Cyclooxygenase-2 Is Involved in S100A2-Mediated Tumor Suppression in Squamous Cell Carcinoma

Wan-Chi Tsai,1 Sen-Tien Tsai,5,6 Ying-Tai Jin,2,7 and Li-Wha Wu3,4

Institutes of 1Basic Medical Sciences, 2Oral Medicine, 3Molecular Medicine, and 4Cardiovascular Research Center College of Medicine, National Cheng Kung University; and Departments of 5Otolaryngology, 6Radiation Oncology, and 7Pathology, National Cheng Kung UniversityHospital, Tainan, Taiwan, Republic of China

Abstract motility, secretion, membrane permeability, protein synthesis, S100A2 is considered a putative tumor suppressor due and extracellular signal transduction, are regulated by members to its loss or down-regulation in several cancer types. of this family (3). Sixteen of the 20 members are tightly However,no mechanism has been described for the clustered on human chromosome 1q21. In this chromosome tumor suppressor role of S100A2. In this study,ectopic region, S100A2 together with 13 other S100 members are expression of S100A2 in the human malignant located within the epidermal differentiation complex. This squamous cell carcinoma cell line KB resulted in a region encodes many other genes that are expressed in significant inhibition of proliferation,migration,and epidermal keratinocytes (4). Among these genes, expression invasion. Moreover,S100A2 significantly reduced the of involucrin is highly correlated with epidermal differentiation number of colonies (z0.5 mm) formed in semisolid agar (5). The location of S100 genes in epidermal differentiation and decreased tumor growth and burden in nude mice. complex has heightened interest in their role in differentiation cDNA microarray analysis was used to compare mRNA of the epidermis. expression profiles of vector- and S100A2-expressing Expression levels of S100 proteins vary considerably in isogenic cells. Among the genes deregulated by S100A2, different tumors and with respect to the progression of the expression of cyclooxygenase-2 (COX-2) mRNA was malignancy (2, 6). Multiple sequence alignments clearly show significantly suppressed by S100A2 (2.4-fold). Western that S100 proteins can be divided into four subgroups (7). blot analysis confirmed that S100A2 reduced the S100A2, S100A3, S100A4, S100A5, and S100A6 are in the expression of COX-2 protein in stably and transiently same subgroup, indicating that these proteins are evolution- transfected KB and RPMI-2650 cells. COX-2 is frequently arily related. Although these proteins display high levels of overexpressed in various types of cancer and plays sequence similarity, their characteristics are distinct. S100A4 an important role in tumor progression. Partial is overexpressed in many types of tumors and is considered a restoration of COX-2 expression attenuated the well-established marker for tumor progression, invasion, antitumor effect of S100A2 both in vitro and in vivo. metastasis, and poor survival prognosis (8). S100A6 is Although the interplay between S100A2 and COX-2 markedly up-regulated in many types of tumor cells, remains to be clarified,these findings first showed a including melanoma, adenocarcinoma, and neuroblastoma potent antitumor role of S100A2 in squamous cell (9). Unlike the relationship of other related S100 members carcinoma partly via reduced expression of COX-2. to carcinogenesis, S100A2 is the first putative tumor (Mol Cancer Res 2006;4(8):539–47) suppressor protein. The S100A2 (formerly called CaN19 or S100L) gene was Introduction first identified via subtractive hybridization screening (10). S100 family (the largest superfamily of EF-hand calcium- Markedly reduced level of S100A2 has been shown in breast binding proteins, composed of at least 20 members) has tumor biopsies (11), melanomas, and esophageal, lung, and multiple functions in various cell types and tissues (1, 2). Many other cancer types (12-14) and related to the prognosis of cellular processes, including cell proliferation, differentiation, certain cancers (15, 16). Reduced nuclear staining of S100A2 in early-stage oral cancers correlates with shorter disease-free survival (17). Therefore, expression of S100A2 was proposed as a valuable prognostic marker. By contrast, S100A2 over- Received 12/30/05; revised 5/1/06; accepted 5/24/06. expression was recently found to correlate with prognosis in Grant support: National Science Council in Taiwan grant NSC-93-2314-B006- ovarian, gastric, and lung cancers (18-20). Taken together, the 041 and Department of Health grant DOH-TD-B-111-004 (S.-T. Tsai); NSC94- role of S100A2 in carcinogenesis remains controversial. 3112-B-006-004-Y (Y.-T. Jin); NSC94-2320-B-006-012 (L.-W. Wu). The costs of publication of this article were defrayed in part by the payment of To clarify this role, we employed stable clones of page charges. This article must therefore be hereby marked advertisement in squamous cell carcinoma (SCC) ectopically expressing accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Requests for reprints: Li-Wha Wu, Institute of Molecular Medicine, College of S100A2. The effects of S100A2 on cell growth, motility, Medicine, National Cheng Kung University, 1 University Road, Tainan 70101, invasion, colony-forming ability, and tumorigenesis in vivo Taiwan, Republic of China. Phone: 886-6-2353535, ext. 3618; Fax: 886-6- were examined. We show that S100A2 could exert its 2095845. E-mail: [email protected] Copyright D 2006 American Association for Cancer Research. antitumor activity by reducing expression of cyclooxyge- doi:10.1158/1541-7786.MCR-05-0266 nase-2 (COX-2).

Results enumeration of cells at the indicated time points) was Decreased Expression of S100A2 mRNA in Oral Cancer examined. Four days after plating, growth rates were Lines significantly slower in S100A2-expressing clones than in both We previously showed down-regulation of S100A2 mRNA parental KB and vector control cells (P < 0.001; Fig. 2B). The in oral cancer cells using cDNA microarray analysis (17). The growth rates of parental KB and vector control cells were not expression of S100A2 mRNA in primary normal oral altered by transfection of vector alone. Fluorescence-activated keratinocytes (NOK) and various cell lines (one precancerous cell sorting analysis of the phase distribution of cells dysplastic oral keratinocyte line DOK and nine established oral synchronized in G1-S by double thymidine treatment and SCC lines) was further examined using semiquantitative reverse harvested at 0, 4, 8, 12, and 22 hours showed no difference at transcription-PCR and found to be significantly down-regulated 0, 4, and 8 hours in the distribution of cycling cells between in precancerous and nine cancer cell lines compared with NOK vector and S100A2-2 clones (Fig. 2C). By 12 hours, 28% and (Fig. 1). KB cells expressed the least amount of S100A2 39% of the vector control and 43% and 48% of the S100A2-2 mRNA (Fig. 1, top) and were used to establish stable clones for cells were in G1 and S phases, respectively, and >30% of vector the following studies. control cells and only 9% of S100A2-2 cells were in G2-M phase. These data indicate that overexpression of S100A2 stalls Ectopically Expressed S100A2 Attenuated Cellular cells in G1 and S phases and delays their entry into G2-M. Growthvia Delayed Entry into G2-M Phase We first established two myc-tagged S100A2-expressing Ectopic Expression of S100A2 Reduced Cell Motility and clones (S100A2-1 and S100A2-2) and a pool of vector Invasive Ability control clones. Western blot analysis confirmed the presence Time-lapse video microscopy with ImageTool software was of S100A2 proteins in S100A2-1 and S100A2-2 clones used to measure the motility of sparsely seeded vector, (Fig. 2A). The protein level of S100A2 in both stable clones S100A2-1, and S100A2-2 clones. Compared with vector cells, was at least 4-fold higher than the endogenous level of S100A2 individual S100A2-1 and S100A2-2 cells showed significantly protein in NOK. No detectable S100A2 protein was detected in decreased (2-fold less) motility over the 6 hours of recording vector control cells. (Fig. 3A). The Matrigel invasion ability of S100A2-expressing To unravel the role of S100A2 overexpression in the cells (relative to that of vector cells) was also significantly malignant phenotype of KB cells, growth rate (direct decreased (Fig. 3B). The inhibitory effect of S100A2 on invasion was more dramatic than that on migration.

S100A2 Reduces Colony Size In vitro and Tumorigenesis In vivo Soft agar assay was used to measure the anchorage- independent growth for 21 days of S100A2-expressing cells. Although the size of colonies formed by vector control was much larger than those of S100A2-expressing clones, there was no significant variation in the total number of colonies individually formed by these three clones (Fig. 4A). However, the number of colony size z0.5 mm was significantly reduced by S100A2 (85-90% reduction; Fig. 4A). Consistent with a negative role of S100A2 in cell cycle progression, increased S100A2 expression reduced the size but not the number of colonies formed in soft agar. An in vivo tumorigenesis assay in tumor-bearing nude mice showed that significantly smaller tumors developed in mice injected with the S100A2-2 clone than in those injected with the vector clone at 28 days after injection (Fig. 4B). S100A2- overexpressing tumors weighed significantly less (reduced by 50-70%) than vector control tumors (Fig. 4C). Western blot analysis confirmed that the expression of human S100A2 protein was maintained in all S100A2-overexpressing tumors but not in vector clone tumors (Fig. 4D). Together, S100A2 attenuates the tumorigenic ability of malignant KB cells in vitro in vivo FIGURE 1. Diminished expression of S100A2 mRNA in oral cancer cell and . lines. Subconfluent cells of NOK and various cell lines were harvested for preparation of total RNA for semiquantitative reverse transcription-PCR. A. Expression level of S100A2 mRNA in NOK was higher than that in one S100A2 Reduced the mRNA and Protein Expression of precancerous DOK cell and nine oral cancer cell lines. B. S100A2/ COX-2 glyceraldehyde-3-phosphate dehydrogenase (GAPDH) ratio of NOK in the indicated cell lines. Columns, mean of three independent experiments; Affymetrix GeneChip Human Genome U133A arrays (with bars, SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001 versus NOK. filtering criteria of z1.8-fold change in expression and all

FIGURE 2. Levels of S100A2 protein in stable clones and their effects on cellular growth and cell cycle progression. A. Western blot analysis of S100A2 protein in NOK and stable clones using anti-S100A2 antibodies. Ectopically expressed myc-tagged S100A2 resulted in increased molecular weight of S100A2 compared with the endogenous S100A2. Level of myc-tagged S100A2 protein in S100A2-1 and S100A2-2 was higher than that of endogenous S100A2 in NOK. No myc-tagged S100A2 could be detected in vector control cells. Actin served as an internal control for protein loading. B. Growth rate of S100A2-1 and S100A2-2 was significantly decreased when compared with the parental and vector control cells. ***, P < 0.001 versus vector control cells. C. Distribution of G1, S, and G2-M for vector and S100A2-2 cells during cell cycle progression after double thymidine block. Following the release, cells were harvested at 0, 4, 8, 12, and 22 hours. DNA histograms for vector and S100A2-2 cells: histogram with percentages of cycling cells in G1, S, and G2-M shows one representative experiment of four independent repeats with similar data.

‘‘present calls’’ in at least one group) were used to compare vector clone, the level of COX-2 protein in S100A2-2-COX-2 the mRNA expression profile in vector and S100A2-express- was at least thrice higher than that in the S100A2-2-control ing isogenic cells. Two hundred seventeen genes were clone (Fig. 7A). Using a growth curve assay, the growth rate of differentially regulated by S100A2 (data not shown). These COX-2-expressing S100A2-2-COX-2 cells was significantly genes were further categorized into at least nine functional increased compared with S100A2-2-control cells (P < 0.01; groups based on their annotation (Fig. 5). The mRNA Fig. 7B). Reintroduction of a COX-2 cDNA expression expression of two up-regulated genes (S100A2 and thrombo- construct also significantly restored the ability of S100A2-2 modulin [TM]) and two down-regulated genes (COX-2 and cells to invade Matrigel (Fig. 7C). The abilities of S100A2-2- cyclin D1) were validated with semiquantitative reverse COX-2 to grow and to invade Matrigel were, however, not as transcription-PCR (Fig. 6A), indicating a robustness of the high as those of control vector cells. Furthermore, the S100A2- microarray analysis used in this study. Western blot analysis 2-COX-2 clone manifested a significant increase in the number further confirmed that the difference in COX-2 mRNA levels of colonies (P < 0.05; Fig. 7D). translated into a significant suppression of COX-2 protein To confirm the positive role of COX-2 in S100A2- (z10-fold) in the S100A2-expressing cells (Fig. 6B). This expressing cells in vivo, we inject nude mice, respectively, decreased expression of COX-2 protein was also recapitulated with S100A2 transfectants ectopically expressing vector control in both KB and RPMI-2650 cells transiently transfected with and COX-2. KB cells expressing vector serves as a negative S100A2-expressing vectors (Fig. 6C and D). Taken together, control. As shown in Fig. 7E and F, COX-2 significantly these data suggest that COX-2 could be one of the attenuated the antitumor effect of S100A2 by slowing down downstream targets of S100A2. tumor growth and burden (P < 0.05).

COX-2 Caused Partial Loss of S100A2-Mediated Anti- Discussion tumor Effects Although many articles suggest that S100A2 may act as a Stable clones, S100A2-2-control and S100A2-2-COX-2, tumor suppressor, none provide direct evidence that S100A2 were established from S100A2-2 cells transfected with either is such a gene. In this report, the human SCC cell line KB empty vector or a COX-2 cDNA expression construct to was used as a model to show a tumor suppressor role for measure the effects of COX-2 restoration on cellular growth, S100A2. Overexpression of S100A2 in KB cells not only invasion, and colony-forming ability in semisolid agar. Western diminished the ability of these cells to grow, migrate, and blot analysis confirmed the ectopic expression of COX-2 invade Matrigel but also suppressed their tumorigenicity protein in the S100A2-2-COX2 stable clone. Although in vitro and in vivo. Comparison between mRNA expression expression of COX-2 was still lower than that in the KB profiles of vector control and S100A2-expressing stable clones

of heterozygosity in chromosome 1q is also frequently observed (22). These observations are consistent with our finding that ectopically expressed S100A2 possesses antitumor effects in vitro and in vivo. The notion that S100A2 is a tumor suppressor gene has also been challenged. Xia et al. reported that (a) S100A is expressed at high levels in hyperplastic perilesional skin and (b) its expression together with that of keratin 6 is induced by calcium, arguing for a role of S100A2 in regenerative differentiation (23). S100A2 was overexpressed in tumor tissues of stomach, lung, and ovary and increased expression positively correlated with poor prognosis (18-20). The tissue-specific effect of S100A2 thus cannot be ruled out. The extensive in vitro and in vivo approaches used in this study support the view that S100A2 behaves as an antitumor gene in human SCC. Although no S100A2 knockout mice have been produced, generation of S100A2 knockout animals in a tissue-specific fashion may be helpful in teasing apart the exact role of S100A2 in carcinogenesis. Overexpression of S100A2 significantly slowed cellular growth both in vitro and in vivo. Because any alteration on the ability of cells to attach substratum, proliferate, or induce apoptosis is likely to affect cellular growth, we also conducted cell adhesion assay and apoptotic index measurement following serum withdrawal in addition to growth curve and cell cycle analyses. There was no significant variation in the ability of vector control and S100A2-expressing cells to attach substratum or in the apoptotic index induced by serum withdrawal (data not shown). Therefore, G1- and S-phase stalling by S100A2 as shown in the cell cycle analysis is likely responsible for the slower growth rate of S100A2-expressing cells in growth curve analysis and in vivo tumor formation. The FIGURE 3. Attenuation of the migration and invasion ability of KB cells involvement of S100A2 in cell cycle regulation was consistent on S100A2 overexpression. A. After cell seeding in duplicates onto culture with previous findings of cell cycle–regulated pattern of dishes for 16 to 24 hours, the distance of individual cell migration was traced for 6 hours. Pixels represent migration distance. The distance of S100A2 mRNA in cycling cells and a biphasic induction of S100A2-expressing clones was significantly decreased by at least 2-fold S100A2 mRNA during cell cycle progression triggered by compared with that of control cells. B. Cells invading Matrigel after growth factor induction (11). Although the interaction of 24 hours were detected as blue stained cells on the opposite side of membrane. Top, representative photomicrographs of vector, S100A2-1, S100A2 with p53 resulting in subsequent activation of p53 and S100A2-2 cells; bottom, quantitative result of invasion ability. transcriptional activity may contribute to slower cell growth, ***, P < 0.001 versus vector control cells. The ability of S100A2- expressing cells to invade Matrigel was significantly attenuated compared little or no p53 could be detected in the KB cells (24, 25). The with vector control cells. Representative of two independent experiments. antitumor effect of S100A2 we reported is likely through a p53- independent pathway. More studies are needed to define the role of S100A2 in the mechanism of cell cycle progression. indicated that 217 genes were differentially regulated by Among 217 deregulated genes, the expression of COX-2 S100A2. Thus, genes involved in a wide spectrum of cell mRNA and protein was consistently repressed by S100A2. behaviors (e.g., growth, apoptosis, migration, invasion, COX-2 is a key enzyme that catalyzes the rate-limiting step in metabolism, and inflammation) participate in the antitumor prostaglandin biosynthesis and is inducible in response to effect of S100A2. inflammatory cytokines, growth factors, and tumor promoters Of the >20 S100 family members, only S100A2 was (26). Expression of COX-2 is increased at inflammatory sites predicted to be a tumor suppressor gene for several reasons. and in various types of human cancer, particularly colorectal, First, expression of S100A2 is down-regulated in many types of breast, prostate, and oral cancers (26, 27). Transgenic mice cancer tissues, including breast, lung, and esophagus (12-14). expressing human COX-2 in mammary glands develop focal Second, treatment of mammary carcinoma cells with the mammary gland hyperplasia, dysplasia, and metastatic tumors inhibitor of DNA methylation 5-aza-2¶-deoxycytidine resulted (28). Conversely, knocking out expression of COX-2 notably in the enhanced expression of S100A2 mRNA, suggesting that reduces the development of intestinal tumors and skin promoter inactivation by methylation is involved in decreased papillomas in mice (29, 30). COX-2 thus plays an important expression of S100A2 (10). Third, S100A2 is located within the role in inflammation and carcinogenesis (31). epidermal differentiation complex on chromosome 1q21 and More than 15% of all malignancies are initiated by inflam- this region is frequently rearranged in cancers (21). Fourth, loss mation. Long-term use of nonsteroidal anti-inflammatory drugs

decreases cancer risk (40-50% reduction in colon cancer; ref. 32). cyclin D1 could not be attenuated by ectopically expressed A link between inflammation and cancer has long been COX-2 in S100A2-expressing cells (data not shown). The suspected, but its molecular nature remained ill-defined until a exact role of other S100A2-deregulated genes, including TM and recent finding that recruitment of host inflammatory cells by cyclin D1 in the antitumor effect of S100A2, remains to be interleukin-8, a transcriptional target of oncogene Ras, plays characterized. a critical role tumor angiogenesis and growth (33). The Expression of COX-2 protein can be regulated by Ras hypothesis that overexpression of S100A2 exerts its tumor oncogene and tumor suppressor p53 in positive and negative suppression function through COX-2 down-regulation is further manners, respectively (34, 35). The mechanisms involved in strengthened by the finding that partial restoration of COX-2 deregulation of COX-2 expression include an alteration of expression attenuated the antitumor effect of S100A2. Incom- mRNA stability and transcriptional promotion via direct plete restoration was probably due to lower COX-2 expression binding to COX-2 promoter. We were not able to detect any levels in S100A2-expressing cells transfected with COX-2 direct modulation of COX-2 promoter activity by S100A2 than in KB vector control cells (Fig. 7A). Alternatively, other using dual-luciferase reporter assays (data not shown). downstream effectors may be also critical for the antitumor Although several upstream regulatory elements for transcription effects mediated by S100A2. Consistent with the latter notion, factors have been characterized in the promoter region of both the induced expression of TMand the reduced expression of COX-2 gene (36), none of these transcription factors was in the

FIGURE 4. Decreased in vitro and in vivo tumorigenicity by ectopically expressed S100A2. A. Numbers of colonies formed by control vector, S100A2-1, and S100A2-2 clones appearing on soft agar after 21 days of growth were enumerated. Top, representative photomicrographs of colonies formed by vector, S100A2-1, and S100A2-2 cells; bottom, quantitative data of colony-forming ability. Gray columns, total number of colonies; black columns, number of colonies (z0.5 mm). **, P < 0.01 versus control vector; N.S., not significant versus control vector. S100A2 significantly reduced the number of colonies (z0.5 mm) but not the total number of colonies on soft agar. B. Size of subcutaneous tumors formed by control vector and S100A2-2 clones was measured for 3 weeks by a caliper every 2 to 3 days. *, P < 0.05; **, P < 0.01; ***, P < 0.001 versus vector control. Increased expression of S100A2 diminished tumor formation in nude mice. C. Tumor sizes (top) and weights (bottom) derived from S100A2-2 clones were significantly reduced compared with those derived from vector control. D. Western blot analysis using anti-S100A2 antibodies confirmed the presence of S100A2 protein in all five tumors isolated from S100A2-2 tumor-bearing nude mice but not those from their matched control mice. Representative of two independent experiments.

FIGURE 5. Functional categories of genes that are significantly altered by S100A2. Gene annotations were ob- tained from Affymetrixand PubMed databases. X axis, names of functional categories; Y axis, numbers of genes in each functional category. Note that the annotation is constantly evolving.

list of differentially regulated genes triggered by overexpression Transduction Laboratories (San Jose, CA). The antibodies to of S100A2. Suppression of COX-2 by overexpression of cyclin D1 and actin were from Santa Cruz Biotechnology S100A2 is probably through an indirect mediation. More (Santa Cruz, CA). Anti-a-tubulin antibodies were from Neo- studies are needed to define the role of reduced COX-2 Marker (Fremont, CA). Chemiluminescence Reagent Plus expression in the S100A2-expressing SCC cells. was from NEN Life Science Products (Boston, MA). RNase In conclusion, S100A2 possesses a profound antitumor effect A, propidium iodine, and other chemicals were purchased from in vitro and in vivo using SCC as a cancer model. Using Sigma (St. Louis, MO). Matrigel was from BD Biosciences oligonucleotide microarray analysis, we found that COX-2 (Bedford, MA). COX-2 cDNA expression vector was a was significantly suppressed in S100A2-expressing stable clones generous gift of Dr. Kan Wai-Ming (Department of Pharma- compared with vector control cells. Ectopic expression of cology, National Cheng Kung University, Tainan, Taiwan). S100A2 also significantly suppressed the expression of COX-2 in transiently transfected SCC cells. Consistent with a role of COX-2 in tumor progression, partial restoration of Semiquantitative Reverse Transcription-PCR COX-2 expression in S100A2-expressing cells attenuated DNA-free total RNA was isolated from cultured cells using A the S100A2-mediated antitumor effect in vitro and in vivo.To TRIzol reagent, and 1 g was reverse transcribed with our knowledge, we are the first to suggest that the antitumor effect oligo(dT12-18) primers. The cDNA mixture was then used as of S100A2 is in part mediated by down-regulation of COX-2. template for gene-specific PCR. The primer sequences were 5¶-ACCACAGTCCATGCCATCAC-3¶ and 5¶-TCCACCACC- CTGTTGCTGTA-3¶ for glyceraldehyde-3-phosphate dehydro- Materials and Methods genase (GAPDH ) and 5¶-TCAAGCTGAGTAAGGGGG-3¶ and Materials 5¶-ATCCATGGCAGGAAGTCAAG-3¶ for S100A2. The PCR Human Caucasian DOK were purchased from and main- conditions were 94jC (15 seconds), 50jC (30 seconds), and tained as described by the European Collection of Cell 72jC (30 seconds). The cycle number for linear amplification Culture. NOK, five Caucasian oral cancer lines (KB, CAL- of GAPDH and S100A2 was 21 and 32, respectively. The 27, SCC-9, SCC-15, and SCC-25), three Taiwanese oral cell experiment was independently repeated thrice. lines (OC-2, OC-3, and OEC-M1), and a human nasal septum SCC line (RPMI-2650) were maintained as described previously (17, 37, 38). HSC-3 cells derived from human tongue Construction of a Mammalian Expression Vector carcinoma with lymph note metastasis were from the Japanese A mammalian expression vector for myc-tagged S100A2 Collection of Research Bioresources Cell Bank of Japan and was generated as follows: the coding region of human S100A2 maintained as described previously (39). All other culture was PCR-amplified and cloned in-frame into the EcoRI and medium, fetal bovine serum, and antibiotics were purchased BamHI sites of pcDNA3.1-myc/His-A(+) (Invitrogen). The from Life Technologies (Grand Island, NY). LipofectAMINE myc-tagged S100A2 cDNA was then subcloned into a 2000 transfection reagent, TRIzol reagents, and reagents for pOPI3CAT-derived mammalian expression vector driven by reverse transcription-PCR were from Invitrogen (Carlsbad, a Rous sarcoma virus promoter. Automatic DNA sequencing CA). Oligonucleotide primers were from MDbio, Inc. (Taipei, analysis verified the direction and coding sequence of Taiwan). Antibodies to S100A2 and COX-2 were from BD S100A2.

Transfection and Establishment of Stable Clones Migration Assay KB or the indicated cells were seeded overnight into 35-mm Culture dishes (35 mm), coated overnight with type I culture dishes in appropriate growth medium supplemented collagen (5 Ag/mL) from rat tails, were seeded with 20,000 with 10% fetal bovine serum before transfection with 1 Ag cells in growth medium. After 16 hours, cells were refed with S100A2-expressing vector or empty vector using Lipofect- MEM containing 1% fetal bovine serum and 20 mmol/L AMINE 2000. For transient transfection, cells were harvested HEPES (pH 7.0), placed on a heated (37jC) microscope for isolation of total RNA or protein 24 to 48 hours after stage, and filmed for 6 hours using a time-lapse video camera. transfection. For stable transfection, stable clones were selected The movement of randomly chosen cells over 15 minutes was 24 hours after transfection over a 2-week period in growth evaluated by exporting the video images to the graphics medium containing 800 Ag/mL G418. S100A2-expressing program, The University of Texas Health Science Center at clones were then expanded into individual cell clones. Stable San Antonio ImageTool software. Only isolated cells and the clones expressing COX-2 were established accordingly. nondividing cells that did not leave the frames during recording were evaluated. For each clone, the migration paths GrowthRate Assay of 10 to 15 cells were measured, summed, averaged, and Parental KB and stable clones (1.6 Â 104 cells per well) represented as pixels. were seeded in duplicate onto 24-well culture dishes. Cells were harvested for direct counting at the indicated days after seeding. Invasion Assay This experiment was independently repeated thrice. Invasion assays were done in 24-well Transwell units with 8-Am-pore polycarbonate membranes (Corning Costar, 5 Cell Cycle Analysis and Double Thymidine Block Cambridge, MA). Cells (1 Â 10 per well) were added to upper Vector and S100A2-2 clones (1 Â 106 cells seeded per chambers (filter coated with 1 mg/mL Matrigel) in 100 ALof 60-mm dish in growth medium and incubated 16 hours) were the growth medium. Lower chambers were filled with 500 AL processed as follows: medium was changed to starvation growth medium. After 24-hour incubation, cells that remained medium containing 2 mmol/L thymidine, incubation for in the Matrigel or attached to the upper side of the filter were 10 hours, cells were washed twice, cells were fed with regular removed with cotton swabs. Cells that had migrated through growth medium for 14 hours, cells were fed with starvation the membrane to the lower surface were stained with Giemsa medium with 2 mmol/L thymidine for another 10 hours, and solution and counted in five different fields under a light both adherent and floating cells harvested for flow cytometry at microscope at Â400 magnification. Each experiment was done 4-hour intervals for 24 hours. Approximately 1 Â 106 cells were in duplicate wells and repeated twice, and results were washed with PBS, fixed with 70% cold ethanol, treated with 100 expressed as mean F SD. Ag/mL RNase A, and stained with 40 Ag/mL propidium iodine. The cell cycle distribution was analyzed by FACSCalibur flow Soft Agar Assay cytometer (Becton Dickinson, San Jose, CA). The percentage of Briefly, 5 Â 103 cells trypsinized to form a single-cell cells present in each phase of the cell cycle was determined suspension were seeded in duplicate onto 60-mm plastic culture using ModFit software (Verity Software House, Topsham, ME). dishes in MEM containing 0.35% agarose overlying a solidified

FIGURE 6. Reduced expression of COX-2 in S100A2-expressing stable clones. Numbers in paren- theses, average fold increase (+) or decrease (À)of mRNA expression in microarray data analysis. h-Actin or a-tubulin served as a loading control. A. RT-PCR analysis indicated that the expression of TM was increased, whereas expression of cyclin D1 was repressed by S100A2. B. Western blot analysis indicates that COX-2 protein was suppressed in both S100A2-expressing clones, S100A2-1 and S100A2-2. C and D. Increased expression of S100A2 is accompanied with the reduced expression of COX-2 in transiently transfected KB and RPMI-2650 cells, respectively.

FIGURE 7. Partial restoration of malignant behaviors of S100A2-expressing cells by reintroduction of COX-2 cDNA S100A2-2 cells containing empty vector (S100A2-control) and the same cells expressing ectopic COX-2 (S100A2-2- COX-2) were established. KB cells with vector only (Vector) served as a negative control for S100A2-expressing cells. A. Western blot analysis confirmed increased expression of COX-2 protein in S100A2- 2-COX2 compared with that in S100A2-2-control cells. B. Direct enumeration of vector, S100A2-2-control, and S100A2-2-COX2 during 4 days of growth after seeding. In- creased expression of COX-2 significantly increased the growth rate of S100A2-2- COX-2 cells compared with the S100A2-2-control cells. **, P < 0.01 versus S100A2-2- control. C. Invasion assay indicates the ectopically expressed COX-2 significantly restored the ability of cells to invade Matrigel compared with their control cells. ***, P < 0.001 versus S100A2-2-control. D. S100A2-2-COX-2 manifested a higher ability to form colonies (z0.5 mm) in soft agar compared with the control cells. *, P < 0.05 versus S100A2-2- control. Representative of three independent experiments with similar data. E and F. S100A2- 2-COX-2 cells not only formed tumors in nude mice at a higher growth rate but also induced larger tumors than S100A2-2- control cells. *, P <0.05. Consistent with the previous finding, S100A2 manifested a significant antitumor effect compared with KB-vector cells. #, P < 0.05. Representative of three independent experiments with similar data.

0.7% agarose underlayer. After the cell-containing layer mice were euthanized 28 days after inoculation, and the solidified, 0.7% agarose was overlaid. Fresh growth medium tumors were excised, weighed, and used for isolation of total was added every 4 days. Plates were incubated at 37jCin5% proteins. This experiment was independently repeated twice to CO2 for 21 days. Following staining with Giemsa solution, blue thrice. colonies (z0.5 and <0.5 mm) were enumerated for each stable clone. Western Blot Analysis Cells were lysed in boiled lysis buffer containing 1% SDS In vivo Tumorigenesis Assay and 10 mmol/L Tris-HCl (pH 7.4). Following centrifugation at The tumorigenic activity of cells was examined in 8-week- top speed for 10 minutes, the protein concentration in each old nude BALB/c nu/nu mice (5-7 per group). Tumor cells lysate was measured (Bradford assay). Equal amounts of total (3.3 Â 106) were suspended in 0.2 mL PBS and subcutane- protein from each sample were fractionated by SDS-PAGE, ously injected into the flank using a 27-gauge needle. Tumors blotted onto a polyvinylidene difluoride membrane, and probed were generally palpable at 5 to 7 days after inoculation and with the indicated antibody and then a secondary antibody. The tumor volumes were measured (using a caliper and calculated hybridized immunocomplex was detected by Renaissance as length Â width2 Â 0.5) twice weekly until day 28. The Chemiluminescence Reagent Plus.

Microarray Hybridization and Data Analysis 15. Lauriola L, Michetti F, Maggiano N, et al. Prognostic significance of the Ca(2+) binding protein S100A2 in laryngeal squamous-cell carcinoma. Int The mRNA expression profiles of vector and S100A2-2 J Cancer 2000;89:345 – 9. clones were analyzed in triplicate independent experiments. 16. Kyriazanos ID, Tachibana M, Dhar DK, et al. Expression and prognostic Raw data were normalized and analyzed using GeneSpring significance of S100A2 protein in squamous cell carcinoma of the esophagus. software version 7.2 (Silicon Genetics, Redwood City, CA). Oncol Rep 2002;9:503 – 10. After data transformation (to convert any negative value to 17. Tsai ST, Jin YT, Tsai WC, et al. S100A2, a potential marker for early recurrence in early-stage oral cancer. Oral Oncol 2005;41:349 – 57. 0.01), two types of normalization were done: a per-chip on 50th 18. Hough CD, Cho KR, Zonderman AB, Schwartz DR, Morin PJ. Coordinately percentile method and a per-gene on median normalization up-regulated genes in ovarian cancer. Cancer Res 2001;61:3869 – 76. method. To identify genes with statistically significant differ- 19. El-Rifai W, Moskaluk CA, Abdrabbo MK, et al. Gastric cancers overexpress ences between vector and S100A2-2 clones in three biological S100A calcium-binding proteins. Cancer Res 2002;62:6823 – 6. repeats, one-way ANOVA nonmultiple testing corrections with 20. Diederichs S, Bulk E, Steffen B, et al. S100 family members and the similarity >90% were done. trypsinogens are predictors of distant metastasis and survival in early-stage non-small cell lung cancer. Cancer Res 2004;64:5564 – 9. 21. Whang-Peng J, Knutsen T, Gazdar A, et al. Nonrandom structural and Statistical Analysis numerical chromosome changes in non-small-cell lung cancer. Genes Chromo- Values are mean F SD of experiments done in duplicate or somes Cancer 1991;3:168 – 88. triplicate. Statistical significance was tested by Student’s t test 22. Li J, Liu Z, Wang Y, Yu Z, Wang M, Zhan Q. Allelic imbalance of chromosome 1q in esophageal squamous cell carcinomas from China: a novel for either paired or unpaired data as appropriate. Statistical region of allelic loss and significant association with differentiation. Cancer Lett significance was defined as P < 0.05, P < 0.01, or P < 0.001. 2005;220:221 – 30. 23. Xia L, Stoll SW, Liebert M, et al. CaN19 expression in benign and malignant hyperplasias of the skin and oral mucosa: evidence for a role in regenerative Acknowledgments differentiation. Cancer Res 1997;57:3055 – 62. We thank Dr. David D. Donner for helpful suggestions and comments about the article and Dr. Tung-Yiu Wong for help in isolation of NOK. 24. Mueller A, Schafer BW, Ferrari S, et al. The calcium binding protein S100A2 interacts with p53 and modulates its transcriptional activity. J Biol Chem 2005;280:29186 – 93. References 25. Chen YJ, Jin YT, Shieh DB, Tsai ST, Wu LW. Molecular characterization of 1. Eckert RL, Broome AM, Ruse M, Robinson N, Ryan D, Lee K. S100 proteins angiogenic properties of human oral squamous cell carcinoma cells. Oral Oncol in the epidermis. J Invest Dermatol 2004;123:23 – 33. 2002;38:699 – 705. 2. Heizmann CW, Fritz G, Schafer BW. S100 proteins: structure, functions and 26. Zha S, Yegnasubramanian V, Nelson WG, Isaacs WB, De Marzo AM. pathology. Front Biosci 2002;7:d1356 – 68. Cyclooxygenases in cancer: progress and perspective. Cancer Lett 2004;215:1 – 20. 3. Donato R. Intracellular and extracellular roles of S100 proteins. Microsc Res 27. Itoh S, Matsui K, Furuta I, Takano Y. Immunohistochemical study on Tech 2003;60:540 – 51. overexpression of cyclooxygenase-2 in squamous cell carcinoma of the oral cavity: its importance as a prognostic predictor. Oral Oncol 2003;39:829 – 35. 4. Marenholz I, Volz A, Ziegler A, et al. Genetic analysis of the epidermal differentiation complex (EDC) on human chromosome 1q21: chromosomal 28. Liu CH, Chang SH, Narko K, et al. Overexpression of cyclooxygenase-2 is orientation, new markers, and a 6-Mb YAC contig. Genomics 1996;37:295 – 302. sufficient to induce tumorigenesis in transgenic mice. J Biol Chem 2001;276:18563–9. 5. Li ER, Owens DM, Djian P, Watt FM. Expression of involucrin in normal, 29. Chulada PC, Thompson MB, Mahler JF, et al. Genetic disruption of Ptgs-1, hyperproliferative and neoplastic mouse keratinocytes. Exp Dermatol 2000;9: as well as Ptgs-2, reduces intestinal tumorigenesis in Min mice. Cancer Res 2000; 431 – 8. 60:4705 – 8. 6. Donato R. S100: a multigenic family of calcium-modulated proteins of the 30. Tiano HF, Loftin CD, Akunda J, et al. Deficiency of either cyclooxygenase EF-hand type with intracellular and extracellular functional roles. Int J Biochem (COX)-1 or COX-2 alters epidermal differentiation and reduces mouse skin Cell Biol 2001;33:637 – 68. tumorigenesis. Cancer Res 2002;62:3395 – 401. 7. Marenholz I, Heizmann CW, Fritz G. S100 proteins in mouse and man: from 31. Chun KS, Surh YJ. Signal transduction pathways regulating cyclooxygenase- evolution to function and pathology (including an update of the nomenclature). 2 expression: potential molecular targets for chemoprevention. Biochem Biochem Biophys Res Commun 2004;322:1111 – 22. Pharmacol 2004;68:1089 – 100. 8. Semov A, Moreno MJ, Onichtchenko A, et al. Metastasis-associated protein 32. Li Q, Withoff S, Verma IM. Inflammation-associated cancer: NF-nBisthe S100A4 induces angiogenesis through interaction with Annexin II and lynchpin. Trends Immunol 2005;26:318 – 25. accelerated plasmin formation. J Biol Chem 2005;280:20833 – 41. 33. Sparmann A, Bar-Sagi D. Ras-induced interleukin-8 expression plays a 9. Lesniak W, Szczepanska A, Kuznicki J. Calcyclin (S100A6) expression is critical role in tumor growth and angiogenesis. Cancer Cell 2004;6:447 – 58. stimulated by agents evoking oxidative stress via the antioxidant response 34. Sheng H, Shao J, Dubois RN. K-Ras-mediated increase in cyclooxygenase 2 element. Biochim Biophys Acta 2005;1744:29 – 37. mRNA stability involves activation of the protein kinase B1. Cancer Res 2001;61: 10. Lee S, Tomasetto C, Sager R. Positive selection of candidate tumor- 2670 – 5. suppressor genes by subtractive hybridization. Proc Natl Acad Sci U S A 1991; 35. Subbaramaiah K, Altorki N, Chung WJ, Mestre JR, Sampat A, Dannenberg 88:2825 – 9. AJ. Inhibition of cyclooxygenase-2 gene expression by p53. J Biol Chem 1999; 11. Lee SW, Tomasetto C, Swisshelm K, Keyomarsi K, Sager R. Down- 274:10911 – 5. regulation of a member of the S100 gene family in mammary carcinoma cells and 36. Kosaka T, Miyata A, Ihara H, et al. Characterization of the human gene reexpression by azadeoxycytidine treatment. Proc Natl Acad Sci U S A 1992;89: (PTGS2) encoding prostaglandin-endoperoxide synthase 2. Eur J Biochem 1994; 2504 – 8. 221:889 – 97. 12. Maelandsmo GM, Florenes VA, Mellingsaeter T, Hovig E, Kerbel RS, 37. Lin SC, Liu CJ, Chiu CP, Chang SM, Lu SY, Chen YJ. Establishment of OC3 Fodstad O. Differential expression patterns of S100A2, S100A4 and S100A6 oral carcinoma cell line and identification of NF-nB activation responses to areca during progression of human malignant melanoma. Int J Cancer 1997;74: nut extract. J Oral Pathol Med 2004;33:79 – 86. 464 – 9. 38. Werner U, Kissel T. In-vitro cell culture models of the nasal epithelium: a 13. Ji J, Zhao L, Wang X, et al. Differential expression of S100 gene family in comparative histochemical investigation of their suitability for drug transport human esophageal squamous cell carcinoma. J Cancer Res Clin Oncol 2004;130: studies. Pharm Res 1996;13:978 – 88. 480 – 6. 39. Kamata N, Chida K, Rikimaru K, Horikoshi M, Enomoto S, Kuroki T. 14. Feng G, Xu X, Youssef EM, Lotan R. Diminished expression of S100A2, a Growth-inhibitory effects of epidermal growth factor and overexpression of its putative tumor suppressor, at early stage of human lung carcinogenesis. Cancer receptors on human squamous cell carcinomas in culture. Cancer Res 1986;46: Res 2001;61:7999 – 8004. 1648 – 53.

Mol Cancer Res 2006;4(8). August 2006 Downloaded from mcr.aacrjournals.org on September 26, 2021. © 2006 American Association for Cancer Research. Cyclooxygenase-2 Is Involved in S100A2-Mediated Tumor Suppression in Squamous Cell Carcinoma

Wan-Chi Tsai, Sen-Tien Tsai, Ying-Tai Jin, et al.

Mol Cancer Res 2006;4:539-547.

Updated version Access the most recent version of this article at: http://mcr.aacrjournals.org/content/4/8/539

Cited articles This article cites 39 articles, 15 of which you can access for free at: http://mcr.aacrjournals.org/content/4/8/539.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/4/8/539.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/4/8/539. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.