Transcriptional activation of the matrix metalloproteinase-9 gene in an. H-ras and v-myc transformed rat embryo cell line. Bruce P Himelstein1, Edward J Lee1, ...
Oncogene (1997) 14, 1995 ± 1998 1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00
SHORT REPORT
Transcriptional activation of the matrix metalloproteinase-9 gene in an H-ras and v-myc transformed rat embryo cell line Bruce P Himelstein1, Edward J Lee1, Hiroshi Sato2, Motoharu Seiki2 and Ruth J Muschel3 1
Division of Oncology, Abramson 902C, The Children's Hospital of Philadelphia, 324 S. 34th Street, Philadelphia, Pennsylvania 19104, USA; 2Department of Molecular Virology and Oncology, Cancer Research Institute, Kanazawa University, Takara-machi, Kanazawa 920, Japan; 3Department of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, John Morgan 269, Philadelphia, Pennsylvania 19104, USA
The 92 kd type IV collagenase/gelatinase (MMP-9) is important in mediating basement membrane and extracellular matrix degradation in metastasis. Because MMP-9 is made in tumor cells, but not in quiescent normal cells, we wished to identify the transcriptional elements responsible for its synthesis in tumor cells. We chose to characterize transcriptional regulation of the MMP-9 gene in a highly metastatic H-ras and v-myc transformed rat embryo cell line which overexpresses MMP-9. Using transient transfection of reporter gene constructs containing either 5'-deleted or mutated MMP9 promoter fragments, as well as electrophoretic mobility shift assays, we have demonstrated that multiple transcription factor consensus binding motifs in the promoter, including those for NFkB, SP-1, Ets, AP-1, and a retinoblastoma binding element, participate in transcriptional regulation of MMP-9 expression in this cell line. Also, deletion of an alternating purinepyrimidine tract in the downstream promoter was found to decrease transcriptional activity, suggesting that promoter conformation may be important in MMP-9 regulation. Thus multiple pathways leading to activation of NFkB, SP-1, Ets, AP-1, and retinoblastoma binding factors in tumor cells all may contribute to MMP-9 transcription and hence to metastasis. Keywords: MMP-9; ras; transcription; purine-pyrimidine repeat
Extracellular matrix and basement membrane degradation are essential steps in tumor metastasis (Goldberg and Eisen, 1991; Liotta et al., 1991). Matrix metalloproteinases (MMPs) are a family of zincdependent enzymes which have been implicated in this process (Birkedal-Hansen et al., 1993). The 92 kd type IV collagenase/gelatinase (gelatinase B or MMP9), for example, has been found to be expressed in a wide variety of human malignancies, including brain tumors (Rao et al., 1993), pancreatic cancer (Gress et al., 1995), bladder cancer (Davies et al., 1993), uveal melanoma (Cottam et al., 1992), basal cell and squamous cell carcinomas of the skin (Karelina et al., 1992; Pyke et al., 1992), squamous cell carcinoma of the lung (Canete-Soler et al., 1994; Urbanski et al., 1992), colon adenocarcinomas (Pyke et al., 1993; Correspondence: BP Himelstein Received 5 August 1996; revised 23 December 1996; accepted 6 January 1997
Zucker et al., 1993), malignant lymphomas (Kossakowska et al., 1993), and breast cancer (Himelstein et al., 1995; Zucker et al., 1993). In several instances tissue expression levels of MMPs have been correlated with stage, tumor progression, or survival. For example, histologic grade of malignant lymphomas or aggressiveness of prostatic adenocarcinomas have been found to be correlated with MMP-9 levels (Hamdy et al., 1994; Kossakowska et al., 1992). In vitro, tumor cell overexpression of MMP-9 has been correlated with the metastatic phenotype. Previous work in our laboratory demonstrated a correlation between MMP-9 expression and the metastatic phenotype in a series of oncogene transformed rat embryo cell lines (Bernhard et al., 1990). Cell line 2.10.10, an H-ras and v-myc transformed rat embryo cell line, was highly metastatic and overexpressed MMP-9 protein and mRNA; ribozymemediated degradation of MMP-9 mRNA in 2.10.10 cells abrogated their metastatic capability (Hua and Muschel, 1996). These data con®rm the importance of MMP-9 expression in the manifestation of the metastatic phenotype. The MMP-9 promoter has several transcription factor binding motifs, as shown schematically in Figure 1, including three TPA response elements or AP-1 sites (TRE-1, -2 and -3), an NFkB site, an Ets site, an SP-1 site, and a GT box or retinoblastoma binding element (RBE) (Fini et al., 1994; Huhtala et al., 1990; Masure et al., 1993; Sato et al., 1993; Sato and Seiki, 1993; Xia et al., 1996). Also, there is a 42 nucleotide CA (purine-pyrimidine) repeat whose function in the promoter is unknown. The proximal 670 bp of the MMP-9 promoter appears to contain sucient response elements for transcriptional activation. Coordinate activation of the NFkB, SP-1, and most proximal TRE-1 were required for promoter activation by tumor necrosis factor-a (TNF-a) or by phorbol ester (TPA), while the RBE and most proximal TRE-1 appeared to be sucient for transcriptional activation by the src oncogene (Sato et al., 1993; Sato and Seiki, 1993). Gum et al. showed that several transcription factor binding sites, including the adjoining upstream Ets and TRE-2 sites, were necessary for MMP-9 upregulation mediated by the ras oncogene (Gum et al., 1996). Given the clear association between MMP-9 expression and metastasis in our rat sarcoma system, we chose to characterize the cis-acting elements in the MMP-9 promoter which are responsible for overexpression in cell line 2.10.10.
Transcriptional activation of MMP-9 BP Himelstein et al
1996
Figure 1 Schematic diagram of potential cis-acting elements in the MMP-9 promoter
First, we sought to de®ne the minimal promoter region necessary for full transcriptional activation in 2.10.10 cells. We transiently transfected 2.10.10 cells with plasmid constructs linking portions of the MMP-9 promoter to the chloramphenicol acetyltransferase (CAT) reporter gene and subsequently measured CAT activity. CAT activity in 2.10.10 cells transfected with a plasmid bearing the full-length human MMP-9 promoter cDNA, 2.2 CAT, was found to be comparable (1.3fold dierence) to those cells transfected with a plasmid bearing the proximal 670 bp promoter fragment, 670 CAT, suggesting that this proximal fragment contains the essential transcriptional machinery, and that the upstream TRE-3 (71651 to 71645) was not required for constitutive MMP-9 expression in 2.10.10 cells (data not shown). We next carried out 5' deletion analysis to further de®ne the minimal MMP-9 promoter needed for constitutive expression of MMP-9 in 2.10.10 cells. We transiently transfected these cells with plasmid constructs linking the CAT reporter gene to 5' deleted fragments of the MMP-9 promoter. As shown in Figure 2, construct 599 CAT, which transects the upstream NFkB consensus binding site, reduced reporter activity fourfold. Further deletion of the promoter 5' to 7531, which deletes the NFkB and SP-1 sites, Ets site and transects TRE-2, greatly reduced reporter activity. Increased activity was seen using promoters deleted 5' to 7241, suggesting the presence of an inhibitory sequence in this region. Deletion of all but 90 bp of proximal promoter, containing a TRE-1 site, a TATA box, and the RBE, reduced activity 18-fold compared to 670 CAT. Activity of the 773 fragment, containing only the RBE and TATA box, was negligible. These results together suggested that the region of bp 7670 to 7478 was a stimulatory region of the promoter required for MMP-9 expression in 2.10.10 cells, as was the region 3' to 7144. They also suggest that activation of src pathways does not contribute to MMP-9 transcription in this setting, since these elements are sucient to lead to promoter activation by src (Sato et al., 1993). An intervening inhibitory sequence was also suggested between these two stimulatory regions. These results prompted us to test the eects of introducing mutations into known transcription factor
Figure 2 The 7670 to 7531 MMP-9 promoter region contains elements essential for MMP-9 transcriptional activation in 2.10.10 cells. 2.10.10 cells were transiently transfected using calcium phosphate coprecipitation with 7.5 mg CAT reporter plasmids containing 5' deleted MMP-9 promoter fragments. Transfected cells were lysed and CAT activity was measured using a liquid scintillation assay (Promega; Madison, WI). Results represent c.p.m. 3H-n-butyrl-chloramphenicol, proportional to CAT activity, and are the average of triplicate samples. Error bars indicate standard deviations. Results are corrected by subtraction of background CAT activity obtained from parallel transfections with a promoterless CAT vector
consensus motifs in the MMP-9 promoter to determine the precise elements required for transcription in tumor cells. First, for these experiments a new parent promoter fragment (670 Xba) was generated following the methods of Gustin and Burk (1993), in which we removed the ATG start site from the MMP-9 coding sequence found in the original 670 CAT construct to decrease the possiblity of `steal' from the ATG start site in the CAT reporter gene. Deletion of this ATG site resulted in a 2.8-fold increase in CAT activity (670 Xba versus 670 CAT) in 2.10.10 cells (data not shown). We next made linker insertion mutations in the recognition sequences for NFkB (7600 to 7591), SP-1 (7565 to 7556), Ets (7543 to 7534), TRE-2 or TRE-1 (7533 to 7524 and 780 to 771), and the RBE (755 to 746). A construct in which the CA repeat sequence (7131 to 790) was deleted was also made. The consensuses in the 670 Xba parent promoter fragment were replaced by a 10 bp BglII cassette using a variation on PCR-mediated splicing by overlap extension (Gustin and Burk, 1993). As shown in Figure 3, mutation of the NFkB site reduced promoter activity approximately fourfold, and mutation of either the SP-1 or Ets sites almost completely abrogated promoter activity. Mutation of either TRE-1 or TRE-2 also reduced promoter function. Similar to the data previously shown for both src and ras activation of the MMP-9 gene, disruption of the RBE abrogated promoter function (Gum et al., 1996; Sato et al., 1993). Finally, replacement of the CA dinucleotide repeat sequence in the downstream promoter also markedly reduced reporter expression. Electrophoretic mobility shifts were performed to determine whether DNA binding activity could be found for the upstream MMP-9 promoter consensuses
Transcriptional activation of MMP-9 BP Himelstein et al
1997
a
Figure 3 Multiple transcription factor recognition sequences and the CA repeat sequence are required for full activation of the MMP-9 gene in 2.10.10 cells. 2.10.10 cells were transiently transfected with 7.5 mg CAT reporter plasmid containing 5' MMP-9 promoter fragments containing mutations in the transcription factor recognition sites shown, or containing a deletion of the CA box, as described in the legend to Figure 2
shown to be required for transcriptional activity by deletion and mutation analysis. Representative gel shifts are shown in Figure 4, which demonstrate that constitutive nuclear binding activities speci®c for the NFkB or Ets consensuses in the MMP-9 promoter were present in 2.10.10 cells. Supershift analysis revealed decreased band intensity in the presence of antibodies speci®c for Ets-1 or for Ets-1/Ets-2, suggesting that the DNA binding activity in 2.10.10 cells is related to Ets-1. Our results in ras-transformed rat embryo cells extend previous studies suggesting that the MMP-9 promoter contains several regulatory regions (Fini et al., 1994; Sato et al., 1993; Sato and Seiki, 1993). There are two stimulatory regions, the upstream region containing consensus sites for binding by NFkB, SP-1, Ets and AP-1, and the downstream region containing another AP-1 site, the TATA box, and the RBE. Sato and Seiki demonstrated that the upstream (NFkB and SP-1) and downstream (AP-1) stimulatory regions cooperated in activation due to TPA or TNF-a, while the downstream AP-1 and RBE acted synergistically in response to the src oncogene; activity of the upstream regulatory unit was not required for src activation, and activity of the RBE was not necessary for TPA activation (Sato et al., 1993; Sato and Seiki, 1993). Gum et al. (1996) recently published a study of the MMP-9 promoter response to the ras oncogene; deletion or mutation of any element of the upstream regulatory region (NFkB, AP-1, Ets, or SP-1) or downstream regulatory region (AP-1 or RBE), signi®cantly impaired promoter activation. Similar to the results of Gum et al. (1996) described above, mutation of either the upstream AP-1 site or adjacent Ets site was found to reduce promoter activation in 2.10.10 cells. This paired site is similar to the TPA- and oncogene-responsive unit (TORU)
1
2
3
4
b
1
2
3
4
5
6
Figure 4 2.10.10 cells contain nuclear protein(s) which bind to the MMP-9 promoter NFkB or Ets consensus motifs. (a) 15 mg of nuclear extract (Schreiber et al., 1989) from 2.10.10 cells was combined with *50 fmol 32P-end-labeled NFkB oligonucleotide (5'-CCAGTGGAATTCCCCAGCCT-3'), 2 mg poly(dIdC), 1 mg salmon sperm DNA, and 3 mg BSA (lane 2), plus a 400-fold molar excess of cold NFkB oligonucleotide (lane 3), or plus a 400-fold molar excess of mutant NFkB oligonucleotide (lane 4, 5'CCAGTTTAATTCCCCAGCCT-3') on ice. Reaction products were separated on a nondenaturing 5% polyacrylamide gel for 3 ± 4 h in 0.56TGE buer at 48C and exposed to autoradiography ®lm. Lane 1 is a probe only control. (b) 15 mg of nuclear extract from 2.10.10 was combined with *50 fmol 32P-end-labeled Ets oligonucleotide (5'-GGAGAGGAAGCT-3'), 2 mg poly(dIdC), 1 mg salmon sperm DNA, and 3 mg BSA (lane 2), plus a 400fold molar excess of cold Ets oligonucleotide (lane 3) or plus a 400-fold molar excess of mutant Ets oligonucleotide (lane 4, 5'GGAGACAAAGCT-3') at room temperature. 2 mg of Ets-1 speci®c antibody was added to the reaction shown in lane 5, and 2 mg of Ets-1/Ets-2 speci®c antibody was added to the reaction shown in lane 6 (Santa Cruz Biotechnology; Santa Cruz, CA). Reaction products were separated on a nondenaturing 5% polyacrylamide gel for 3 ± 4 h in 0.56TGE buer at 48C and exposed to autoradiography ®lm. Lane 1 is a probe only control
found in the interstitial collagenase promoter (Gutman and Wasylyk, 1990) and in the urokinase plasminogen activator promoter (Nerlov et al., 1991; Rorth et al., 1990). In both cases, the PEA3 and AP-1 components were found to act synergistically in promoter activation. In the MMP-9 promoter, this TORU is completely conserved in the mouse, human, and rabbit genes (Fini et al., 1994). Fini et al. found that loss of the TORU signi®cantly reduced basal and TPAstimulated promoter activation in corneal ®broblasts, less so in corneal epithelial cells, and Gum et al. (1996) also showed that mutation of either the PEA3 or AP-1 elements in the TORU abrogated the promoter response to ras. They also showed that overexpression of PEA3 could drive the MMP-9 promoter, a response which could be turned o by mutation of the PEA3 consensus. Higashino et al. also recently described activation of the MMP-9 promoter by the ets-related transcription factor E1A-F (Higashino et al., 1995). Our data con®rm that Ets is transcriptionally active in the MMP-9 promoter.
Transcriptional activation of MMP-9 BP Himelstein et al
1998
Transcriptional activity of the MMP-9 promoter was reduced with deletion of bp 7144 to 790, the region containing the alternating purine-pyrimidine (APP) tract, suggesting the presence of a transcriptional activation domain in this region. This APP tract is found in both the human and mouse MMP-9 promoter, but is absent in the recently cloned rabbit promoter (Fini et al., 1994). Similar APP tracts have been described in both the urokinase plasminogen activator promoter as well as the rat MMP-2 promoter, yet their function is unknown (Harendza et al., 1995; Verde et al., 1988). APP sites have been found in the 5'-¯anking regions, introns, and 3'-untranslated regions of many genes. Long APP tracts are known to alter DNA conformation with a tendency towards formation of Z-DNA (Tripathi and Brahmachari, 1991). APP tracts have been found to modulate transcriptional activity. For example, Wu et al. (1994) showed by deletion analysis that a 48 bp CA repeat in the human 85 kd cytosolic phospholipase A2 gene may have an inhibitory eect on transcription. APP sequences in the mouse alcohol dehydrogenase gene Adh-1 were found to be a tissue-speci®c positive
regulatory element (Yu and Edenberg, 1995), and a shorter repeat sequence was found to transactivate the rat liver glycogen phosphorylase gene (Herrick et al., 1993). Our data provide novel evidence that the APP tract may be a transcriptional activating domain in the MMP-9 gene. In sum, we have found that multiple transcription factors as well as the conformational state of the promoter may be important in regulating overexpression of MMP-9 in a ras and myc transformed metastatic rat embryo cell line. These results con®rm and expand upon previous studies exploring the regulation of metalloproteinase genes, and suggest potential methods for blocking gene expression at the transcriptional level.
Acknowledgements The authors wish to thank Dan Dilks for expert technical assistance. This work was supported by NIH CA63312 to BPH and NIH CA46830 and NIH CA64805 to RJM.
References Bernard E, Muschel R and Hughes E (1990). Cancer Res., 50, 3872 ± 3877. Birkedal-Hansen H, Moore W, Bodden M, Windsor L, Birkedal-Hansen B, DeCarlo A and Engler J. (1993). Crit. Rev. Oral. Biol. Med., 4, 197 ± 250. Canete-Soler R, Litsky L, Lubensky I and Muschel R. (1994). Am. J. Pathol., 144, 518 ± 527. Cottam D, Rennie I, Woods K, Parsons M, Bunning R and Rees R. (1992). Invest. Opthal. Vis. Sci., 33, 1923 ± 1927. Davies B, Waxman J, Wasan H, Abel P, Williams G, Krausz T, Neal D, Thomas D, Hamby A and Balkwill F. (1993). Cancer Res., 53, 5365 ± 5369. Fini M, Bartlett J, Matsubara M, Rinehart W, Mody M, Girard M and Rainville M. (1994). J. Biol. Chem., 269, 28620 ± 28628. Goldberg G and Eisen A. (1991). In: Regulatory mechanisms in breast cancer. Lippmann M and Dickson R. (eds.) Kluwer Academic Publishers: Boston. pp. 421 ± 440. Gress TM, Muller PF, Lerch MM, Friess H, Buchler M and Adler G. (1995). Int. J. Cancer, 62, 407 ± 413. Gum R, Lengyel E, Juarez J, JH C, Sato H, Seiki M and Boyd D. (1996). J. Biol. Chem., 271, 10672 ± 10680. Gustin K and Burk R. (1993). BioTechniques, 14, 22 ± 23. Gutman A and Wasylyk B. (1990). EMBO J., 9, 2241 ± 2246. Hamdy F, Fadlon E, Cottam D, Lawry J, Thurrell W, Silcocks P, Anderson J, Williams J and Rees R. (1994). Br. J. Cancer, 69, 177 ± 182. Harendza S, Pollock A, Mertens P and Lovett D. (1995). J. Biol. Chem., 270, 18786 ± 18796. Herrick K, Gorin F, Park E and Tait R. (1993). Gene, 126, 203 ± 211. Higashino F, Yoshida K, Noumi T, Seiki M and Fujinaga K. (1995). Oncogene, 10, 1461 ± 1463. Himelstein BP, Canete-Soler R, Bernhard EJ, Dilks DW and Muschel RJ. (1995). Invasion Metast., 1 ± 6, 246 ± 258. Hua J and Muschel RJ. (1996). Cancer Res., 56, 5279 ± 5284. Huhtala P, Chow L and Tryggvason K. (1990). J. Biol. Chem., 265, 11077 ± 11082. Karelina T, Hruza G, Goldberg G and Eisen A. (1992). J. Invest. Dermatol., 100, 159 ± 165.
Kossakowska A, Urbanski S, Huchcroft S and Edwards D. (1992). Oncol. Res., 4, 233 ± 240. Kossakowska A, Urbanski S, Watson A, Hayden L and Edwards D. (1993). Oncology Res., 5, 19 ± 28. Liotta L, Steeg P and Stetler-Stevenson W. (1991). Cell, 64, 327 ± 336. Masure S, Nys G, Fiten P, Van DJ and Opdenakker G. (1993). Eur. J. Biochem., 218, 129 ± 141. Nerlov C, Rorth P, Blasi F and Johnsen M. (1991). Oncogene, 6, 1538 ± 1592. Pyke C, Ralfkiaer E, Huhtala P, Hurskainen T, Dano K and Tryggvason K. (1992). Cancer Res., 52, 1336 ± 1341. Pyke C, Ralfkiaer E, Tryggvason K and Dano K. (1993). Am. J. Pathol, 142, 359 ± 365. Rao J, Steck P, Mohanam S, Stetler-Stevenson W, Liotta L and Sawaya R. (1993). Cancer Res., 53, 2208 ± 2211. Rorth P, Nerlov C, Blasi F and Johnsen M. (1990). Nucleic Acid Res., 18, 5009 ± 5017. Sato H, Kita M and Seiki M. (1993). J. Biol. Chem., 268, 23460 ± 23468. Sato H and Seiki M. (1993). Oncogene, 8, 395 ± 405. Schreiber E, Matthias P, Muller MM and Schaner W. (1989). Nucleic Acids Res., 17, 6419. Tripathi J and Brahmachari S. (1991). J. Biomol. Struct. Dyn., 9, 387 ± 397. Urbanski S, Edwards D, Maitland A, Leco K, Watson A and Kossakowska A. (1992). Br. J. Cancer, 66, 1188 ± 1192. Verde P, Boast S, Franze A, Robbiati F and Blasi F. (1988). Nucleic Acids Res., 16, 10699 ± 10716. Wu T, Ikezono T, Angus C and Shelhamer J. (1994). Nucleic Acids Res., 22, 5093 ± 5098. Xia Y, Garcia G, Chen S, Wilson C and Feng L. (1996). FEBS Lett., 382, 285 ± 288. Yu M and Edenberg H. (1995). Arch. Biochem. Biophys., 316, 407 ± 412. Zucker S, Lysik R, Zarrabi M and Moll U. (1993). Cancer Res., 53, 140 ± 146.
