Over-expression of COX-2 mRNA in colorectal cancer

Background Cyclooxygenase-2 (COX-2, PTGS2) is an enzyme involved in the synthesis of prostaglandins and thromboxanes, which are regulators of biologic processes such as inflammation, cell proliferation and angiogenesis. COX-2 over-expression was reported in many (pre) malignant tissues, but data strongly vary and seem to depend on the methodology used. Methods Normal colorectal mucosa and paired cancerous tissue from 60 patients with colorectal cancer was investigated for the levels of COX-2 mRNA by real-time quantitative Polymerase Chain Reaction (qPCR). COX-2 levels were expressed relative to either: tissue weight or levels of the housekeeping genes beta-2 microglobulin (B2M) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Results COX-2 mRNA levels, normalized with respect to tissue weight or mRNA levels of the housekeeping genes B2M or GAPDH, were over-expressed in 80%, 70% and 40% of the colorectal tumor tissues, as compared to the paired adjacent normal colorectal mucosa samples, respectively. Highest mRNA COX-2 ratios tumor/normal were measured when expressed per mg tissue (mean ratio 21.6). When normalized with respect to the housekeeping genes B2M or GAPDH, mean tumor/normal ratios were 16.1 and 7.5, respectively. Conclusion Expression of COX-2 mRNA levels per mg tissue is most simple in comparison to normalization with respect to the housekeeping genes B2M or GAPDH. Levels of COX-2 mRNA are found over-expressed in almost 80% of the colorectal tumors, compared to paired adjacent normal colorectal mucosa, suggesting a role of COX-2 as a potential biomarker for cancer risk, whereas inhibitors of COX-2 could be of value in chemoprevention of colon cancer.

Conclusion: Expression of COX-2 mRNA levels per mg tissue is most simple in comparison to normalization with respect to the housekeeping genes B2M or GAPDH. Levels of COX-2 mRNA are found over-expressed in almost 80% of the colorectal tumors, compared to paired adjacent normal colorectal mucosa, suggesting a role of COX-2 as a potential biomarker for cancer risk, whereas inhibitors of COX-2 could be of value in chemoprevention of colon cancer.

Background
The development of colorectal cancer is a multistep process, where mutations may occur in the oncogenes K-ras and APC, or in the tumor suppressor gene p53, causing cell degeneration and uncontrolled cell proliferation. Cell proliferation is pivotal in tumorigenesis and cyclooxygenases (COXs) are important regulatory enzymes in this process. Cyclooxygenases catalyze the conversion of free arachidonic acid into prostaglandin H2, which is the precursor of other prostaglandins and thromboxanes. These regulatory compounds play a role in various biological processes such as cell proliferation, angiogenesis, immune function and inflammation, which are all crucial in the development or progression of neoplasms [1].
The human COX family consists of three members, COX 1-3. COX-1 is found in most tissues and plays a role in homeostasis of many physiologic processes. COX-3 is an alternative splice product of COX-1 and is involved in the regulation of pain and fever. COX-2 is an inducible enzyme, whose expression can be influenced by pro-inflammatory and mitogenic stimuli like cytokines and growth factors. COX-2 plays an important role in the development of metaplastic and dysplastic tissue, as well as in the development and progression of cancer, by involvement in the regulation of cell proliferation, cell transformation, tumor growth, tumor metastasis and invasion [1,2].
Over-expression of COX-2 has been associated with various premalignant and malignant lesions of epithelial origin, in particular in organs of the gastrointestinal tract [3][4][5][6][7][8]. Tumors with high levels of COX-2 seem to be more aggressive [5] and patients bearing those tumors had a significantly reduced survival [9]. In addition, after drug-induced suppression of COX-2 levels in laboratory animals subjected to entero-esophageal reflux, less animals developed esophageal adenocarcinoma [10]. This is of particular interest since more or less specific inhibitors of COX-2, such as various non-steroidal antiinflammatory drugs (NSAIDs), have been or are being developed, that could have a role in the chemoprevention of gastrointestinal neoplasms [6]. Evidence is accumulating that a considerable reduction in the development of adenomatous polyps or colorectal cancer could be achieved in individuals regularly taking NSAIDs [6]. Although most NSAIDs do not specifically inhibit COX-2, the findings of over-expression of COX-2 in many (pre) malignant tissues, in combination with the reported beneficial effects of NSAIDs on cancer prevention, suggests that inhibition of COX-2 may be crucial in this process.
Since alterations in COX-2 levels may be pivotal in influencing the development of colorectal cancer [11], it is crucial that COX-2 levels can be estimated in a reliable way. Different techniques were used to investigate the COX-2 levels in colorectal neoplasms, yielding a large variety in outcomes. Semi-quantitative immunohistochemical staining for COX-2 was applied, thereby using different COX-2 antibodies [3,[12][13][14][15]. Quantitative assay of COX-2 mRNA levels was performed [4,[16][17][18], whereas both techniques were combined in some studies [12][13][14][15]. Large variations in the colorectal COX-2 expression levels were reported. In most studies, higher levels of COX-2 mRNA were demonstrated in the tumor, compared to the normal colorectal tissue [4,14,15,17,18], however sample sizes generally were small. In two larger studies, no elevated COX-2 mRNA levels in the tumor were reported [12,16]. Furthermore, in the studies where mRNA levels were assayed, most often no paired samples of normal and cancerous tissue from the same patient were investigated. In addition COX-2 mRNA levels were normalized with respect to different housekeeping genes.
The aim of the present study was to quantify the COX-2 mRNA levels in 60 paired samples of non-necrotic tumor tissue and corresponding normal colorectal mucosa, taken at least 5 cm distant from the tumor. The COX-2 mRNA levels will be expressed per mg tissue, as well as relative to the levels of two different housekeeping genes, beta-2 microglobulin (B2M) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH).

Methods
Normal colorectal mucosa and tumor tissue of 60 patients with sporadic colorectal cancer was included in this study. These tissues were obtained as rest material after surgical intervention and tissues were processed and assayed anonymously. Patients did not receive any treatment just before surgery. Rest material after resection for colorectal cancer was collected at the Department of Gastroenterology, Radboud University Nijmegen Medical Center (RUNMC), in the period 1999-2002. Immediately after resection, the tissue was cooled on ice and transported to the Pathology Department, where the pathologist inspected the tissue within 30 min. Representative specimens of macroscopically vital, non-necrotic tumor tissue and normal colorectal mucosa, taken at least 5 cm distant from the tumor, were obtained. Specimens were transported on ice to the laboratory and the tissue was washed with ice-cold phosphate buffered saline (PBS), cut in small pieces, frozen in liquid nitrogen and stored at −80°C until use. Ethical approval for the anonymous use of colorectal rest material was obtained by the RUNMC medical ethical review committee.

Assay of COX-2 mRNA levels
Tissue was kept frozen while small samples (10-20 mg tissue) were cut off for mRNA analyses. From normal colon tissue, only mucosa was scraped off from the frozen samples. The normal-and tumor tissue samples were put into separate vials, weighed, and immediately thereafter 0.5 ml TRIzol (Life Technologies, Pailey, UK) was added. The within day variation of weighing tissues in the 10-50 mg range, using a Mettler AC100 precision weighing device (Mettler Instruments, Tiel, the Netherlands), was estimated to be 0.2-0.6%. Tissue was homogenized by 10 strokes with a Teflon pestle. RNA was isolated according to the instructions of the manufacturer (Life Technologies). Subsequently, 1 μg RNA was converted into cDNA, using the Roche Transcriptor cDNA synthesis kit, according to the instructions of the manufacturer (Roche Diagnostics, Almere, the Netherlands). Detection and quantification of mRNAs was performed by real-time quantitative PCR (qPCR) using the CFX96 Real-Time PCR Detection System (Bio-Rad Laboratories, Veenendaal, the Netherlands). Sequences of the primers were designed using Primer3 software (http://primer3.ut.ee/). Primers are: COX-2, 5'-CCGGGTACAATCGCACTTAT-3' and 5'-GGCGCT CAGCCATACAG-3' (Isogen Life Science, Maarssen, The Netherlands); GAPDH, 5'-GAAGGTGAAGGTCGGAGT CA-3' and 5'-TTGAGGTCAATGAAGGGGTC-3' (Isogen Life Science); B2M, 5'-ATGAGTATGCCTGCCGTGTG-3' and 5'-CCAAATGCGGCATCTTCAAAC-3' (Biolegio, Nijmegen, The Netherlands). Pairs of primers are designed to localize on different exons.
PCR products of COX-2 and GAPDH were detected with SYBR Green (Molecular Probes, Life Technologies, Bleiswijk, the Netherlands). For detection of B2M, a specific molecular beacon 5'-FAM-CGCGTCGTGGGATGGAGA CATGTAAGCAGACGCG-Dabcyl-3' was used (Biolegio). The annealing temperatures of the COX-2, GAPDH and B2M primers were 65.0˚C, 59.0˚C and 60.5˚C, respectively. Specificity of the PCR products of COX-2, GAPDH and B2M was confirmed by melting curve analysis and agarose gel electrophoresis. All real time qPCR analyses were performed in triplicate.

Statistics
Real time qPCR data were converted to linear data by calculating the 2 -Ct values when data are expressed per mg tissue, and by calculating the 2 -ΔCt values for B2M and GAPDH normalized data. Data are expressed as arbitrary units or as ratios of COX-2 expression (tumor/ normal mucosa). Statistical analyses were performed using GraphPad Prism, version 4.00 (GraphPad Software Inc., La Jolla, CA, USA).

Results
Clinical data of the patients with colorectal cancer, who's tissue was used in this study, are summarized in Table 1. Sixty paired samples of normal colorectal mucosa and corresponding tumor tissue were analyzed for COX-2 mRNA expression. Levels of COX-2 mRNA, measured in triplicate, were expressed per mg tissue weight or the expression was normalized with respect to mRNA levels of the housekeeping genes B2M and GAPDH. Mean expression of COX-2 mRNA was higher in colon tumor tissue, compared to the corresponding normal colorectal mucosa. However, COX-2 mRNA ratios tumor/normal clearly were dependent on the standardization used. Highest mean COX-2 ratios tumor/normal were observed when normalized with respect to tissue weight (mean ratio 21.6), followed by normalization with respect to the housekeeping genes B2M (mean ratio 16.1) or GAPDH (mean ratio 7.5); results are shown in Table 2. The COX-2 mRNA expression levels in normal colorectal mucosa and corresponding tumor tissue, normalized with respect to either tissue weight or the housekeeping genes B2M and GAPDH, are depicted in Figure 1.

Discussion
COX-2 catalyzes the conversion of free arachidonic acid into prostaglandin H2, the precursor of other prostaglandins and thromboxanes. These compounds are pivotal in the regulation of cell proliferation, angiogenesis, immune function and inflammation, which may all contribute to the development and progression of neoplasms [1].
An important item in the measurements of COX-2 mRNA expression is the standardization. Standardization against tissue weight, with a variation of 0.2-0.6% in the weighing results when using a precision weighing device, is more accurate as compared to the reported variation of 0.7-5.3% for a quantitative real time PCR assay, by performing 3 repeats within one run [19].  B2M, beta-2 microglobulin. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. *Note that GAPDH was not detectable or the analysis did not succeed in 5 cases. **Ratio > 1 means that COX-2 expression is higher in colorectal tumor tissue compared to paired normal mucosa. ***Ratio < 1 means that COX-2 expression is lower in colorectal tumor tissue compared to paired normal mucosa.
Several studies reported that the COX-2 expression was elevated in colorectal tumors, in comparison to normal colorectal tissue. However, different techniques were used for estimating the COX-2 levels; semi-quantitative immunohistochemical staining of COX-2, making use of different COX-2 antibodies [3,[12][13][14][15], quantitative PCR assay of COX-2 mRNA [4,[16][17][18], or both techniques were applied simultaneously [12,14,15]. Immunohistochemical studies generally showed more staining in the tumor as compared to the normal colorectal mucosa, however staining positivity in the tumor varied from 35 to 62% in sporadic colorectal cancers [12,14] to 100% staining positivity in colorectal tumors from patients with familial adenomatous polyposis [13]. Variations in results on COX-2 mRNA analyses were also large. Again, in most studies COX-2 mRNA was found elevated in the tumor, however sample numbers were small [4,14,15,17,18]. In two larger studies, investigating 99 and 46 tumor samples respectively, no elevated COX-2 mRNA levels in the tumor were reported [12,16]. In most mRNA studies however, no paired samples of normal and cancerous tissue, obtained from the same patient, were investigated. In addition, COX-2 mRNA levels were normalized relative to the expression of one housekeeping gene, whereas the choice of which housekeeping gene was used also differed between the various studies.
Recently Carvalho et al. showed that COX-2 mRNA was up-regulated in an age-dependent fashion in colorectal carcinoma as compared to adenoma [20]. However, it needs to be further elucidated, whether the COX-2 levels in colorectal adenomas are elevated compared to normal mucosa. Eberhart et al. [4] reported elevated COX-2 levels in 3 of 6 colorectal adenomas compared with paired normal mucosa. Nosho et al. [14] found an over-expression of COX-2 mRNA in only 42% of the colorectal tumors investigated, being 63 adenomas and 27 carcinomas. In comparison with our data, where over-expression in 77% of colorectal carcinomas is noticed, this suggests that overexpression of COX-2 is lower in colorectal adenomas than in carcinomas, and that expression is increasing somewhere in the stage between adenoma and carcinoma. This could plead for the potential use of COX-2 mRNA levels in feces, as a potential non-invasive marker for colorectal cancer screening, as suggested by Kanaoka and co-workers [21,22].
For interpretation of quantitative mRNA measurements in clinical samples, a proper normalization is an essential component of reliable qPCR assays, which is necessary to correct for sample differences in cellular input, RNA quality and real-time efficiency [23]. According to recently published guidelines, normalization against a single reference gene is not acceptable [24], since large variations in the expression of commonly used housekeeping genes have been described [23][24][25][26][27]. For instance, in normal human colon samples, a more than 100 fold variation in GAPDH levels was reported; values varied between 8 × 10 4 and 9.6 × 10 6 copies of GAPDH per μg RNA [28]. In our study, COX-2 mRNA tumor/normal ratios normalized with respect to GAPDH showed the lowest correlation compared with normalization for tissue weight.
As stated by de Kok et al. in [23], ideally normalization could best be done based on the number of cells analyzed. In solid tissue, the amount of tissue is directly associated with cell numbers. Strikingly however, analysis based on cell numbers or amount of tissue was not applied in any of the recent studies on COX-2 mRNA expression in colorectal tissue [12,[14][15][16][17][18].
We quantified the COX-2 mRNA levels per mg tissue, as well as relative to the levels of two different housekeeping genes GAPDH and B2M, in 60 paired samples of normal colorectal mucosa and corresponding nonnecrotic tumor tissue. Detectable levels of COX-2 mRNA were found in all 60 paired specimens. After normalizing of the COX-2 levels relative to tissue weight, the COX-2 levels were higher in 77% of the carcinoma tissue compared to the adjacent normal mucosa, whereas the mean COX-2 levels in the tumor were 21.6 times higher compared to the normal mucosa levels. Since it is very simple and accurate to determine the tissue weight just before RNA extraction, this may be the method of choice, since there is only a small weighing error (0.2-0.6%) when using a precision weighing device.
Comparing the mRNA expression of the two commonly used housekeeping genes B2M and GAPDH, B2M seems the second best choice for normalization in colorectal tissue, since there is very much variation in the expression of GAPDH in normal colorectal tissue, as already outlined above [28].

Conclusions
COX-2 mRNA levels as quantified by qPCR and normalized with respect to tissue weight, are over-expressed in 77% of colorectal carcinomas compared to adjacent normal colorectal mucosa, and mean carcinoma levels are almost 22 times higher than normal mucosa levels. Normalization with respect to tissue weight is simple and accurate, compared to normalization based on the levels of the housekeeping genes B2M or GAPDH. The common over-expression of COX-2 in colorectal carcinoma suggests a role for COX-2 as a colorectal carcinoma risk biomarker, whereas attention should also be focussed on COX-2 inhibitors, as potential promising chemopreventive drugs for colorectal cancer.