Non-small cell lung cancer cyclooxygenase-2-dependent invasion is mediated by CD44.

Elevated tumor cyclooxygenase (COX-2) expression is associated with increased angiogenesis, tumor invasion, and suppression of host immunity. We have previously shown that genetic inhibition of tumor COX-2 expression reverses the immunosuppression induced by non-small cell lung cancer (NSCLC). To assess the impact of COX-2 expression in lung cancer invasiveness, NSCLC cell lines were transduced with a retroviral vector expressing the human COX-2 cDNA in the sense (COX-2-S) and antisense (COX-2-AS) orientations. COX-2-S clones expressed significantly more COX-2 protein, produced 10-fold more prostaglandin E(2), and demonstrated an enhanced invasive capacity compared with control vector-transduced or parental cells. CD44, the cell surface receptor for hyaluronate, was overexpressed in COX-2-S cells, and specific blockade of CD44 significantly decreased tumor cell invasion. In contrast, COX-2-AS clones had a very limited capacity for invasion and showed diminished expression of CD44. These findings suggest that a COX-2-mediated, CD44-dependent pathway is operative in NSCLC invasion. Because tumor COX-2 expression appears to have a multifaceted role in conferring the malignant phenotype, COX-2 may be an important target for gene or pharmacologic therapy in NSCLC.

Tumor metastasis is a complex series of events in which cells migrate beyond tissue compartments and spread to distant organ sites. Cell surface CD44, the receptor for hyaluronate, has an important role in regulating tumor growth and metastasis because it mediates cellular adhesion to extracellular matrix, which is prerequisite for tumor cell migration (25,26).
While COX-2 expression has previously been linked to enhanced matrix metalloproteinase (MMP) expression and invasion (27), the role of CD44 in this COX-2-induced invasion has not been defined. Here, we report that stable overexpression of COX-2 in NSCLC results in up-regulation of CD44. Furthermore, we demonstrate a CD44-dependent increase in invasion in Matrigel matrix assays. In contrast, abrogation of tumor-COX-2 expression results in decreased PGE 2 production, diminished CD44 expression and decreased invasion. This is the first report documenting the critical role of tumor COX-2 expression in the regulation of CD44-dependent invasion by human NSCLC.

EXPERIMENTAL PROCEDURES
Transfection Protocol-A 2.0-kilobase pair cDNA fragment of human COX-2 (generously provided by Dr. Harvey Herschman, UCLA) was cloned into the PmeI site of the retroviral vector pLNCX (CLONTECH, Palo Alto, CA). In this vector, transcription of the COX-2 cDNA is controlled by the cytomegalovirus promoter, and transduced cells can be enriched under G418 selection. Sense (COX-2-S) and antisense (COX-2-AS) oriented expression vectors were prepared. A549 (human lung adenocarcinoma) and H157 (squamous cell carcinoma) cells were obtained from American Type Culture Collection (ATCC, Manassas, VA) and the National Cancer Institute, respectively. Tumor cells were transfected with COX-2-S, COX-2-AS-expressing vectors, or pLNCX (vector alone) using calcium-phosphate-mediated transfection (Promega, Madison, WI). The transfected cells were grown in an atmosphere of 5% CO 2 in air at 37°C in cell culture medium consisting of RPMI 1640 (Mediatech, Herndon, VA) supplemented with 10% fetal bovine serum (Gemini Biological Products, Calabasas, CA), 100 units/ml penicillin, 0.1 mg/ml streptomycin, and 2 mM glutamine (Life Technologies, Inc.). For selection, the medium also contained 0.5 mg/ml active G418 (Life Technologies, Inc.). Following G418 selection, the polymerase chain reaction (PCR) was used to confirm the positive COX-2-S and COX-2-AS clones. The PCR was done using the pLNCX 5Ј primer (AGCTTCGTTTAGTGAACCTCAGATCG) and pLNCX 3Ј primer (ACCTACAGGTGGGGTCTTTCATTCCC). PCR positive clones were then screened by Western blot and EIA analysis for COX-2 ex-* This work was supported by National Institutes of Health Grants 1P50 CA90388, RO1 CA71818, and RO1 CA78654; the American Lung Association; Merit Review Research Funds from the Department of Veterans Affairs; and by the Research Enhancement Award Program in Cancer Gene Medicine. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ** To whom correspondence and reprint requests should be addressed: UCLA School of Medicine, 37- pression and PGE 2 production, respectively. For each tumor cell line a high COX-2-expressing and PGE 2 -producing clone for COX-2-S, and a low COX-2-expressing and PGE 2 -producing COX-2-AS clone, were identified from a survey of 25 clones. These clones were then expanded for further studies. Measurement of Prostaglandin E 2 -Control, COX-2-S, and COX-2-AS cells (A549 and H157) were stimulated with IL-1␤ (200 units/ml, Genzyme, Cambridge, MA) for 24 h. PGE 2 concentration in each group (with or without IL-1␤ stimulation) were measured by EIA using a PGE 2 EIA kit (Cayman Chemical, Ann Arbor, MI) as reported previously (28). All measurements were made in triplicate and repeated in at least three separate experiments.
Western Blot Analysis for COX-2 and CD44 Expression-The cells from each treatment group were lysed at 4°C for 15 min in lysis buffer (10 mM HEPES, pH 7.9, 10 mM KCl, 1.5 mM MgCl 2 , 0.1 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and 0.6% Nonidet P-40). The cell lysates were centrifuged at 13,000 rpm for 10 min and the supernatant collected. Total protein was measured with a protein assay reagent (Bio-Rad), and 20 g of cell lysate protein were separated on 10% SDS-polyacrylamide gels. Following separation, the proteins were transferred to Hybond nitrocellulose membranes (Amersham Pharmacia Biotech) and the filters probed with anti-human COX-2 antibody (Cayman Chemical). For CD44 detection, the filters were probed with anti-human CD44 antibody 4A4 (29). The membranes were developed by the ECL chemiluminescence system (Amersham Pharmacia Biotech) and exposed to x-ray film (Fujifilm, Fuji Medical Systems Inc., Stamford, CT). Equal loading of samples was confirmed by probing the membranes with ␤-actin antibody.
Invasion Assay-To quantify invasion, the membrane invasion assay was carried out in Matrigel-coated invasion chambers (Becton Dickinson Labware, Franklin Lakes, NJ). Control and COX-2 transfectants were cultured in RPMI 1640 supplemented with 10% fetal bovine serum. Tumor cells in log phase growth were detached by trypsin-EDTA (Mediatech) and resuspended in RPMI 1640 with 0.1% bovine serum albumin. Serum-free A549-and H157-conditioned medium was obtained by incubation of these cells for 24 h. This tumor cell-conditioned medium was added in the lower chamber as a chemoattractant, and the resuspended cells (5 ϫ 10 4 ) were plated in the upper chamber. Following 18-h incubation at 37°C in a humidified 5% CO 2 atmosphere, the cells in the upper chamber and on the Matrigel were mechanically removed with a cotton swab. The cells adherent to the outer surface of the membrane were fixed with methanol and stained with hematoxylin/ eosin. The invading cells were examined, counted, and photographed by microscopy (Nikon Labphot-2 Microscope with an attached Spot Digital Camera, A. G. Heinz, Lake Forest, CA) at ϫ 50 magnification. Six fields were counted per filter in each group, and the experiment was repeated five separate times. To assess the role of CD44 in mediating Matrigel invasion, the cells from COX-2-S-expressing clones were plated in the presence of 500 ng of anti-CD44 antibody (The Binding Site, Inc., San Diego, CA) or control mouse IgG (Dako Corp., Carpenteria, CA) and the cell numbers determined as described.

Expression of COX-2 and PGE 2 Production in COX-2-transduced NSCLC Cells-
To evaluate the role of COX-2 expression in mediating the lung cancer invasiveness, two NSCLC cell lines were stably transduced with a retroviral vector encoding COX-2 and selected for G418 resistance. The COX-2-S clones (A549-S and H157-S) exhibited enhanced constitutive COX-2 protein expression by immunoblot (Fig. 1, A and B). In contrast, COX-2 was not detectable in COX-2-AS clones (Fig. 1, A and B). The COX-2 protein expression in the sense clones was found to be significantly higher (3-fold for H157 and 10-fold for A549 cells), as measured by densitometry, than that of parental or vector-transduced cells (data not shown). Compared with parental and vector-transduced cells, COX-2-S clones (H157-S and A549-S) exhibited a 5-12-fold increase in PGE 2 production. In contrast, COX-2-AS clones showed decreased constitutive PGE 2 production (Fig. 2, A and B). IL-1␤ is one of several cytokines known to potently up-regulate COX-2 expression in a variety of cells (30). Consistent with our previous findings (28),

CD44-mediated COX-2-dependent NSCLC Invasion 20810
parental and vector-transduced control cells exhibited a 5-10fold increase in PGE 2 production in response to IL-1␤ (Fig. 2, A  and B). In contrast, a similar induction of PGE 2 secretion was not observed in COX-2-AS clones in the presence of IL-1␤ (Fig.  2, A and B).
Overexpression of COX-2 Enhances the Invasive Capacity of NSCLC Cells-COX-2 overexpression may be associated with enhanced invasiveness and metastasis in a variety of malignancies (27,31). To examine its role in NSCLC, we tested cells overexpressing COX-2 in Matrigel matrix invasion assays. As depicted in Fig. 3A, COX-2-S cells were significantly more invasive than parental or antisense-transfected A549 and H157 cells. The invasiveness was found to be similar for both of the COX-2-S cell lines. By counting cells in random microscopic fields, we determined that the sense clones of both A549 and H157 were 3-fold more invasive than parental or vector-transfected controls (Fig. 3B). Furthermore, consistent with their undetectable COX-2 expression, COX-2-AS cells demonstrated a decreased invasiveness through Matrigel (Fig. 3, A and B).
NSCLC COX-2-dependent Invasion Is Mediated through CD44 Expression-Because tumor CD44 expression is known to play an important role in tumor invasion and metastasis (26,32), we hypothesized that CD44 expression may be important for NSCLC COX-2-dependent invasion. Indeed, CD44 expression was augmented in COX-2-S clones as determined by Western blot analysis (Fig. 4, A and B). In contrast, CD44 was not detectable in COX-2-AS cells (Fig. 4, A and B). Consistent with the expression of COX-2 and PGE 2 production, CD44 expression was also found to be up-regulated in parental tumor cells in the presence of IL-1␤. This suggested that related signaling events could be mediating increased COX-2 and CD44 expression in NSCLC. Because IL-1␤ could not induce COX-2 activity or CD44 expression in the COX-2-AS transfectants, COX-2 appears to be a proximal regulator of CD44 expression in NSCLC.
Consistent with the findings of Western blot analysis, FACS analysis of the COX-2 transfected cells with a fluorescein isothiocyanate-labeled CD44 antibody showed an increase in the expression of CD44 in COX-2-S but decreased expression in COX-2-AS cells (data not shown). To determine the importance of CD44 expression in tumor COX-2-dependent invasion, we performed antibody blocking experiments with anti-CD44 antibody. The enhanced invasiveness of the COX-2-S clones was found to be inhibited with anti-CD44 antibody, whereas control monoclonal antibody had no effect (Fig. 3, A and B). Anti-CD44 antibody did not completely decrease COX-2-S invasion to the level of COX-2-AS cells, suggesting that CD44-independent mechanisms may also be operative. DISCUSSION Mounting evidence from several studies indicates that tumor COX-2 activity has a multifaceted role in conferring the malignant and metastatic phenotypes. The data in the current study implicates COX-2 overexpression as a proximal mediator of CD44-dependent invasion in human NSCLC. We demonstrate that augmenting COX-2 expression leads to increased invasion by NSCLC cells. Importantly, this enhanced invasion is associated with increased CD44 expression, and blocking CD44 abrogates the COX-2-mediated enhancement of NSCLC cell mobilization through Matrigel.
Although multiple genetic alterations are necessary for lung cancer invasion and metastasis, COX-2 may be a central element in orchestrating this process (13)(14)(15)(16)(17). Studies indicate that overexpression of COX-2 is associated with apoptosis resistance (21)(22)(23)33), angiogenesis (19,20,23,34), decreased host immunity (13,24), and enhanced invasion and metastasis (27). We reported previously that COX-2 is overexpressed in human NSCLC and the resultant high level PGE 2 productionmediated dysregulation of host immunity by altering the balance of interleukins 10 and 12 (13). Indeed, specific inhibition of COX-2 led to significant in vivo tumor reduction in murine lung cancer models (24). Recently, other studies have corroborated and expanded on our initial findings documenting the importance of COX-2 expression in lung cancer (12, 14 -16, 35, 36). COX-2 activity can be detected throughout the progression of a premalignant lesion to the metastatic phenotype (14). Markedly higher COX-2 expression was observed in lung cancer lymph node metastasis compared with primary adenocar- CD44-mediated COX-2-dependent NSCLC Invasion 20811 cinoma (14). These reports, together with studies documenting an increase in COX-2 expression in precursor lesions (15,16), suggest the involvement of COX-2 overexpression in the pathogenesis of lung cancer. Epidemiological studies that indicate a decreased incidence of lung cancer in subjects who regularly use aspirin have been interpreted as supporting this hypothesis (37).
In addition to regulating immune responses, tumor COX-2 has been implicated in inhibiting apoptosis (21,38) and angiogenesis (3,20). The inhibition of programmed cell death in COX-2-expressing cells is found to be associated with an increase in Bcl2 expression and a decrease in the expression of both the TGF-␤2 receptor and E-cadherin protein (21). Experimental evidence suggests that ligation of the cell surface matrix adhesion receptor CD44 by anti-CD44 antibody induces cell detachment and triggers apoptosis in a variety of cells (39). Thus, inhibiting anchorage dependence mediated by CD44 may contribute to induction of apoptosis (40).
Overexpression of COX-2 also enhances tumor invasiveness and thus may increase metastatic potential (3,27). Tumor cell invasion involves the active movement of cells across the extracellular matrix (41). Adhesion to extracellular matrix, a critical initial step in the metastatic process, has been found to be CD44-dependent in several tumors (26,32,42,43). CD44 is a receptor for hyaluronate, a major glycosaminoglycan component of the extracellular matrix. In this capacity, CD44 also serves to induce co-clustering with MMP-9 and can therefore promote MMP-9 activity, tumor invasion, and angiogenesis (25,44). Our findings indicate that tumor COX-2 overexpression in human NSCLC constitutes an important driving force for CD44 induction. Thus, COX-2 expression may form the basis for an important tumor-induced invasive pathway. The fact that CD44-induced MMP-2 and -9 have the capacity to activate latent TGF-␤ suggests an autocrine and paracrine pathway in which collagen deposition and further invasion may be enhanced (44,45). The activation of latent TFG␤ also provides an additional pathway for tumor-induced immune suppression (44,46,47). In addition, recent studies by Sun et al. (48) demonstrate that hyaluronate fragments have the capacity to up-regulate COX-2 by a CD44-dependent pathway. Thus, hyaluronate itself may serve to further enhance tumor COX-2 and CD44 expression leading to maintenance of the COX-2-dependent invasive phenotype.
Lung cancer is the leading cause of cancer death in men and women in the United States (49). Despite therapeutic efforts, 5-year survival in lung cancer patients is less than 15% (50). Defining new molecular targets will lead to more effective therapeutic strategies. Here we document for the first time a pathway whereby COX-2 overexpression leads to CD44-dependent invasion in NSCLC. These findings suggest that therapies targeting COX-2 may diminish the propensity for invasion and metastases in NSCLC.