Cannabidiol and Palmitoylethanolamide are anti-inflammatory in the acutely inflamed human colon

DG Couch, JN Lund and S O’Sullivan were responsible for overall content of the article. A C C E P T E D M A N U S C R IP T 10.1042/CS20171288 . Please cite using the DOI 10.1042/CS20171288 http://dx.doi.org/ up-to-date version is available at encouraged to use the Version of Record that, when published, will replace this version. The most this is an Accepted Manuscript, not the final Version of Record. You are : Clinical Science


INTRODUCTION
In health, the gut absorbs nutrients from the lumenal environment into the sterile submucosa without absorbing noxious material, such as bacteria and lipopolysaccharide. The barrier between lumen and submucosa is formed by epithelial cells, which allow selective absorption of particles into the enteric circulation via paracellular and transcellular pathways, whilst preventing bacterial translocation (1-5). Inflammation causes this barrier to become compromised (6). Inflammation is caused by conditions such as diverticulitis, infective colitis and appendicitis, and also inflammatory bowel disease (IBD), such as ulcerative colitis and Crohn's disease (7). When inflammation occurs, lumenal bacteria and lipopolysaccharide are able to translocate into the submucosal space and beyond, resulting in secondary complications such as endotoxaemia, sepsis and death (there are 30,000 deaths from sepsis per year in the UK alone) (8,9). Currently, there are no clinical treatments to counter permeability changes seen in systemic inflammation and sepsis. Development of an agent to prevent bacterial and lipopolysaccharide translocation across the gut wall with the intention of reducing or preventing the triggering of sepsis is therefore of high clinical importance.
We have previously demonstrated, in Caco-2 culture models, that inflammation stimulated by tumour necrosis factor alpha (TNFα) and interferon gamma (IFNγ) increases epithelial permeability, shown by falls in trans-epithelial electrical resistance (TEER) (10). The nonpyschotrophic constituent of cannabis sativa, cannabidiol (CBD), and the endogenous fatty acid amide palmitoylethanolamide (PEA) prevented these changes when used prophylactically, and restored membrane resistance when given therapeutically, acting via the CB1 and PPARα receptors respectively (10,11). Others have also observed a protective effect of PEA and CBD on the gut barrier during inflammation (12,13).
What is not clear is the mechanism by which PEA and CBD modify permeability during inflammation. Recently we showed that PEA may act through modification of the actin cytoskeleton by inducing FAK production and down regulating aquaporins 3 and 4, but it was not clear if this was secondary to an anti-inflammatory effect of PEA or due to direct action of PEA (14). PEA and CBD exert an anti-inflammatory effect on enteric glial cells, which can modify the immune response in vivo (15)(16)(17)(18), therefore any in vivo permeability effects may indeed be secondary to an anti-inflammatory effect rather than direct action on cellular structures which contribute to regulation of permeability.
Previous studies have shown both PEA and CBD to blunt increases in permeability in experimentally inflamed explant colonic tissue (17)(18)(19). As the extent of the antiinflammatory effect of PEA and CBD in vitro and in vivo has not yet been quantified, we examined the effect of PEA and CBD on the local inflammatory response in cultured Caco-2 cells and explant human colonic tissue. We hypothesised that CBD and PEA cause changes in intestinal permeability through suppression of the local immune response. We therefore examined the effect of CBD and PEA on the intracellular signalling phosphoproteins in response to inflammation, and on down-stream production of inflammatory cytokines. This allowed us to compare the effects of PEA and CBD on the epithelium alone, to their effect on whole tissue. In order to assess the effect of CBD and PEA on clinically inflamed colonic tissue we examined the effects of these drugs on explant colon from patients with established inflammation caused by acute appendicitis and inflammatory bowel disease.

MATERIALS AND METHODS
All experiments and procedures received prior approval of the University of Nottingham Ethics Committee and local NHS Research Ethics Committee.

Caco-2 Cell Culture
Caco-2 cells were purchased from European Collection of Cell Culture (Wiltshire, UK; passages 21-42). Cells were cultured in Eagle's minimum essential medium supplemented with L-glutamine, 10% foetal bovine serum (FBS) 1% penicillin/streptomycin and 1% nonessential amino acids mixture (all Sigma-Aldrich). Cells were kept at 37°C in 5% CO2 and 95% humidity. Cells were seeded at 4x10 5 cells per well in polystyrene 12 well plates (Corning Incorporated, USA), and grown for 2 weeks until fully differentiated. Cells were used for experimentation at day 14-16. The medium was changed on alternate days.
Randomly assigned wells (n=8) were treated with the following drug treatments: vehicle, an inflammatory protocol of IFNγ (10ng.ml -1 , Sigma-Aldridge) for 8 hours, followed by TNFα (10ng.ml -1 , Sigma-Aldridge) for 16 hours, inflammation and PEA (10µM, PEA was added simultaneously with IFNγ at the start of the 24 hour inflammatory period), inflammation and CBD (10µM, CBD was added simultaneously with IFNγ at the start of the 24 hour inflammatory period), PEA (10µM) alone, or CBD (10µM) alone. PEA and CBD were purchased from Tocris Bioscience (Bristol, UK). At the end of the 24 hour experimental period, media was collected and stored at -80 °C until analysis. Cells were washed twice with ice-cold phosphate-buffered saline (PBS), and treated with radioimmunoprecipitation assay (RIPA) buffer supplemented with phosphatase and protease inhibitors (Sigma-Aldridge) at 4°C for one hour on a rocking platform to cause cell lysis. Cell lysates were then collected and stored at -80 °C until analysis.

Human Colon Experimentation
Experiments on ex vivo human tissue were performed by obtaining colonic samples from patients having elective bowel resections for bowel cancer (n=13), planned resections for quiescent inflammatory bowel disease (n=6) or emergency appendectomies for acute appendicitis (n=6) at Derby Teaching Hospitals NHS Trust, Derbyshire, UK. Samples of normal colon at least 10cm proximal to right sided bowel tumours (in the case of bowel cancer resections), sections of inflamed colon (in inflammatory bowel disease resection), or sections of inflamed appendix (in the case of emergency appendicectomy) were obtained immediately after resection in the operating theatre. Sections of tissue 2cm x 2cm were removed from the resected specimen and transferred on ice to the laboratory within ten minutes, in pre-chilled Eagle's minimum essential medium supplemented with 1% FBS 1% penicillin/streptomycin and 1% non-essential amino acids mixture (Sigma-Aldrich). The remaining operative specimen was sent to pathology for routine analysis. Once in the laboratory samples were pinned on Stylgard plates. Mucosa with submucosa was dissected free from the underlying muscularis layer. Mucosal samples were then further dissected into 2mm x 2mm sections and placed in individual wells of 24-well polystyrene plates (Corning Incorporated, USA), each containing 1ml of media. Samples of colonic tissue were then treated with TNFα and IFNγ, in the absence or presence of PEA and CBD as described above within Caco-2 experiments.
Experiments were carried out in triplicate, with final values derived from the mean result of three measurements. Samples were incubated for 24 hours at 37°C in 5% CO2 and 95% humidity. At the end of the 24 hour experimental period media was collected and stored at -80 °C until analysis. Explant tissue was washed with ice cold PBS and stored frozen at -80 °C until homogenisation and analysis. Prior to analysis colonic samples were thawed on ice and cryohomogenised using the method described by von Ziegler (20). Collected homogenates were then dissolved in 215µl of RIPA buffer, incubated on an oscillating thermomixer for 30 minutes at 60 °C, then centrifuged at 10,000G for 15 minutes. Supernatant was collected, vortexed for 20 seconds and then analysed.

Cytokine production
We measured specific proteins induced as a consequence of TNFα stimulation. To quantify the effects of PEA and CBD on the inflammatory response we measured media concentrations from cell or colonic cultures of seven cytokines at the end of the 24 hour experimental period using ELISA. Cytokines measured were interleukin-8 (IL-8), monocyte chemoattractant protein-1 (MCP-1), intercellular adhesion molecule-1 (ICAM-1), matrix metallopeptidase-3 (MMP-3); DUOSET ELISA kits R&D Systems Minneapolis, US. Interleukin-17 (IL-17), granulocyte-macrophage-colony stimulating factor (GM-CSF), interleukin-6 (IL-6); ready-setgo ELISA kits, Affymetrix eBioscience, San Diego, CA. Cytokine concentrations were normalised for protein content as previously using BCA assay determination of cell lysate protein concentration against a standard curve. Experimental conditions were averaged from triplicate readings, as above.

Target sites of action of PEA & CBD
To identify target sites of action of PEA and CBD we co-applied the following antagonists to

Effects of PEA and CBD on cytokine production in response to inflammation
Stimulation of Caco-2 cells with IFNγ and TNFα caused an increase in the secretion of IL-8 and IL-6 although did not increase in the secretion of IL-17 (figure 3). In human colonic tissue IFNγ and TNFα caused and increase in the production of IL-8, IL-6 and IL-17 ( figure   3). Both PEA and CBD did not affect the production of these cytokines in Caco-2 cultures, but did prevent their increased in human colonic explants.
Stimulation of Caco-2 cells with the inflammatory protocol did not increase the production of GM-CSF, but both PEA alone and CBD alone increased GM-CSF production in the absence of inflammation (figure 4). In human tissue IFNγ and TNFα stimulation increased the production of GM-CSF which was prevented by CBD but not PEA (figure 4).
IFNγ and TNFα markedly increased MCP-1 production by Caco-2 cells and human colonic tissue (figure 4). CBD and PEA had no effect on Caco-2 production of MCP-1, but did significantly reduce production in human tissue.
Stimulation of Caco-2 cells with IFNγ and TNFα increased the production of ICAM-1 (figure 5, A). CBD and PEA did not prevent this increase in production, but PEA alone caused a marked increase in ICAM-1 production compared to vehicle. In human colonic tissue, an increase in ICAM-1 caused by IFNγ and TNFα was prevented by the administration of PEA (figure 5, C). Treatment of inflammation-stimulated colonic tissue with CBD also decreased production, though did not reach significance. PEA alone, CBD in the presence of inflammation and CBD alone had no effect on ICAM-1 production in colonic tissue compared to vehicle.
Stimulation of Caco-2 cells with IFNγ and TNFα had no effect on the production of the enzyme MMP-3 ( figure 5, B). In human tissue the inflammatory protocol did not significantly increase the production of MMP-3, however both PEA and CBD did significantly reduce its production in the presence of IFNγ and TNFα ( figure 5, D).

Antagonist studies
When investigating for a receptor mechanism for PEA and CBD we found again that stimulation of colonic explant tissue with the inflammatory protocol caused a significant rise in IL-8, IL-6 and MCP-1 production compared to baseline, whilst treatment of inflamed colon with simultaneous PEA or CBD prevented these increases in cytokine production (figure 6, A to F). The anti-inflammatory effects of PEA on IL-8, IL-6 and MCP-1 production were prevented by the addition of the PPARα antagonist GW6471. The antiinflammatory effects of CBD on IL-8, IL-6 and MCP-1 production were prevented by the CB2 antagonist AM630. The anti-inflammatory effects of CBD on IL-8 and MCP-1 production were also inhibited by the addition of the TRPV1 antagonist SB366791. SB366791 had no effect on the anti-inflammatory effect of CBD in the presence of inflammation. GW6471, AM630 and SB366791 had no effect on cytokine production in the presence of IFNγ and TNFα alone (data not shown).

Effects of PEA and CBD on cytokine production in response to inflammation in IBD colonic explants
Because we found PEA and CBD had an anti-inflammatory effect on experimentally inflamed tissue we collected samples of inflamed colon from patients with IBD and acute appendicitis.

DISCUSSION
The aim of this study was to examine the anti-inflammatory properties of PEA and CBD in Caco-2 cell lines and explant human colonic tissue. We demonstrate that under inflammatory conditions, PEA and CBD supress the phosphorylation of several intracellular proteins in Caco-2 cells, however this does not supress the secretion of pro-inflammatory cytokines.
Conversely, in explant human colonic tissue stimulated with IFNγ and TNFα, PEA and CBD both supressed the phosphorylation of intracellular proteins which were up-regulated by inflammation, and also prevented the increased secretion of pro-inflammatory cytokines.
Additionally we have shown that PEA and CBD have an anti-inflammatory effect in explant IBD and appendicitis tissue.

Caco-2 cell cultures
We previously demonstrated that CBD and PEA prevent changes in the permeability of Caco-2 monolayers under inflammatory conditions. Prior to this study we had hypothesized that this effect on permeability was secondary to a local anti-inflammatory action, as it had been seen in work from other centres that CBD and PEA supress the inflammatory response in the inflamed colon of mice and humans (17,18,(21)(22)(23). For this hypothesis to be correct we would expect PEA and CBD to have an anti-inflammatory action in Caco-2 monolayers.
IFNγ and TNFα, caused a significant increase in all measured intracellular signalling proteins, in line with similar experiments using TNFα stimulation on Caco-2 cell lines (24).
We found that increases were suppressed by PEA and CBD. It is possible that both of these effects could be due to increased anandamide (AEA) tone. PEA has been shown to increase the action of local AEA either by preventing hydrolysis of AEA through substrate competition or FAAH inhibition (25), or by enhancing AEA potency at villanoid receptors (26). Secondly CBD has been shown to prevent AEA uptake and catabolism (27). AEA itself has been shown to down regulate NF-KB and exert anti-inflammatory properties through IL-10, and may therefore been at least partly responsible for these effects on signal phosphorylation (28). We then quantified the effect of PEA and CBD on the extracellular inflammatory response, measuring seven pro-inflammatory cytokines representing five aspects of immune activation. As an indicator of pro-inflammatory cytokine production we assayed IL-8, IL-6, and IL-1 (29)(30)(31). As a marker of leucocyte recruitment and activation we assayed MCP-1 and GM-CSF (32,33). As a marker of extracellular matrix remodelling we measured MMP-3 concentrations (34), and as a marker of cell-to-cell adhesion we measured concentrations of ICAM-1. We found that stimulation of Caco-2 cultures with IFNγ and TNFα caused an increase in all measured cytokines, except MMP-3. Surprisingly, neither PEA nor CBD prevented the increased secretion of these cytokines. We may suggest therefore that the effects of PEA and CBD on signal phosphorylation and permeability are distinct from their effects on inflammation, and not due to suppression of a local extracellular immune response.

Ex vivo Colonic Tissue
In explant human tissue we found that the local inflammatory response caused by stimulation with IFNγ and TNFα was inhibited by treatment with PEA and CBD. This finding, in contrast to results in cultured Caco-2 cells, could be explained by the presence of submucosal immunocytes such as dendritic cells and macrophage activity. Explant colonic samples contain lymphoid aggregates and innate immune cells, as opposed to monolayers of Caco-2 cells (35). It has previously been described that within the gut the CB2 receptor, AEA and 2-AG are found in highest abundance within these submucosal immune and also nervous tissues such as enteric glial cells, and that this may be the primary site of endocannabinoid activation during inflammatory episodes (12,36). Within our colonic explant tissue PEA and CBD may be acting on such macrophage and dendritic cell colonies, thus supressing the local inflammatory response (35,37). This would explain why cytokine secretion was decreased in the presence of IFNγ and TNFα, but no decrease was found in tissue treated with PEA or CBD alone, compared to vehicle treated tissue. Further work therefore could examine the effect of CBD and PEA on Caco-2 cells in a co-culture model exploring possible receptor mechanisms in these specialised tissues. in both dextran-sulphate sodium treated mice and colonic biopsies from patients with ulcerative colitis were inhibited by blockade of the PPARα receptor (18). As mentioned above, it has been suggested that PEA may exert its effects through increasing the local concentration or potency of a second agent, such as anandamide, and therefore antagonising receptor targets of PEA or the second agent would inhibit their biologic effects (40). In view of this we examined the anti-inflammatory effect of PEA in the presence of six receptor antagonists at which cannabinoid agonists have been postulated to act; CB1, CB2, PPARα, PPARγ, TRPV1 and GPR55. We found that across three measured cytokines the effect of PEA was prevented by antagonism of PPARα. This is the second human colonic study to demonstrate PEA action at PPARα, suggesting that this PEAs primary site of action in colonic mucosa (14).
Similarly, multiple studies have suggested various sites of action of CBD in the mammalian gut. Two murine studies demonstrating the beneficial effect of CBD on inflammationinduced effects on gut transport showed that CB1 rather than CB2 was the target receptor of CBD (41,42). However two independent studies examining the effect of CBD on the immune response in human and murine colonic tissue suggested that PPARα was the dominant receptor target of CBD, with a possible role for CB1, but again suggesting that CB2 was not a receptor target (17,43). However our data show a role for CB2 and TRPV1. It is possible than these data differ from pre-existing literature because of site of colonic sampling and mode of inflammation used for simulating colitis. It has been demonstrated that the distribution of endocannabinoid receptors and differs across the colon, and that these receptors are activated by inflammation. Within our study we collected colonic samples from right sided colonic resections, whereas mouse-colonic studies previously cited have used whole colonic homogenates, and the pre-existing human colon studies have used left-sided (sigmoid) colonic biopsies. It is possible therefore the sites of CBD activation differ throughout the gut, hence in the proximal gastrointestinal tract PPARα is the pre-dominant site of activation, whereas distally CB2, TRPV1 and then CB1 become more important. This may be supported by evidence showing that in established inflammation of the gut and other organ systems CBD has epithelial protective effects medicated by CB2 and TRPV1 (44,45).
One study in particular found that during a murine model of colitis-induced sepsis, CBD prevented peripheral organ oxidation, further demonstrating the effect of CBD on epithelial barriers during sepsis (46). Further work could compare the receptor targets of CBD across the gut, using a similar model of inflammation.
We then compared the anti-inflammatory effect of PEA and CBD in experimentally inflamed normal colonic tissue, explant IBD tissue, and explant appendicitis. We found that although appendicitis tissue had higher baseline levels of cytokine production, similar increases in cytokine secretion was caused by IFNγ and TNFα treatment in all three tissue types. These differences in baseline cytokine production are likely to represent the acute inflammatory nature of acute appendicitis, versus the chronic low-levels of inflammation in long-standing IBD, compared with healthy tissue. CBD and PEA were strongly anti-inflammatory in acute appendicitis tissue, but not paralleled in IBD tissue. We also did not find that PEA and CBD were effective in preventing increased cytokine production in cytokine-treated IBD and appendicitis tissue. This may suggest that the receptor profile in acute inflammation differs from that in long-standing chronic inflammation, and therefore any benefit seen in IBD may not be secondary to an anti-inflammatory effect, and may be secondary to effects on mucosal permeability. This is supported by a clinical study from Naftali et al (2013) who showed an improvement in disease activity scores and quality of life in IBD patients receiving cannabis sativa cigarettes, though did not find any change in biochemical markers of inflammation using serum CRP levels as a marker (13). Furthermore a study from Di Sabatino (2011) conducted in a similar explant manner found differences in the expression of endocannabinoid ligands between control and inflamed IBD explants, the inflammatory response of which was down regulated with the addition of methanandamide (47). A recent study from the same centre used low dose CBD in inflammatory bowel disease, though did not find any benefit in improving quality of life scores (48). Before drawing conclusions however regarding the efficacy of CBD in IBD it is important to highlight that this study may have been hampered by small group sizes and the ultra-low doses of CBD employed.
This study provides further evidence that PEA and CBD may play a role in the treatment of acute inflammation of the gut, including Crohn's and ulcerative colitis. Our data are limited by using explant tissue, rather than any clinical in vivo data, and the generic nature of the explant models used. Furthermore we did not carry out tissue viability assays on explant tissue to ensure the mucosa had not been damaged by dissection, or had become necrotic.
Further clinical work examining the use of these two drugs in the treating inflammatory disease of the gut should now be considered. Additionally we have hypothesized that the effects of CBD and PEA on signal phosphorylation and the inflammatory response may be due to increased efficacy or presence of AEA. Further work should now be conducted within 21 both Caco-2 models and healthy colon, appendicitis and inflammatory bowel disease explants to quantify the effect of CBD and PEA on endocannabinoid production. Lastly within this study we did not examine for any additive effects of CBD together with PEA on the immune response, which may have been positive.
In summary we have demonstrated that CBD and PEA prevent cytokine production in human colonic explant tissue via PPARα, CB2 and TRPV1, but not in cultured epithelial cells.
These effects extend into chronic inflammatory processes such as IBD, but also acute inflammatory conditions such as appendicitis. Further clinical work must examine the effects of these two drugs at higher doses, and clarify their clinical role. The effects of PEA and CBD on the intracellular levels of phosphorylated nuclear signalling proteins in response to an inflammatory protocol in cultured Caco-2 monolayer, measured by multiplex. Data is presented as percentage change from vehicle per plate +/-SEM, n=8 per condition. Data was analysed by one-way ANOVA comparing against the vehicle control or inflammation (*<0.05, **<0.01 and ***<0.001).

Figure 2
The effects of PEA and CBD on the intracellular levels of phosphorylated signalling proteins in response to 24hr exposure to TNFα and IFNγ in cultured human colonic explants, measured by multiplex. Data is presented as percentage change from vehicle +/-SEM, n=13 per condition. Data was analysed by repeated measures ANOVA comparing against the vehicle control (*), *<0.05, **<0.01.

Figure 3
The effects of PEA and CBD on the secreted cytokine response to an inflammatory protocol in cultured Caco-2 monolayers (A, C and E, column 1, n=8, percentage change from vehicle, one way ANOVA) and human colonic tissue (B, D and F, column 2, n=7, percentage change compared to vehicle, repeated measures ANOVA), measured by ELISA. Error bars represent +/-SEM per condition. Asterixes (*) represent significant difference from vehicle, *<0.05, **<0.01, ***<0.001.

Figure 6
The effects of PEA (A,C,E) and CBD (B,D,F) on the secretion of IL-8, MCP-1 and IL-6 in response to an inflammatory protocol in explant human colonic tissue in the presence of receptor antagonists, measured by ELISA (compared by repeated measures ANOVA, n=7). Data presented as mean +/-SEM per condition. Asterixes (*) represent significant difference from vehicle, *<0.05, **<0.01.