat fibroblast transin system, in which EGF was more effective than TPA (10).

Transcription Induced by Il-1 and Repressed by Dexamethasone
Treatment of rabbit synovial fibroblasts with cholera toxin in the presence of IBMX, which elicits a large accumulation of CAMP (31), induced the stromelysin gene to about the same extent as EGF, implicating CAMP-dependent protein kinase activation as one possible mode of signal transduction for the gene.
Because glucocorticoids have been shown to suppress extracellular matrix degradation in cell culture systems (9) and in vivo (17), and because the collagenase gene is one target of glucocorticoid repression (9), we examined the effect of dexamethasone on stromelysin gene expression. We found that dexamethasone diminished stromelysin induction by all inducing agents tested (by 5-fold for TPA, 5-fold for EGF, 7.5fold for 11-1, 14-fold for cholera toxin; Fig. la); in fact, even basal stromelysin expression was repressed. To exclude the possibility that dexamethasone elicited a nonspecific repression of transcription, RNA from control and dexamethasonetreated cells was translated in vitro, and the translation products were analyzed by SDS-polyacrylamide gel electrophoresis (Fig. 16). Dexamethasone neither had an effect on the amount of total translation product synthesized in vitro nor on the synthesis of the 43-kDa translation product corresponding to actin (compare lanes 1 to 2 , 3 to 4 , 5 to 6, and 7 to 8.) Also, the dexamethasone effect was not due to a general inhibition of protein synthesis, as determined by an [":'S]methionine pulse-labeling experiment (data not shown).
Mapping of the Transcription Start Site-Stromelysin genomic clones were isolated by screening a rabbit genomic library using a 1.2-kbp stromelysin cDNA clone (7), and one of the five unique genomic clones was further characterized by hybridization with probes generated from pAct3, a bluescript plasmid containing an 1822-bp stromelysin cDNA insert (from a Xgtll clone provided by Dr. E. Fini, M.I.T.).2 The position of the transcriptional start site of the stromelysin gene within a 1.1-kbp fragment (which hybridized with a probe representing 146 nucleotides of the extreme 5'-end of pAct3) was determined by primer extension analysis using a 433-nucleotide penultimate cDNA fragment as primer, and total RNA from TPA-induced rabbit synovial fibroblasts as template for reverse transcription. The major primer extension product was -590 nucleotides (Fig. 2a, determined from a shorter exposure), indicating that the start site was approximately 11 nucleotides upstream of the 5'-end of the cDNA. Alignment of the cDNA with the genomic fragment was accomplished by sequencing the latter and comparing it with the cDNA sequence, yielding the approximate start position indicated in Fig. 3. Verification of the approximate start site was provided by ribonuclease protection hybridization, using a T 3 polymerase-generated RNA transcript representing +115 to -500 (with respect to the presumed start site). When hybridized with total RNA from TPA-induced rabbit synovial fibroblasts, and digested with ribonucleases, the major product (Fig. 2b) was about 110 nucleotides, closely agreeing with the primer extension result and with the location of a "TATA box" (in this case TATAAAATT) at 20-20 nucleotides upstream of the indicated start site (Fig. 3).
Comtruction and Transfection of Stromelysin Promoter-Chloramphenicol Acetyltransferase Fusions-DNA fragments spanning from the transcription start site (Sac at position -10) to various distances upstream were fused upstream of a chloramphenicol acetyltransferase coding sequence as described under "Materials and Methods." When constructs containing either a 700-bp fragment (Fig. 4a)   a, primer extension analysis was performed as described under "Materials and Methods" using a 433-bp cDNA primer whose 5'-end was 146 nucleotide downstream of the 5'-end of the cDNA. The templates for reverse transcription were: yeast tRNA (y) or total RNA from TPA-induced rabbit synovial fibroblasts ( r ) . p shows the unreacted primer. b, ribonuclease protection analysis was performed as under "Materials and Methods" using a T3 polymerase transcript spanning +115 to -500 relative to the presumed start site, hybridized to total RNA from TPA-induced rabbit synovial fibroblasts ( r ) or yeast tRNA (Y). ovial fibroblasts, the basal expression level was very low, but could be induced by about 20-fold by treatment with TPA, suggesting that a phorbol ester-inducible sequence element resides within the 700-bp fragment. (As a control, TPA was found to induce RSVCAT by only -%fold; data not shown.) 11-1 induction of chloramphenicol acetyltransferase activity was also conferred by the 700-bp SL fragment, although magnitude of the induction (8-fold) was smaller than that observed in Northern blot analysis. EGF and cholera toxin    (upper lines). Sequence was determined as described under "Materials and Methods" and aligned with the rat transin promoter sequence, which was available from GenRank. Matches are indicated by pairs of dots. Some features of the stromelysin sequence are indicated: the TATA box is underlined, the position of the start site is arrowed, and the region homologous to the TPAinducihility element described under "Discussion" is boxed. The Sac1 site a t position -10 was used to create stromelysin-chloramphenicol acetyltransferase fusion plasmids. also induced chloramphenicol acetyltransferase activity. The induction by each factor was diminished significantly by cotreatment with dexamethasone (Fig. 46). (By contrast, chloramphenicol acetyltransferase activity driven by the Rous sar-b .' * * * e coma virus long terminal repeat in RSV-CAT or driven by the SV40 enhancer in pSV2CAT were diminished only slightly by dexamethasone treatment (Fig. 46).) Total &galactosidase activity present in the same protein extracts used for the chloramphenicol acetyltransferase assays (endogenous 8-galactosidase plus @galactosidase from the plasmid RSVPgal which was cotransfected with the stromelysin plasmids) was affected by less than 20% by any of these treatments (data not shown).

DISCUSSION
Induction of stromelysin gene expression by 11-1 demonstrated above is consistent with the role of 11-1 as an inflammatory mediator. For example, 11-1 induces the synthesis of collagenase and prostaglandins in human synovial cells (31). Because of the extreme substrate specificity of collagenase for a single cleavage site in collagen, collagenase synthesis is probably necessary but insufficient for the massive cartilage degradation observed in rheumatoid arthritic synovia (where 11-1 is found in elevated concentration compared to normal synovia (32)) or for the cartilage degradation resulting from experimental 11-1 treatment of synovia (14). In at least three systems, rabbit chondrocytes (34, 35), endothelial cells (36), and synovia (14), secretion of a proteoglycan-degrading (or in uitro caseinolytic) activity results from 11-1 treatment. (One problem in the interpretation of these reports has been that assays of enzyme activity, in vivo or in uitro, can be heavily influenced by the levels of metalloproteinase inhibitors, as in Ref. 37, for example, where large amounts of endothelial cell collagenase are masked by tissue inhibitor of metalloproteinases.) Our present observation that the stromelysin gene is induced by 11-1 identifies the stromelysin gene product, at least in the synovial cell system, as an important component of the tissue-degradative process. We have also demonstrated that 11-1 inducibility can be conferred upon a heterologous coding sequence (chloramphenicol acetyltransferase) by a 700-bp 5"flanking fragment of the stromelysin gene, which allows the mapping of 11-1 inducibility sequence elements.:' The smaller magnitude of 11-1 induction of the transfected stromelysin promoter as compared to the endogenous gene could be due to limiting amounts of a transcriptional factor which mediates the 11-1 response, in light of the expected high :I S. M. Frisch and H. E. Ruley, work in progress. copy number of the transfected gene. For example, in Ref. 46, the TPA induction of sV40 enhancer is observed only at low DNA input per plate. Alternatively, additional 11-1 inducibility elements may reside in regions other than the 700-bp fragment tested.
Partial protection by glucocorticoids against inflammationinduced tissue degradation (17) may be partly attributed to repression of collagenase synthesis, as has been shown for synovial fibroblasts (18) and macrophages (19). The repression of stromelysin transcription presently reported implicates an additional mechanism of protection.
Both an increase in matrix-degrading proteinase activity (6) and a decrease in fibronectin accumulation (38) have been observed in a wide variety of tumor cells. Treatment of several transformed cell lines having both of these properties leads to restoration of nearly normal fibronectin levels ( e g . Ref. 21). If the level of fibronectin in transformants is abnormally low because of increased degradation by proteinases such as stromelysin, then the glucocorticoid repression of stromelysin synthesis may partly explain restoration of normal fibronectin levels.
Our observation that dexamethasone repressed stromelysin expression under the influence of a variety of inducers (TPA, EGF, 11-1, cholera toxin) suggests two possible mechanisms: (i) the blockage is in a late step of signal transduction (which would hypothetically be a common signal generated by all of the above inducers, since the early responses to these inducers are diverse); or (ii) the glucocorticoid receptor-dexamethasone complex binds to the stromelysin 5'-flanking region so as to repress transcription. Thus far, we have insufficient evidence to distinguish between these possibilities. Although there is no precedent in the literature, to our knowledge, for direct inhibitory effect of glucocorticoid receptor on transcription, we have searched for glucocorticoid-regulatory sequence elements in the stromelysin 5"flanking region and have found three of them (assuming the consensus sequence to be AGAA/ TCA(G)A/T (39)), AGAACA at position -643, AGATCA at position -530, and GAACAT a t position -287.
TPA induction of the collagenase and SV40 enhancer has led to the discovery of a consensus sequence element TGAGT-CAG (40), appearing at position -73 to -64 in the human collagenase gene; this sequence is protected by transcription factor AP-1 and can confer TPA-inducibility upon heterologous promoters such as thymidine kinase. The rabbit stromelysin 5"flanking region has a sequence closely fitting this consensus a t position -65 to -57 (Fig. 3). In the case of the human proenkephalin gene, which also has a homology to this consensus in its 5'-flanking region (41), the CAMP and TPA inducibility conferred by this element are apparently nondissociable.
Extensive sequence homology between the promoter-proximal -370-bp of the stromelysin 5'-flanking region and its rat transin counterpart (Fig. 3) suggests conservation of a series of binding sites for trans-acting regulatory proteins. It will be of considerable interest to determine whether discrete or overlapping sequences (or trans-acting proteins) confers responsiveness to the various inducers used in this study, given the multitude of potential interactions among signal transduction systems involved in stromelysin induction. For example, interactions between protein kinase C and EGF receptor (42) and protein kinase C with the adenylate cyclase system (43) have been demonstrated. In addition, 11-1 stimulates phospholipid turnover, possibly leading to protein kinase C activation (44). It is also conceivable that the repression of 11-1 synthesis by dexamethasone (45) interrupts an autocrine cycle that could be involved in stable activation of stromelysin gene expression.