Possible predisposition for colorectal carcinogenesis due to altered gene expressions in normal appearing mucosa from patients with colorectal neoplasia

Background Investigations of colorectal carcinogenesis have mainly focused on examining neoplastic tissue. With our aim of identifying potentially cancer-predisposing molecular compositions, we chose a different approach by examining endoscopically normal appearing colonic mucosa of patients with and without colorectal neoplasia (CRN). Directed by this focus, we selected 18 genes that were previously found with altered expression in colorectal cancer affected mucosa. Methods Biopsies of colonic mucosa were sampled from 27 patients referred for colonoscopy on suspicion of colorectal disease. Of these, 14 patients had present or previous CRN and the remaining 13 patients served as controls. Using qPCR and Western blot technique, we investigated mRNA and protein expressions. Expressions were investigated for selected kinases in the extracellular signal-regulated kinase/mitogen activated protein kinase (ERK/MAPK), the phosphoinositide 3-kinase/Akt, and the Wnt/β-catenin pathways as well as for selected phosphatases and several entities associated with prostaglandin E2 (PGE2) signaling. Colonic mucosal contents of PGE2 and PGE2 metabolites were determined by use of ELISA. Results We found up-regulation of ERK1, ERK2, Akt1, Akt2, PLA2G4A, prostanoid receptor EP3 and phosphatase scaffold subunit PPP2R1B mRNA expression in normal appearing colonic mucosa of CRN patients compared to controls. Conclusion Present study supports that even normal appearing mucosa of CRN patients differs from that of non-CRN patients at a molecular level. Especially expression of ERK1 mRNA was increased (p = 0.007) in CRN group. ERK1 may therefore be considered a potential candidate gene as predictive biomarker for developing CRN. Further validation in larger cohorts are required to determine such predictive use in translational medicine and clinics.


Investigating endoscopically normal appearing mucosa
Most investigations of tumorigenesis focus on alterations in neoplastic tissues or transformed cells in culture. Due to progressive derangement in neoplastic tissue of the genome, growth factor-signaling, and metabolic networks, screening of colorectal neoplasia (CRN) tissue itself demonstrates many alterations. Thus, investigations on CRN tissues provide limited information regarding possible predisposing factors and their role in transforming normal appearing mucosa to CRN. Screening of endoscopically normal appearing mucosa from patients with CRN and comparing it to mucosa of non-CRN patients may generate important new insights into identifying subjects with increased risk of developing CRN (risk stratification, predictive biomarker). As such, we and a few others have communicated altered expression of several genes and proteins in normal appearing mucosa from CRN patients [1][2][3][4].

Wnt/β-catenin-, ERK/MAPK-and PI3K/Akt pathways in colorectal cancer
Among the frequently disrupted signaling pathways, perturbation of the Wnt/β-catenin pathway has been identified in early lesions of colorectal epithelium. Concurrently, Wnt/β-catenin perturbation is regarded a major initiating event in development of CRN as its tumor-suppressor gene adenomatous polyposis coli (APC) is inactivated in about 80% of sporadic colorectal cancers, CRC. APC inactivation leads to increased nuclear transfer of β-catenin with formation of a constitutively active β-catenin-T-cell factor complex [5]. In addition to perturbation of Wnt/βcatenin pathway, two other major pathways, the extracellular signal-regulated kinase/mitogen activated protein kinase (ERK/MAPK) and phosphoinositide 3-kinase/Akt (PI3K/Akt)-signaling pathways, are frequently overactive in CRC. Activation of these pathways stimulate cell growth, proliferation and survival [6,7]. As such, constitutive activation of the ERK/MAPK pathway occurs via mutations in KRAS in about 50% of CRCs [8]. Less frequently, dysregulation of PI3K/Akt pathway is observed following mutations in the CA subtype of the phosphoinositide-3-kinase gene (PIK3CA) or in the PI3K/Akt-related phosphatase and tensin homolog (PTEN) [9,10].

Prostaglandin E2 in colorectal carcinogenesis
During carcinogenesis the regulatory system of cell cycle is affected by both internal genetic alterations and external signaling. In particular, cytokines, growth factors and eicosanoids trigger signal transduction cascades through receptor tyrosine kinases and G protein-coupled receptors e.g., epidermal growth factor via its receptor EGF and PGE 2 through prostanoid receptors type 1-4. Among the prostanoids, prostaglandin E 2 (PGE 2 ) has been identified as the principal entity promoting cell growth and survival in CRN. It is believed that PGE 2 execute these effects via the PI3K/Akt-, ERK/MAPKand Wnt/β-catenin pathways [11]. Induction of PGE 2 signaling is also well-established as an early and critical step in development of CRN and in tumor progression [12,13]. All these pathways and PGE 2 metabolism are schematized in Fig. 1.

Hypothesis and aim of study
We hypothesized that up-regulation of some specific CRC-associated kinases and PGE 2 -related proteins occur in normal appearing colonic mucosa. The aim of this study was to test this 0-hypothesis by screening colonic mucosa for a panel of 18 specific genes, all proven altered and involved in CRC development. Tissue samples were endoscopic biopsies obtained from patients with and without CRN. Indeed, identifying perturbation of some of these signaling networks might give further insight into colorectal carcinogenesis per se and also identify potential predictive biomarkers.

Study population
Patients ≥ 50 years of age, referred for colonoscopy on suspicion for colorectal disease, were screened for enrollment in the study. Patients with a history of CRN, or CRN detected during colonoscopy, were included in CRN group. Patients with no present or history of CRN represents the control group. Patients were excluded if they received weekly medications in the form of nonsteroidal anti-inflammatory drugs, paracetamol, systemic corticosteroids and/or cytostatics. Furthermore, incomplete colonoscopy, inflammatory bowel disease, malabsorption or previous sigmoid resection and exposure to radiation or chemotherapy within the last year precluded enrollment. Finally, patients were also excluded if diagnosed with any of the genetic CRN syndromes (e.g. adenomatous polyposis coli and hereditary nonpolyposis colorectal cancer), as the focus of this study was sporadic CRN. Risk stratification of adenomas was performed based on number, size and histology grade in accordance with the European guidelines for quality assurance in CRC screening and diagnosis [14].

Biopsy procedure
Six biopsies from each patient were obtained during endoscopy from endoscopically normal appearing mucosa using standard biopsy forceps (Boston Scientific, Radial Jaw 4, outside diameter of 2.2 mm). Biopsies were obtained approximately 30 cm orally from the anal verge on retraction of the endoscope and at least 10 cm from endoscopically abnormal appearing mucosa.

Statistical analysis
Values were expressed as the mean ± SEM and as fold change. Data was analyzed using two-tailed unpaired t-test. A p-value less than 0.05 was considered statistically significant. Bonferroni correction was applied.  The engagement of the Wnt receptor, Frizzled, leads to the inhibition of the β-catenin destruction complex, composed of APC, axin and GSK3β. β-catenin thereby avoids ubiquitination and subsequent degradation, thus allowing it to translocate to the nucleus to activate an array of regulatory genes. B) The RAS/RAF/MEK/ERK MAPK pathway. Stimulation of the receptor tyrosine kinase (RTK) or G-protein coupled receptors (GPCRs) leads to sequential activation of RAS, RAF, MEK, and ERK causing modification of substrates promoting cell survival and proliferation. C) In the PI3K/Akt pathway, activation of the RTK or GPCRs leads to sequential modification of phosphatidyl inositol residues of the phospholipid bilayer. In this process, PI3K generates PIP3. PIP3 in association with PDK1 activates Akt. Akt then modulates the activity of downstream substrates including mTOR, thus promoting proliferation and cell survival. D) PGE 2 -metabolism. PGE 2 -synthesis begins with catalytic hydrolysis of membrane phospholipids by cytoplasmic phospholipase A2 (cPLA 2 ), thus releasing arachidonic acid (AA). By the action of the COX-1 and COX-2, AA is converted to prostaglandin H 2 (PGH 2 ). PGH 2 is then converted to PGE 2 by prostaglandin E synthase (PTGES). The main exporter of PGE 2 is thought to be multi-drug resistance related polypeptide 4 (MRP4). Removal of PGE 2 from the extracellular compartment around target cells occurs by diffusion to the blood stream and subsequent uptake and degradation in lung, liver or kidney endothelial cells or by import to colonic epithelial cells through the prostaglandin transporter (PGT) and subsequent degradation by 15-prostaglandin dehydrogenase (15-PGDH). Through autocrine and paracrine signaling, extracellular PGE 2 stimulates the prostaglandin receptors EP1-4. The EPs are GPCRs with EP1 being Gαq-coupled while EP2 and EP4 are Gαs-coupled. EP3 is capable of coupling with different G-proteins including Gαi, Gαs and Gαq

Homogenization and solubilization of colonic biopsies for Western blot analysis
Snap-frozen biopsies were homogenized for 20 s at 5000 rpm in 150 μl buffer (50 mM TrisHCl, pH 8.5, 5 mM EDTA, 150 mM NaCl, 10 mM KCl, 1% Triton X-100, 5 mM NaF, 5 mM β-glycerophosphate, 1 mM sodiumorthovanadate and Complete Protease Inhibitor Cocktail Tablet (Roche, Hvidovre, Denmark)) using a PreCellys 24 (Bertin Instruments, Montigny-le-Bretonneux, France) and ceramic beads (mix of 1.4 mm and 2.8 mm beads, VWR, Soeborg, Denmark). Solubilization was continued for 2 h in the same buffer at 4°C with rotation. The solubilized samples were spun for 15 min at 15000 g and the resulting supernatant collected for further protein determination and Western blot analysis.

Western blotting
Twenty-four biopsies were analyzed by Western blotting (12 CRN, 12 controls). 12.5 μg protein was separated on Table 1 Quantitative real-time PCR primer sequences

Patient characteristics
Twenty-seven patients were included. Fourteen patients in CRN group and 13 patients in control group. One additional patient was excluded from analysis due to histologically verified hyperplastic polyps, which are not classified as neoplasia. Median age was 64 years in CRN group and 62 years in control group. CRN group contained 7 (7/14) women versus 8 (8/13) in control group.  [14]. Eight patients (8/14) from CRN group and 6 (6/13) patients from control group had one or more comorbidities such as anemia, diabetes, hypothyroidism, cardiovascular disease, asthma, osteoporosis, psoriasis, chronic dermatitis, allergic rhinitis and/or a psychiatric diagnosis. Two patients in CRN group had a history of primary prostate cancer without metastasis. In the control group 4 patients had a history of primary non-colorectal cancer without metastasis: Melanoma (n = 1/14), testis seminoma (n = 1/14), uterus cancer (n = 1/14) and breast cancer (n = 1/14). Nine patients (9/14) in CRN group and 10 (10/13) patients in control group received medications e.g. anti-thrombotics, angiotensin-converting-enzyme inhibitors, angiotensin-IIreceptor antagonists, β-blockers, calcium-blockers, statins, diuretics, β2-agonists, anti-histamines, insulin, levothyroxine, proton pump inhibitors, laxantia, loperamid, antidepressants, estrogen or bisphosphonates. There were no apparent differences between the CRN and control groups in use of medications nor in co-morbidity profiles. Level changes for ERK, Akt, β-catenin and PPP2R1B expression. We screened colonic mucosa for a panel of 18 specific genes, all proven altered and involved in CRC development. All 18 targeted mRNAs were detected. Perturbed expression profile was observed in CRN patients with significantly higher mRNA-expressions measured for the following gene transcripts: PLA2G4A, EP3, ERK1, ERK2, Akt1, Akt2 and PPP2R1B, Table 2 and Fig. 2.
For the CRN group we detected upregulation of ERK1 (p = 0.007) and ERK2 (p = 0.02) which are ubiquitous regulators of cellular proliferation, differentiation, survival and transformation. ERK1 and ERK2 were expressed 1.87 and 2.0 times higher, respectively, in CRN patients. ERK1 expression was higher compared to ERK2 in both patient groups. Akt1 and Akt2 were both significantly up-regulated in CRN group (p = 0.02 and p = 0.041, respectively). β-catenin was highly expressed in both groups and with substantial variability in CRN group. The phosphatase subunit PPP2R1B, which is considered a negative regulator of both ERK and Akt activation and a stimulator of Wnt/β-catenin signaling, was significantly up-regulated in CRN group (p = 0.03).

Level changes for PLA2G4A and EP3
Analysis of entities in PGE 2 metabolism by qPCR demonstrated expression of PLA2G4A, which encodes a major enzyme involved in arachidonic acid mobilization, was significantly increased in CRN group (p = 0.02). With respect to the EP-receptors, EP4 had by far the highest expression, followed by EP2 and EP3, while EP1 showed the lowest expression. Only EP3 was significantly up-regulated in CRN group compared to controls (p = 0.016). Of all investigated mRNAs, 15-PGDH (the major enzyme involved in PGE 2 degradation), showed the highest expression in both groups and 1.9-fold higher expression in CRN group compared to controls (p = 0.066). Expression of COX-1 was higher than COX-2 in both groups (CRN 2.4-fold; controls 2.5-fold), while none of the two COX enzymes were significantly altered between patient groups. When Bonferroni correction was applied none of the 18 investigated genes reached the required significance level of p < 0.0028.

Trend towards increase in β-catenin, COX-1, COX-2 and ERK1 protein expression
To determine whether the changed mRNA expression profile observed in CRN patients was associated with changes in the protein expression level of these targets, we next performed Western blot analysis on colonic biopsies from CRN patients and controls. We successfully detected COX-1, COX-2, 15-PGDH, ERK1/2 and the active phosphorylated ERK1/2 (pERK1/2) proteins as well as AKT (using an antibody that detects AKT1-3) and β-catenin. We analyzed the protein expression level for each of these targets in the control group (12 biopsies) and compared them to the expression levels in the CRN group (12 biopsies). Relative protein expressions for the detected target proteins are shown in Fig. 3. The Western blot results suggest a trend towards a moderate increase in β-catenin as well as minor increases in COX-1, COX-2 and ERK1. The observed changes were not, however, statistically significant.

Localization of COX-1, 15-PGDH and β-catenin
We next sought to investigate the localization of target proteins by immunohistochemistry. We attempted to detect cPLA2A, COX-1, COX-2, EP3, PGT, 15-PGDH, ERK1/2, AKT, β-catenin and PPP2R1B in colonic biopsies from CRN patients and controls. The targets were selected to map localizations of proteins involved in different aspects of the PGE 2 pathway and to study any differences in protein localizations and/or possible expressions in CRN patients compared to controls. Possibly due to inadequate (human) specificity of the tested antibodies or paucity of target proteins, we failed to reliably detect cPLA2A, COX-2, PGT, EP3, ERK1/2, AKT and PPP2R1B in IHC. Only COX-1, 15-PGDH and β-catenin were detected. The results are shown in Fig. 4. 15-PGDH displayed strong cytoplasmic staining in epithelial cells at the crypt apex. COX-1 was selectively expressed in a small subset of cells in the epithelial cell layer where the enzyme displayed a reticular, intracellular localization associated with the perinuclear area. We believe that these COX-1 expressing cells represent epithelial tuft cells [15,16]. β-catenin showed expression throughout the epithelial layer and was mainly associated with lateral membranes of the epithelial cells. No apparent differences in protein localization were observed between CRN and control groups.

Discussion
Normal appearing, tumor-remote colonic mucosa in CRN patients has so far been scarcely explored. Here we document significant up-regulation of mRNA expression for the ERK1, ERK2, Akt1, Akt2, PLA2G4A, the prostanoid receptor EP3 as well as for PPP2R1B in normal appearing colonic mucosa from patients with CRN, Table 2. The range in fold change of these gene expressions was 1.46-2.34. The fact that our observed changes in gene expression were somewhat modest compared to changes reported in cancer tissues was to be expected since we examined normal appearing colon mucosa distant from neoplastic tissue. Furthermore, CRN group encompassed a wide range of stages of CRN and both patients with current CRN as wells as patients with a history of CRN. Since we perceive CRN development as a spectrum from low-grade adenomas to CRC, all CRN data were pooled into one group, tacitly acknowledging that different stages of CRN are likely to contribute differentially to the observed alterations of gene expression. Thus, it was a heterogenous CRN group with a limited number of observations and hence the observed changes between groups were not strong enough to reach the required significance level of a Bonferroni correction. However, our results point towards perturbed gene expression in the normal colon mucosa of CRN patients. These alterations in mRNA expression did not translate into significant alterations at protein level, Fig. 3. Although, the mRNA expressions of the above stated gene transcripts have previously been shown perturbed in CRC affected mucosa, we do not know the precise biological impact of the altered gene expressions in the normal appearing mucosa.

ERK and Akt signaling pathways
We observed significant up-regulation of mRNA expression for ERK1, ERK2, Akt1 and Akt2 in normal appearing mucosa of CRN patients. Contrary, examining protein abundance, we found a marginal increase of ERK1 (p = 0.09) in CRN-group, while phospho-ERK1, ERK2 and panAkt were similar between groups. One might argue that protein expression, especially for phosphorylated proteins, is a more direct marker for cell function compared to mRNA. However, Western blot measurements of protein expression are far less accurate than the qPCR method for measuring mRNA. The observed up-regulation of both ERK and Akt mRNAs are quite interesting findings given the fact that ERK/MAPK and PI3K/Akt pathways are well established important pathways in regulating proliferation and human carcinoma survival [17,18]. ERK1 and ERK2 are ubiquitous regulators of multiple cellular processes and dysregulated nuclear accumulation of activated ERKs (pERK) can lead to genomic instability and subsequent CRN progression [19][20][21][22][23]. It is difficult to assign a precise role of ERK signaling in human carcinogenesis due to its complexity and dependence on signaling intensity. Still, ERKs may play oncogenic and/or tumor suppressing roles in normal appearing colonic mucosa [24][25][26]. Our findings of markedly lifted expression levels for ERK1 points to this kinase as a possible useful predictive biomarker for CRN development. In addition to ERK and Akt, we found enhanced mRNA expression of the structural subunit Aβ, PPP2R1B, which is an isoform of the scaffold subunit of protein phosphatase 2A (PP2A). PP2A is considered a tumor suppressor and a negative regulator of both ERK and Akt activation and stimulator of Wnt/β-catenin signaling [27,28]. In the present study, PPP2R1B mRNA was up-regulated in CRN group and we speculate that this is a compensatory up-regulation to counteract up-regulation of Akt and ERK.
Prostaglandin E2 signaling Prostaglandin E2 synthesis and mucosal content As mentioned PLA2G4A mRNA expression is upregulated in normal appearing colonic mucosa from CRN patients, Table 2. The PLA2G4A gene encodes the enzyme cPLA2A which is a key enzyme involved in arachidonic acid mobilization and upstream release of many lipid mediators including lysophospholipids, prostaglandins, leukotrienes and lipoxins [29,30]. Understanding how the cPLA2A enzyme regulates tumorigenesis is hampered by the entwined effects of its many interacting products of downstream eicosanoid mediators.  [32].

Prostaglandin E2 receptors
Investigating EP receptors, we found EP3 to be upregulated in CRN patients, Table 2. Interestingly, a study of paired colonic normal and tumor tissues from CRN patients, reported the tumor tissue had downregulated expression of EP3 mRNA [33]. Of note, the EP3 receptor is unique among EP receptor subtypes, in that there are multiple isoforms generated through alternative mRNA splicing in the carboxyl tail of the EP3 gene resulting in isoform specific differences in G-protein coupling and signaling [34]. Our EP3 qPCR-primer did not differentiate between EP3 isoforms. Nonetheless, the major EP3 isoform is thought to couple to an inhibitory G protein (Gi), and hence the major outcome of PGE 2 -EP3 receptor signaling is inhibition of adenylate cyclase and activation of the ERK/MAPK pathway [35]. Thus, the role of the EP3 receptor in tumorigenesis seems to be multifaceted and isoform-dependent. As for cPLA2A, azoxymethane-induced colon cancer development is enhanced in EP3 receptor knockout mice, suggesting an antitumorigenic function for the EP3 receptor [33,36]. Further studies of the mRNA and protein expressions of individual EP3 isoforms in normal appearing colonic mucosa from humans are warranted.

Prostaglandin E2 influx and degradation
In terms of cellular PGE 2 influx and subsequent elimination we found no change in mRNA expressions of the major and specific PGE 2 influx transporter PGT in CRN patients. The expression of PGT, an organic anion polypeptide transporter (OATP) has been reported downregulated in CRC tissue [37]. However, other specific PGE 2 OATP-transporters, potentially involved in regulatory removal of PGE 2 , were not investigated in the present study. However, our group reported compensatory increase in the level of two OATP PGE 2 influx transporters, OATP2B1 and OATP4A1, located in the basolateral membrane of human colonic epithelia from CRN patients [38]. Since lowered expression of the PGE 2 inactivating enzyme, 15-PGDH, has been observed in cancer cells [39], we hypothesized a decrease in 15-PGDH expression in CRN patients compared to controls. Meanwhile, this hypothesis was not supported by a nearly 2-fold (p = 0.066) increase in 15-PGDH mRNA expression in CRN patients. The mechanism(s) for the observed elevation of 15-PGDH mRNA expression in the normal appearing colonic mucosa of CRN patients is unresolved. We speculate that a negative feedback balance between increased PGE 2 production and subsequent increase in removal/inactivation is present in the normal appearing mucosa of CRN patients.
Is our finding of perturbed mRNA expression a CRN predisposition?
Our suggestions of perturbed mRNA expressions as possible predisposing factors for the individual development of CRN could also be explained by the extant tumor tissue per se. Thus, at play could be local paracrine inducers or more systemic neuro-, endo-or immunocrine signaling from tumor to neighboring colonic mucosal areas. Yet, another pathway might be special cancer-inducing environmental stimulants as products of food digestion or microbial activity affecting gene expression in non-tumor mucosa. A conclusive settlement of this question between inborn predisposing constitutions versus tumor activity or environmental factors in normal appearing epithelium will require larger population studies. Regardless of the underlying mechanisms, our observation of up-regulation of several genes in normal appearing colonic mucosa suggests that normal appearing mucosa of CRN patients differs from non-CRN patients at a molecular level. Since up-regulated genes in this study have been found perturbed in CRC studies on colonic biopsies of affected mucosa, they could indicate possible predispositions for CRC development. The exploratory nature of our relatively small study cohort points to a need for further confirmation in larger prospective studies in order to determine if the observed aberrant marker genes may be useful predictive biomarkers.

Conclusion
We observed significant up-regulation of ERK1, ERK2, Akt1, Akt2, PLA2G4A, prostanoid receptor EP3 and phosphatase scaffold subunit PPP2R1B mRNA expression in normal appearing colonic mucosa of patients with CRN. Accordingly, normal appearing mucosa of CRN patients differs from non-CRN patients at a molecular level. Most notably, mRNA expression of ERK1 was lifted with high significance of p = 0.007 and may therefore be considered a potential candidate gene as predictive biomarker for developing CRN.
Our observations need to be validated in larger prospective studies.

Funding
This work was kindly supported by The Danish Cancer Society (jr. no. R141-A8964-15-S7), Else and Mogens Wedell-Wedellborgs Foundation (jr. no. 30-17-1), and Civilingeniør Bent Bøgh og Hustru Inge Bøghs Fond. The above listed funding supporters have contributed with financial support exclusively and did not have any influence on the design, experimental data, data interpretation or writing of paper.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
The study protocol was approved by The Scientific Ethical Committee of Copenhagen (H32013107) and The Danish Data Protection Agency (BBH-2013-024, I-Suite nr: 02342) and conducted in accordance with the Helsinki declaration. All patients participating gave prior informed written consent.

Consent for publication
Not applicable.