Title: Overexpression of Cyclooxygenase-2 Is Sufficient to Induce Tumorigenesis in Transgenic Mice
Abstract: The cyclooxygenase (COX)-2 gene encodes an inducible prostaglandin synthase enzyme that is overexpressed in adenocarcinomas and other tumors. Deletion of the murine Cox-2 gene inMin mice reduced the incidence of intestinal tumors, suggesting that it is required for tumorigenesis. However, it is not known if overexpression of Cox-2 is sufficient to induce tumorigenic transformation. We have derived transgenic mice that overexpress the human COX-2 gene in the mammary glands using the murine mammary tumor virus promoter. The human Cox-2 mRNA and protein are expressed in mammary glands of female transgenic mice and were strongly induced during pregnancy and lactation. Female virgin Cox-2 transgenic mice showed precocious lobuloalveolar differentiation and enhanced expression of the β-casein gene, which was inhibited by the Cox inhibitor indomethacin. Mammary gland involution was delayed in Cox-2 transgenic mice with a decrease in apoptotic index of mammary epithelial cells. Multiparous but not virgin females exhibited a greatly exaggerated incidence of focal mammary gland hyperplasia, dysplasia, and transformation into metastatic tumors. Cox-2-induced tumor tissue expressed reduced levels of the proapoptotic proteins Bax and Bcl-xL and an increase in the anti-apoptotic protein Bcl-2, suggesting that decreased apoptosis of mammary epithelial cells contributes to tumorigenesis. These data indicate that enhanced Cox-2 expression is sufficient to induce mammary gland tumorigenesis. Therefore, inhibition of Cox-2 may represent a mechanism-based chemopreventive approach for carcinogenesis. The cyclooxygenase (COX)-2 gene encodes an inducible prostaglandin synthase enzyme that is overexpressed in adenocarcinomas and other tumors. Deletion of the murine Cox-2 gene inMin mice reduced the incidence of intestinal tumors, suggesting that it is required for tumorigenesis. However, it is not known if overexpression of Cox-2 is sufficient to induce tumorigenic transformation. We have derived transgenic mice that overexpress the human COX-2 gene in the mammary glands using the murine mammary tumor virus promoter. The human Cox-2 mRNA and protein are expressed in mammary glands of female transgenic mice and were strongly induced during pregnancy and lactation. Female virgin Cox-2 transgenic mice showed precocious lobuloalveolar differentiation and enhanced expression of the β-casein gene, which was inhibited by the Cox inhibitor indomethacin. Mammary gland involution was delayed in Cox-2 transgenic mice with a decrease in apoptotic index of mammary epithelial cells. Multiparous but not virgin females exhibited a greatly exaggerated incidence of focal mammary gland hyperplasia, dysplasia, and transformation into metastatic tumors. Cox-2-induced tumor tissue expressed reduced levels of the proapoptotic proteins Bax and Bcl-xL and an increase in the anti-apoptotic protein Bcl-2, suggesting that decreased apoptosis of mammary epithelial cells contributes to tumorigenesis. These data indicate that enhanced Cox-2 expression is sufficient to induce mammary gland tumorigenesis. Therefore, inhibition of Cox-2 may represent a mechanism-based chemopreventive approach for carcinogenesis. cyclooxygenase non-steroidal anti-inflammatory drug(s) peroxisomal proliferator-activated receptor murine mammary tumor virus prostaglandin The cyclooxygenase (Cox)1 enzymes, Cox-1 and -2, catalyze the rate-limiting steps in the biosynthesis of prostaglandins and thromboxanes, collectively known as prostanoids. Regulation of expression of Cox-1 and -2 is distinct; Cox-1 is expressed ubiquitously, whereas Cox-2 is induced as an immediate-early gene in most cells (1Smith W.L. DeWitt D.L. Garavito R.M. Annu. Rev. Biochem. 2000; 69: 145-182Crossref PubMed Scopus (2477) Google Scholar, 2Vane J.R. Bakhle Y.S. Botting R.M. Annu. Rev. Pharmacol. Toxicol. 1998; 38: 97-120Crossref PubMed Scopus (2620) Google Scholar, 3Hla T. Bishop-Bailey D. Liu C.H. Schaefers H.J. Trifan O.C. Int. J. Biochem. Cell Biol. 1999; 31: 551-557Crossref PubMed Scopus (191) Google Scholar). Various extracellular stimuli, including growth factors, cytokines, tumor promoters, peroxisomal proliferators, and carcinogens, induce Cox-2 expression (1Smith W.L. DeWitt D.L. Garavito R.M. Annu. Rev. Biochem. 2000; 69: 145-182Crossref PubMed Scopus (2477) Google Scholar, 2Vane J.R. Bakhle Y.S. Botting R.M. Annu. Rev. Pharmacol. Toxicol. 1998; 38: 97-120Crossref PubMed Scopus (2620) Google Scholar, 3Hla T. Bishop-Bailey D. Liu C.H. Schaefers H.J. Trifan O.C. Int. J. Biochem. Cell Biol. 1999; 31: 551-557Crossref PubMed Scopus (191) Google Scholar, 4Prescott S.M. Fitzpatrick F.A. Biochim. Biophys. Acta. 2000; 1470: M69-M78PubMed Google Scholar, 5Taketo M.M. J. Natl. Cancer Inst. 1998; 90: 1529-1536Crossref PubMed Scopus (496) Google Scholar, 6Taketo M.M. J. Natl. Cancer Inst. 1998; 90: 1609-1620Crossref PubMed Scopus (501) Google Scholar). Although transcriptional regulation of Cox-2 has been studied extensively, post-transcriptional mechanisms are also important for Cox-2 expression (1Smith W.L. DeWitt D.L. Garavito R.M. Annu. Rev. Biochem. 2000; 69: 145-182Crossref PubMed Scopus (2477) Google Scholar, 2Vane J.R. Bakhle Y.S. Botting R.M. Annu. Rev. Pharmacol. Toxicol. 1998; 38: 97-120Crossref PubMed Scopus (2620) Google Scholar, 3Hla T. Bishop-Bailey D. Liu C.H. Schaefers H.J. Trifan O.C. Int. J. Biochem. Cell Biol. 1999; 31: 551-557Crossref PubMed Scopus (191) Google Scholar). A strong correlation has been established between the use of non-steroidal anti-inflammatory drugs (NSAIDs) and the decreased incidence of colorectal, breast, and lung cancers (4Prescott S.M. Fitzpatrick F.A. Biochim. Biophys. Acta. 2000; 1470: M69-M78PubMed Google Scholar, 5Taketo M.M. J. Natl. Cancer Inst. 1998; 90: 1529-1536Crossref PubMed Scopus (496) Google Scholar, 6Taketo M.M. J. Natl. Cancer Inst. 1998; 90: 1609-1620Crossref PubMed Scopus (501) Google Scholar, 7Thun M.J. Gastroenterol. Clin. North Am. 1996; 25: 333-348Abstract Full Text Full Text PDF PubMed Scopus (131) Google Scholar, 8Williams C.S. Mann M. DuBois R.N. Oncogene. 1999; 18: 7908-7916Crossref PubMed Scopus (1268) Google Scholar). In addition, NSAIDs and Cox-2 inhibitors suppress carcinogen-induced tumorigenesis in animal models (9Reddy B.S. Tokumo K. Kulkarni N. Aligia C. Kelloff G. Carcinogenesis. 1992; 13: 1019-1023Crossref PubMed Scopus (151) Google Scholar, 10Nakatsugi S. Ohta T. Kawamori T. Mutoh M. Tanigawa T. Watanabe K. Sugie S. Sugimura T. Wakabayashi K. Jpn. J. Cancer Res. 2000; 91: 886-892Crossref PubMed Scopus (115) Google Scholar, 11Harris R.E. Alshafie G.A. Abou-Issa H. Seibert K. Cancer Res. 2000; 60: 2101-2103PubMed Google Scholar). Such data have prompted the examination of expression of Cox-1 and -2 in human cancer tissues. Cox-1 is expressed in both normal and malignant cells; however, Cox-2 is up-regulated in a high percentage of tumors (4Prescott S.M. Fitzpatrick F.A. Biochim. Biophys. Acta. 2000; 1470: M69-M78PubMed Google Scholar, 5Taketo M.M. J. Natl. Cancer Inst. 1998; 90: 1529-1536Crossref PubMed Scopus (496) Google Scholar, 6Taketo M.M. J. Natl. Cancer Inst. 1998; 90: 1609-1620Crossref PubMed Scopus (501) Google Scholar, 7Thun M.J. Gastroenterol. Clin. North Am. 1996; 25: 333-348Abstract Full Text Full Text PDF PubMed Scopus (131) Google Scholar, 8Williams C.S. Mann M. DuBois R.N. Oncogene. 1999; 18: 7908-7916Crossref PubMed Scopus (1268) Google Scholar, 9Reddy B.S. Tokumo K. Kulkarni N. Aligia C. Kelloff G. Carcinogenesis. 1992; 13: 1019-1023Crossref PubMed Scopus (151) Google Scholar, 10Nakatsugi S. Ohta T. Kawamori T. Mutoh M. Tanigawa T. Watanabe K. Sugie S. Sugimura T. Wakabayashi K. Jpn. J. Cancer Res. 2000; 91: 886-892Crossref PubMed Scopus (115) Google Scholar). This finding was established originally in gastrointestinal tumors but has been extended to carcinomas of diverse origins (4Prescott S.M. Fitzpatrick F.A. Biochim. Biophys. Acta. 2000; 1470: M69-M78PubMed Google Scholar, 5Taketo M.M. J. Natl. Cancer Inst. 1998; 90: 1529-1536Crossref PubMed Scopus (496) Google Scholar, 6Taketo M.M. J. Natl. Cancer Inst. 1998; 90: 1609-1620Crossref PubMed Scopus (501) Google Scholar, 7Thun M.J. Gastroenterol. Clin. North Am. 1996; 25: 333-348Abstract Full Text Full Text PDF PubMed Scopus (131) Google Scholar, 8Williams C.S. Mann M. DuBois R.N. Oncogene. 1999; 18: 7908-7916Crossref PubMed Scopus (1268) Google Scholar, 12Sano H. Kawahito Y. Wilder R.L. Hashiramoto A. Mukai S. Asai K. Kimura S. Kato H. Kondo M. Hla T. Cancer Res. 1995; 55: 3785-3789PubMed Google Scholar, 13Hwang D. Scollard D. Byrne J. Levine E. J. Natl. Cancer Inst. 1998; 90: 455-460Crossref PubMed Scopus (609) Google Scholar, 14Soslow R.A. Dannenberg A.J. Rush D. Woerner B.M. Khan K.N. Masferrer J. Koki A.T. Cancer. 2000; 89: 2637-2645Crossref PubMed Scopus (856) Google Scholar). Indeed, recent studies have indicated that high levels of Cox-2 are expressed in human mammary tumor tissues compared with the adjacent normal tissue (13Hwang D. Scollard D. Byrne J. Levine E. J. Natl. Cancer Inst. 1998; 90: 455-460Crossref PubMed Scopus (609) Google Scholar, 14Soslow R.A. Dannenberg A.J. Rush D. Woerner B.M. Khan K.N. Masferrer J. Koki A.T. Cancer. 2000; 89: 2637-2645Crossref PubMed Scopus (856) Google Scholar). Gene deletion studies support the concept that Cox-2 plays a critical role in the development of intestinal tumors (15Oshima M. Dinchuk J.E. Kargman S.L. Oshima H. Hancock B. Kwong E. Trzaskos J.M. Evans J.F. Taketo M.M. Cell. 1996; 87: 803-809Abstract Full Text Full Text PDF PubMed Scopus (2286) Google Scholar). This work has been confirmed not only in intestinal tumors but also in carcinogen-induced model of skin cancer (16Chulada P.C. Thompson M.B. Mahler J.F. Doyle C.M. Gaul B.W. Lee C. Tiano H.F. Morham S.G. Smithies O. Langenbach R. Cancer Res. 2000; 60: 4705-4708PubMed Google Scholar, 17Langenbach R. Loftin C.D. Lee C. Tiano H. Ann. N. Y. Acad. Sci. 1999; 889: 52-61Crossref PubMed Scopus (140) Google Scholar). Surprisingly however, deletion of the murine Cox-1 gene also reduced the incidence of both intestinal and skin tumors (16Chulada P.C. Thompson M.B. Mahler J.F. Doyle C.M. Gaul B.W. Lee C. Tiano H.F. Morham S.G. Smithies O. Langenbach R. Cancer Res. 2000; 60: 4705-4708PubMed Google Scholar, 17Langenbach R. Loftin C.D. Lee C. Tiano H. Ann. N. Y. Acad. Sci. 1999; 889: 52-61Crossref PubMed Scopus (140) Google Scholar). The Cox-2 gene was expressed in the mesenchymal compartment in the neoplastic tissue in the mice, which is in contrast with both mesenchymal and epithelial expression in human tumors (15Oshima M. Dinchuk J.E. Kargman S.L. Oshima H. Hancock B. Kwong E. Trzaskos J.M. Evans J.F. Taketo M.M. Cell. 1996; 87: 803-809Abstract Full Text Full Text PDF PubMed Scopus (2286) Google Scholar). These genetic loss-of-function studies suggest that expression of Cox-1 and -2 at the site of transformation or at some distal site is necessary for tumorigenesis. In contrast, other studies show that inhibition of Cox-2 by NSAIDs cannot account for their anti-tumor effects. For example, sulindac sulfone, a derivative of sulindac which lacks Cox-1 or -2 enzymatic inhibitory activity, induced epithelial cell apoptosis and inhibited carcinogenesis in animal models (18Piazza G.A. Rahm A.K. Finn T.S. Fryer B.H. Li H. Stoumen A.L. Pamukcu R. Ahnen D.J. Cancer Res. 1997; 57: 2452-2459PubMed Google Scholar, 19Boolbol S.K. Dannenberg A.J. Chadburn A. Martucci C. Guo X.J. Ramonetti J.T. Abreu-Goris M. Newmark H.L. Lipkin M.L. DeCosse J.J. Bertagnolli M.M. Cancer Res. 1996; 56: 2260-2556Google Scholar). Furthermore, various NSAIDs, albeit at high concentrations, inhibited growth and tumorigenicity of transformed mouse embryonic fibroblasts lacking both Cox-1 and Cox-2 (20Zhang X. Morham S.G. Langenbach R. Young D.A. J. Exp. Med. 1999; 190: 451-459Crossref PubMed Scopus (259) Google Scholar). Several other targets of NSAID action have also been proposed. First, the nuclear receptor peroxisomal proliferator-activated receptor (PPAR)∂ was induced by the APC mutation, which resulted in the transcriptional enhancement of PPAR∂ via the β-catenin pathway. High doses of NSAIDs inhibited the action of PPAR∂ on transcriptional regulation of downstream genes, suggesting that it might be one of the targets for NSAID inhibition of carcinogenesis (21He T.C. Chan T.A. Vogelstein B. Kinzler K.W. Cell. 1999; 99: 335-345Abstract Full Text Full Text PDF PubMed Scopus (1036) Google Scholar). Second, NSAIDs at high doses inhibited the activity of the nuclear factor κB pathway and thus may promote cell death in a prostanoid independent manner (22Yin M.J. Yamamoto Y. Gaynor R.B. Nature. 1998; 396: 77-80Crossref PubMed Scopus (1438) Google Scholar). These studies have raised questions about the causal role of Cox-2 in tumorigenesis and have identified potential non-Cox-2 targets of NSAID action. The question of whether Cox-2 overexpression is sufficient to induce tumorigenesis has not been addressed. In this report, we describe a genetic gain-of-function approach to examine the direct role of Cox-2 in tumorigenesis. The human COX-2gene (23Appleby S.B. Ristimaki A. Neilson K. Narko K. Hla T. Biochem. J. 1994; 302: 723-727Crossref PubMed Scopus (460) Google Scholar) was cloned behind the murine mammary tumor virus (MMTV) promoter (24Lane T.F. Leder P. Oncogene. 1997; 15: 2133-2144Crossref PubMed Scopus (122) Google Scholar). The linearized construct was used to derive transgenic CD1 mice (Charles River Laboratories) in the UCHC transgenic core facility. Transgenic mice were identified by Southern blot analysis of tail DNA using a human Cox-2 3′-untranslated region specific probe. Three founder animals (two males and one female) were used to derive F1 and F2 hemizygotic transgenic mice. Gene expression and phenotypic analysis were done on female F1 and F2 mice derived from all three founders. For the analysis of mammary glands during involution, lactating mice at 7 days postpartum were used, and the pups were removed to induce mammary gland involution. Mammary glands were dissected, total RNA was purified, and Northern analysis for human Cox-2 (25Hla T. Neilson K. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 7384-7388Crossref PubMed Scopus (1488) Google Scholar) and mouse β-casein (26Sympson C.J. Talhouk R.S. Alexander C.M. Chin J.R. Clift S.M. Bissell M.J. Werb Z. J. Cell Biol. 1994; 125: 681-693Crossref PubMed Scopus (349) Google Scholar) was done as described previously (25Hla T. Neilson K. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 7384-7388Crossref PubMed Scopus (1488) Google Scholar). For analysis of Cox-2 polypeptide expression, mammary glands were homogenized in the extraction buffer containing 1% Tween 20, and Western analysis was conducted as described previously (25Hla T. Neilson K. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 7384-7388Crossref PubMed Scopus (1488) Google Scholar). The Cox-2 monoclonal antibody was described previously (27Jang B.C. Sanchez T. Schaefers H.J. Trifan O.C. Liu C.H. Creminon C. Huang C.K. Hla T. J. Biol. Chem. 2000; 275: 39507-39515Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar). Polyclonal antibodies for Bax, Bcl-XL, and Bcl-2 were purchased from Santa Cruz. Mammary glands were dissected, cut into 1-mm3 sections, washed extensively, and incubated with 12 μm[1-14C]arachidonic acid (PerkinElmer Life Sciences) for 30 min at 37 °C. Medium was acidified, extracted, and prostanoids were separated and identified by TLC/autoradiography procedures as described previously (25Hla T. Neilson K. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 7384-7388Crossref PubMed Scopus (1488) Google Scholar). In some experiments, cold arachidonic acid was used, and medium was analyzed for PGE2 synthesis by radioimmunoassay as described (25Hla T. Neilson K. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 7384-7388Crossref PubMed Scopus (1488) Google Scholar, 27Jang B.C. Sanchez T. Schaefers H.J. Trifan O.C. Liu C.H. Creminon C. Huang C.K. Hla T. J. Biol. Chem. 2000; 275: 39507-39515Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar). Mammary glands were dissected and processed for whole mount analysis as described (24Lane T.F. Leder P. Oncogene. 1997; 15: 2133-2144Crossref PubMed Scopus (122) Google Scholar, 26Sympson C.J. Talhouk R.S. Alexander C.M. Chin J.R. Clift S.M. Bissell M.J. Werb Z. J. Cell Biol. 1994; 125: 681-693Crossref PubMed Scopus (349) Google Scholar). Some glands were fixed and analyzed by histological methods (28Narko K. Ristimaki A. MacPhee M. Smith E. Haudenschild C.C. Hla T. J. Biol. Chem. 1997; 272: 21455-21460Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar). Immunohistochemistry for human Cox-2 was conducted as described (12Sano H. Kawahito Y. Wilder R.L. Hashiramoto A. Mukai S. Asai K. Kimura S. Kato H. Kondo M. Hla T. Cancer Res. 1995; 55: 3785-3789PubMed Google Scholar, 28Narko K. Ristimaki A. MacPhee M. Smith E. Haudenschild C.C. Hla T. J. Biol. Chem. 1997; 272: 21455-21460Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar) using the Cox-2 monoclonal (27Jang B.C. Sanchez T. Schaefers H.J. Trifan O.C. Liu C.H. Creminon C. Huang C.K. Hla T. J. Biol. Chem. 2000; 275: 39507-39515Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar) and polyclonal antisera (Cayman Chemical). Specificity of immunostaining was confirmed by competition with excess peptide antigen as described (12Sano H. Kawahito Y. Wilder R.L. Hashiramoto A. Mukai S. Asai K. Kimura S. Kato H. Kondo M. Hla T. Cancer Res. 1995; 55: 3785-3789PubMed Google Scholar, 28Narko K. Ristimaki A. MacPhee M. Smith E. Haudenschild C.C. Hla T. J. Biol. Chem. 1997; 272: 21455-21460Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar). Terminal nucleotidyl transferase staining was performed as described (29Iizuka T. Sasaki M. Koike M. Adv. Exp. Med. Biol. 2000; 478: 369-370Crossref PubMed Google Scholar) on formallin-fixed, paraffin-embedded sections (4 μm) with a in situ cell death detection kit (Roche, Indianapolis) according to the manufacturer's instructions. Sections were counterstained with methyl green (Vector Laboratories, Burlingame, CA). The percentage of apoptotic cells was calculated from randomly selected fields (total of 4,000 cells/section) at a magnification of ×60). Statistical analysis was done using Student'st test. The MMTV promoter (24Lane T.F. Leder P. Oncogene. 1997; 15: 2133-2144Crossref PubMed Scopus (122) Google Scholar) was used to direct expression of the humanCOX-2 gene in transgenic mice (Fig.1 A). Three independent founder lines, obtained after pronuclear microinjection of the linearized construct, were bred with CD1 mice, and the hemizygous F1 and F2 mice were characterized for the studies described below. By Northern analysis, human Cox-2 transgenic mRNA was detected only in mammary glands and not in liver, lung, heart, brain, spleen, or muscle tissues (data not shown). RNA samples were isolated from the mammary gland during progressive stages of maturation. Northern blot analysis was conducted to determine the expression of endogenous murine Cox-1 and Cox-2 as well as transgenic human Cox-2 mRNA. High stringency hybridization conditions (20% formamide, 65 °C) were used to discriminate between mouse and human Cox-2 mRNAs. Only endogenous murine Cox-1 mRNA was expressed in the mammary glands of virgin mice. During lactation, Cox-1 expression was induced and returned to base line after the involution of the mammary gland in weaned mothers (Fig. 1 B). Murine Cox-2 mRNA was not detected by Northern analysis of total RNA preparations, suggesting that endogenous Cox-2 is expressed at extremely low levels during the development and involution of the mammary gland. Reverse transcription-polymerase chain reaction analysis for murine Cox-2 confirmed the Northern data and suggests that it is expressed at extremely low levels in normal mammary tissue (data not shown). In the human Cox-2 transgenic mice, a low level of Cox-2 mRNA was detected in mature virgin mice, but significantly higher levels were induced during pregnancy and lactation. Concomitant with the regression of the mammary gland in weaning females, the transgenic Cox-2 expression was attenuated after removal of nursing young (Fig. 1 B). The highest level of Cox-2 expression was achieved during lactation, which is consistent with the activity of the MMTV promoter (24Lane T.F. Leder P. Oncogene. 1997; 15: 2133-2144Crossref PubMed Scopus (122) Google Scholar). The Cox-2 transcript was not detected by Northern analysis in non-transgenic normal mice. Expression of the human Cox-2 mRNA did not alter the expression of endogenous murine Cox-1 and -2. Immunoblot analysis of mammary tissues indicates that the human Cox-2 polypeptide accumulates concomitantly with the transcript (Fig.1 C). Immunohistochemistry of mammary tissue sections was conducted using the human Cox-2-specific monoclonal antibody (Fig.1 D). The expression of the Cox-2 transgene was observed in the mammary epithelial cells of developing alveoli. In contrast, stromal cells in the mammary gland do not express the transgenic Cox-2. Prominent perinuclear immunostaining was observed, consistent with previous studies that demonstrated endoplasmic reticulum localization of the Cox-2 polypeptide (30Morita I. Schindler M. Regier M.K. Otto J.C. Hori T. DeWitt D.L. Smith W.L. J. Biol. Chem. 1995; 270: 10902-10908Abstract Full Text Full Text PDF PubMed Scopus (512) Google Scholar). The functionality of the transgenic Cox-2 polypeptide was demonstrated by thin layer chromatorgraphic analysis of radioactive arachidonic acid metabolites secreted from mammary gland explants. As shown in Fig.1 E, enhanced synthesis of PGE2, 6-keto-PGF1α, PGD2, and PGF2αwas observed in Cox-2 transgenic mammary glands compared with the age-matched normal counterparts. Quantitative analysis of PGE2 synthesis by mammary gland explants (Fig.1 F) was conducted next in mammary glands from normal as well as Cox-2 transgenic mice. Mammary glands from non-transgenic mice also exhibited an increase in Cox activity during lactation, which is consistent with the induction of endogenous Cox-1 mRNA expression. The human Cox-2 transgenic mice exhibited a higher level of PGE2 synthesis, which correlated with the expression of the Cox-2 mRNA and polypeptide. These data suggest that overexpression of Cox-2 was achieved in the mammary glands of transgenic mice, particularly during pregnancy and lactation. The mammary glands of Cox-2 transgenic mice were analyzed by morphological methods. Analysis of whole mount preparations of virgin mammary glands demonstrated abnormal and hyperplastic alveolar development (Fig. 2 A). Histological analysis of mammary sections confirmed precocious development of alveolar glands in Cox-2 transgenic mice (Fig.2 B). Northern blot analysis indicated that the expression of the β-casein, a gene normally expressed during late pregnancy and lactation, was markedly up-regulated in virgin mammary glands of Cox-2 transgenic mice (Fig. 2 C). To determine if secreted prostanoids are involved in this phenomenon, virgin mice were treated with indomethacin, a potent inhibitor of prostanoid synthesis (2Vane J.R. Bakhle Y.S. Botting R.M. Annu. Rev. Pharmacol. Toxicol. 1998; 38: 97-120Crossref PubMed Scopus (2620) Google Scholar). As shown in Fig. 2, A, C, and D, indomethacin treatment suppressed PGE2 synthesis, β-casein gene expression, and precocious alveolar differentiation. These data strongly suggest that enhanced Cox-2 expression and the concomitant secretion of prostanoids mediate the precocious mammary gland differentiation in Cox-2 transgenic mice. The Cox-2 transgenic mice lactated normally and were able to nurse their young. Histological and whole mount analysis of mammary glands from pregnant and lactating animals showed no gross abnormalities (Fig.3 A). These data suggest that exaggerated Cox-2 expression did not interfere with the physiological functions of the mammary gland. We next studied the effect of Cox-2 on mammary gland involution after weaning. As shown in Fig. 3 A, the expression of Cox-2 polypeptide decayed rapidly after weaning in Cox-2 transgenic mice. Histological analysis of mammary tissue (Fig.3 B) indicates that a delay in mammary gland involution occurred in Cox-2 transgenic mice. Mammary glands from control mice showed expected rates of epithelial cell death and collapse of the alveolar glands at 1–2 days postweaning, resulting in the regression of the mammary gland. Although mammary glands from Cox-2 transgenic mice regressed at 7 days, the rate was delayed at early stages (days 1–2), no signs of involution were apparent (Fig. 3 B). 2 weeks after weaning, the mammary glands appeared similar in both Cox-2 transgenic and normal mice. Thus enhanced Cox-2 expression at 0–2 days postweaning may have maintained the viability and function of the mammary epithelial cells. Because cellular apoptosis contributes to the involution of the mammary gland, we measured the rates of apoptosis in mammary glands of normal and Cox-2 transgenic mice after weaning using the terminal nucleotidyl transferase method (29Iizuka T. Sasaki M. Koike M. Adv. Exp. Med. Biol. 2000; 478: 369-370Crossref PubMed Google Scholar). As shown in Fig. 3 C, significant decreases in apoptotic cells were observed in Cox-2 transgenic mice compared with non-transgenic normal mice at 2 days after weaning. These observations are consistent with the previous findings that enhanced Cox-2 expression inhibits cellular apoptosis in transfected epithelial cells (31Tsujii M. DuBois R.N. Cell. 1995; 83: 493-501Abstract Full Text PDF PubMed Scopus (2140) Google Scholar). Because pregnancy and lactation induced the Cox-2 transgene to very high levels, we compared the mammary glands from multiparous females. Focal areas of hyperplasia, also known as hyperplastic alveolar nodules, are present with high frequency in human Cox-2 transgenic mice that have undergone three or four cycles of pregnancy and lactation (Fig. 4 A). Histologic analyses of such sections indicate that alveolar hyperplasia, squamous metaplasia, and adenomatous carcinomas of the mammary epithelial cells contributed to these lesions. In contrast, age-matched, multiparous non-transgenic female mice did not show such tumors (Fig.4 A). The incidence of mammary tumors in Cox-2 transgenic mice and non-transgenic mice were analyzed by a combination of whole mount analysis and histological sections. As shown in Fig. 4 B, > 85% of Cox-2 transgenic mice that have undergone multiple cycles of pregnancy and lactation show the presence of tumors in mammary glands. In contrast, a very low incidence of tumors was seen in non-transgenic mice with similar age and multiparosity. Transgenic female mice derived from all three founders showed this phenomenon, suggesting that the effects are caused by the expression of the human COX-2 gene rather than integration-dependent events. Cox-2 transgenic mice that did not undergo pregnancy and lactation did not show the increased mammary tumorigenesis. Three female Cox-2 transgenic mice (the founder female and two F1 females from a different founder) developed large mammary tumors with metastatic spread. One such animal at necropsy is shown in Fig. 4 C, as an example of the presence of large metastatic tumors in and outside of the mammary glands. Histological analyses of mammary tissue from non-transgenic mice and from Cox-2-induced tumors are shown in Fig.5, A and B, respectively. In Fig. 5 B, one can observe numerous hyperplastic glandular structures, keratinizing squamous cells (indicating metaplasia), stromal proliferation, and numerous blood vessels. In Fig. 5 C, adenomatous carcinoma was observed in the peripheral lymph node. Well differentiated tumors appeared to have metastasized into the lymph nodes of the Cox-2 transgenic mice. In Fig.5 D, an example of a mammary tumor with squamous metaplasia is shown. In Fig. 5 E, an example of a hyperplastic and invasive ductal structure with hemosiderin (indicating bleeding into the ducts) is shown. These data suggest that Cox-2 overexpression induced various types of invasive mammary tumors, including ductal and lobuloalveolar carcinomas. Mammary tumors in the Cox-2 transgenic mice were stained with the Cox-2 antibody using immunohistochemical procedures (Fig. 5 F). Numerous epithelial cells in glandular structures exhibited strong immunoreactivity in the perinuclear locale. Immunoreactivity was competed by coincubation of the antibody with the Cox-2 antigenic peptide (data not shown). These data suggest that Cox-2 is expressed in the mammary tumors of Cox-2 transgenic mice. To determine if decreased apoptosis of mammary epithelial cells contributed to the Cox-2-induced tumorigenesis, we analyzed the expression of apoptotic regulatory proteins Bcl-2, Bax, and Bcl-xL. Expression of Bcl-2 was induced by Cox-2 overexpression in intestinal epithelial cells (31Tsujii M. DuBois R.N. Cell. 1995; 83: 493-501Abstract Full Text PDF PubMed Scopus (2140) Google Scholar), and Bax is known to regulate NSAID-induced apoptosis (32Zhang L., Yu, J. Park B.H. Kinzler K.W. Vogelstein B. Science. 2000; 290: 989-992Crossref PubMed Scopus (797) Google Scholar). As shown in Fig.6, high levels of Cox-2 polypeptide were observed in the mammary tumors and lower levels were seen in the adjacent normal tissue. In contrast, Cox-2 polypeptide was not expressed in mammary glands of multiparous, non-transgenic mice. Interestingly, expression of the proapoptotic proteins Bax and Bcl-xL was reduced in Cox-2-expressing tumors. In contrast, Bcl-2 expression was up-regulated in the tumor tissue. These data suggest that regulation of expression of apoptosis regulatory factors Bcl-2, Bax, and Bcl-XL may be a mechanism via which Cox-2 overexpression contributes to tumorigenesis. Together, these observations support the concept that overexpression of Cox-2 alone is sufficient to transform the mammary epithelium into a tumorigenic state. In this report, we describe a gain-of-function approach to evaluate the direct role for Cox-2 in tumorigenesis. Although studies utilizing various approaches, namely, epidemiologic, expression studies and genetic loss-of-function approaches, have implicated the functional role of Cox-2 in tumorigenesis, questions remain regarding the requirement and sufficiency of Cox-2 (4Prescott S.M. Fitzpatrick F.A. Biochim. Biophys. Acta. 2000; 1470: M69-M78PubMed Google Scholar, 5Taketo M.M. J. Natl. Cancer Inst. 1998; 90: 1529-1536Crossref PubMed Scopus (496) Google Scholar, 6Taketo M.M. J. Natl. Cancer Inst. 1998; 90: 1609-1620Crossref PubMed Scopus (501) Google Scholar, 7Thun M.J. Gastroenterol. Clin. North Am. 1996; 25: 333-348Abstract Full Text Full Text PDF PubMed Scopus (131) Google Scholar, 8Williams C.S. Mann M. DuBois R.N. Oncogene. 1999; 18: 7908-7916Crossref PubMed Scopus (1268) Google Scholar, 18Piazza G.A. Rahm A.K. Finn T.S. Fryer B.H. Li H. Stoumen A.L. Pamukcu R. Ahnen D.J. Cancer Res. 1997; 57: 2452-2459PubMed Google Scholar, 19Boolbol S.K. Dannenberg A.J. Chadburn A. Martucci C. Guo X.J. Ramonetti J.T. Abreu-Goris M. Newmark H.L. Lipkin M.L. DeCosse J.J. Bertagnolli M.M. Cancer Res. 1996; 56: 2260-2556Google Scholar). Furthermore, NSAIDs were shown to inhibit various animal models of tumorigenesis; however, issues of nonspecificity have prevented the unequivocal interpretation of these studies (18Piazza G.A. Rahm A.K. Finn T.S. Fryer B.H. Li H. Stoumen A.L. Pamukcu R. Ahnen D.J. Cancer Res. 1997; 57: 2452-2459PubMed Google Scholar, 19Boolbol S.K. Dannenberg A.J. Chadburn A. Martucci C. Guo X.J. Ramonetti J.T. Abreu-Goris M. Newmark H.L. Lipkin M.L. DeCosse J.J. Bertagnolli M.M. Cancer Res. 1996; 56: 2260-2556Google Scholar, 20Zhang X. Morham S.G. Langenbach R. Young D.A. J. Exp. Med. 1999; 190: 451-459Crossref PubMed Scopus (259) Google Scholar, 21He T.C. Chan T.A. Vogelstein B. Kinzler K.W. Cell. 1999; 99: 335-345Abstract Full Text Full Text PDF PubMed Scopus (1036) Google Scholar, 22Yin M.J. Yamamoto Y. Gaynor R.B. Nature. 1998; 396: 77-80Crossref PubMed Scopus (1438) Google Scholar). In this study we provide evidence that overexpression of Cox-2 alone is sufficient to induce tumorigenic transformation in a tissue-specific manner. Overexpression of Cox-2 expression was achieved using the entire humanCOX-2 gene (23Appleby S.B. Ristimaki A. Neilson K. Narko K. Hla T. Biochem. J. 1994; 302: 723-727Crossref PubMed Scopus (460) Google Scholar) under the control of the MMTV promoter (24Lane T.F. Leder P. Oncogene. 1997; 15: 2133-2144Crossref PubMed Scopus (122) Google Scholar). Previously, we attempted to derive transgenic mice using the keratin K14- driven human Cox-2 cDNA (33Byrne C. Tainsky M. Fuchs E. Development. 1994; 120: 2369-2383Crossref PubMed Google Scholar) and cytomegalovirus promoter-driven human Cox-2 cDNA in a tetracycline-regulated system (34Shockett P.E. Schatz D.G. Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 5173-5176Crossref PubMed Scopus (120) Google Scholar). Both approaches were unsuccessful. In the former case, embryonic lethality appears to be induced by the developmental expression of the transgene. In the second case, the Cox-2 cDNA was not expressed after induction in two independent lines of transgenic mice. Therefore, we used the MMTV promoter, which is known to be induced postpartum in a mammary gland-enriched manner (24Lane T.F. Leder P. Oncogene. 1997; 15: 2133-2144Crossref PubMed Scopus (122) Google Scholar). In addition, we utilized the entire human COX-2 gene to include all of the post-transcriptional signals required for efficient expression of the gene (23Appleby S.B. Ristimaki A. Neilson K. Narko K. Hla T. Biochem. J. 1994; 302: 723-727Crossref PubMed Scopus (460) Google Scholar, 25Hla T. Neilson K. Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 7384-7388Crossref PubMed Scopus (1488) Google Scholar). As shown in Fig. 1, MMTV-hCox-2 transgenic mice expressed high levels of Cox-2 transcript, protein, and enzymatic activity in the mammary glands. Specifically, expression of the transgenic Cox-2 polypeptide was induced during mammary gland development, achieved very high levels during pregnancy, and was maximal during lactation. Although the expression of Cox-2 is low in normal mammary glands, it can be induced by a high fat diet and carcinogen exposure (35Badawi A.F. Archer M.C. Prostaglandins Other Lipid Mediat. 1998; 56: 167-181Crossref PubMed Scopus (27) Google Scholar, 36Badawi A.F. El-Sohemy A. Stephen L.L. Ghoshal A.K. Archer M.C. Carcinogenesis. 1998; 19: 905-910Crossref PubMed Scopus (68) Google Scholar). However, significant expression of Cox-1 was observed in murine mammary tissue during all phases of mammary gland development. In addition, Cox-1 activity was induced during lactation. Cox-2 enzymatic activity followed a kinetics of induction similar to that of the mRNA and the protein, resulting in the production of PGE2, 6-keto-PGF1α, PGD2, and PGF2α in the human Cox-2 transgenic mammary glands. After weaning, Cox-2 expression decayed rapidly, consistent with the short half-life of the mRNA and the protein (1Smith W.L. DeWitt D.L. Garavito R.M. Annu. Rev. Biochem. 2000; 69: 145-182Crossref PubMed Scopus (2477) Google Scholar, 2Vane J.R. Bakhle Y.S. Botting R.M. Annu. Rev. Pharmacol. Toxicol. 1998; 38: 97-120Crossref PubMed Scopus (2620) Google Scholar, 3Hla T. Bishop-Bailey D. Liu C.H. Schaefers H.J. Trifan O.C. Int. J. Biochem. Cell Biol. 1999; 31: 551-557Crossref PubMed Scopus (191) Google Scholar). Thus the MMTV-hCox-2 transgenic mouse model provides an animal model in which persistent low level expression of Cox-2 is achieved in mammary epithelial cells, and exaggerated expression of Cox-2 can be induced during pregnancy and lactation. Interestingly, MMTV-hCox-2 mice exhibited precocious development of the mammary glands as virgins. This is most likely the result of the secretion of prostanoids because indomethacin treatment reversed the changes. Mammary development is controlled by hormonal factors as well as by local signals (37Cardiff R.D. J. Mamm. Gland Biol. Neoplasia. 1996; 1: 61-73Crossref PubMed Scopus (36) Google Scholar). It is known that PGE2 induces mammary epithelial cell proliferation (38Imagawa W. Bandyopadhyay G.K. Wallace D. Nandi S. J. Cell. Physiol. 1998; 135: 509-515Crossref Scopus (21) Google Scholar). In addition, PGE2 is known to induce aromatase, an enzyme involved in the local biosynthesis of estrogen, a potent inducer of mammary gland development (39Zhao Y. Agarwal V.R. Mendelson C.R. Simpson E.R. Endocrinology. 1996; 137: 5739-5742Crossref PubMed Scopus (364) Google Scholar). Thus, low level persistent expression of Cox-2 may induce precocious mammary gland development via the secretion of PGE2. MMTV-hCox-2 transgenic mice exhibited a delayed mammary gland involution after weaning. Mammary gland involution is accompanied by rapid decay of Cox-2 expression, consistent with the knowledge that Cox-2 mRNA has a rapid turnover rate (23Appleby S.B. Ristimaki A. Neilson K. Narko K. Hla T. Biochem. J. 1994; 302: 723-727Crossref PubMed Scopus (460) Google Scholar, 27Jang B.C. Sanchez T. Schaefers H.J. Trifan O.C. Liu C.H. Creminon C. Huang C.K. Hla T. J. Biol. Chem. 2000; 275: 39507-39515Abstract Full Text Full Text PDF PubMed Scopus (72) Google Scholar). Cox-2 expression in epithelial cells of the gastrointestinal tract is associated with inhibition of apoptosis (31Tsujii M. DuBois R.N. Cell. 1995; 83: 493-501Abstract Full Text PDF PubMed Scopus (2140) Google Scholar). Mammary gland involution is brought about in part by apoptosis of epithelial cells (40Furth P.A. J. Mammary Gland Biol. Neoplasia. 1999; 4: 123-127Crossref PubMed Scopus (45) Google Scholar). Indeed, analysis of apoptotic rates by terminal nucleotidyl transferase assay indicates that Cox-2 overexpression resulted in decreased apoptotic cells. However, the effect on inhibition of mammary gland involution was transient because the Cox-2 expression decayed rapidly. These data suggest that Cox-2 overexpression may dysregulate the normal homeostatic mechanisms in the mammary gland by inhibition of apoptosis. Mammary tumorigenesis occurred with high frequency in MMTV-hCox-2 transgenic mice after multiple rounds of pregnancy. The tumor incidence was high in all three lines of mice, suggesting that Cox-2 overexpression, rather than integration-dependent effects, were involved. The tumors were invasive and malignant, and secondary metastases were observed. The lag period of tumorigenesis may suggest the requirement for exaggerated induction of Cox-2, which occurred only during pregnancy and lactation. Indeed, low level expression of Cox-2 in virgin mammary gland did not result in spontaneous development of tumors. Alternatively, it is possible that a high level of Cox-2 induction and a second mutation are required to transform the mammary epithelium completely. Indeed, proliferating epithelial cells in the pregnant mammary gland may be more susceptible to mutagenic events of critical tumor suppressor genes. This observation is similar to the phenotype of other transgenes such as MMTV-cyclin D1 and MMTV-stromelysin (41Sternlicht M.D. Bissell M.J. Werb Z. Oncogene. 2000; 19: 1102-1113Crossref PubMed Scopus (234) Google Scholar, 42Wang T.C. Cardiff R.D. Zukerberg L. Lees E. Arnold A. Schmidt E.V. Nature. 1994; 369: 669-671Crossref PubMed Scopus (896) Google Scholar). Nevertheless, these findings point out that overexpression of Cox-2 alone is sufficient to induce tumorigenic transformation. The molecular mechanisms responsible for Cox-2 induction of tumorigenesis are unclear at present. Secreted prostanoids may interact with plasma membrane-localized G protein-coupled receptors to induce mitogenic or anti-apoptotic signals (1Smith W.L. DeWitt D.L. Garavito R.M. Annu. Rev. Biochem. 2000; 69: 145-182Crossref PubMed Scopus (2477) Google Scholar, 2Vane J.R. Bakhle Y.S. Botting R.M. Annu. Rev. Pharmacol. Toxicol. 1998; 38: 97-120Crossref PubMed Scopus (2620) Google Scholar, 3Hla T. Bishop-Bailey D. Liu C.H. Schaefers H.J. Trifan O.C. Int. J. Biochem. Cell Biol. 1999; 31: 551-557Crossref PubMed Scopus (191) Google Scholar). Alternatively, nuclear receptors such as PPARγ and PPARδ, which are activated by prostanoids, may be involved (1Smith W.L. DeWitt D.L. Garavito R.M. Annu. Rev. Biochem. 2000; 69: 145-182Crossref PubMed Scopus (2477) Google Scholar, 2Vane J.R. Bakhle Y.S. Botting R.M. Annu. Rev. Pharmacol. Toxicol. 1998; 38: 97-120Crossref PubMed Scopus (2620) Google Scholar, 3Hla T. Bishop-Bailey D. Liu C.H. Schaefers H.J. Trifan O.C. Int. J. Biochem. Cell Biol. 1999; 31: 551-557Crossref PubMed Scopus (191) Google Scholar, 21He T.C. Chan T.A. Vogelstein B. Kinzler K.W. Cell. 1999; 99: 335-345Abstract Full Text Full Text PDF PubMed Scopus (1036) Google Scholar). The possibility that the peroxidase action of Cox-2 is involved in tumorigenesis also cannot be ruled out at present (1Smith W.L. DeWitt D.L. Garavito R.M. Annu. Rev. Biochem. 2000; 69: 145-182Crossref PubMed Scopus (2477) Google Scholar, 2Vane J.R. Bakhle Y.S. Botting R.M. Annu. Rev. Pharmacol. Toxicol. 1998; 38: 97-120Crossref PubMed Scopus (2620) Google Scholar, 3Hla T. Bishop-Bailey D. Liu C.H. Schaefers H.J. Trifan O.C. Int. J. Biochem. Cell Biol. 1999; 31: 551-557Crossref PubMed Scopus (191) Google Scholar). The animal model described in this study may provide a useful system to begin to address these issues. However, Cox-2-induced tumors expressed high levels of Cox-2, the anti-apoptotic protein Bcl-2, and reduced levels of the pro-apoptotic proteins Bcl-XL and Bax. This suggests that Cox-2 may regulate expression of these death-regulatory proteins and thus inhibit apoptosis and contribute to tumorigenesis. In agreement, the Cox-2 transgenic mice exhibit reduced apoptotic rates after mammary gland involution. Many risk factors for human breast cancer induce Cox-2 expression. For example, dietary n-6 fatty acids, activation of HER/Neu signaling, Wnt signaling, as well as carcinogen exposure induceCox-2 gene expression (5Taketo M.M. J. Natl. Cancer Inst. 1998; 90: 1529-1536Crossref PubMed Scopus (496) Google Scholar, 8Williams C.S. Mann M. DuBois R.N. Oncogene. 1999; 18: 7908-7916Crossref PubMed Scopus (1268) Google Scholar, 36Badawi A.F. El-Sohemy A. Stephen L.L. Ghoshal A.K. Archer M.C. Carcinogenesis. 1998; 19: 905-910Crossref PubMed Scopus (68) Google Scholar, 43McPherson K. Steel C.M. Dixon J.M. Br. Med. J. 2000; 321: 624-628Crossref PubMed Scopus (1073) Google Scholar, 44Howe L.R. Subbaramaiah K. Chung W.J. Dannenberg A.J. Brown A.M. Cancer Res. 1999; 59: 1572-1577PubMed Google Scholar, 45Vadlamudi R. Mandal M. Adam L. Steinbach G. Mendelsohn J. Kumar R. Oncogene. 1999; 18: 305-314Crossref PubMed Scopus (207) Google Scholar). In addition, data from carcinogen-induced rodent models of mammary cancer suggest that Cox-2 inhibition reduces the incidence of mammary tumors (10Nakatsugi S. Ohta T. Kawamori T. Mutoh M. Tanigawa T. Watanabe K. Sugie S. Sugimura T. Wakabayashi K. Jpn. J. Cancer Res. 2000; 91: 886-892Crossref PubMed Scopus (115) Google Scholar, 11Harris R.E. Alshafie G.A. Abou-Issa H. Seibert K. Cancer Res. 2000; 60: 2101-2103PubMed Google Scholar). Our data are consistent with the concept that overexpression of Cox-2 in the mammary epithelium is sufficient to induce mammary carcinogenesis. Coupled with the strong epidemiological evidence of NSAID usage and cancer incidence in humans (7Thun M.J. Gastroenterol. Clin. North Am. 1996; 25: 333-348Abstract Full Text Full Text PDF PubMed Scopus (131) Google Scholar), these findings strongly support the notion that dysregulated and exaggerated expression of Cox-2 are strong carcinogenic risk factors. Whether inhibition of Cox-2 in high risk populations would confer an anti-cancer benefit in human breast cancer warrants further study. However, efforts to define chemopreventive agents that work via Cox-2 inhibition will be greatly facilitated by this human Cox-2 transgenic model. Furthermore, mechanistic studies and identification of downstream genetic targets of Cox-2 action may be readily approached in this model. We thank M. Keough for help with mouse procedures, Dr. D. DeWitt for the gift of mouse Cox probes, Dr. Steven Clark for transgenesis protocols, Dr. Cristophe Creminon for the Cox-2 antibody, and Dr. Gilbert Jay for helpful comments.