Ulcerative Colitis Induces Changes on the Expression of the Endocannabinoid System in the Human Colonic Tissue

Background Recent studies suggest potential roles of the endocannabinoid system in gastrointestinal inflammation. Although cannabinoid CB2 receptor expression is increased in inflammatory disorders, the presence and function of the remaining proteins of the endocannabinoid system in the colonic tissue is not well characterized. Methodology Cannabinoid CB1 and CB2 receptors, the enzymes for endocannabinoid biosynthesis DAGLα, DAGLβ and NAPE-PLD, and the endocannabinoid-degradating enzymes FAAH and MAGL were analysed in both acute untreated active ulcerative pancolitis and treated quiescent patients in comparison with healthy human colonic tissue by immunocytochemistry. Analyses were carried out according to clinical criteria, taking into account the severity at onset and treatment received. Principal Findings Western blot and immunocytochemistry indicated that the endocannabinoid system is present in the colonic tissue, but it shows a differential distribution in epithelium, lamina propria, smooth muscle and enteric plexi. Quantification of epithelial immunoreactivity showed an increase of CB2 receptor, DAGLα and MAGL expression, mainly in mild and moderate pancolitis patients. In contrast, NAPE-PLD expression decreased in moderate and severe pancolitis patients. During quiescent pancolitis, CB1, CB2 and DAGLα expression dropped, while NAPE-PLD expression rose, mainly in patients treated with 5-ASA or 5-ASA+corticosteroids. The number of immune cells containing MAGL and FAAH in the lamina propria increased in acute pancolitis patients, but dropped after treatment. Conclusions Endocannabinoids signaling pathway, through CB2 receptor, may reduce colitis-associated inflammation suggesting a potential drugable target for the treatment of inflammatory bowel diseases.


Introduction
The endocannabinoid system (ECS) has been described in the gastrointestinal tract in the epithelial, immune and neural compartments. It is involved in many physiological and physiopathological actions (peristalsis/contraction, secretion, gastric emptying, emesis, satiety and immunomodulation/inflammation and pain). [1][2][3][4][5][6] ECS roles comprise main facets of the pathogenesis of Inflammatory Bowel Disease (IBD) in humans, a disease that is likely to result from multiple factors, especially a disregulation of intestinal immune system and an inappropriate response to comensal bacteria or other luminal antigens. [7][8][9] Components of ECS include cannabinoid CB 1 and CB 2 receptors, their endogenous lipid ligands (2-arachidony glycerol-2-AG; anandamide -AEA) and enzymes involved in their biosynthesis and release (DAGLa and DAGLb for 2-AG; NAPE-PLD for AEA) [10][11][12][13][14][15], as well as mechanisms for cellular uptake and degradation, such as fatty acid amide hydrolase (FAAH) for AEA and monoacylglycerol lipase (MAGL) for 2-AG. [16,17] The role of endocannabinoids and its derivatives in IBD is not completely known [18][19][20][21][22], although cannabinoid CB 1 receptors have been proposed to participate in the epithelial wound healing during intestinal inflammation. [1][2][3][4]20] Additionally, cannabinoid CB 2 receptors are expressed in intestinal lamina propria suggesting a role in immunomodulation. [19,20,22] Data from animal model and human studies have suggested an upregulation of the ECS in inflammation processes either by increased receptor expression or by an enhancement of endocannabinoid production. [23][24][25][26][27] Treatment with CB 1 agonists, FAAH antagonists, inhibitors of endocannabinoid membrane transport, or genetic ablation of FAAH reduced inflammation. [23,25,28] Additionally, cannabinoid CB 2 agonists cause inhibition of proinflammatory citokines such as tumoral necrosis factor alfa (TNFa) and IL8. [29] Thus, ECS is positioned to exert a protective role in many of the points where homeostasis breaks in IBD, although this antiinflammatory role of the ECS remains to be conclusively determined in humans. [25,30] The aim of the present study is to analyse, by immunocytochemistry, the expression of components of the endocannabinoid system such as cannabinoid CB 1 and CB 2 receptors and the enzymes involved in cannabinoid degradation (FAAH and MAGL) and biosynthesis (DAGLa, DAGLb and NAPE-PLD), in normal human colonic tissue in comparison with untreated active ulcerative pancolitis at disease onset and after achieving remission, according to clinic and endoscopic criteria, and depending on severity of flare and treatment received.

Ethics statement
Biopsies and colonic resection samples were obtained after a written inform consent from all the patients, as requested by the clinical guides of Hospital del Mar. Research procedures were approved by the Hospital del Mar Clinical Research and Ethics Committee and were conducted according to the principles expressed in the Declaration of Helsinki.

Subjects
Human colonic endoscopic biopsies were selected from 24 patients with a first ever flare of extensive Ulcerative Colitis (UC) diagnosed by clinical, endoscopic and pathological criteria (E3, Montreal classification). [31] In each patient rectal mucosal samples were obtained at onset, at first colonoscopy, before any treatment (acute group) and after achieving clinical (Truelove and Witts score ,6 points) [32] and endoscopic remission (Mayo clinic score 0) [33], (quiescent group).
Twenty-two rectal samples were removed from colonic tissue of patients underwent colonic resections for colorectal cancer, at least 10 cm from the tumour (control group). In the control group, we confirmed histopathologically the absence of microscopic alterations. The analysis of the immunostaining patterns was carried out at transmural planes of the normal colonic tissue by comparing it with H&E staining.
Colonic samples were retrieved from tissue bank of Pathology Service at the Hospital del Mar from Barcelona, Spain. Data from each patient were collected retrospectively from medical records including age, sex, smoke and alcohol history, Body Mass Index (BMI) and comorbidity. In UC patients we recorded date of diagnosis, disease location (Montreal classification), endoscopic (Mayo clinic score) and clinical score (Truelove and Witts score: mild, moderate and severe) at onset, histological features and treatments received since initial diagnostic (5-aminosalicilates (5-ASA); corticosteroids; and/or the immunomodulators (Cyclospor-ineA and/or Azathioprine). Table 1 shows some of these records that characterize each UC patients.
Digital high-resolution microphotographs were taken under the same conditions of light and brightness/contrast by an Olympus BX41 microscope equipped with an Olympus DP70 digital camera (Olympus Europa GmbH, Hamburg, Germany). Digital images were mounted and labelled using Adobe PageMaker (San Jose, CA, USA).

Western blotting
We collected prospectively 8 rectal samples of control patients underwent colonic resection biopsies, processed as previously described [34,35], to evaluated the presence of CB 1 and CB 2 receptors, FAAH, MAGL, DAGLa, DAGLb and NAPE-PLD by Western blotting. Samples from were immediately snap frozen in liquid nitrogen and stored at 280uC until use. Membrane extracts of colon tissue were prepared in HEPES 50 mM (pH 8)-sucrose 0.32 M buffer by using a homogenizer. The homogenate was centrifuged at 800 xg for 10 minutes at 4uC and the supernatant was centrifuged at 40000 xg for 30 minutes. The pellet was suspended in HEPES 50 mM buffer and potterized using a homogenizer.
For immunoblotting, equivalent amounts of membrane proteins (20 mg) were separated by 10% sodium dodecyl sulphatepolyacrylamide gel electrophoresis (SDS-PAGE), electroblotted onto nitrocellulose membranes, and controlled by Ponceau red staining. Blots were preincubated with a blocking buffer containing PBS, 0.1% Tween 20 and 2% albumin fraction V from bovine serum (Merck, Whitehouse Station, NJ, USA) for 1 h at room temperature. For protein detection, each blotted membrane lane was incubated separately with the specific CB 1 Healthcare, Buckinghamshire, UK). Membranes were subjected to repeated washing in PBS-T and the specific protein bands were visualized using the enhanced chemiluminiscence technique (ECL, Amersham) and Auto-Biochemi TM Imaging System (LTF Labortechnik GmbH, Wasserburg/Bodensee, Germany). Western Blots showed that each primary antibody detects a protein of the expected molecular size.

Quantification of mucosa immunostaining
One immunostaining batch contained 70 tissue sections of all experimental groups, thus slices corresponding to the three experimental groups were stained simultaneously. For each primary antibody and for each subject, 2-3 different batches were run. On each tissue section we focussed on epithelium and lamina propria of the mucosa. For epithelium, we carried out a densitometrical quantification for each component of the ECS. For lamina propria, we evaluated the type and the number of immunostained immune cells for each 100 cells observed by hematoxylin-eosin (H&E) staining. In addition, ECS quantification was segregated depending on UC severity scored to mild, moderate and severe (Truelove and Witts score), and by the treatment received (5-ASA, corticosteroids, and/or the immunomodulators).
Digital high-resolution microphotographs were taken with the 106 objective of an Olympus BX41 microscope under the same

Statistical analysis
Data were analyzed using SPSS 15.0 software (Statistical Package for the Social Sciences Inc., Chicago, Illinois, USA). Results are expressed as mean6SEM. Differences between groups were evaluated using U Mann Witney and Wilcoxon tests for non parametric observations. A P value of P,0.05 was considered statistically significant.

Results
Presence of the endocannabinoid system in the normal human colonic tissue: Western blot analysis Immunohistochemical distribution of the endocannabinoid system in the normal human colonic tissue Results for the immunohistochemical distribution were summarized in a rating scale (Table 3)   immunoreactivity in the Paneth cells, at the bottom of the crypts, than in the remaining colonic epithelium ( fig. 2G inset). A number of subepithelial CB 2 + plasma cells and probably some macrophages were detected in the lamina propria ( fig. 2H arrow, inset). We also observed weak CB 2 immunoreactivity in the muscularis mucosae and muscularis externa whereas intense staining was located in the endothelial cells of the blood vessels ( fig. 2I arrow, inset). Numerous CB 2 + ganglion cells and fibers were evident in the myenteric plexus ( fig. 2I) and the submucosal plexi. FAAH immunostaining disposed in the epithelial cells, being prominent in the apical one third and perinuclear portions of the absorptive cells ( fig. 2J, K inset, asterisk). The brush border of the microvilli was nearly absent of staining ( fig. 2K inset, arrowheads). We detected few scattered FAAH+ immune plasma cells in the lamina propria. No staining was observed neither in the muscularis mucosae, muscularis externa or serosa, whereas intense staining was observed in some ganglion cells and fibers of the myenteric plexus ( fig. 2L).
MAGL immunoreactivity was located in the central portion of the epithelial cells, thus, apical to the nucleus of the absorptive cells and basal to the mucus droplets of the goblet cells ( fig. 2M, N, inset n0). A number of immunoreactive polymorphonuclear cells was distinguished in the lamina propria ( fig. 2N, inset n9). No staining was detected in both muscularis mucosae and externa. The myenteric plexus was characterized by a meshwork of MAGL+ fibers that disposed surrounding unstained parasympathetic nervous cells (fig. 2O).
Strong NAPE-PLD immunoreactivity in the apical surface of the epithelial border of the crypts ( fig. 3A) and numerous positive plasma cells was observed ( fig. 3B, inset). Intense NAPE-PLD immunostaining characterized both layers of muscularis externa ( fig. 3C). Numerous immunoreactive fibers, but not cell bodies, disposed in the myenteric plexus ( fig. 3C).
We observed a similar DAGLa staining pattern in the colonic tissue to that of CB 1 and NAPE-PLD proteins ( fig. 3D). An intense immunoreactivity characterized the apical surface of epithelial border facing to lumen (arrows in fig. 3E, inset e9). We observed some DAGLa+ plasma cells in the lamina propria ( fig. 3E, inset  e0). Muscularis mucosae and externa showed an intense DAGLa immunoreactivity ( fig. 3F) in a similar granular aspect to that of CB 1 immunoreactivity. Numerous DAGLa+ fibers disposed surrounding unstained ganglion cells in the myenteric plexus ( fig. 3F, inset).
Intense DAGLb expression was mainly located surrounding the nucleus of the epithelial cells ( fig. 3G, H, inset h0). A number of scattered plasma cells also showed intense DAGLb staining   fig. 3H, inset h9). Muscularis mucosae appeared positive, but strongly DAGLb expression was evident in both layers of the muscularis externa, mainly in the inner one ( fig. 3I). The myenteric plexus was characterized by strongly DAGLb+ ganglion cells and a dense fibre network ( fig. 3I).

Densitometrical quantification of ECS immunoreactivity in the colonic epithelium
Microphotographs showing qualitative differences of the immunoreactivity for each ECS component in the epithelium of control, acute and quiescent groups are shown in figure 4.
Quantification of epithelial immunoreactivity for ECS components is shown in figure 5. CB 1  These data suggest a dysregulation of the AEA balance in the acute inflammatory process that recovers to control level after treatment.

Percentage of the ECS immunoreactive cells in the lamina propria
We found pronounced changes in the number of FAAH+ and MAGL+ cells, but not to the remaining ECS components (fig. 6). FAAH+ cell number rose in acute group compared with control one (11.2%61.9% vs 1.29%60.3%; p,0.001). Besides, a decrease in the number of FAAH+ cells was evidenced in quiescent group compared with acute group (4.8%60.6% vs 11.2%61.9%; p,0.001) but was still notably higher than in controls (p,0.001).

Quantification of epithelial ECS immunoreactivity depending on the severity of the UC disease
We compared ECS in acute group depending on the severity of the disease and after remission (quiescent group) vs control tissue ( fig. 7). CB 1 expression did not change in acute samples at any clinic score. In quiescent samples, CB 1  In mild UC these levels were even higher in quiescent stage than in acute one (p,0.05). No differences were seen in severe colitis among the three groups.

Quantification of epithelial ECS immunoreactivity depending on treatment
We analyzed ECS immunoreactivity in quiescent samples depending on the treatment received: 5-ASA (3 cases), 5-ASA and corticosteroids (15 cases), or 5-ASA, corticosteroids and immunomodulators (6 cases) ( fig. 8). Regarding CB 1 levels, there was a decrease in patients treated with 5-ASA+corticosteroids

Discussion
Our data are consistent with previous studies on the expression of CB 1 and CB 2 receptors in human and rodent colon. [20,21,36,37] A novelty of our study is the finding of CB 1 staining in the goblet cells. Interestingly, the previous human study [20] did not report CB 1 staining in the goblet cells probably as a result of mucus-blocking antibody binding. Casu and collaborators [21] described non-specific labelling in the murine colonic epithelial cells of the large intestine because it persisted in preabsorption and omission controls. In contrast, we observed faintly CB 1 immunoreactivity in the submucosal and myenteric ganglion plexi, with the exception of some fibers. The welldescribed presynaptic localization of CB 1 receptor contrasts with the presence of this receptor into submucosal ganglion cell bodies, as was described in the human and mouse colon. [20,21,37] Our results revealed similar CB 2 expression in the mucosal epithelial cells from normal patient samples in a previous human colonic study that, using different CB 2 antibodies, supports our immunohistochemical data. [20] Of note, we observed strong CB 2 expression in the Paneth cells at the bottom of the crypts. CB 2 + subepithelial plasma cells and macrophages in the lamina propria was described previously by Wright and collaborators. [20,22] A novelty data was the finding of CB 2 staining in the submucosal and myenteric plexi of the normal human colonic tissue. Recently, CB 2 expression was observed in the enteric nervous system in rodent and human ileum [19,22], and in the rat ileum containing longitudinal muscle and myenteric plexus. [38] Taking together Figure 5. Quantification of ECS component immunoreactivity in the colonic epithelium. A: Untreated acute UC at disease onset showed increases in CB 2 , DAGLa and MAGL immunoreactivity, and decreases in NAPE-PLD immunostaining. After achieving remission (quiescence), CB 1 and CB 2 receptor immunoreactivity dropped, MAGL immunostaining maintained the same levels than acute group and NAPE-PLD immunoreactivity reverted to control levels. B: CB 2 /CB 1 ratio increased in both groups. However, CB 2 immunoreactivity increased in acute patients, while in quiescent patients there was a decrease of CB 1 receptor and a reverted restoration of CB 2 level. NAPE-PLD/FAAH ratio dropped in acute group, but rose to control levels in quiescent one. Histograms represent the mean6SEM. U Mann Witney and Wilcoxon tests: *P,0.05, **P,0.01 and ***P,0.001 versus control group; #P,0.05 and ##P,0.01 versus acute group. N = 22, 24 and 24 for control, acute and quiescent groups respectively. doi:10.1371/journal.pone.0006893.g005 these results point to a differential role of cannabinoid CB 1 and CB 2 receptors in human colonic tissue. CB 1 could be modulating colonic neuronal input and secretion while CB 2 may participate in colonic immunomodulation.
Other important novelty is the presence of the two endocannabinoid-degradating enzymes (FAAH and MAGL) in the epithelial cells of human colonic tissue. We have clearly detected FAAH expression in plasma cells of the lamina propria and in ganglion cells of the enteric nervous system. These results are related to the fact that FAAH blockers like URB597 reduce significantly the inflammation in the mouse colon [28], and selective FAAH inhibitors like AA-5-HT inhibited intestinal motility. [39] MAGL localization into epithelial cells is in agreement with the presence of MAGL activity in the soluble and membrane cellular fractions. [40] Of note the immunoreactive polymorphonuclear cells in the lamina propria, a fact that has not been observed previously. In contrast to Duncan and collaborators [40], we did not observed MAGL immunoreactivity in the human smooth muscle and mucosal layers, but we detected MAGL expression in fibers of the enteric nervous system.
We have reported the first analysis of the presence of DAGLa, DAGLb and NAPE-PLD in the human colonic tissue. Although 2-AG is considered a full cannabinoid receptor agonist, it is also an intermediate in triacyl/diacylglycerol metabolism as well as a prominent molecule linking the cannabinoid signalling with lysophospholipids and diacilglycerol-PKC signalling system. However, although we cannot strictly consider both DAGLa and DAGLb as pure endocannabinoid-synthesizing enzymes, we will focus on their potential role in the endocannabinoid system.
On the other hand, NAPE-PLD is another recently characterized cannabinoid biosynthesis enzyme that mediates the release of Nacyl ethanolamides (including AEA) from a phospholipid precursor (N-acyl-phosphatidylethanolamide, NAPE). [15,41] Our results are compatible with an active synthesis of ECs, i.e. AEA and 2-AG, in healthy human colonic tissue.
There are higher levels of cannabinoid CB 2 receptors (but not CB 1 receptors) in the mucosa epithelium of UC, mainly in mild and moderate-scored patients. These data suggest a dysregulated AEA tone in the colon of these patients, in agreement with previous findings. [20,25] However, we observed low NAPE-PLD expression, mainly in moderate and severe-scored pancolitis patients, and no changes in the AEA-degrading enzyme FAAH, suggesting a decrease of AEA levels, as deduced by the NAPE-PLD/FAAH ratio, while D-Argenio et al. found high AEA levels in biopsy samples of colons from untreated UC patients. [25] This discrepancy may be explained by the fact that NAPE-PLD is not the only source for AEA, as others enzymes are also capable of generating AEA from NAPE, such as a/b hydrolase 4, lyso-PLD, lyso-PLC, and phosphatases such as PTPN22. [42][43][44] Thus, although we detect a dysregulated AEA tone, the whole changes of AEA-related enzymes could lead to an increased level of this EC.
Regarding 2-AG, we observed an increase of DAGLa and MAGL expression in the colonic epithelium of acute UC patients, suggesting an increase of 2-AG turnover during the inflammation, but not a dysbalance of 2-AG levels, as suggest the DAGL/MAGL ratio. The maintenance of DAGL/MAGL ratio is in agreement with the absence of 2-AG variations observed in the mucosa of TNBS-treated rats, DNBS-treated mice and UC patients. [25] The high DAGLa and DAGLb expression detected in the human colonic epithelium may be partially related with the high 2-AG levels described in colonic mucosa of untreated rats, in contrast to that of control patients. [25] Interestingly, severe clinic score patients showed no significant increase in CB 2 receptors, and this fact correlates with a lack of increased 2-AG turnover (no increases of synthesizing-and degrading enzymes), thus suggesting a diminished ECS response to the inflammatory insult. In light of these findings, we could speculate that ECS-related drugs potentiating ECs turnover could be useful in managing the disease in this subpopulation of patients. Regarding the cannabinoid receptors in treated UC, the acute CB 2 increase in UC patients is reverted in the chronic state, irrespective of the treatment. This fact suggests a putative role of CB 2 receptor in mediating acute inflammatory response. In addition, the treatments, mainly the 5-ASA+corticosteroids one, lead to a chronic down-regulation of CB 1 receptor (not displayed acutely), probably reflecting a diminished colonic functionality in the chronic state of the disease, since CB 1 receptor have been implicated in colonic motility and secretion. [27,39] Thus, cannabinoid CB 1 receptor could be a biological marker of UC progression. Interestingly, while the high MAGL expression is maintained in quiescent patients, NAPE-PLD expression recovered to control levels, suggesting a partial recovery of the ECS dysregulation after treatments.
In summary, these data indicate that endocannabinoid signaling pathway is altered in UC, acting probably through cannabinoid Figure 7. Quantification of epithelial ECS immunoreactivity depending on the UC severity score. Main changes were observed mainly in mild and moderate acute UC. Histograms represent the mean6SEM. U Mann Witney and Wilcoxon tests: *P,0.05, **P,0.01 and ***P,0.001 versus control group; #P,0.05 versus acute group. N = 6, 13 and 5 for mild, moderate and severe groups respectively. doi:10.1371/journal.pone.0006893.g007 CB 2 receptor as a counteregulatory system aimed to reduce colitisassociated inflammation. In addition, the changes observed in the remaining ECS components, both acutely and after treatment, suggest that drugs acting at the ECS could be potential therapeutic approaches that need to be explored in more depth, for the treatment of inflammatory bowel diseases.

Supporting Information
Supporting Information S1 Generation of NAPE-PLD-, DAGLa-, DAGLb-specific antibodies. We have generated poly-clonal rabbit antibodies against proteins of the cannabinoid machinery. Immunizing peptides were 1) a 13-amino-acid (aa) peptide comprising part of both the C-terminal and the Nterminal regions of NAPE-PLD (MDENSCDKAFEET); 2) a 16aa peptide from the C-terminal region of DAGL alpha (CGASPTKQDDLVISAR); 3) a 16-aa peptide from an internal sequence of DAGL beta (SSDSPLDSPTKYPTLC). We employed a chimeric sequence peptide as immunogen for NAPE-PLD antibody generation. The aim of this chimeric construction was to obtain two distant epitopes exposed in the native protein because one of them belongs to the N-terminal and the other to the C- Figure 8. Quantification of epithelial ECS immunoreactivity depending on the treatment received. As a relevant finding, treatment is associated with changes in the expression of cannabinoid receptors and EC-production and -degradation enzymes, suggesting that these proteins can be considered as biomarkers of active disease/response to treatment. Histograms represent the mean6SEM. U Mann Witney and Wilcoxon tests: *P,0.05 and **P,0.01 versus control group; #P,0.05 versus acute group. N = 3, 15 and 6 for 5-ASA, 5-ASA+corticosteroids and 5-ASA+corticosteroids+immunomodulators respectively. doi:10.1371/journal.pone.0006893.g008 terminal region of the protein, both regions having random coil structures. NAPE-PLD, DAGL alpha and DAGL beta peptides were synthesized and coupled to keyhole limpet hemocyanin (KLH, JPT Peptide Technologies, Berlin, Germany). The three peptides were injected to rabbits (two animals per peptides), according to standard protocols for generation of antisera, with the IgG fraction subsequently purified by means of a protein A column (Sigma, St. Louis, MO, USA).