Introduction

Celiac disease is a permanent intolerance to wheat gluten and related proteins in rye, barley, and possibly, also oats. In genetically predisposed individuals, ingestion of gluten may lead to inflammatory injury of the small intestinal mucosa, characterized by villous atrophy, which resolves on a strict gluten-free diet [1]. Diagnosis of celiac disease is based on histological examination of a small intestinal biopsy specimen demonstrating mucosal lesions, while the patient is on a normal, gluten-containing diet [2]. The serological tests most commonly used in clinical practice are the detection of either tissue transglutaminase (tTG) or antiendomysial (EMA) antibodies [3]. In older children and adults, the symptoms are often vague, and consequently, the disorder can be initially neglected [4]. It has become apparent that celiac disease is a public health problem in many European countries and the USA [5]. Recent findings have clarified many aspects of celiac disease pathogenesis, particularly the role of tTG in favoring the binding of gluten peptides to HLA-DQ2 and their recognition by disease-specific intestinal T cells [6]; nevertheless, many issues remain to be clarified. Research on these issues has also been greatly hampered by the fact that no well-established animal model for celiac disease exists. However, it is possible to induce in rodents a celiac-like atrophy with methotrexate (MTX). This is a compound used for neoplastic treatment regimens that inhibits the enzyme dihydrofolate reductase. The reduced availability of intracellular folate impairs DNA synthesis [7]. MTX-induced transient intestinal mucositis in rats is characterized histologically by crypt loss, villus fusion and atrophy, capillary dilatation, and a mixed inflammatory cellular infiltrate [8, 9], very much resembling the gut appearance commonly evidenced in celiac patients [10].

A signalling system comprising (1) G-protein-coupled receptors for Cannabis psychoactive principle, Δ9-tetrahydrocannabinol, i.e., the cannabinoid CB1 and CB2 receptors, (2) the endogenous ligands of these receptors, i.e., the “endocannabinoids” anandamide and 2-arachidonoyl-glycerol (2-AG), and (3) enzymes for the biosynthesis and metabolism of the endocannabinoids has been shown to be present and tonically active in the gastrointestinal system [11, 12]. This “endocannabinoid” system plays important roles under physiological conditions, such as the control of both gastric and intestinal motility and secretion [13]. An increasing body of recent evidence suggests that an overactive endocannabinoid system in the gut might intervene to reduce the extent and consequences of intestinal inflammation in several animal models of colitis [14]. In these models, the levels of the endocannabinoids (in most cases, anandamide rather than 2-AG) or the expression of their receptors (namely, CB1 receptors), or both, are elevated in the colon or small intestine to exert a protective action against some of the consequences of inflammation, i.e., increased motility and secretion or epithelial wounding. This protective action of anandamide can be exerted by activating not only CB1 receptors but also CB2 receptors and transient receptor potential vanilloid type 1 (TRPV1) channels. Although strong evidence for the overactivity of the endocannabinoid system has been found also in human inflammatory bowel diseases (IBDs) [15, 16] as well as during other pathological conditions where this signalling system seems to play a protective role, such as in human colorectal carcinoma [17], no study has addressed to date whether intestinal endocannabinoid and/or cannabinoid receptor levels are upregulated in human celiac disease. With this background in mind, we have undertaken the present investigation where we measured the levels of endocannabinoids in duodenal biopsies from celiac patients, both during active disease with atrophy and after remission, and in the jejunum of rats at different times after treatment with MTX. We report that elevated intestinal endocannabinoid levels peaking with atrophy is a hallmark not only of celiac disease but also of MTX-treated rats, which might, therefore, be suitable for future experimental studies on the role of the endocannabinoid system in this intestinal inflammatory disorder.

Materials and methods

Sampling and analysis of biopsies from patients with celiac disorder

Biopsy specimens were obtained at the Azienda Ospedaliera “Rummo” from the distal duodenum of 13 celiac disease patients at first diagnosis internalized at the same Hospital (5 men and 8 women; median age 22.3 years, range 8–45) and were snap frozen in liquid nitrogen. Diagnosis was established by the finding of typical mucosal lesions with crypt cell hyperplasia and total villous atrophy, according to Marsh’ criteria [18] and by positivity for EMA antibodies. After obtaining informed consent, duodenal biopsies were also taken from seven celiac disease patients (four men and three women; median age 32.3 years, range 26–42) on total remission and with normal mucosal architecture with long villi and short crypts. Biopsies were also obtained from 11 non-celiac control subjects (4 men and 7 women; median age 35.1 years, range 18–61) who underwent gastrointestinal endoscopy for dyspepsia. Both celiac disease patients on a gluten-free diet and control subjects were negative for EMA antibodies, whereas in the former, the remission of the disease was documented also by a normal mucosal architecture on histological evaluation. All patients and controls recruited gave their informed consent to the study. All human studies were approved by the appropriate ethics committee and have been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki.

Immunofluorescence

In experiments aimed at detecting CB1 receptors, immunofluorescence with confocal microscopy was used. Cryosections were fixed in acetone and incubated for 1 h at room temperature with a rabbit polyclonal anti-CB1 receptor (Cayman, 1:50), followed by horse anti-rabbit flourescein isothiocyanate conjugated second antibody (Vector, 1:200). These antibodies were applied for 45 min in the dark followed by incubation (20 min) with ToPro-3 (Molecular Probes) for counterstaining the nuclei. Finally, the sections were mounted in phosphate buffered saline/glycerol (1:1 by vol.). To confirm the specificity of the primary antibody, controls included preadsorption of the primary antibody with the corresponding synthetic blocking peptide (100 μg/ml, for 1 h at room temperature) or omission of the primary antibody. CB1 receptor immunofluorescent cells were imaged with a Leica SP confocal microscope (Germany).

Experiments in methotrexate-treated rats

Intestinal atrophy was induced in male adult Wistar rats by i.v. injection of a single dose of MTX through the femoral vein (30 mg/kg body weight in 0.4 ml saline) [10]. Rats (n = 4 per group) were fed ad libitum up to day 7 after the injection and were killed under ether anesthesia at 2, 3, and 7 days after drug administration. Rats without MTX injection (n = 4) were used as normal controls (day 0). Specimens from the proximal jejunum were freshly collected, and aliquots were either snap frozen at −80°C or fixed in formalin fixative for 12 h before routine processing for paraffin embedding. On the other part of jejunum, the mucosa was gently scraped off the submucosal layer, and tissue specimens were then kept at −80°C until assayed. All these procedures were approved by the Animal Ethics Committee of the Federico II University Hospital, Naples, Italy.

Histopathological analysis of intestinal damage and repair

Transverse sections (4 μm in thickness) of paraffin-embedded proximal jejunal specimens were stained with hematoxylin and eosin (H and E) and examined with a light microscope. Quantitative histological analyses were conducted to measure the changes in crypt depth and villous height in the jejunal specimens over the time course. Microscopic images were acquired and analyzed. For each animal, the height and depth of ten villi and crypts, respectively, were measured at three tissue levels, each separated by 100 μm, and means of 30 measurements were calculated for each animal.

Measurement of myeloperoxidase activity

To measure the presence of inflammatory infiltration in jejunal tissue, tissue myeloperoxidase (MPO) activity was determined by a standard enzymatic procedure as described previously [19] with slight modifications. Briefly, the tissue specimen was homogenized in buffer (0.5% hexadecyltrimethylammonium bromide in 50 mmol/l potassium phosphate buffer, pH 6.0) using a Polytron-type homogenizer three times for 30 s each on ice. Tissue homogenate was then sonicated for 10 s. The sample was centrifuged at 20,000×g for 20 min at 4°C, and the supernatant was collected. The sample (100 μl) was added to 2.9 ml of 50 mmol/l phosphate buffer (pH 6.0) containing 0.167 mg/ml O-dianisidine hydrochloride and 0.0005% hydrogen peroxide, and the kinetics of absorbance at 460 nm and at 25°C were measured by using a spectrophotometer. Protein concentration of the supernatant was determined by using Lowry assay. The values were standardized using MPO purified from human leukocytes (SigmaChemical, St. Louis, MO).

Endocannabinoid measurements

Biopsy samples were weighed, immersed into liquid nitrogen, and stored at −70°C until extraction of endocannabinoids. Tissues were extracted with chloroform/methanol/Tris HCl 50 mM (2:1:1, by volume) containing 100 pmol/l each of d8-anandamide, d4-PEA, and d5-2-AG (Cayman Chemicals, Ann Arbor, MI, USA). The lipid extracts were purified by silica column chromatography, as described previously [15], and the fractions containing anandamide, PEA, and 2-AG, were analyzed by isotope dilution liquid chromatography–atmospheric pressure chemical ionization mass spectrometry in the selected ion monitoring mode, as described in detail elsewhere [20]. Results were expressed as picomoles or nanomoles per milligram of extracted lipids. Data were compared by one-way analysis of variance (ANOVA) followed by Bonferroni’s post hoc analysis.

Results

Endocannabinoid levels in duodenum biopsies from celiac patients

As shown in Fig. 1, the levels of both the endocannabinoid anandamide and of its congener PEA, which possesses anti-inflammatory activity not mediated by cannabinoid receptors, were significantly elevated in the duodenum of patients with untreated celiac disease at the highest degree of atrophy (n = 13) as compared to those in non-celiac patients (n = 11). The levels of the other endocannabinoid, 2-AG, were also elevated but not significantly, perhaps because of the strong variability between individual values. Importantly, after reduction in atrophy following gluten-free diet, the levels of anandamide, but not PEA, in n = 7 patients returned to those observed in non-celiac patients. The levels of 2-AG also appeared to go back to normal, although again, in a nonstatistically significant way.

Fig. 1
figure 1

Endocannabinoid levels in duodenum biopsies from celiac patients. Data are means±SE of n = 11 for healthy patients (controls), n = 13 for non-treated celiac patients, and n = 7 for celiac patients after remission from atrophy after a gluten-free diet. Single asterisk p < 0.05; double asterisks p < 0.01; triple asterisks P < 0.001 vs controls; plus sign p < 0.05 vs non-treated patients, as assessed by ANOVA followed by the Bonferroni’s test

Full size image

Cannabinoid CB1 receptors in duodenum biopsies from celiac patients

To look at the presence of CB1 receptors, we performed immunohistochemical analysis of duodenal sections from celiac patients at first diagnosis and from control patients using a CB1 receptor polyclonal antibody. As shown in Fig. 2, CB1 immunofluorescence was detected mostly in elongated, fiber-like structures that correspond possibly to myenteric neuronal fibers, in agreement with previous studies [21]. CB1 immunofluorescence appeared to be strongly increased in biopsies from patients with active celiac disease as compared to healthy patients and patients after remission. Importantly, we found that most of the immunostained fibers expressing CB1 receptors were located in the subepithelial region, where gluten-reactive pro-inflammatory Th1 cells are well known to be present. The specificity of CB1 staining was demonstrated by the lack of signal in sections incubated in the presence of the CB1 blocking peptide or in the absence of the first antibody (data not shown).

Fig. 2
figure 2

Cannabinoid CB1 receptor immunofluorescence (green or light gray in black and white) by confocal analysis in bioptic duodenal samples from non-celiac (a) and celiac (b) patients. Each panel is representative of similar results obtained in three distinct patients from each group. (Original magnification ×400)

Full size image

Histopathological analysis and myeloperoxidase activity of the jejunum of methotrexate-treated rats

MTX treatment of rats induced typical intestinal histological damage profiles, with the most pronounced effect in the duodenum and proximal jejunum. In the proximal jejunum, MTX induced apparent damage by day 2 (Fig. 3b) and maximal damage on day 3 (Fig. 3c), followed by a rapid recovery on day 7 (Fig. 3d). By day 2, a slight inflammatory cellular infiltrate peaking on day 3 was also present. On day 3, jejunal histology was characterized by increased crypt depth, villous fusion, and shortening (Fig. 3c). The villous returned to normal height on day 7. Crypt depth also normalized by day 7 (Fig. 4a). To eliminate the possibility that the above histological changes were a result of a reduced food intake after MTX injection, groups of pair-fed rats without MTX injection were killed on day 3 and day 7 for comparison. Histological analysis revealed normal intestinal villous height and crypt depth in the pair-fed rats on days 3 and 7 (not shown), indicating that the intestinal histological changes in the MTX-injected animals did not result from a reduction in food intake but rather were a result of a direct effect of MTX on the intestinal mucosa. Myeloperoxidase assay on tissue specimens confirmed the histopathologic observation that jejunal inflammation was present on days 2 and 3 (Fig. 4b).

Fig. 3
figure 3

Histopathological (H and E staining) features of proximal jejunum of rats during MTX-induced damage and repair. a Normal jejunal histology on day 0. b A partial damage of the jejunum in day 2 rats. c Severe mucosal damage in the jejunum on day 3. d Nearly normal histology on day 7. (Original magnification ×200)

Full size image
Fig. 4
figure 4

a Structure of small bowel mucosa in rats injected with methotrexate (MTX, 30 mg/kg) and untreated (control) rats. Asterisk p < 0.05 vs control rats. b Myeloperoxidase activity (MPO) in the small bowel after MTX injection and untreated rats. Asterisk p < 0.05 vs control rats

Full size image

Endocannabinoid levels are elevated in the jejunum of methotrexate-treated rats

Samples from the jejunum of MTX-treated rats were divided into mucosa and muscle/serosa layers to investigate the possible cellular origin of putative changes in endocannabinoid levels. As shown in Fig. 5a, in the muscle/serosa samples, a peak of anandamide, 2-AG, and PEA levels was found 3 days after treatment and returned to basal levels after 7 days. In the mucosa samples (Fig. 5b), 2-AG and PEA levels after 3 days were significantly different from those of control rats and also returned to normal after 7 days. Anandamide levels at 3 days were significantly different from those at 2 and 7 days (p < 0.05), but not from those of control rats.

Fig. 5
figure 5

Endocannabinoid levels in the jejunum of MTX-treated rats. a Levels in muscle/serosa. b Levels in mucosa. Data are means±SE of n = 3 different determinations. Single asterisk p < 0.05; double asterisks p < 0.01 vs controls; single plus sign p < 0.05; double plus signs p < 0.01 vs 3 days; number sign p < 0.05 vs 2 and 5 days; as assessed by ANOVA followed by the Bonferroni’s test

Full size image

Discussion

The present study was undertaken with two main aims. First, we wanted to determine if the elevated intestinal levels of endocannabinoids observed in IBDs and in animal models of colitis also occur during active celiac disease. Secondly, we wanted to assess if the rat model of MTX-induced intestinal atrophy, which has been described to share both biochemical and histological features with atrophic celiac intestinal mucosa [8, 9, 22], could be used as an experimental model for the study of the role of the endocannabinoid system in this impairing disorder. We report that, indeed, significantly elevated anandamide levels are found in bioptic samples from the duodenum of patients with active celiac disease as compared to the levels found in non-celiac individuals. This selective increase in anandamide levels is similar to what was found previously in biopsies from patients with untreated ulcerative colitis [15] or diverticular disease [23] and was accompanied by the upregulation of anandamide’s most important cannabinoid receptor target, the CB1 receptor (anandamide is almost functionally inactive at cannabinoid CB2 receptors [11]) in the subepithelial region of the duodenum. This suggests that anandamide is the endocannabinoid that is most involved in human inflammatory conditions of the gastrointestinal tract. However, in celiac patients, we also observed a strong increase in 2-AG levels, which could have become statistically significant if more biopsies had been analyzed or if lesser variability among samples had been found. Therefore, also 2-AG might play a role in this particular human inflammatory disorder. At any rate, the present data represent the first example of a human gastrointestinal disorder where it has been possible to correlate directly the levels of an endocannabinoid with the occurrence of the most active manifestation of the inflammatory condition. In fact, we have shown here that anandamide levels in the duodenum of celiac patients peak in correspondence with the atrophic phase and then recede to normal levels in patients after remission following a gluten-free diet. This finding provides evidence that the levels of anandamide, and hence the activity of its most important molecular targets, the cannabinoid CB1 receptors, are strictly related to active inflammation. This phenomenon might represent an adaptive response to counteract the consequences of inflammation, for example via the reduction in the Th1-mediated immune response. Indeed, adaptive mechanisms aimed at counteracting the gluten-dependent inflammation, in which regulatory T cells inhibiting the proliferation of gluten-reactive Th1 cells are found in celiac intestinal mucosa, have been recently reported [24], and activation of CB1 receptors can also suppress the Th1 response [25]. Upon remission from atrophy, this enhanced endocannabinoid tone might not be needed anymore, and anandamide levels would return to normal. However, CB1 receptors were found here to be present in subepithelial nervous fibers, and at least in the human colon, are expressed also in epithelial cells, subepithelial plasma cells, and smooth muscle cells [21]. Therefore, anandamide, by acting at these receptors, might re-establish intestinal homeostasis after inflammation in several additional ways, i.e., by (1) reducing the output of cholinergic myenteric neurons thereby reducing hypersecretion and hypermotility and (2) directly contributing to epithelial wound healing [14, 21, 26]. Indeed, it was shown that celiac patients display marked alterations of gastrointestinal motility, and it is possible that the endocannabinoid system plays a role by inhibiting also intestinal peristalsis [26] (see [27] for review). Evidence for a cause–effect relationship between endocannabinoid overactivity and recovery from inflammation has been provided previously in mice by showing that agents that inhibit endocannabinoid (and particularly anandamide) inactivation, thereby enhancing colon anandamide levels, can induce a complete recovery from dinitro-benzenesulfonic acid-induced colitis [15].

We have also shown here that endocannabinoid levels are elevated also in the jejunum of MTX-treated rats, again in an atrophy-dependent way. In this case, we could determine that both mucosa and muscle/serosa layers exhibit elevated levels of endocannabinoids in correspondence with the strongest histopathological damage, i.e., 3 days from MTX injection, although with some differences between the two tissues (e.g., the levels of anandamide were increased with respect to controls only in the muscle/serosa, although they did peak at 3 days also in mucosa). This observation allows us to speculate that putative protective effects of endocannabinoids during this pathological condition are exerted both at the muscle level, perhaps again by reducing the hyperactivity of cholinergic myenteric neurons, and at the epithelial level, perhaps contributing to wound healing. Indeed, the presence of CB1 receptors in the myenteric plexus of the rat small intestine has been described previously [28], although no evidence exists to date for the presence and regulation of these receptors in rat intestinal epithelial cells and after inflammation. However, CB1 receptors were shown to be upregulated in mice during experimental inflammatory conditions, in both the small intestine and colon, and at the level of both myenteric neurons and endothelial cells [16, 29, 30]. If one keeps into account that slight differences between our results in celiac patients and MTX-treated rats might be due to species differences and to the different types of biological sample analyzed (only in rats, we could distinguish between mucosa and muscle/serosa tissues), these findings allow us to propose that the MTX-treated rat might be a good experimental model for a future more detailed study of the role of the endocannabinoid system in celiac disease. For example, it would be of interest to use MTX for induction of atrophy in celiac disease-associated, HLA-DQ2 or -DQ8, transgenic mice [19] and to investigate the effect of endocannabinoids in this model. It was not possible in this study to investigate the cellular source of elevated endocannabinoid levels in biopsies of celiac patients or in jejunal samples from MTX-treated rats, as the sensitivity of the technique used for quantitative measurements requires no less than 5–10 mg of wet tissue to detect 2-AG and, particularly, anandamide. Future studies will have to be carried out to investigate the distribution in the human and rodent small intestine of the recently cloned endocannabinoid biosynthesizing enzymes [14] by the use of immunohistochemical or in situ hybridization techniques.

We have presented here also data indicating that the levels of the cannabinoid receptor–inactive anandamide congener, PEA, which is known to be endowed with strong anti-inflammatory actions [31] possibly mediated by either vanilloid TRPV1 receptors or peroxisome proliferator-activated receptor α [32], are also elevated in both the duodenum of celiac patients and the jejunum of MTX-treated rat. Earlier studies have shown that PEA can inhibit small intestine motility [33] and that its intestinal levels are also increased in patients with untreated ulcerative colitis [34]. In this study, we have found that the levels of this compound return to normal after recovery from atrophy only in MTX-treated rats, and therefore, we cannot conclude, as in the case of anandamide, that PEA actively participates in intestinal inflammation in celiac disease.

In conclusion, we have reported here that the levels of the endocannabinoids are elevated in the duodenum of celiac patients and in the jejunum of MTX-treated rats in correspondence of the atrophic phase induced by these conditions, to return to normal after atrophy remission. Our data support the hypothesis that endocannabinoid overactivity represents an adaptive reaction to intestinal inflammation aimed at counteracting its consequences [14, 27] and suggest that also PEA might play a role in this context.