PGE 2 Production in Oral Cancer Cell Lines is COX-2-dependent

Abstract

It has been suggested that epithelial cyclooxyge-nase-2 (COX-2) promotes oral carcinogenesis and carcinoma malignancy through increased prostaglandin E₂ (PGE₂) production. Although oral squamous cell carcinomas (OSCC) often express COX-2, they may also produce PGE₂ in a COX-1-dependent manner. We used 6 isolated cell lines to investigate which COX isoforms OSCC may use for PGE₂ production. COX-1 and -2 expression patterns divided the 6 OSCC cell lines into 3 distinct groups: both COX isoforms low, only COX-1 high, or both COX isoforms high. Multicolor immunohistofluorescence staining confirmed the COX-expression profiles in organotypic 3D cultures and the COX-2 dominance in OSCC tumors. Epidermal growth factor (EGF) stimulation induced COX-2 (but not COX-1) expression and increased PGE₂ production, which was attenuated by COX-2 (but not COX-1) specific inhibition or siRNA-mediated COX-2 gene knockdown. Thus, PGE₂ production in OSCC cell lines was COX-2-dependent.

INTRODUCTION

Cyclooxygenase (COX) catalyzes the rate-limiting step in prostaglandin (PG) synthesis from aracidonic acid (Fig. 1). Whereas the constitutively expressed COX-1 is important for homeostasis, COX-2 is up-regulated in inflammation and diseases. Increased epithelial COX-2 expression and PGE₂ production may promote carcinogenesis through cross-activation of the epidermal growth factor receptor (EGFR) (Fig. 1 ), and regular non-steroidal anti-inflammatory drug (NSAID) use reduces carcinoma incidence (Thun et al., 1991; Bosetti et al., 2006). Both oral dysplasia and oral squamous cell carcinomas (OSCC) express increased COX-2 levels (Chan et al., 1999; Renkonen et al., 2002; Atula et al., 2006). NSAIDs or COX-2 selective blockers may therefore prevent oral carcinogenesis (Lin et al., 2002) or be used as adjuvant treatment for oral carcinomas (Dannenberg et al., 2005), since they may inhibit PGE₂-dependent OSCC cell proliferation (Minter et al., 2003). Whether COX-2 expression has an independent prognostic value is still debated (Lim et al., 2004; Atula et al., 2006; Sakurai et al., 2007; Soland et al., 2008). However, if COX-2-negative OSCC produced PGE₂ via COX-1, as shown for ovarian cancer cell lines (Kino et al., 2005), it may explain some of this controversy, since OSCC express increased COX-1 levels (Shibata et al., 2005). Moreover, the COX-2 selective inhibitor NS-398 may inhibit both COX-1(IC₅₀ = 10.75 μM; Laufer et al., 1999) and the PGE₂ synthase (PGES; IC₅₀ = 20 μM; Thoren and Jakobsson, 2000). There are 3 PGES (Fig. 1), of which microsomal (m)PGES-1 is functionally coupled to COX-2 (Murakami et al., 2000), cytosolic (c)PGES to COX-1 (Tanioka et al., 2000), and mPGES-2 to both COX-isoforms (Murakami et al., 2003).

The aim of this study was to examine whether OSCC could produce PGE₂ through COX-1 and to elucidate which of the PGE₂ synthases the cell lines expressed.

MATERIALS & METHODS

Human OSCC Cell Lines

Six isolated human OSCC cell lines—PE/CA-PJ15 clone B11 (male, 45 yrs), PE/CA-PJ46 clone B5 (male, 63 yrs), and PE/CA-PJ49 clone D12 and clone E10 (male, 55 yrs), all established from tongue tissue, PE/CA-PJ34 clone C12 (male, 60 yrs) from the oral cavity and PE/CA-PJ41 clone D2 (female, 68 yrs) from oral squamous epithelium (a kind gift from Dr. Berndt and Dr. Kosmehl, Friedrich-Schiller University, Jena, Germany)—were cultured in standard medium (IMDM, Sigma-Aldrich, St. Louis, MO, USA), supplemented with 10% fetal bovine serum (FBS) (Cambrex, Verviers, Belgium), 2 mM L-glutamine, and 1% penicillin-streptamycin-fungizone (PSF) (Cambrex).

Culture of Primary Fibroblasts and Organotypic Cancer Models

Normal oral mucosal biopsies were incubated in dispase (0.25% w/v, Invitrogen, Carlsbad, CA, USA) for 24 hrs at 4°C, and then in collagenase (2 mg/mL, 30 min). Fibroblasts were collected by centrifugation, and grown in Dulbecco’s modified Eagle’s medium (DMEM) (Sigma) with supplements as described for cancer cell line culture.

Organotypic cultures were made by the growing of all 6 oral carcinoma cell lines on reconstituted collagen I biomatrices supplemented with fibroblasts for 10 days (Costea et al., 2003).

Oral Carcinoma Samples

Primary OSCC were snap-frozen in liquid nitrogen, mounted in OCT (Tissue Tek, Tokyo, Japan), and used for multicolor immunohistofluorescence staining to illustrate the COX-isoform expression pattern in situ.

Ethics Approval

All donors gave their written consent, and the study was approved by the regional ethical committee.

Immunohistofluorescence

We performed multicolor immunohistofluorescence staining on 4-μm cryosections by applying mixtures of pre-titrated antibodies to COX-1 (IgG2b), COX-2 (IgG1, both Cayman, Tallinn, Estonia), vimentin (IgG1), and rabbit anti-keratin (Dako, Glostrup, Denmark), followed by Alexa Fluor-594 and -488-conjugated subclass-specific goat anti-mouse IgG (Molecular Probes, Eugene, OR, USA) or Alexa Fluor-405-conjugated goat anti-rabbit IgG (Molecular Probes). The results were recorded by means of a Zeiss Axioplan 2 microscope (Carl Zeiss, Göttingen, Germany) equipped with a plan-Neofluar^® x40/1.3 oil lens and appropriate fluorochrome filters, including single-, double-, and triple-color filter blocks for the simultaneous examination of green, red, and blue emissions. Negative controls included substitution of primary antibodies with non-immune mouse and rabbit sera at 1/1000. Western blotting of recombinant COX isoforms ensured COX-2 antibody specificity (not shown).

siRNA-mediated COX-2 Knockdown

Cell lines B5, D2, and D12 were re-suspended in medium, then incubated in IMDM with 1% FBS, transfected (DharmaFECT, Dharmacon, Lafayette, CO, USA) with 50 or 60 nM COX-2 specific siRNA (5′GGA-UUC-UAU-GGA-GAA-AAC-Utt3′) (Ambion, Austin, TX, USA) or a scrambled control siRNA (Dharmacon) overnight, re-suspended in standard medium, and harvested after 24 (RNA) and 48 hrs (protein isolation).

Real-time RT-PCR

Total RNA was extracted (RNAeasy, QIAGEN, Valencia, CA, USA), and cDNA was synthesized (RT-RTCK-05, Eurogentec, Berlin, Germany) and stored at −20°C. A standard PCR reaction was performed with the following primer sequences: (COX-1) forward, 5′-CGA-TGA-GCA-GCT- TTT-CCA-GAC-3′, reverse, 5′-TTG-GAACTG-GAC-ACC-GAA-CAG-3′; (COX-2) forward, 5′-TGA-AAT-ATC-AGA-TAA-TTG-ATG-GAG-AGA- TG-3′, reverse, 5′-ACC-AGG-CAC-CAG-CC-AAA-G-3′; (COX-1 Intron 1 retained) forward, 5′-CAT-GAG-CCG-TGA-GTG-CGA-3′, reverse, 5′-GAG-CAA-GAG-ACT-CCC-TGC-AGA-3′ (Nurmi et al., 2005); (cPGES) forward, 5′-GAT-GAT-GAA-CAA-CAT-GGG-TGG-T-3′, reverse, 5′-TCT-CAA-AAT-CCA-GGT-GAT-GAC-AA-3′; (mPGES-1) forward, 5′-GGA-ACG-TGG-AGA-CCA-TCT-AC-3′, reverse, 5′-TCC-AGG-CGA-CAA-AAG-GGT-TA-3′; (mPGES-2) forward, 5′-AGC-CGA-ATC-TCG- CTG-ATT-T-3′, reverse, 5′-GCT-GCA-TCA-GGT-CAT-CGA-A-3′; (HPRT) forward, 5′-TTG-ACA-CTG-GCA-AAA-CAA-TGC-3′, reverse, 5′-GCT-TGC-GAC-CTT-GAC-CAT-CT-3′; and (TBP) forward, 5′-CGT-GGC-TCT-CTT-ATC-CTC-ATG-A-3′, reverse, 5′-GCC-CGA-AAC-GCC-GAA-TAT-A-3′.

Duplicate samples in SYBR-I PCR mix (RT-SN2X, Eurogentec) were run for 40 cycles (MX4000, Stratagene, La Jolla, CA, USA) with 60°C annealing for 1 min. Dissociation curves ensured product uniformity. EGF-stimulation (100 ng/mL) for 2, 4, 6, 8, 12, 16, and 24 hrs revealed a COX-2 mRNA peak at 4 hrs.

Western Blotting

Cells underwent lysis in MAPK lysis buffer (Zhang et al., 2006) with protease and phosphatase inhibitor cocktails (Pierce Biotechnology, Rockford, IL, USA). From 10 to 30 μg protein (BioRad, Munich, Germany) were separated by SDS-PAGE, electroblotted onto nitrocellulose membranes (BioRad), and probed with antibodies (Cayman) to COX-1 (mAb, 1 μg/mL), COX-2 (rabbit, 0.27 μg/mL), cPGES (rabbit, 0.5 μg/mL), mPGES-1 (rabbit, 1 μg/mL), and mPGES-2 (rabbit, 0.5 μg/mL), then visualized with alkali phosphatase-conjugated anti-rabbit IgG (1:10,000) (Sigma) or mouse IgG (1:10,000) (Dako), with ECF substrate (GE Healthcare, Uppsala, Sweden) in a scanner (Storm), and analyzed by ImageQuantTL (both GE Healthcare). Re-probing with anti-GAPDH (0.05 μg/mL, Abcam, Cambridge, UK) was used as loading and transfer control.

Six-, 12-, 24-, 48-, and 72-hour EGF-stimulated samples (100 ng/mL) showed a peak in COX-2 protein expression at 12 hrs.

PGE₂ ELISA

Cell culture supernatant was used undiluted in a PGE₂ immunoassay (R&D systems, Abingdon, UK), read in a Victor2 (EG&G Wallac, Perkin-Elmer, Wellesley, MA, USA), and analyzed by WorkOut 2.0 (DazDaq Limited, Brighton, UK).

RESULTS

COX Isoform Expression in OSCC Cell Lines

All unstimulated OSCC cell lines expressed variable COX-1 and COX-2 mRNA levels (Fig. 2A). Although COX-1 and COX-2 protein was detected in all cell lines, only one had high expression of both isoforms (Fig. 2B ). Thus, the protein expression pattern divided the cell lines into 3 groups: either low (B11, D2) or high expression of both COX isoforms (D12), and high for COX-1 (B5, C12, E10) only.

EGF stimulation induced COX-2 mRNA and protein expression in all cell lines, particularly in C12, D2, and D12, whereas the EGF-induced COX-2 protein increase in B5, B11, and E10 was visible only after image analysis. In contrast, COX-1 mRNA and protein expression was unaffected by EGF stimulation. The intron 1 retained COX-1 splice variant (Chandrasekharan et al., 2002) detected in human tissues (Qin et al., 2005) was expressed only at low levels in all cell lines (not shown).

Prostaglandin E₂ Synthase Isoform Expression

Cytosolic (c)PGES mRNA was the dominant isoform in all 6 cell lines, but its expression levels varied up to ten-fold (not shown). All cell lines expressed high cPGES and mPGES-2 protein levels, while only one cell line had high mPGES-1 protein expression (Fig. 2B). EGF stimulation did not influence PGES isoform expression, analyzed by pixel intensity.

EGF-induced PGE₂ Production

All cell lines except B11 increased PGE₂ production after EGF stimulation (Fig. 2C ).

Immunohistofluorescence Expression of COX-1 and -2 in Oral Carcinomas

We used cryosections from 3 available OSCCs to illustrate in situ COX isoform expression. All carcinomas expressed medium to high COX-2 positivity at the periphery of the cancer (Fig. 2D ). Although most of the tumor cells did not express COX-1, some scattered COX-2-expressing tumor cells expressed weak COX-1 (not shown).

COX-1 and -2 Protein Expression in OSCC Organotypic Models

Immunohistofluorescence staining for COX-1 and COX-2 in the 3D model revealed differences in COX expression patterns. The objective of this staining was to investigate expression patterns in cell lines grown in an in vivo-like manner, and possibly to illustrate whether the differences in expression patterns were due to variable cellular expression patterns, similar to what was observed in carcinoma samples. Not all cells within one cell line had the same COX-isoform expression pattern (Fig. 2E ), and COX-1 and COX-2 were rarely co-expressed in the same cells. Intra-epithelial COX-1-positive cells often expressed vimentin as well, suggesting fibroblast origin (not shown).

PGE₂ Production was Reduced by COX-2 Inhibitors

To investigate which COX isoform was responsible for PGE₂ production, we treated 4 cell lines (B5, C12, D2, and D12), representing the 3 different basal COX-isoform protein expression patterns (B5 and C12, COX-1 high; D2, both isoforms low; D12, both isoforms high), with various COX-1 and/or COX-2 inhibitors.

The COX-2 selective inhibitor (NS-398) reduced PGE₂ production in both unstimulated and EGF-stimulated cell lines with high COX-2 expression, while the COX-1 inhibitor (SC-560) had an effect only at the highest concentration (Figs. 3A–3D ), except in EGF-stimulated D2, where SC-560 seemed to reduce PGE₂ production weakly at all concentrations. Unstimulated B5 did not produce measurable PGE₂ in this assay (Fig. 3A ).

siRNA-mediated COX-2 Knockdown Reduced COX-2 Expression and PGE₂ Production

Three selected cell lines were treated with COX-2-specific siRNA before and after EGF stimulation. While basal COX-2 mRNA expression levels were reduced by 59% ± 23 (± SD) (cell line B5), 33% ± 21 (D2), and 63% ± 12 (D12) relative to scrambled controls (Fig. 4A ), COX-2 protein was almost undetectable (not shown). PGE₂ production was reduced by 65% in D2 and 34–74% in D12 (Fig. 4C ). Unstimulated B5 did not produce measurable PGE₂.

COX-2-specific siRNA gene knockdown reduced EGF-induced COX-2 mRNA expression levels by 74% ± 9 (B5), 75% ± 14 (D12), or 41% ± 21 (D2) relative to the scrambled controls (Fig. 4A ). EGF-induced COX-2 protein expression was practically abolished in siRNA-treated B5 and reduced in the cell lines D2 (46%) and D12 (86%) relative to scrambled controls (Fig. 4B ). Similarly, COX-2-specific knockdown reduced EGF-induced PGE₂ production by 58–70% (B5), 37–45% (D2), and 53–71% (D12) (Fig. 4C ), which paralleled the reduction in COX-2 protein expression.

DISCUSSION

Whereas COX-2 expression and increased PGE₂ production have been linked to carcinogenesis in many different carcinoma types (reviewed in Zha et al., 2004), their association with OSCC has been controversial (Lim et al., 2004; Atula et al., 2006; Sakurai et al., 2007). However, if PGE₂ was produced via COX-1 (Kino et al., 2005), which is over-expressed in both pre-malignant lesions and in oral carcinomas (Shibata et al., 2005), it could explain why COX-2 expression is less associated with oral carcinogenesis and prognosis. The 4 selected OSCC cell lines with different COX-isotype expression patterns revealed, however, that basal and EGF-induced PGE₂ production was COX-2-dependent. Not only did the COX-2 selective inhibitor, NS-398, reduce PGE₂ production in all cell lines, but also the claimed COX-1 selective production minimally, and mainly inhibitor, SC-560, reduced PGE₂ at the highest concentrations. Although SC-560 inhibits the isolated COX-1 enzyme at concentrations 700 times lower than in the isolated COX-2 (Smith et al., 1998), SC-560 was recently shown to be almost non-selective in cellular systems (Brenneis et al., 2006). This COX-2 cross-inhibition may explain why SC-560 inhibited PGE₂ production slightly in both the high (B5) and the low COX-1 expressing cell line (D2).

The absence of COX-1-induced PGE₂ production was further supported by siRNA-mediated COX-2-specific knockdown in unstimulated and EGF-stimulated cell lines, where reduction in COX-2 mRNA levels paralleled reduced COX-2 protein expression and PGE₂ production. EGF-induced COX-2 expression may explain why COX-2 has been predominantly observed at the periphery or at the rim of cancer nests (Shibata et al., 2005), where EGF may be released from extracellular matrix and/or EGF-producing fibroblasts. Interestingly, EGF-induced COX-2-mediated PGE₂ production may facilitate cancer invasion activation through integrin α_vβ₆ (Nystrom et al., 2006), and/or increased immune survival (Baratelli et al., 2005). This may explain why the combined inhibition of COX-2 and EGFR activity prevented metastatic lung colonization in a breast cancer mouse model (Gupta et al., 2007).

Tumor cells grown in 3D organotypic cultures on a fibroblast-containing collagen I matrix may mimic the in vivo situation better than monolayer cultures, since such cultures allow for fibroblast-epithelial interactions (Costea et al., 2003), which are essential for the differentiation, polarization, and growth of adjacent adherent cells (reviewed in Ingber, 2002). Multicolor immunohistofluorescence staining of such 3D cultures revealed variability in both the intensity and proportion of COX-positive carcinoma cells. The organotypic models also revealed interestingly low COX-1 expression in comparison with the Western blots from monolayer cultures, perhaps due to differences in culture conditions and/or interaction with COX-1-expressing fibroblasts, similar to the strong COX-1-expressing subepithelial fibroblasts observed in the solid OSCC. In fact, most of the COX-1-positive cells in the 3D model expressed the mesoderm-specific protein vimentin, as do fibroblasts. Although matrix fibroblasts may have migrated into the OSCC cells, we cannot exclude the possibility of epithelial-mesenchymal transition (EMT) (Guarino et al., 2007).

PGE₂ is produced by 3 PGES isoforms, of which cPGES is functionally coupled to COX-1 (Tanioka et al., 2000), mPGES-1 to COX-2 (Murakami et al., 2000), and mPGES-2 to both COX isoforms (Murakami et al., 2003). Although the COX-1 high- and COX-2 low-expressing cell lines expressed high levels of the COX-1-associated PGES’s, they produced PGE₂ through COX-2.

In summary, both basal and EGF-induced PGE₂ production in the OSCC cell lines was COX-2-dependent. The inconsistency between COX-2 expression level and oral cancer malignancy in OSCC is therefore presumably not due to additional COX-1-induced PGE₂ production.

Figure 1.

The cyclooxygenases (COXs) catalyze the rate-limiting step in prostaglandin production from arachidonic acid. The COX-2 isoform is induced by different stimulations such as epidermal growth factor receptor (EGFR) activation, through downstream signaling cascades. The COX product prostaglandin H₂ (PGH₂) is further processed to prostaglandin E₂ (PGE₂) by 3 known prostaglandin synthase isoforms: cytosolic PGE₂ synthase (cPGES), microsomal PGE₂ synthase-1 (mPGES-1), and microsomal PGE₂ synthase-2 (mPGES-2). Secreted PGE₂ may transactivate EGFR, lowering threshold or intensifying receptor activity, potentially facilitating a signaling loop.

Figure 2.

COX-isoform expression and PGE₂ production. (A) COX mRNA expression in 6 OSCC cell lines. The cell lines expressed variable COX isoform levels. Whereas EGF stimulation did not induce COX-1 expression, it clearly induced COX-2 mRNA expression in C12, D2, D12, and E10, but only slightly in B5 and B11. Data from 3 independent experiments in duplicate, median ± SD. (B) COX- and PGES-isoform expression in 6 OSCC cell lines analyzed by Western blotting. Unstimulated cell lines displayed variable COX-1 and COX-2 expression. EGF stimulation induced COX-2 only. While cPGES and mPGES-2 were highly expressed in all cell lines, mPGES-1 was highly expressed only in B11. No isoforms were affected by EGF stimulation. GAPDH was the loading control. The illustrated result is from 1 of 3 independent experiments. (C) All cell lines produced increasing PGE ₂ levels after EGF stimulation (100 ng/mL), except B11. Median ± SD. Immunohistofluorescence triple-staining of OSCC (D) and 3D organotypic culture (cell line D2) (E) for COX-1 (green), COX-2 (red), and keratin (blue). (D) Although the OSCC tumor cells predominantly expressed COX-2, there was occasionally weak COX-1 co-expression in some parts of the sections (data not shown). Additionally, COX-1-expressing cells were scattered throughout the stroma. (E) The organotypic models revealed intercellular variation in COX-isoform expression, in that only a few cells in COX-1- and COX-2-expressing cell lines expressed both isoforms simultaneously. Most intra-epithelial COX-1 high-expressing cells expressed vimentin, as did fibroblasts (data not shown).

Figure 3.

Unstimulated and EGF-induced PGE₂ production after COX-isoform-specific inhibitions. We selected 4 OSCC cell lines, representing the different COX-isoform-expressing patterns, to identify the COX isoform responsible for PGE₂ production. ▪□ represent COX-1 inhibitor (SC-560), •○ represent COX-2 inhibitor (NS-398), and open □○ symbols represent un-stimulated cells, while filled ▪• symbols represent EGF-stimulated cells. Both basal and EGF-induced PGE₂ production was reduced after COX-2-selective inhibition, whereas only the highest concentration of the COX-1 inhibitor reduced PGE₂ production. COX-1 inhibitor concentrations are in nM, whereas the COX-2 inhibitor is in μM. The Fig. illustrates median values ± SD from 1 experiment with 3 parallels in duplicate. The EGF-stimulated D2 and D12 are represented by 1 experiment in duplicate. Unit scale is percentage of PGE₂ produced after COX-selective inhibition, with uninhibited controls as 100%.

Figure 4.

COX-2 expression and PGE₂ production after COX-2-specific siRNA-mediated gene knockdown. (A) siRNA treatment reduced EGF-induced mRNA level by 41–75% and basal COX-2 mRNA level by 33–63%. COX-2 mRNA expression is normalized to the housekeeping gene TBP and displayed as % expression of non-specific (scrambled) controls. Bars represent median value, while circles represent datapoints. Results from 6–8 independent samples run in duplicate. (B) Decreased COX-2 protein expression was detected by Western blotting. COX-2 protein levels were detectable only in EGF-stimulated samples, and the low COX-2-expressing cell line B5 had to be loaded with 3 times the total protein concentrations for COX-2 protein to be visualized. COX-2-specific siRNA reduced the COX-2 protein expression by > 90% in B5, by 46% in D2, and by 86% in D12. All 3 independent experiments produced similar results. (C) COX-2-specific siRNA reduced EGF-induced PGE₂ production by 45–71%. The basal PGE₂ production was reduced by 65% in D2 and 34–74% in D12, while B5 had no detectable basal production. Results from 3 independent experiments, with one technical failure in the unstimulated D2 assay. Bars represent median value, and circles individual datapoints.

Footnotes

Acknowledgements

The authors acknowledge Andreas Karatsaidis and Daniela Elena Costea for help with organotypic cultures, Tine M. Søland for discussion, and Solveig Stig and Hanne Weidemann for technical assistance. Studies were supported with grants from the University of Oslo and the Norwegian Cancer Society. A preliminary report was presented at the 5th European Workshop on Basic Biology of Head and Neck Cancer, Poznan, Poland, 2006.

References

Atula T, Hedstrom J, Ristimaki A, Finne P, Leivo I, Markkanen-Leppanen M, et al. -2006-. Cyclooxygenase-2 expression in squamous cell carcinoma of the oral cavity and pharynx: association to p53 and clinical outcome. Oncol Rep 16:485–490.

Baratelli F, Lin Y, Zhu L, Yang SC, Heuze-Vourc’h N, Zeng G, et al. -2005-. Prostaglandin E2 induces FOXP3 gene expression and T regulatory cell function in human CD4+ T cells. J Immunol 175:1483–1490.

Bosetti C, Gallus S, La Vecchia C -2006-. Aspirin and cancer risk: an updated quantitative review to 2005. Cancer Causes Control 17:871–888.

Brenneis C, Maier TJ, Schmidt R, Hofacker A, Zulauf L, Jakobsson PJ, et al. -2006-. Inhibition of prostaglandin E2 synthesis by SC-560 is independent of cyclooxygenase 1 inhibition. FASEB J 20:1352–1360.

Chan G, Boyle JO, Yang EK, Zhang F, Sacks PG, Shah JP, et al. -1999-. Cyclooxygenase-2 expression is up-regulated in squamous cell carcinoma of the head and neck. Cancer Res 59:991–994.

Chandrasekharan NV, Dai H, Roos KL, Evanson NK, Tomsik J, Elton TS, et al. -2002-. COX-3, a cyclooxygenase-1 variant inhibited by acetaminophen and other analgesic/antipyretic drugs: cloning, structure, and expression. Proc Natl Acad Sci USA 99:13926–13931.

Costea DE, Loro LL, Dimba EA, Vintermyr OK, Johannessen AC -2003-. Crucial effects of fibroblasts and keratinocyte growth factor on morphogenesis of reconstituted human oral epithelium. J Invest Dermatol 121:1479–1486.

Dannenberg AJ, Lippman SM, Mann JR, Subbaramaiah K, DuBois RN -2005-. Cyclooxygenase-2 and epidermal growth factor receptor: pharmacologic targets for chemoprevention. J Clin Oncol 23:254–266.

Guarino M, Rubino B, Ballabio G -2007-. The role of epithelial-mesenchymal transition in cancer pathology. Pathology -Phila- 39:305–318.

10.

Gupta GP, Nguyen DX, Chiang AC, Bos PD, Kim JY, Nadal C, et al. -2007-. Mediators of vascular remodelling co-opted for sequential steps in lung metastasis. Nature 446:765–770.

11.

Ingber DE -2002-. Cancer as a disease of epithelial-mesenchymal interactions and extracellular matrix regulation. Differentiation 70:547–560.

12.

Kino Y, Kojima F, Kiguchi K, Igarashi R, Ishizuka B, Kawai S -2005-. Prostaglandin E2 production in ovarian cancer cell lines is regulated by cyclooxygenase-1, not cyclooxygenase-2. Prostaglandins Leukot Essent Fatty Acids 73:103–111.

13.

Laufer S, Zechmeister P, Klein T -1999-. Development of an in-vitro test system for the evaluation of cyclooxygenase-2 inhibitors. Inflamm Res 48:133–138.

14.

Lim SC, Zhang S, Ishii G, Endoh Y, Kodama K, Miyamoto S, et al. -2004-. Predictive markers for late cervical metastasis in stage I and II invasive squamous cell carcinoma of the oral tongue. Clin Cancer Res 10-1 Pt 1-:166–172.

15.

Lin DT, Subbaramaiah K, Shah JP, Dannenberg AJ, Boyle JO -2002-. Cyclooxygenase-2: a novel molecular target for the prevention and treatment of head and neck cancer. Head Neck 24:792–799.

16.

Minter HA, Eveson JW, Huntley S, Elder DJ, Hague A -2003-. The cyclooxygenase 2-selective inhibitor NS398 inhibits proliferation of oral carcinoma cell lines by mechanisms dependent and independent of reduced prostaglandin E2 synthesis. Clin Cancer Res 9:1885–1897.

17.

Murakami M, Naraba H, Tanioka T, Semmyo N, Nakatani Y, Kojima F, et al. -2000-. Regulation of prostaglandin E2 biosynthesis by inducible membrane-associated prostaglandin E2 synthase that acts in concert with cyclooxygenase-2. J Biol Chem 275:32783–32792.

18.

Murakami M, Nakashima K, Kamei D, Masuda S, Ishikawa Y, Ishii T, et al. -2003-. Cellular prostaglandin E2 production by membrane-bound prostaglandin E synthase-2 via both cyclooxygenases-1 and -2. J Biol Chem 278:37937–37947.

19.

Nurmi JT, Puolakkainen PA, Rautonen NE -2005-. Intron 1 retaining cyclooxygenase-1 splice variant is induced by osmotic stress in human intestinal epithelial cells. Prostaglandins Leukot Essent Fatty Acids 73:343–350.

20.

Nystrom ML, McCulloch D, Weinreb PH, Violette SM, Speight PM, Marshall JF, et al. -2006-. Cyclooxygenase-2 inhibition suppresses {alpha}v{beta}6 integrin-dependent oral squamous carcinoma invasion. Cancer Res 66:10833–10842.

21.

Qin N, Zhang SP, Reitz TL, Mei JM, Flores CM -2005-. Cloning, expression, and functional characterization of human cyclooxygenase-1 splicing variants: evidence for intron 1 retention. J Pharmacol Exp Ther 315:1298–1305.

22.

Renkonen J, Wolff H, Paavonen T -2002-. Expression of cyclo-oxygenase-2 in human tongue carcinoma and its precursor lesions. Virchows Arch 440:594–597.

23.

Sakurai K, Urade M, Noguchi K, Hashitani S, Takaoka K, Segawa E, et al. -2007-. Prognostic significance of cyclooxygenase-2 and DNA topoisomerase IIalpha expression in oral carcinoma. Head Neck 29: 1002–1009.

24.

Shibata M, Kodani I, Osaki M, Araki K, Adachi H, Ryoke K, et al. -2005-. Cyclo-oxygenase-1 and -2 expression in human oral mucosa, dysplasias and squamous cell carcinomas and their pathological significance. Oral Oncol 41:304–312.

25.

Smith CJ, Zhang Y, Koboldt CM, Muhammad J, Zweifel BS, Shaffer A, et al. -1998-. Pharmacological analysis of cyclooxygenase-1 in inflammation. Proc Natl Acad Sci USA 95:13313–13318.

26.

Soland TM, Husvik C, Koppang HS, Boysen M, Sandvik L, Clausen OP, et al. -2008-. A study of phosphorylated ERK1/2 and COX-2 in early stage -T1–T2- oral squamous cell carcinomas. J Oral Pathol Med 37:535–542.

27.

Tanioka T, Nakatani Y, Semmyo N, Murakami M, Kudo I -2000-. Molecular identification of cytosolic prostaglandin E2 synthase that is functionally coupled with cyclooxygenase-1 in immediate prostaglandin E2 biosynthesis. J Biol Chem 275:32775–32782.

28.

Thoren S, Jakobsson PJ -2000-. Coordinate up- and down-regulation of glutathione-dependent prostaglandin E synthase and cyclooxygenase-2 in A549 cells. Inhibition by NS-398 and leukotriene C4. Eur J Biochem 267:6428–6434.

29.

Thun MJ, Namboodiri MM, Heath CW Jr -1991-. Aspirin use and reduced risk of fatal colon cancer. N Engl J Med 325:1593–1596.

30.

Zha S, Yegnasubramanian V, Nelson WG, Isaacs WB, De Marzo AM -2004-. Cyclooxygenases in cancer: progress and perspective. Cancer Lett 215:1–20.

31.

Zhang Q, Adiseshaiah P, Kalvakolanu DV, Reddy SP -2006-. A phosphatidylinositol 3-kinase-regulated Akt-independent signaling promotes cigarette smoke-induced FRA-1 expression. J Biol Chem 281:10174–10181.