Oncogene (1999) 18, 7755 ± 7764 ã 1999 Stockton Press All rights reserved 0950 ± 9232/99 $15.00 http://www.stockton-press.co.uk/onc Ets transcription factors cooperate with Sp1 to activate the human Tenascin-C promoter

Fumiaki Shirasaki1,5, Huda A Makhluf1,4,5, Carwile LeRoy2, Dennis K Watson3 and Maria Trojanowska*,1

1Department of Medicine, Division of Rheumatology and Immunology, Hollings Cancer Center, Medical University of South Carolina, Charleston, South Carolina, SC 29425-2229, USA; 2Department of Microbiology and Immunology, Hollings Cancer Center, Medical University of South Carolina, Charleston, South Carolina, SC 29425-2229, USA; 3Center for Molecular and Structural Biology, Hollings Cancer Center, Medical University of South Carolina, Charleston, South Carolina, SC 29425-2229, USA

Tenascin-C (TN-C), an extracellular matrix glycoprotein upregulated in some physiological and pathological is expressed during embryonic development, but is conditions including wound healing, in¯ammatory present only at low levels in normal adult tissues. TN- lesions, ®brosis and in cancer tissues (Chiquet- C is re-expressed during wound healing, ®brotic diseases Ehrismann et al., 1995; Crossin, 1996; Mackie, 1997). and in cancer. To better understand the mechanisms that Studies have demonstrated elevated TN-C expression control TN-C gene expression, we examined the in various human cancers, including breast (Shoji et al., regulation of the human TN-C promoter in human 1992; Yoshida et al., 1997), lung (Kusagawa et al., ®broblasts. We demonstrate that a short segment of the 1998), colon (Sakai et al., 1993; Hanamura et al., TN-C promoter between bp 7133 and 727 contains 1997), and liver (Yamada et al., 1992). TN-C is three evolutionarily conserved Ets binding sites (EBS). expressed in both cancer and stromal cells, suggesting These three EBSs bind in vitro expressed Fli1 protein a possible role in invasive properties of the cancer cells and mediate transactivation of the TN-C gene by Fli1. themselves as well as in the remodeling of cancer Furthermore, two proximal EBSs contribute signi®cantly stroma. Strong positive TN-C staining is usually to basal activity of the TN-C promoter. GABP, which is associated with stromal ®broblasts surrounding tumor present in human ®broblast nuclear extracts, interacts cells, suggesting that TN-C is a stromal marker for with the two proximal EBSs. In addition, several Sp1 epithelial malignancy. Furthermore, positive staining in and Sp3 binding sites have been located in close cancer cells has been associated with unfavorable proximity to the EBSs within this promoter region. disease outcome (Yoshida et al., 1995). The transcrip- The studies performed in Drosophila cells demonstrate tional factors that regulate TN-C re-expression in that either Fli1 or GABPa+b1 functionally interact with tumors are presently unknown. Sp1 resulting in a synergistic stimulation of the TN-C The transcriptional regulation of the TN-C gene has promoter activity. In conclusion, this study shows for the been investigated in several species including human, ®rst time that the TN-C gene is regulated by Ets mouse, and chicken (Jones et al., 1990; Copertino et proteins, which together with Sp1 act as potent al., 1995; Gherzi et al., 1995a). The nucleotide sequence activators of TN-C expression. of the promoter region encompassing about 250 base pairs (bp) upstream of the transcriptional start site is Keywords: Tenascin-C; transcriptional regulation; Fli1; highly conserved among these three species. This GABP; Sp1 promoter region also confers the highest promoter activity in transient transfection assays. In contrast, human and mouse promoter regions upstream of 7250 contain putative negative regulatory elements which may cause repression of TN transcription (Copertino et Introduction al., 1995; Gherzi et al., 1995a). In addition, the ®rst untranslated exon of the human TN gene was shown to Tenascin-C (TN-C) is a hexameric extracellular matrix contain additional positive and negative regulatory glycoprotein, whose expression is precisely controlled elements (Gherzi et al., 1995b). Moreover, several temporally and spatially during embryonic develop- regulatory elements including a GCN4/AP-1 like site at ment (Erickson and Bourdon, 1989). It is expressed at position 7146 and an NF-kb element at position sites where cells undergo extensive epithelial-mesench- 7210 have been characterized by transfection of the ymal transformation such as the central and peripheral 7220 to +79 human promoter construct into primary nervous system, lung, early heart, kidney and teeth rat embryo ®broblasts (Mettouchi et al., 1997). Recent (Erickson and Bourdon, 1989). TN-C is generally studies of the mouse TN promoter revealed the found at low levels in the normal adult tissues, but is presence of four distinct protein binding sites, including sites for Krox24 at position 7257, octamer (N-Oct3, N-Oct5A and N-Oct5B) at position 7200, nuclear factor one at position 7180, and an uncharacterized *Correspondence: M Trojanowska protein that binds to a site called the tenascin control 4Current address: Department of Orthopedic Research, Brigham and element at position 7247 (Copertino et al., 1997). Women's Hospital, Harvard Medical School, Boston, MA, USA 5The ®rst two authors contributed equally to this paper, Interestingly, di€erent combinatorial e€ects of these Received 18 May 1999; revised 8 November 1999; accepted elements were found in cells of di€erent lineage 8 November 1999 (Copertino et al., 1997). Fli1 and Sp1 activate TN-C promoter F Shirasaki et al 7756 Since TN-C expression is highly regulated during development, transcription factors involved in deter- mining speci®c expression patterns are of particular interest. It has been suggested, based on the tissue culture experiments, that TN-C gene is a target of the homeodomain proteins which can either activate (e.g. Evx-1) or repress (e.g. OTX2) TN-C promoter activity (Jones et al., 1992; Gherzi et al., 1997). On the other hand, in vivo studies by Meyer et al implicated Fli1 (a member of the Ets family of transcription factors) as a candidate transcription factor regulating TN-C expres- sion during embryogenesis (Meyer et al., 1995). It was demonstrated by in situ hybridization of Xenopus and avian embryos that expression pattern of Fli1 coincides with sites of cell migration and/or epithelial-mesench- Figure 1 Functional analysis of the human TN-C promoter in ymal transformation (Meyer et al., 1995; Mager et al., human dermal ®broblast. Subcon¯uent cells were transiently transfected with 20 mg of the indicated recombinant plasmids and 1998). These sites include migrating cranial neural crest 2.5 mg of p-SV-b-galactosidase. The diagram on the left shows the cells, endothelial cells of the heart, and kidney/ extents of TN-C promoter deletions. The bar graph on the right pronephric duct, suggesting that Fli1 may regulate a represents per cent activity relative to that of the 7248 promoter, set of genes that are involved in cellular adhesion. which was arbitrarily set at 100%. Means+s.e.mean of six independent experiments are presented. *indicate signi®cantly Consistent with this notion, putative Fli1 binding sites di€erent values compared to 7248 and +75 construct, (P50.05, were found in the promoter regions of av and b3 Mann-Whitney U-test) integrins, vitronectin and tenascin. However, the role of Fli1 in regulating expression of these genes has not been directly investigated. end points at 7516, 71239 and 72100 showed The Ets family of transcription factors shares a gradual decrease of the promoter activity as compared highly conserved DNA-binding motif termed the Ets with the 7248 deletion. These data indicate that TN-C domain, which recognizes purine-rich DNA core motif expression is controlled by combination of cis-elements GGAA/T in the promoter/enhancer regions of many in a similar manner as reported in other cell types genes. The Ets family consists of over 30 related (Copertino et al., 1995; Gherzi et al., 1995a). members which have been found throughout the metazoan world including human, mouse, chicken, Mapping of Fli1 response elements in the TN-C promoter sea urchin, Drosophila and C. elegans (Graves and Petersen, 1998; Gysdael and Boureux, 1997; Sharrocks The expression patterns of TN-C and Fli1 during et al., 1997). The Ets proteins, except for GABPa, bind development has suggested that TN-C gene may be to their recognition sites as monomers. However, in regulated by Fli1 (Meyer et al., 1995; Mager et al., 1998). many cases, the anity of the Ets proteins to their To investigate this possibility we focused our studies on DNA targets is increased by association with other the mechanism of TN-C gene regulation by Fli1 in transcription factors including, among others, AP1, human dermal ®broblasts. To identify the speci®c Sp1, CBF, C/EBP and SRF (Graves and Peterson, regions in TN-C promoter that mediate transactivation 1998; Ghysdael and Boureux, 1997; Sharrocks et al., by Fli1, a series of previously generated 5'-deletion 1997; McNagny et al., 1998). Ets genes have been constructs was cotransfected with the Fli1 expression implicated in cellular proliferation and di€erentiation, vector. Overexpression of Fli1 strongly activated all the as well as in tumor formation and metastasis (Graves deletion constructs with end-points between 72100 and and Peterson, 1998; Ghysdael and Boureux, 1997; 7133 with similar potency, whereas 727 deletion was Sharrocks et al., 1997). not responsive to Fli1 (Figure 2a). These results indicate In this study, we have investigated the regulation of that the TN-C promoter region between bp 7133 and the human TN-C promoter and report for the ®rst time 727 mediates transactivation of this gene by Fli1. The that TN-C is regulated by Ets transcription factors in a stimulatory e€ect of Fli1 on TN-C promoter is not the synergistic interaction with Sp1 and Sp3 in human result of nonspeci®c modulation of transcription appa- dermal ®broblasts. ratus due to high level of overexpression, since cotransfection of Fli1 did not have any e€ect on the activities of vector alone and early SV40 promoter (Figure 2a). Inspection of the sequence revealed four Results GGAA/T motifs which are potential binding sites for Ets family proteins. They are henceforth denoted EBSs (Ets Functional analysis of the TN-C promoter Binding Sites) 1 ± 4 (Figure 2b). To identify potential regulatory elements of the human TN-C gene, a series of 5'-deletions of the TN-C Ets1 and Ets2 are modest activators of the TN-C promoter linked to the chloramphenicol acetyltransfer- promoter ase reporter gene was generated. These constructs were tested by transient transfection assays in human dermal The co-transfection experiments presented in Figure 2 ®broblasts (Figure 1). The 7248 construct showed the indicated that Fli1 is a potent activator of the TN-C highest promoter activity. Subsequent deletions to bp promoter in our system. To test whether other Ets- 7133 and 727 decreased promoter activity to 39 and family members can a€ect the human TN-C promoter 19%, respectively. The longer constructs with deletion activity, the 7248 TN-C construct was co-transfected Fli1 and Sp1 activate TN-C promoter F Shirasaki et al 7757

Figure 2 Identi®cation of the TN-C promoter region mediating Fli1 stimulation. (a) The indicated TN-C promoter deletion constructs or the pSV2-CAT construct were co-transfected with 1 mg of Fli1 expression vector or 1 mg of control vector. The bar graph on the right represents fold stimulation of the promoter activity of each construct co-transfected with Fli1 relative to the activity of the promoter co-transfected with pSG5, which was arbitrarily set at 1. The average of at least four independent experiments are shown. *indicate statistically signi®cant results (P50.05, Mann-Whitney U test). (b) Nucleotide sequence of the TN-C promoter region from position 7133 to +1. The four putative Ets binding sites (EBS) are boxed. The probes (A ± D) used in EMSA are underlined

Fli1 stimulated TN-C mRNA expression levels To test whether Fli1 is capable of inducing endogenous gene expression, we measured TN-C mRNA levels in ®broblasts transiently transfected with the Fli1 expres- sion vector. Total RNA was prepared from either Fli1 or pSG5 transfected dermal ®broblasts and used for the preparation of cDNA. PCR was performed using di€erent cDNA input, with primer pairs speci®c for TN-C, Fli1 and GAPDH. As shown in Figure 4, signi®cantly higher expression of TN-C mRNA was detected in cells overexpressing Fli1.

Fli1 interacts in vitro with the GGAA motifs in TN-C promoter

Figure 3 Fli1 is a potent transactivator of the TN-C promoter. We next asked whether any of the four putative EBSs Subcon¯uent dermal ®broblasts were co-transfected with 10 mgof in the TN-C promoter are capable of binding Fli1 in the 7248 TN-C promoter construct and indicated amounts of the vitro. To test this, we synthesized Fli1 protein in vitro Ets1 (-~-), Ets2 (-&-), and Fli1 (-&-) expression vectors (or using TNT Coupled Reticulocyte Lysate System equal amounts of control vector pSG5). The total amounts of (Promega). The size of the in vitro translated Fli1 DNA was kept constant by adding pSG5. The graph depicts the TN-C promoter activities relative to the activity of promoter co- protein was con®rmed by an immunoblot analysis with transfected wtih pSG5, which was arbitrarily set at 1. Means+ the anti-Fli1 antibody (data not shown). The in vitro s.e.mean from four independent experiments are presented translated Fli1 protein was reacted with the various 32P-labeled fragments containing individual EBSs (probes A ± D, as underlined in Figure 2b) in the absence or presence of the antibody to Fli1. The DNA- protein complexes were analysed by electrophoretic with increasing amounts of Ets1, Ets2 or Fli1 mobility shift assay (EMSA) (Figure 5 left panel). No expression vectors. As shown in Figure 3, Fli1 strongly binding was observed with unprogrammed reticulocyte activated the TN-C promoter in a dose dependent lysate (data not shown). The Fli1 protein bound manner (over 15-fold), whereas Ets1 and Ets2 had weakly to the TN-C promoter fragments containing more modest stimulatory e€ects (3 ± 5-fold). EBS1 (probe A) or EBS3 (probe 3); and relatively Fli1 and Sp1 activate TN-C promoter F Shirasaki et al 7758

Figure 4 Expression of Fli1 induces the endogenous TN-C promoter. Total RNA was harvested from either Fli1 or pSG5 transfected dermal ®broblasts and used for RT ± PCR. Dilutions of the reverse-transcribed products (1/10, 1/20 and 1/50 volume) were ampli®ed with primers for TN-C, Fli1, and GAPDH for 24, 28 and 21 cycles, respectively (as indicated). M, 1 kb DNA ladder; N, negative control assay carried out in the absence of reverse transcriptase

Figure 5 Interactions of in vitro translated Fli1 protein with the putative EBSs. (Left) EMSA using probes A ± D containing EBS 1 ± 4.2 mlofin vitro translated Fli1 protein (see Materials and methods) were used in binding reactions with 5'-end-labeled oligonucleotides containing EBS1-4 in the absence (lanes 2, 4, 6, 8, 11 and 13) or presence (lanes 3, 5, 7, 9, 12 and 14) of 1 mlof anti-Fli1 antibody. Lanes 1 and 10 are probe D only. The probes used are indicated on top of the gel. The nucleotide sequence of the EBS probes is shown in Figure 2b. (Right) EMSA with probe D or mutated probe D (CCATTACAGATTAAGGAGCTCGC). Speci®c DNA-protein complexes and supershifted bands are indicated by arrows

stronger binding was observed with the fragment EBS1, 3 and 4 function as Fli1 response elements containing EBS4 (probe D). By contrast, no detectable binding occurred with the fragment containing EBS2 Since Fli1 could bind to EBS1, 3 and 4 in in vitro (probe B). The DNA-protein complexes were super- binding assays, these three sites may mediate transacti- shifted with anti-Fli1 antibody indicating that the vation of the TN-C promoter by Fli1. To examine this, binding was speci®c. Interestingly, Fli1 binding to the 7133 TN-C promoter constructs carrying the DNA was stabilized in the presence of speci®c individually mutated GGAA motifs (as described in monoclonal anti-Fli1 antibody. This is likely due to Figure 5) were tested in transient transfection assays unmasking of the Fli1 DNA-binding domain by the (Figure 6, left panel). Substitution mutations in EBS1, antibody (Watson et al., 1992; Hodge et al., 1996). To 3 and 4 signi®cantly reduced transactivation by Fli1. determine whether the GGAA core motifs are Mutation in EBS4 had the greatest e€ect (2.4-fold responsible for the observed binding, we introduced reduction) followed by those in EBS1 (twofold) and substitution mutations into these motifs, changing EBS3 (1.7-fold). Mutating EBS2 did not signi®cantly GGAtoTTA in the EBS1, 3 and 4 probes. In all a€ect Fli1 transactivation. These data correlate well cases, mutating the GG motif prevented formation of with the relative anity of those sites to Fli1 (Figure 5) the DNA-Fli1 complexes (representative data for probe and suggest that EBS1, 3 and 4 are the functional Fli1 D are shown in Figure 5 right panel). response elements in the TN-C promoter. Fli1 and Sp1 activate TN-C promoter F Shirasaki et al 7759

Figure 6 Identi®cation of the functional Fli1 response elements in the human TN-C promoter. The diagram on the left indicates mutated EBSs. The mutants were constructed by replacing two guanosines in the EBS core sequence with two thymidines using the 7133 TN-C promoter construct. Mutated plasmids were cotransfected with either 1 mg of Fli-1 expression vector or 1 mg of control plasmid. The bar graph represents fold stimulation of the promoter activity of each construct co-transfected with Fli1 relative to the activity of the promoter cotransfected with pSG5 which was arbitrarily set at 1. The numbers on the right show the basal promoter activities (i.e. without exogenous Fli1) of each mutant construct relative to the 7133 promoter, which was arbitrarily set at 100%. Means+s.e.mean of ®ve independent experiments are shown. *indicate statistically signi®cant results (P50.05, Mann-Whitney U- test)

To determine whether these response elements also by the addition of the consensus Ets oligonucleotide contribute to the basal promoter activity, the e€ects (lane 5), suggesting that the binding proteins are of substitution mutations in EBS1 ± 4 were investi- related to the Sp1 and Ets factors. To test directly gated (Figure 6, right panel). Mutating either EBS3 whether the GGAA motif is involved in formation of or EBS4 resulted in the most signi®cant reduction of the observed DNA-protein complexes, the 23 bp TN-C the basal promoter activity, while mutating EBS1 promoter segment containing mutated EBS4 motif was has no e€ect on the basal level. Modest, but used either as a cold competitor (lane 6) or as a probe consistent reduction of the basal level was observed (lane 7). No competition was observed with the with the EBS2 mutant. Thus, the four EBSs in the mutated oligo, and no speci®c binding occurred with TN-C promoter may have di€erent functions: EBS3 the mutated probe, suggesting that the GGAA motif is and 4 mediate transactivation by Fli1 and contribute responsible for the formation of these speci®c DNA- to basal promoter activity, EBS1 mediates transacti- protein complexes. vation by Fli1, but has little e€ect on the basal To further characterize the nature of the proteins activity, and EBS2 contributes modestly to the basal interacting with this probe, we employed antibodies activity but does not function as Fli1 response against two members of the Sp1 family (Sp1 and Sp3) element. and four members of the Ets family (Ets1, Ets2, Fli1 and GABPa) as well as GABPb1. As shown in Figure 7b, addition of the anti-Sp1 antibody caused a Nuclear proteins from human dermal fibroblasts interact supershift of complex 1 (lane 3), whereas addition of with the TN-C promoter the anti-Sp3 antibody abolished formation of complex EMSA were performed to determine which transcrip- 2 (lane 4). When anti-Sp1 and anti-Sp3 antibodies tion factors may interact with the TN-C promoter in were added simultaneously, formation of complexes 1 vivo. Speci®cally, we sought to determine the nature of and 2 were a€ected (lane 5). Addition of the anti-Ets1, the transcription factors expressed by dermal ®bro- anti-Ets2, and anti-Fli1 antibodies did not a€ect blasts that interact with EBSs. Four probes, A ± D, formation of the DNA-protein complexes (lanes 6 ± containing EBS 1 ± 4 respectively, were used for this 8). However, addition of the anti-GABPa or b1 analysis (as described in Figure 2b). Representative antibodies prevented formation of complex 3 (lanes results of EMSA using a 23 bp TN-C promoter 9 and 10). segment containing EBS4 (probe D) and nuclear Similar EMSA analyses were carried out with extracts derived from human dermal ®broblasts are probes A, B and C. These studies showed that GABPa presented in Figure 7. Three speci®c DNA-protein interacts with probe C containing EBS3, but does not complexes (denoted 1 ± 3 in Figure 7a, lane 2) were interact with probes A or B. Furthermore, EMSA observed. The two complexes below the complex 3 with mutated oligonucleotides and antibodies have were not competed by an excess of unlabeled wild-type determined that sequences present at 761 and 7111 probe (lane 3), suggesting nonspeci®c binding. Com- are critical for Sp1/Sp3 interaction. The EMSA results plexes 1 ± 3 were all competed by an excess of are summarized in Figure 7c. Taken together these unlabeled probe D. Furthermore, complexes 1 and 2 results indicate that Sp1, Sp3 and GABP interact with were competed by 200-fold excess of a consensus Sp1 several sites within the 7133 to 727 TN-C promoter oligonucleotide (lane 4) while complex 3 was completed region. Fli1 and Sp1 activate TN-C promoter F Shirasaki et al 7760

Figure 7 Sp1, Sp3 and GABP in human ®broblasts nuclear extracts bind to the probe D containing EBS4. (a) EMSA with competitor oligonucleotides. The wild-type probe D (lanes 1 ± 6) or the mutated probe D (lane 7) were incubated with human ®broblasts nuclear extracts (5 mg/lane) in the absence (lanes 2 and 7) or presence (lanes 3 ± 6) of the unlabeled competitor oligonucleotides (200-fold molar excess). The competitor used is indicated on the top of the gel: wild-type probe D (WT, CCATTACAGAGGAAGGAGCTCGC), Sp1 consensus recognition sequence (ATTCGATCGGGGCGGGGCGAGC), Ets consensus recognition sequence (GATCCTCGAGCCGGAAGTTCGA) and probe D mutant probe (CCATTACAGATTAAG- GAGCTCGC). DNA-protein complexes are indicated by arrows. (b) EMSA with antibodies. The probe D was incubated with 5 mg nuclear extract and 1 mg of speci®c antibodies as indicated on the top. Speci®c DNA-protein complexes and supershifted complexes are indicated by arrows. (c) Sp1, Sp3 and GABPa binding sites within the 7133/+1 bp region of the TN-C promoter. The data were obtained by EMSA with speci®c competitors and antibodies using wild-type and mutated binding site probes (a and b and unpublished observations). The mutated nucleotides are bold and EBSs and TATA box are boxed

the TN-C promoter dose dependently (Figure 8b). Ets factors cooperate with Sp1 to activate the TN-C Likewise, GABPa+b1 modestly transactivates the TN- promoter C promoter (Figure 8c). Simultaneous addition of Sp1 It has been well documented that Ets protein cooperate (or Sp3) and Fli1 resulted in synergistic activation of with other transcription factors to regulate expression the TN-C promoter activity (Figure 8d). Similarly, of target genes (Graves and Petersen., 1998; Ghysdael synergistic stimulation was observed when GABPa+b1 and Boureux, 1997; Sharrocks et al., 1997; McNagny et were added together with Sp1 or Sp3 (Figure 8e). al., 1998). Our analysis of the TN-C promoter raised Similar data were obtained using the 7248 TN-C the possibility that a cooperation between Ets and Sp1 promoter (data not shown). We conclude from these factors may also modulate transcriptional activity of experiments that Ets factors and Sp1 cooperate to the TN-C promoter. Such experiments are not feasible transactivate the TN-C promoter. in human dermal ®broblasts, because of the high endogenous levels of Sp1 in mammalian cells. By contrast, Drosophila SL2 cells have been used exten- Discussion sively for testing the activity of ubiquitous human transcription factors such as Sp1 (Courey and Tjian, The present study of the 5' ¯anking region of the 1988). The results of the synergistic analyses are human TN-C gene has identi®ed and characterized presented in Figure 8. The empty pPac Drosophila four GGAA motifs necessary for its function. Full expression vector did not increase promoter activity. promoter activity depends on the presence of intact Ets Sp1 (25 ± 250 ng) potently activates the 7133 TN-C and Sp1/Sp3 motifs. These motifs are located within a promoter in a dose-dependent manner (Figure 8a). Sp3 minimal promoter region between bp 7133 and 727. activates the TN-C promoter with similar potency The signi®cance of these motifs can be further inferred (data not shown). Fli1 (50 ± 500 ng) also transactivates from the previous observations that the proximal 250- Fli1 and Sp1 activate TN-C promoter F Shirasaki et al 7761

Figure 8 Fli1 and GABPa+b1 cooperate with Sp1 to transactivate the TN-C promoter. Drosophila cells were transfected with 10 mg of the 7133 TN-C promoter plasmid and increasing amounts of (a) pPacSp1 (25 ± 250 ng); (b) pPacFli1 (50 ± 500 ng) or (c) pPacGABPa and pPacGABPb1 (1.0 ± 5.0 mg of each). Control transfections with increasing amounts of pPacSp0 or pPacP1 had no e€ects on promoter activity. To ensure that cells in di€erent experiments were exposed to equal amount of transfecting DNA, appropriate amounts of pPacSp0 or pPacP1 were added to individual cotransfections. The bar graphs depict means+s.e.mean of promoter activity relative to the activity of promoter cotransfected with pPacSp0 (A) or pPacP1 (B, C) which were arbitrarily set as 1. (d) Synergistic e€ects of Sp1 and Fli1 on TN-C promoter activity. Drosophila cells were cotransfected wtih 250 ng of pPacSp1 and/or 250 ng of pPacFli1, as indicated. (e) Synergistic e€ects of Sp1 and GABPa+b1 on the TN-C promoter. Drosophila cells were cotransfected with 250 ng of pPacSp1 and/or 2.5 mg of pPacGABPa and pPacGABPb1, as indicated

bp region of this promoter is highly conserved among expression of Sp1 or Sp3 in ®broblasts has no human, mouse and chicken (Chiquet-Ehrismann et al., stimulatory e€ects (Ihn and Trojanowska, unpublished 1995). Signi®cantly, the most proximal GGAA motif observations). In contrast to GABP, overexpression of (EBS4) and its ¯anking sequences are highly conserved other members of Ets family stimulated the TN-C in all three species. EBS1 and 3 are fully conserved in promoter activity. Ets1 and 2 caused a ®vefold human and mouse promoters, while EBS2 is conserved increase, while overexpression of Fli1 caused the most in human and chicken promoters. Such high level of dramatic increase, over 15-fold (Figures 2 and 3). conservation suggests that these sites play important Stimulatory e€ects of Fli1 were mediated by EBS1, 3 roles in TN-C gene regulation. and 4. These three motifs were shown to interact with This study was conducted in human dermal Fli1 in vitro and substitution mutations in each of these ®broblasts which are known to have relatively high sites signi®cantly reduced transactivation of the TN-C expression levels of TN-C (Rettig et al., 1994). We promoter by Fli1 (Figures 5 and 6). We have also have demonstrated that GABPa, a ubiquitously demonstrated that Sp1 and Sp3 binding sites are expressed member of the Ets family, contributes to located in close proximity to the EBSs in the TN-C the activity of the TN-C promoter in dermal ®bro- promoter. Furthermore, both Sp1 and Sp3 strongly blasts. Speci®cally, this study shows that GABPa/b1 stimulated the activity of the TN-C promoter when bind to the EBS3 and 4 and that substitution tested in Drosophila cells (Figure 8). Although Sp3 mutations in each of these two sites signi®cantly often functions as a transcriptional repressor (Hagen et decrease the basal promoter activity. Interestingly, al., 1994), other extracellular matrix genes have been overexpression of the GABPa and b1 in ®broblasts previously shown to be stimulated by this transcription did not increase the promoter activity any further, factor (Ihn and Trojanowska, 1997). The role of Sp1 suggesting that endogenous GABP is expressed at the and Sp3 in regulation of the TN-C promoter has yet to saturation levels (data not shown). Similarly, over- be investigated fully. Fli1 and Sp1 activate TN-C promoter F Shirasaki et al 7762 Cotransfection experiments performed in Drosophila endogenous TN-C mRNA (Figure 4). Thus, Fli1 is able to cells, which lack Sp1-like factors, clearly indicate that interact with the endogenous promoter. Dermal ®bro- Sp1 is a potent inducer of TN-C promoter activity, blasts unstimulated by cytokines or growth factors which probably re¯ects its interaction with multiple express low levels of Ets1 (Gilles et al., 1996) of Fli1 sites within the TN-C promoter. The formation of (Shirasaki and Trojanowska, unpublished observations). multimeric complexes via protein ± protein interaction This may explain why in the antibody super-shift involving Sp1 molecules and other transcription factors experiment (Figure 7b), endogenous Fli1 protein was as well as Sp1 and components of the basal not detected as a binding factor to EBS4. Another transcription apparatus have been shown to facilitate possibility is that the endogenous Fli1 in dermal transcription of numerous other genes (Pascal and ®broblasts is in an inactive state under our growth Tjian, 1991; Verrijzer and Tjian, 1996). In addition to conditions. However, under certain conditions, when the Sp1 binding sites identi®ed in this study, additional elevated levels of active Fli1, Ets1, Ets2 or other family putative sites are present both downstream from bp members are present, these factors may contribute to the 727 and upstream of bp 7133 (Gherzi et al., 1995a). TN-C gene expression. For example, it has been shown Future studies are planned to characterize in more that Ets1 is induced by TNFa, PDGF and bFGF in detail the role of Sp1 and Sp3 in the TN-C gene human dermal ®broblasts (Gilles et al., 1996). Our regulation. preliminary observations suggest that Fli1 is also An important feature of the transcriptional regula- stimulated by TNFa (Shirasaki and Trojanowska, tion by the Ets proteins is their requirement for unpublished data). Interestingly, these cytokines have cofactors. The Ets factors that bind to DNA as been shown to induce TN-C expression (Rettig et al., monomers are rather weak activators of transcription 1994; Makhluf et al., 1996). TNFa and other growth and protein ± protein interaction with other factors has factors are present in lesional tissues such as rheumatoid been shown to strengthen their transactivating poten- arthritis, granulation tissue of healing wounds, and tial, although in some cases it can also lead to certain tumors, all of which show increased TN-C transcriptional repression (Sharrocks et al., 1997). expression. It is possible that under these conditions Ets Cooperation between Ets and other factors facilitates factors may play a role in TN-C induction. The nature of binding to sites not matching preferred consensous the speci®c Ets family member that regulates TN-C binding site (Halle et al., 1997). For example, the expression in vivo may depend on the speci®c tissues or GABP complex consists of an Ets family member, speci®c stimuli. GABPa, and an ankyrin repeat-containing protein, This study demonstrates that TN-C is a target gene GABPb. The GABPa subunits bind weakly to DNA, for Ets transcription factors, thus corroborating the but binding is signi®cantly increased after association previous in vivo observation by Meyer et al. that with its partner protein, GABPb. The heterotetramer implicated Fli1 in regulation of TN-C during embryo- containing two GABPa/b heterodimers binds to two nic development (Meyer et al., 1995; Mager et al., tandem repeats of the GGA motif (Thompson et al., 1998). A recent work from the same laboratory 1991; Batchelor et al., 1998). In our study, binding of provides additional support for the role of Fli1 during GABP to the EBS3 and EBS4 which contain two the process of cell migration (Mager et al., 1998). closely positioned tandem repeats of the GGA motifs Using quail embryo, it was shown that Fli1 is was demonstrated (Figure 7). Other Ets proteins have expressed in neural crest cells giving rise to mesench- evolved di€erent mechanisms to recruit speci®c partner yme and in mesoderm-derived endothelial cells of proteins (Graves and Petersen, 1998; Ghysdael and forming heart and blood vessels (Mager et al., 1998). Boureux, 1997; Sharrocks et al., 1997; McNagny et al., While the expression and function of Fli1 in the adult 1998). Signi®cantly, one such partner protein is Sp1. tissues has been primarily investigated in the hemato- Some of the examples include synergistic activation of poietic cells (Rao et al., 1993; Klemsz et al., 1993; the HTLV1 long terminal repeat by Ets1 and Sp1 Zhang et al., 1995a), the observation that Fli1 is (Ge gonne et al., 1993) or activation of CD18 promoter expressed in embryonic tissues at the sites of cell by Sp1 and GABP (Rosmarin et al., 1998). In a recent migration and epithelial-mesenchymal transformation study of the E-cadherin promoter binding sites for Sp1 raises the possibility that it may be reexpressed in and Erg-Fli1 were shown to contribute to the basal certain pathological conditions that are associated with promoter activity but possible synergistic interaction tissue remodeling. Other Ets proteins have been shown between these factors have not been investigated (Gory to participate in regulation of genes related to tissue et al., 1998). Our study of the TN-C promoter provides remodeling such as metalloproteases (MMPs) (ButticeÁ evidence for a cooperation between Ets factors and Sp1 et al., 1996; Westermarck et al., 1997; Gutman and in activation of the TN-C promoter activity. The Wasylyk, 1990; Wasylyk et al., 1991; Buttice and studies performed in Drosophila cells demonstrate that Kurkinen, 1993) and tissue inhibitors of metallopro- either Fli1 or GABPa+b1 functionally interact with teases (Edwards et al., 1992; Logan et al., 1996). TN-C Sp1, resulting in a greater than additive stimulation of is also expressed at sites of epithelial-mesenchymal the TN-C promoter activity. Taken together, this study transformation (Erickson and Bourdon, 1989). Inter- shows for the ®rst time that the TN-C gene is regulated estingly, similar distribution patterns of TN-C and by Ets proteins, which together with Sp1 act as potent MMPs were observed in the stroma of some tumors, activators of this gene. suggesting that TN-C may play a role in tissue In the context of cultured human dermal ®broblasts, we restructuring and may in¯uence the expression of genes have shown that Ets proteins transcriptionally activate involved in tissue remodeling and repair (Alexander the TN-C promoter to various degree, with Fli1 having and Werb, 1991; Tremble et al., 1994). It is therefore the most potent e€ects. Signi®cantly, exogenous expres- formally possible that Ets proteins may be the sion of Fli1 resulted in the increased expression of the conserved developmental regulators of the genes Fli1 and Sp1 activate TN-C promoter F Shirasaki et al 7763 related to extracellular matrix formation. Results of oligo d(T)16. Di€erent amounts of the reverse-transcribed our study provide the basis for the future investigation products (1/10, 1/20 and 1/50 of the total cDNA products) of this intriguing possibility. were ampli®ed. The following primer pairs were used for ampli®cation: TN-C forward primer, CAGCTCCACACTC- CAGGTAC; TN-C reverse primer, CTTTCGCTGG- GCTCTGAAGG: Fli1 forward primer, CCTGGAGGGG- Materials and methods CACAAACGAT; Fli1 reverse primer, GCTTCTAGTAG- TAGCTGCCTAAGTG; GAPDH forward primer, TGA- Cell culture AGGTCGGAGTCAACGGATTTGGT; GAPDH reverse primer, CATGTGGGCCATGAGGTCCACCAC. The pri- Human foreskin ®broblasts were obtained from foreskins of mers were mRNA speci®c in that the recognition sites of the healthy newborns (following institutional approval and upstream and downstream primers resided in separate exons informed consent). Primary explant cultures were established in genomic sequence. The sizes of PCR products for TN-C, as previously described (Tamaki et al., 1995). Drosophila Fli1, and GAPDH were 448, 638 and 980 bp, respectively. Schneider line 2 (SL2) cells were obtained from Dr T Hsu The PCR products were separated by agarose gel electro- (MUSC) and propagated in Schneider medium (Life phoresis and stained with ethidium bromide. Technologies, Inc.) supplemented with 10% heat-treated FCS. Preparation of nuclear extracts Plasmid constructions Nuclear extracts were prepared according to Andrews and Faller (1991) with minor modi®cations as previously pTN-I-CAT construct was kindly provided by Dr R Gherzi described (Tamaki et al., 1995). In vitro transcribed and (Gherzi et al., 1995a). The deletion constructs were generated translated human Fli1 protein was prepared using TNT by PCR using pTN-I-CAT as a template. Substitution Coupled Reticulocyte Lysate System (Promega). Protein mutations were generated using Quick Change site-directed production was con®rmed by immunoblotting assay using mutagenesis kit (Stratagene) and con®rmed by sequencing. anti-Fli1 antibody (Zhang et al., 1995b). pSG5Fli1, pSG5Ets1, and pSG5Ets2 has been described previously (Watson et al., 1992; Seth et al., 1993). Plasmids used in transfections of Drosophila cells were generated as Electrophoretic mobility shift assay (EMSA) follows: Fli1 cDNA was excised by SspI and BamHI from Oligonucleotides used as probes and competitors were pSG5Fli1. After ®lling-in with Klenow polymerase, the 1386- purchased from Medical University of South Carolina Core bp fragment was subcloned at blunt-ended XbaI site of pPac- Facility. Radioactive double-stranded probes for electro- pl (provided by Dr T Hsu, MUSC). The 1951-bp GABPa phoretic mobility shift assays were 5'-end labeled using T4 fragment was excised by XbaI from GABPa/Bluescript polynucleotide kinase and [g-32P]ATP. DNA mobility shift (kindly provided by SL McKnight, Tularik Inc, South San assay was performed as described previously (Tamaki et al., Francisco, CA, USA). After ®lling-in, this fragment was 1995). Brie¯y, 5 mg of nuclear extracts or 2 mlofin vitro subcloned at blunt-ended XbaI site of pPac-pl. The 2663-bp translated proteins were incubated for 30 min on ice in 24 ml GABPb1 fragment was excised with BamHI and KpnI from binding bu€er (10 mM HEPES-KOH, pH 7.9, 50 mM NaCl, the GABPb1/Bluescript (kindly provided by SL McKnight) 1mM MgCl , 0.5 mM EDTA, 1 mM dithiothreitol, 0.7 mM and subcloned into the BamHI/KpnI site of pPac-pl. The 2 phenylmethyl-sulfonyl ¯uoride, 10% glycerol, 10 mg/ml expressed proteins were veri®ed by immunoblot analysis aprotinin, 2 mg/ml leupeptin, and 2 mg/ml pepstatin) contain- using speci®c antibodies. pPacSp1 and pPacSp3 cloned into ing 50 000 c.p.m. labeled probe and 0.2 ± 1 mg of poly(dI-dC)- pPacSp0 have been described previously (Ihn and Troja- poly(di-dC). In some assays, double-stranded competitors nowska, 1997). Plasmids used in transient transfection assays (200-fold molar excess) or antibodies were preincubated with were puri®ed by double cesium chloride gradients. nuclear extracts 30 min prior to addition of radioactive oligonucleotide probe. Polyclonal anti-Sp1 and anti-Sp3 Transient transfection and chloramphenicol acetyltransferase antibodies were purchased from Santa Cruz Biotechnology. assays Monoclonal anti-Ets1, anti-Ets2 and anti-Fli1 antibodies were previously characterized (Zhang et al., 1995b; Koizumi Human foreskin ®broblasts were transfected as described et al., 1990; Fujiwara et al., 1990; Hodge et al., 1996). previously by the calcium phosphate technique (Tamaki et Polyclonal anti-GABPa and anti-GABPb1 were kindly al., 1995) with 20 mg of various TN-C promoter-chloram- provided by Dr SL McKnight (Tularik Inc). Separation of phenicol acetyltransferase constructs and various amounts of free radiolabeled DNA from DNA-protein complexes was Ets expression vectors. PSV-b-galactosidase control vector carried out on a 5% nondenaturing polyacrylamide gel. (Promega) was co-transfected to normalize for transfection Electrophoresis was performed in 0.256Tris-borate electro- eciency. Drosophila Schneider cells were transfected as phoresis bu€er at 250 V at 48C. The gels were dried and previously described (Ihn and Trojanowska, 1997). Transfec- exposed to X-ray ®lm at 7808C. tions were performed in duplicates and repeated at least four times using two di€erent plasmid preparations. The Mann- Whitney U-test was used to determine statistical signi®cance.

Acknowledgments Reverse transcriptase PCR analysis We thank Dr R Gherzi for providing pTN-I-CAT con- RT ± PCR was performed using GeneAmp RNA PCR Core struct and Dr SL McKnight for providing GABPa and b1 kit (Perkin-Elmer) according to the manufacturer's protocol. cDNAs and antibodies. We also thank Dr T Hsu for Brie¯y, total RNA was harvested (Chomczynski and Sacchi, critically reading the manuscript and many helpful sugges- 1987) from either Fli1 or pSG5 transfected dermal ®bro- tions. This work was supported by National Institutes of blasts. One mg of total RNA was reverse-transcribed using Health grant AR42334 and by RGK Foundation. Fli1 and Sp1 activate TN-C promoter F Shirasaki et al 7764 References

Alexander CM and Werb Z. (1991). Cell Biology of KoizumiS,FisherRJ,FujiwaraS,JorcykC,BhatNK,Seth Extracellular Matrix. Hay ED (ed). Plenum Press: New A and Papas TS. (1990). Oncogene, 5, 675 ± 681. York, pp. 255 ± 302. Kusagawa H, Onoda K, Namikawa S, Yada I, Okada A, Andrews NC and Faller DV. (1991). Nucleic Acids Res., 19, Yoshida T and Sakakura T. (1998). Br.J.Cancer,77, 98 ± 2499. 102. Batchelor AH, Piper DE, de la Brousse FC, McKnight SL Logan SK, Garabedian MJ, Campbell CE and Werb Z. and Wolberger C. (1998). Science, 279, 1037 ± 1041. (1996). J. Biol. Chem., 271, 774 ± 782. ButticeÁ G, Duterque-Coquillaud M, Basuyaux JP, CarreÁ re S, Mackie EJ. (1997). Int. J. Biochem. Cell Biol., 29, 1133 ± Kurkinen M and Ste helin D. (1996). Oncogene, 13, 2297 ± 1137. 2306. Mager AM, Grapin-Botton A, Ladjali K, Meyer D, Wol€ ButticeÁ G and Kurkinen M. (1993). J. Biol. Chem., 268, CM, Stiegler P, Bonnin MA and Remy P. (1998). Int. J. 7196 ± 7204. Dev. Biol., 42, 561 ± 572. Chiquet-Ehrismann R, Hagios C and Schenk S. (1995). Makhluf HA, Stepniakowska J, Ho€man S, Smith E, LeRoy Bioessays, 17, 873 ± 878. EC and Trojanowska M. (1996). J. Invest. Dermatol., 107, Chomczynski P and Sacchi N. (1987). Anal Biochem., 162, 856 ± 859. 156 ± 159. McNagny KM, Sieweke MH, DoÈ derlein G, Graf T and Copertino DW, Edelman GM and Jones FS. (1997). Proc. Nerlov C. (1998). EMBO J., 17, 3669 ± 3680 Natl. Acac. Sci. USA, 94, 1846 ± 1851. Mettouchi A, Cabon F, Montreau N, Dejong V, Vernier P, Copertino DW, Jenkinson S, Jones FS and Edelmann GM. Gherzi R, Mercier G and Binetruy B. (1997). Mol. Cell. (1995). Proc. Natl. Acad. Sci. USA, 92, 2131 ± 2135. Biol., 17, 3202 ± 3209. Courey AJ and Tjian R. (1988). Cell, 55, 887 ± 896. MeyerD,StieglerP,HindelangC,MagerA-MandRemyP. Crossin KL. (1996). J. Cell. Biochem., 61, 592 ± 598. (1995). Int. J. Dev. Biol., 39, 909 ± 919. Edwards DR, Rocheleau H, Sharma RR, Wills AJ, Cowie A, Pascal E and Tjian R. (1991). Genes. Dev., 5, 1646 ± 1656. Hassell JA and Heath JK. (1992). Biochim. Biophys. Acta, Rao VN, Ohno T, Prasad DD, Bhattacharya G and Reddy 1171, 41 ± 55. ES. (1993). Oncogene, 8, 2167 ± 2173. Erickson HP and Bourdon M. (1989). Annu. Rev. Cell. Biol., Rettig WJ, Erickson HP, Albino AP and Garin-Chesa P. 5, 71 ± 92. (1994). J. Cell Sci., 107, 487 ± 497. Fujiwara S, Koizumi S, Fisher RJ, Bhat NK and Papas TS. Sakai T, Kawakatsu H, Hirota N, Yokoyama T, Sakakura T (1990). Hybridoma, 9, 559 ± 571. and Saito M. (1993). Br. J. Cancer, 67, 1058 ± 1064. Ge gonne A, Bosselut R, Bailly R-A and Ghysdael J. (1993). Seth A, Robinson L, Thompson DM, Watson DK and Papas EMBO J., 12, 1169 ± 1178. TS. (1993). Oncogene, 8, 1783 ± 1790. Gherzi R, Briata P, Boncinnelli E, Ponassi M, QuerzeÁ G, Viti Sharrocks AD, Brown AL, Ling Y and Yates PR. (1997). Int. F, Corte G and Zardi L. (1997). DNA Cell Biol., 16, 559 ± J. Biochem. Cell Biol., 29, 1371 ± 1387. 567. Shoji T, Kamiya T, Tsubura A, Hatano T, Sakakura T, Gherzi R, Carnemolla B, Siri A, Ponassi M, Balza E and Tamamoto M and Morii S. (1992). Virchows Arch. A Zardi L. (1995a). J. Biol. Chem., 270, 3429 ± 3434. Pathol. Anat. Histopathol., 421, 53 ± 56. Gherzi R, Ponassi M, Gaggero B and Zardi L. (1995b). Tamaki T, Ohnishi K, Hartl C, LeRoy EC and Trojanowska FEBS letts, 369, 335 ± 339. M. (1995). J. Biol. Chem., 270, 4299 ± 4304. Ghysdael J and Boureux A. (1997). Oncogenes as Transcrip- Thompson CC, Brown TA and McKnight SL. (1991). tional Regulators. Yaniv M and Ghysdael J. (ed). Science, 253, 762 ± 768. BirkhaÈ user Verlag: Basel, pp. 29 ± 87. Tremble P, Chiquet-Ehrismann R and Werb Z. (1994). Mol. Gilles F, Raes MB, Stehelin D, Vandenbunder B and Fafeur Biol. Cell, 5, 439 ± 453. V. (1996). Exp. Cell Res., 222, 370 ± 378. Verrijzer CP and Tjian R. (1996). Trends Biochem. Sci., 21, Gory S, Dalmon J, Prandini M-H, Kortulewski T, de 338 ± 342. Launoit Y and Huber P. (1998). J. Biol. Chem., 273, Wasylyk C, Gutman A, Nicholson R and Wasylyk B. (1991). 6750 ± 6755. EMBO J., 10, 1127 ± 1134. Graves BJ and Petersen JM. (1998). Adv. Cancer Res., 75, 1± Watson DK, Smyth FE, Thompson DM, Cheng JQ, Testa 55. JR, Papas TS and Seth A. (1992). Cell Growth Di€er., 3, Gutman A and Wasylyk B. (1990). EMBO J., 9, 2241 ± 2246. 705 ± 713. Hagen G, Muller S, Beato M and Suske G. (1994). EMBO J., Westermarck J, Seth A and KaÈ haÈ ri V-M. (1997). Oncogene, 13, 3843 ± 3851. 14, 2651 ± 2660. Halle J-P, Haus-Seu€ert P, Woltering C, Stelzer G and Yamada S, Ichida T, Matsuda Y, Miyazaki Y, Hatano T, Meisterernst M. (1997). Mol. Biol. Cell, 17, 4220 ± 4229. Hata K, Asakura H, Hirota N, Geerts A and Wisse E. Hanamura N, Yoshida T, Matsumoto E, Kawarada Y and (1992). Liver, 12, 10 ± 16. Sakakura T. (1997). Int. J. Cancer, 73, 10 ± 15. Yoshida T, Ishihara A, Hirokawa Y, Kusakabe M and Hodge DR, Robinson L, Watson D, Lautenberger J, Zhang Sakakura T. (1995). Cancer letters, 90, 65 ± 73. XK, Venanzoni M and Seth A. (1996). Oncogene, 12, 11 ± Yoshida T, Matsumoto E, Hanamura N, Kalembeyi I, 18. Katsuta K, Ishihara A and Sakakura T. (1997). J. Pathol., Ihn H and Trojanowska M. (1997). Nucleic Acids Res., 25, 182, 421 ± 428. 3712 ± 3717. ZhangL,EddyA,TengY-T,FritzlerM,KluppelM,MeletF Jones FS, Chalepakis G, Gruss P and Edelman GM. (1992). and Bernstein A. (1995a). Mol. Cell. Biol., 15, 6961 ± 6970. Proc. Natl. Acad. Sci. USA, 89, 2091 ± 2095. Zhang XK, Papas TS, Bhat NK and Watson DK. (1995b). Jones FS, Crossin KL, Cunningham BA and Edelman GM. Hybridoma, 14, 563 ± 569. (1990). Proc. Natl. Acad. Sci. USA, 87, 6497 ± 6501. Klemsz MJ, Maki RA, Papayannopoulou T, Moore J and Hromas R. (1993). J. Biol. Chem., 268, 5769 ± 5773.