Correction: Nanomolar EP4 receptor potency and expression of eicosanoid-related enzymes in normal appearing colonic mucosa from patients with colorectal neoplasia


 BackgroundAberrations in cyclooxygenase and lipoxygenase (LOX) pathways in non-neoplastic, normal appearing mucosa from patients with colorectal neoplasia (CRN), could hypothetically qualify as predisposing CRN-markers. To test this hypothesis, biopsies were obtained during colonoscopy from macroscopically normal colonic mucosa from patients with and without CRN. Prostaglandin E2 (PGE2) receptors, EP1-4, were examined in Ussing-chambers by exposing biopsies to selective EP receptor agonists, antagonists and PGE2. Furthermore, mRNA expression of EP receptors, prostanoid synthases and LOX enzymes were evaluated using qPCR technology.Results Data suggest that PGE2 binds to high and low affinity EP receptors. In particular, PGE2 demonstrated EP4 receptor potency in the low nanomolar range. Similar results were detected using EP2 and EP4 agonists. In CRN patients, mRNA-levels were higher for EP1 and EP2 receptors and for enzymes prostaglandin-I synthase, 5-LOX, 12-LOX and 15-LOX. ConclusionIn conclusion, normal appearing colonic mucosa from CRN patients demonstrates deviating expression in eicosanoid pathways, indicating a likely predisposition for early CRN development. Moreover, PGE2 potency activates high affinity EP4 receptor subtypes, supporting relevance of testing EP4 antagonists in colorectal cancer management.


Results
Data suggest that PGE 2 binds to high and low a nity EP receptors. In particular, PGE 2 demonstrated EP4 receptor potency in the low nanomolar range. Similar results were detected using EP2 and EP4 agonists.

Conclusion
In conclusion, normal appearing colonic mucosa from CRN patients demonstrates deviating expression in eicosanoid pathways, indicating a likely predisposition for early CRN development. Moreover, PGE 2 potency activates high a nity EP4 receptor subtypes, supporting relevance of testing EP4 antagonists in colorectal cancer management.

Background
Colorectal cancer (CRC) is the third most common type of cancer worldwide and the second leading cause of cancer related deaths (1). CRC carcinogenesis is a multifactorial process, in which an accumulation of mutations leads to the formation of colorectal neoplasia (CRN) initially as adenomas and later adenocarcinomas (2). Genetics and chronic colonic in ammation are known risk factors for developing CRC (3) and an altered activity of the arachidonic acid (AA) metabolism including prostaglandins is likely involved. The speci c mechanisms, however, are poorly understood.
Aspirin (acetylsalicylic acid) is a non-steroid anti-in ammatory drug (NSAID) and non-selective cyclooxygenase (COX) inhibitor, which ameliorates CRC development (4,5). NSAIDs attenuate the in ammatory response mainly by inhibiting enzyme activity of COX isozymes, COX-1 and COX-2, which convert AA into the prostanoids PGD 2 , PGE 2 , PGF 2α , PGI 2 and thromboxane A 2 (TXA 2 ), Fig. 1, (3). COX-2 expression is elevated in human adenomas as well as in adenocarcinomas, which is why COX-2 is believed to be central to CRN and CRC pathogenesis (6). Accordingly, the protective effect of NSAIDs on CRC development is likely due to a reduced COX-activity and associated PGE 2 production (3,5,7). PGE 2 elicits tumorigenic effects by binding to either of its 4 G-protein coupled surface receptors termed EP1-4, Fig. 1 (8). These effects include proliferation, migration, invasion and angiogenesis (8). Each of the receptor subtypes has been linked to CRC tumorigenesis using knock-out mice (9)(10)(11). In particular, EP4 is suspected to be of special tumorigenic importance due to its activation of several central kinases (12,13).
We hypothesize that changes in eicosanoid signaling is an early tumorigenic mechanism detectable in even macroscopically normal appearing tissue. Here, we examine eicosanoid-related enzymes and receptors in non-neoplastic colonic mucosa from patients with and without CRN. Speci cally, we characterize function and expression of the EP receptor subtypes and examine the expression levels of prostaglandin D2 synthase (PTGDS), prostaglandin I2 synthase (PTGIS) and the PGF 2 α-reductase AKR1B1 (an aldo-keto reductase), all as indicators for altered levels of their respective prostanoids (18). Finally, we determine expression levels of 5-, 12-, and 15-LOX enzymes.

Study population
Patients (45-80 years of age) referred for colonoscopy, were screened for participation. Exclusion criteria included history of in ammatory bowel disease, conditions of intestinal malabsorption (e.g. coeliac disease and lactose intolerance), familiar risk of CRC (hereditary nonpolyposis colorectal cancer and familial adenomatous polyposis), pregnancy and/or continuous treatment with NSAID, anti-coagulant or phosphodiesterase inhibitor. Furthermore, incomplete examination of the entire colon resulted in exclusion.

Page 4/23
Patients were divided into 2 groups based on endoscopic ndings and medical history: patients with present or history of CRN (termed CRN patients) and patients without present nor history of CRN (termed and served as controls, CTRL patients). A total of 73 patients were enrolled. Patient group characteristics are shown in Table 1.  Selection of receptor agonists and antagonists was based on a thorough search of available literature, with a preference for compounds tested on human tissue.

Biopsy extraction
Six endoscopic biopsies were obtained from each patient using standard biopsy forceps (Boston Scienti c, Radial Jaw 4, large capacity). Biopsies were taken from macroscopically normal appearing sigmoid mucosa on retraction of the endoscope; about 30 cm orally from the anal verge and at least 10 cm from macroscopically abnormal appearing tissue.

Experimental methods
Two experimental methods were employed: functional studies in modi ed air-suction Ussing (MUAS) chambers measuring short circuit current (SSC) and quantitative real-time polymerase chain reaction (qPCR).

Functional studies in MUAS-chambers
Four biopsies were mounted and oxygenated in MUAS-chambers after extraction as described by Larsen et al (19). Biopsies were bathed on both sides with 10 mL Ringer, supplemented with 5.5 mM D-glucose. Temperature was maintained at 37.2 ºC by water jackets. An automated voltage-clamp device continuously recorded SCC and slope conductance (19).
Experiments began after a stable basal SCC was obtained within 10 min. All experiments were initiated by addition of amiloride (20 µM, mucosal side) to inhibit electrogenic sodium absorption mediated through epithelial sodium channels and followed by theophylline (400 µM, serosal side) to inhibit phosphodiesterase-dependent cyclic adenosine monophosphate (cAMP) degradation. Finally, to eliminate endogenous cAMP synthesis, indomethacin (13 µM, serosal side) was added and incubated for 40 min.
Biopsies from 47 patients were treated with PGE 2 and selective EP receptor agonists to investigate receptor function, Table 2. A single agonist was added in increasing concentrations (1 nM to 5 µM, serosal side) to each MUAS-chamber. The nal agonist concentration step was followed by the addition of 5 µM PGE 2 , to elicit a maximal PGE 2 -induced response. Experiments were terminated by the addition of acetazolamide, a carbonic anhydrase inhibitor (250 µM, serosal side), to measure HCO 3 -/H + -secretion, followed by bumetanide (25 µM, serosal side), to inhibit Na-K-Cl cotransporters and chloride secretion, and nally the Na + /K + -ATPase inhibitor ouabain (0.2 mM, serosal side) to assess and ensure tissue viability and data quality.

RNA isolation
Twenty biopsies, 10 CRN and 10 CTRL were matched according to gender and used for further qPCR investigations. RNA was extracted from the biopsies using RNeasy Mini Kit (Qiagen, Copenhagen, Denmark). Following extraction, RNA samples were placed on ice and quanti ed using a Nanodrop Spectrophotometer (LabTech International) in accordance with the MIQE guidelines (20). to 20 μL with nuclease-free water. The following cycling conditions were used: initial activation at 95 °C for 10 min., followed by 40 cycles of 95 °C for 15 sec., and 60 °C for 1 min. and data was collected during each cycling phase. Melt curve analysis, to ensure each primer set ampli ed a single, speci c product, completed the protocol. Quanti cation cycle (Cq) values were determined using Bio-Rad CFX96 Manager 3.0 software and the single threshold mode.
The geNorm reference gene selection kit (Primerdesign Ltd.) was used to identify the most stable reference genes and to determine optimal number of reference genes required for reliable normalization of qPCR data in these tissue samples (21). ß-actin and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were validated as the most stable reference genes in samples. The expression levels of genes of interest are expressed relative to the mean Cq value of the reference genes in each sample.
Primers were designed, synthesized and quality controlled by Primerdesign Ltd., Additional le 1. The sequences for the reference genes ß-actin and GAPDH are commercially sensitive and therefore unavailable.

Data analyses
The present study is exploratory and therefore not statistically powered for speci c endpoints. If identical experiments were performed on several biopsies from the same patient, a mean value of parameter results was used. A comparison of parameter values between patient groups was performed by an unpaired t-test when standard deviations were equal, and a Welch's t-test if unequal. Data are presented as mean ± SEM.
To assess agonists and receptors, data obtained from dose-response curves were analyzed with either a single-Michaelis-Menten model (srm) or a two-Michaelis-Menten receptor/site model (trm) using The EP4 agonist produced a similar sensitivity, demonstrating high potency in the low nanomolar range, Fig. 2. Concentrations of 30 nM or higher were necessary to induce SCC increases when stimulating with the other selective EP-agonists, data not shown. Moreover, 4 out of 22 biopsies exposed to the selective EP1 agonist showed no increase in SCC.
When applying Michaelis-Menten models (srm and trm) to data, a trm provided a better t than the srm in most analyses of data from experiments with PGE 2 , and agonists for EP2 and EP4 receptor subtypes, Fig.   3. Accordingly, at least 2 types of EP receptors appear activated, a high and a low a nity receptor, with different EC 50 s separated by a factor up to 200 in single experiments, Fig. 3. Average separation factors of the receptors were 64 for PGE 2 stimulation and 15 for the EP4 agonist, Fig. 4

Selective EP antagonists are unsuitable for determining EP receptor subtypes
Forty-one biopsies from 26 patients were exposed to EP antagonist cocktails, intended to inhibit all but one of the 4 EP receptor subtypes, followed by increasing PGE 2 concentrations. Unexpectedly, we recorded sizable SCC increases upon PGE 2 stimulation, even in the low nanomolar range, regardless of antagonist combination as well as in the presence of all 4 EP receptor antagonists, data not shown. These data indicate a lack of irreversible and competitive inhibition by all the 4 selective EP antagonists. Thus, with the present study design and protocol, none of the employed selective antagonists quali ed.

Competitive antagonism between EP4 receptor antagonist GW-X and PGE 2
Additional experiments were performed with only the selective EP4 antagonist GW-X, added prior to stimulation with PGE 2 , Table 3. Figure 8 shows  Table 3. To run a t-test for reliable judgement of differences in mean EC 50 s for GW-X between patient groups, more experiments are required.

Enzymes related to the COX and LOX pathways are upregulated in CRN patients
All investigated LOX enzymes (5-LOX, 12-LOX, and 15-LOX) demonstrated elevated levels of mRNA in CRN patients compared to CTRLs, Fig. 10. Moreover, the expression of PTGIS was signi cantly upregulated in the CRN group, whereas expression levels of PTGDS and ARK1B1 were unaltered, Fig.10.

Discussion
In the present study, we identi ed several differences in normal-appearing colonic mucosa from CRN patients, supporting the hypothesis of aberrations in enzymes and receptors of the eicosanoid pathway.
Independently of CRN history, we demonstrate that EP receptors bind PGE 2 with 2 different a nities indicating the presence of high and low a nity EP receptor subtypes. Furthermore, we observed similar mucosal responses to selective EP2 and EP4 receptor agonists. Assuming selectivity of these compounds towards their receptors, our data suggest presence of both a high a nity EP4 and a low a nity EP2 receptor subtype (22,23). High and low a nity EP receptors in human colonic mucosa have only been reported twice previously, but not investigated further (24,25).
Our experiments identi ed the EP4 receptor to be the EP receptor subtype with the highest secretory response in the colon, which is consistent with existing reports (25,26). Furthermore, based on experiments with the highly selective EP4 receptor agonist TC 2510 (23), our data suggest a presence of both high and low a nity EP4 receptors with associated higher mean potencies and lower mean e ciencies compared to PGE 2 . Meanwhile, the existence of 2 EP4 receptors was not corroborated by experiments with the selective EP4 receptor antagonist, GW-X, which was effective in human colonic mucosa previously (25). GW-X eliminated the biphasic PGE 2 dose-response curve, resulting in a single receptor dose-response curve. This may be explained as a surmountable rightward potency-shift for a single EP4 high a nity receptor, moving it closer to the potency of the low a nity receptor(s) in the presence of GW-X, maintaining a combined e ciency at high concentrations of PGE 2 with no antagonist present.
Stimulation of the EP4 subtype receptor is well documented as an important immunosuppressive trigger in the CRC microenvironment (27). Within the last few years, several interventional clinical phase-1 studies have been initiated with newly developed EP4 antagonists. Of these, 3 studies focus on CRC (28). Furthermore, another study points to a carcinogenic mechanism involving pericryptal COX-2-expressing broblasts, which exert paracrine control over tumor-initiating stem cells via a COX-2 and PGE2-EP4-Yap signaling pathway (29,30).
Taken together and respecting the relative few subjects in the present study, our ndings support presence of a high sensitivity for PGE 2 in even normal appearing colonic mucosa.
Separate additions of single selective EP antagonists did not change the ensuing PGE 2 -induced SCC.
Whether the PGE 2 -induced SCC increases re ect remaining secretion of incompletely inhibited EP receptor subtype(s) or resemble PGE 2 -induced secretion by other prostanoid receptors cannot be ascertained.
Surprisingly, employed EP receptor antagonists, except for GW-X, were not useful in the present study.
Our mRNA expression studies revealed increased expressions of receptor subtypes EP1 and EP2 in CRN patients. We, as others, have investigated EP receptor expression levels in human colonic tissue previously (31,32). We reported increased expression of EP3 in CRN patients, whereas expression levels of the remaining receptors were unaltered (31). Since identical primers against the subtype receptors were used in the 2 studies, the only recognized difference in study design were the number of reference genes.
Taken together, our studies indicate expressional alterations in at least 3 out of 4 EP receptor subtypes present in normal appearing colonic mucosa from CRN patients. EP4 expression levels were similar in CRN and CTRL patients. We speculate that, the previously reported elevated levels of EP4 found in human colonic adenoma and adenocarcinoma cell lines most likely re ect neoplastic time dependent differences (33).
We found PTGIS expression to be upregulated in CRN patients. Previous expression studies of PTGIS/PGI 2 in CRC patients have been ambiguous. One study found decreased PGI 2 levels using radioimmunoassay in CRC patients (34). Conversely, Lichao et al. found weak or no staining of PTGIS in normal tissue (corresponding to our biopsies from CRN patients) in microarray expression studies, while PTGIS expression was detected in CRC patients and increased in CRC patients with liver metastasis (35). Merging results, we hypothesize a stepwise increase relationship in PTGIS expression and the degree of colonic mucosa dysplasia and risk for liver metastasis.
All tested LOX enzymes had higher mRNA expression levels in colonic mucosa from CRN patients. For 5-LOX and 12-LOX, this is consistent with the bulk of literature. Both enzymes elicit key pro-in ammatory and pro-tumorigenic downstream functions and are upregulated in human colon adenomas and adenocarcinomas (16, 36, 37). Our results suggest that an upregulation of the LOX pathway is already present in normal appearing colonic mucosa from CRN patients. As such, 5-LOX and/or 12-LOX, enzyme expression might possess the potential of becoming an early predictive biomarker of CRN development.
Both 15-LOX isoforms are considered anti-tumorigenic and especially 15-LOX-1 and its product 13(S)-HODE appear tumor protective and downregulated in CRC tissue (17,37). Our employed 15-LOX primer unfortunately did not differentiate between the 2 isoforms. In contrast to previous studies, we observed increased 15-LOX expression in the mucosa of CRN patients. Given that we only investigated normalappearing mucosa, the observed upregulated expression of 15-LOX might be a compensatory effect before mucosal cells become neoplastic. It would be interesting to further track the expression of 15-LOX, to determine whether the expression is suppressed as the cells become carcinogenic.

Conclusions
Normal appearing colonic mucosa from patients with history of CRN demonstrates altered enzymatic expression of the eicosanoid pathway. Our data indicate a likely gene-based predisposition for early disease development. Furthermore, PGE 2 did activate EP receptors with different a nity including a high a nity EP4 receptor with nanomolar potency to PGE 2 . Whether this highly sensitive EP4 receptor is tumorigenic and as such could be targeted in CRN management remains to be clari ed.

Declarations
Ethics approval and consent to participate The study protocol was approved by the Scienti c Ethical Committee of Copenhagen (H-3-2013-107) and the Danish Data Protection Agency (BBH-2013-024, I-Suite no: 02342). The study was conducted in accordance with the Helsinki declaration. All participating patients gave written informed consent.
Authors' contributions URF was the principal investigator, took part in every aspect of this study and was major contributor in the writing of the manuscript. SKH was a major contributor in performing analyses of functional data and contributed in writing the manuscript. MABH contributed in generating functional data. TAJ contributed as an expert in performing and analyzing the expressional data. NB contributed as an expert in the functional part of the study, its study design, in data analysis and contributed in writing the manuscript. MBH served as the supervisor of the project and contributed in writing the manuscript. All authors read and approved the manuscript.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Model of the metabolization of arachidonic acid (AA). AA is metabolized by 3 different groups of enzymes: cyclooxygenases (COX), lipoxygenases (LOX) and epoxygenases (cyto-chrome P450). The COX pathway consists of 2 isozymes: COX-1 and COX-2. Both iso-zymes metabolize AA into PGG2 and then into PGH2, which is further converted to the pros-taglandins (PGs) PGD2, PGE2, PGF2α, PGI2 and thromboxane A2, (TXA2) by their respective synthases (3). Each product binds to its speci c membrane receptor. The CYP-450 pathway converts AA by epoxygenases and ω-hydroxylase into other downstream products not shown. The LOX pathway consists of 3 main enzymes termed 5-LOX, 12-LOX and 15-LOX (isozymes 15-LOX-1 and 15-LOX-2). They metabolize AA into hydroperoxydeico-satetraenoic acids (HPETEs), which are further reduced to hydroxyeicosatetraenoic acids (HETEs). The 5-LOX enzyme differs by also metabolizing 5-HPETE into leukotriene A4 by means of 5-lipoxygenase-activating protein (FLAP). * Enzymes already investigated in our laboratories, data published.

Figure 3
Dose-response curves of PGE2 and EP4 agonist, TCS 2510, experiments. X-axis: ligand con-centrations scaled logarithmically. Y-axis: changes in SCC. Triangles (black) indicate increas-es in SCC as a response to increasing PGE2 concentrations. Large dots (black) show increases in SCC as a response to increasing EP4 agonist concentrations. Dotted and long dotted lines (in blue colors) resemble single (srm) and two receptor model (trm) tting for PGE2 respec-tively. The unbroken and the medium dotted lines (in red colors) show trm and srm respec-tively for EP4 agonist. The trm ts data points more closely.    Calculated mean EC50 values of PGE2 and EP receptor agonists using single receptor model (srm) equations. Numbers under the graph show N/n, N = number of patients, n = number of biopsies, NA = not applicable due to too few N/n. Data are presented as means ± SEM. *p < 0.05.

Figure 6
Calculated mean RMax values displayed as µA·cm-2 from single receptor models upon biopsy stimulation with PGE2 or a selective EP receptor subtype agonist. Numbers under the graph show N/n, N = number of patients, n = number of biopsies, NA = not applicable due to too few N/n. Data are presented as means ± SEM. Calculated mean RMax values displayed as µA·cm-2 from two receptor models upon biopsy stimulation with PGE2 or a selective EP receptor agonist. Numbers under the graph show N/n, N = number of patients, n = number of biopsies, NA = not applicable due to too few N/n. Data are presented as means ± SEM. *p < 0.05.

Figure 8
Dose-response curves of PGE2 stimulation with and without EP4 antagonist GW627368X (GW-X). X-axis: PGE2 concentrations scaled logarithmically. Y-axis: changes in SCC. Trian-gles (black) show increases in SCC as a response to PGE2 doses without the addition of GW-X. Big dots (black) show increases in SCC in the presence of EP4 antagonist GW-X fol-lowed by PGE2 stimulation. The small dotted and the unbroken line (blue colors) resemble single (srm) and two receptor model (trm) tting. Long dotted line (red) show srm for exper-iments with GW-X, trm could not be calculated.

Figure 9
Expression levels of EP receptors. Expression of EP1 and EP2 are signi cantly higher in CRN patients.
Expression levels are relative to ß-actin and GAPDH. Data are presented as means ± SEM. *p < 0.05.