Oleoylethanolamine and palmitoylethanolamine modulate intestinal permeability in vitro via TRPV1 and PPARα

Cannabinoids modulate intestinal permeability through cannabinoid receptor 1 (CB1). The endocannabinoid‐like compounds oleoylethanolamine (OEA) and palmitoylethanolamine (PEA) play an important role in digestive regulation, and we hypothesized they would also modulate intestinal permeability. Transepithelial electrical resistance (TEER) was measured in human Caco‐2 cells to assess permeability after application of OEA and PEA and relevant antagonists. Cells treated with OEA and PEA were stained for cytoskeletal F‐actin changes and lysed for immunoassay. OEA and PEA were measured by liquid chromatography‐tandem mass spectrometry. OEA (applied apically, logEC50 25.4) and PEA (basolaterally, logEC50 24.9; apically logEC50 25.3) increased Caco‐2 resistance by 20–30% via transient receptor potential vanilloid (TRPV)‐1 and peroxisome proliferator‐activated receptor (PPAR)‐a. Preventing their degradation (by inhibiting fatty acid amide hydrolase) enhanced the effects of OEA and PEA. OEA and PEA induced cytoskeletal changes and activated focal adhesion kinase and ERKs 1/2, and decreased Src kinases and aquaporins 3 and 4. In Caco‐2 cells treated with IFNg and TNFa, OEA (via TRPV1) and PEA (via PPARα) prevented or reversed the cytokine‐induced increased permeability compared to vehicle (0.1% ethanol). PEA (basolateral) also reversed increased permeability when added 48 or 72 h after cytokines (P <0.001, via PPARα). Cellular and secreted levels of OEA and PEA (P <0.001–0.001) were increased in response to inflammatory mediators. OEA and PEA have endogenous roles and potential therapeutic applications in conditions of intestinal hyperpermeability and inflammation.—Karwad, M. A., Macpherson, T., Wang, B., Theophilidou, E., Sarmad, S., Barrett, D. A., Larvin, M., Wright, K. L., Lund, J. N., O'Sullivan, S. E. Oleoylethanolamine and palmitoylethanolamine modulate intestinal permeability in vitro via TRPV1 and PPARα. FASEB J. 31, 469–481(2017). www.fasebj.org

The human gastrointestinal (GI) tract forms the largest interface between the external environment and internal milieu (1). Aside from its digestive functions, it constitutes the most complex and evolved element of immune defense. Intestinal epithelial cells, together with their mucous coatings, constitute a protective barrier, across which paracellular permeation is selectively regulated by transmembrane protein contractility within the intercellular tight junctions (2,3), thus preventing the loss of water and solutes from the gut, while simultaneously permitting the absorption of water and nutrients, but preventing the ingress of  (4)(5)(6). Impaired intestinal barrier function leading to hyperpermeability is associated with a wide variety of human diseases and conditions, for example acutely in shock and multiple organ-system dysfunction with splanchnic ischemia (3), sepsis (2), or, more gradually, including inflammatory bowel disease (7)(8)(9)(10), celiac disease (11), irritable bowel syndrome (12,13), and a range of other conditions (14,15). Family studies have demonstrated that increased intestinal permeability can precede the clinical presentation of inflammatory bowel disease (16)(17)(18). The regulation of intestinal permeability is poorly understood, and improved understanding is necessary for the development of therapeutic interventions specifically targeted at restoring normal permeability (19).
Cannabis sativa plant extracts have been used anecdotally for over 5 millennia for the treatment of GI disorders including nausea, vomiting, anorexia, intestinal inflammation, and diarrhea (20). Endocannabinoids are intercellular lipid signaling molecules derived from arachidonic acid and synthesized on demand from cell membrane precursors. Examples were found to be expressed in the gut only 20 yr ago (21), and subsequently endocannabinoids and their receptors were shown to be key regulators of a variety of GI functions, including emesis (22), intestinal motility (23), and secretion (24). Endocannabinoids play significant roles in inflammation and apoptosis (25,26) and specifically in intestinal inflammation (27), opening up the possibility of new therapeutic options (28).
Endocannabinoids exert their effects by activation of cannabinoid receptor 1 and 2 (CB 1 and CB 2 ) (29) and other target sites of action, such as transient receptor potential ion channels (TRPs) (30), peroxisome proliferator-activated receptors (PPARs) (31), and the orphan GPCRs GPR119 (32) and GPR55 (33). All of these target sites are expressed in the GI tract. These receptors, together with endocannabinoid ligands and the enzymes responsible for their metabolism, are collectively referred to as the endocannabinoid system (ECS). The ECS is involved in modulating GI motility and intestinal inflammation and is up-regulated in intestinal inflammation. Our group has reported that cannabinoids modulate intestinal permeability in vitro in Caco-2 intestinal cells (34,35), which has also been shown in vivo (36). The plant-derived cannabinoids D 9 tetrahydrocannabinol and cannabidiol reverse the increased permeability caused by EDTA or cytokines via CB 1 activation (34,35). By contrast, the endocannabinoids anandamide (AEA) and 2-arachidonoylglycerol (2-AG) increase permeability of the Caco-2 monolayer via CB 1 (34,35), and inhibiting their synthesis improves the effects of inflammation on permeability, suggesting that the endogenous production of these compounds in response to inflammation plays a role in promoting permeability changes at the epithelium.
Oleoylethanolamine (OEA) is an endocannabinoid-like compound that does not bind to cannabinoid receptors (37). It is produced on demand in enterocytes, and its production is stimulated by food intake (38) or reduced by food deprivation (39). It is a PPARa agonist (40) that activates TRPV1 channels (41), and the orphan GPCRs GPR55 and GPR119 (42). Administration of OEA suppresses food intake, inhibits body weight gain (43), and induces satiety via PPARa activation (44). It also has a role in regulating lipid metabolism (45) and reduces cholesterol levels in mice via PPARa (40,46). Palmitoylethanolamine (PEA) is another endocannabinoid-like compound found at high levels in the upper GI tract compared with other organs and tissues (39). It reduces intestinal injury and inflammation in mice via PPARa (47,48). More recently, oral or intraperitoneal administration of PEA has been found to reduce inflammation and damage in dinitrobenzene sulfonic acid-induced colitis in mice, mediated by PPARa, CB 2 , and GPR55 (49), and inhibition of the enzyme responsible for PEA degradation also reduces inflammation in 2 mouse models of colitis (50).
In the present study, we hypothesized that OEA and PEA, which often have opposing physiologic actions and pharmacology that differ from that of AEA and 2-AG, may also modulate intestinal permeability and play a role in intestinal inflammation. We hypothesized that these compounds would have a beneficial effect on intestinal permeability based on their positive effects in vivo in simulated inflammation.

Effects of OEA and PEA on Caco-2 monolayer permeability
The cells were seeded at 20,000 cells on 6.4 mm diameter, 0.4 mm pore size polyethylene terephthalate inserts (BD Falcon Biosciences, Oxford, United Kingdom) and grown for 14-18 d. Transepithelial electrical resistance (TEER) was measured with an epithelial volt-ohm meter (EVOM 2 ; World Precision Instruments, Sarasota, FL, USA) as an indicator of cellular permeability. Caco-2 cell monolayers with TEER value .1000 V · cm 2 were used. Caco-2 cell monolayers were washed twice in HBSS (+N-2hydroxyethylpintestinal permeabilityerazine-N9-2-ethanesulfonic acid, or HEPES and penicillin/streptomycin), and the baseline TEER was measured. Increasing concentrations of OEA or PEA (1 nM-10 mM) or vehicle (0.1% ethanol) were applied in prewarmed EMEM to the apical or basolateral compartment of inserts, and TEER was measured over the next 48 h.
The following target sites of action were investigated (receptor antagonist and concentrations shown in brackets): To simulate inflammatory conditions, 10 ng/ml IFNg was added basolaterally. After 8 h, 10 ng/ml TNFa was added for another 16 h. OEA and PEA were added to the apical or basolateral compartment at various time points, either at the same time as IFNg (time 0 h, to potentially block the development of inflammation) or after the induction of inflammation or at 24, 48, or 72 h, to potentially limit the inflammatory increase in permeability). In some experiments, this protocol was performed in the presence of antagonists. For prolonged (chronic) inflammatory studies, repeated applications of 3 ng/ml of IFNg and TNFa were used.

Cell viability assays
To test the effects of OEA and PEA on cell viability in fully differentiated Caco-2 cells, the cells were brought to confluence and maintained in complete medium for up to 18 d in 96-well plates. A concentration response to OEA and PEA was then performed in complete medium over 48 h, after which PrestoBlue Reagent (Thermo Fisher Scientific, Waltham, MA, USA) was added directly to the cell culture (1:10). After 10 min, absorbance was measured with excitation at 570 nm with 600 nm as the reference wavelength for normalization. Data were calculated as the mean percentage change from untreated control.
To test the effects of OEA and PEA on cell viability in proliferating Caco-2 cells, we seeded 5 3 10 3 cells in quadruplicate into a 96-well microplate in standard medium with 8% serum. A series dilution of OEA and PEA from 10 mM was performed across the plate. Cells were incubated for 72 h. Growth medium was carefully removed, and 50 ml of 13 CyQuantNF dye reagent (Thermo Fisher Scientific) was added to each well. The microplate was incubated at 37°C for 30 min, and the fluorescence intensity of each sample was measured with a fluorescence microplate reader (Tecan, Männedorf, Switzerland) with excitation at ;485 nm and emission detection at ;530 nm. Data were calculated as the mean of the percentage of untreated control.

Phalloidin staining
Phalloidin is an F-actin stain that allows for the visualization of the structure and inferred function of this cytoskeletal filament protein. Cells are fixed first so the image is a snapshot of the given timepoint. Linear actin fibers (mostly parallel) are the polymerized F-actin in the cytoplasm and can be quite pronounced at the cell boundaries or at focal adhesion plaques. Changes in the network can be a result of disruption or required changes for movement and changes to adhesion. For this experiment, cells were grown on 8-well chamber slides (BD Bioscience) for 16-18 d and fixed in 4% paraformaldehyde for 45 min at room temperature after the various treatments. The cells were then washed in PBS before permeabilization with 0.5% Triton X-100, blocked in 5% bovine serum albumin in PBS for 20 min at room temperature, and incubated with phalloidin tetramethylrhodamine B isothiocyanate (TRITC)-conjugated solution 5 U/ml (a kind donation from Dr. Alan Shirras, Faculty of Health and Medicine, Lancaster University) for 20 min at room temperature. The cells were washed 3 times with PBS and mounted with Vectashield Mounting Medium containing DAPI (Vector Laboratories, Ltd., Peterborough, United Kingdom). Stained cells were viewed on a confocal microscope (Zeiss, Jena, Germany) at 363 magnification.

Immunoassays
For all immunoassay experiments, Caco-2 ells were grown in 6or 12-well culture plates and treated with OEA and PEA at the apical membrane. After treatment protocols, the cells were washed in PBS and lysed in RIPA buffer [150 mM NaCl, 1.0% IGEPAL CA-630, 0.5% sodium deoxycholate, 0.1% SDS, and 50 mM Tris (pH 8.0)] with protease inhibitor cocktail (P8340; Sigma-Aldrich, Poole, United Kingdom). Protein content of lysates was determined with Bradford reagent.

Potassium channel activation
To test the ability of OEA and PEA to modulate potassium channels in Caco-2 cells, the FluxOR Potassium ion channel assay (Thermo Fisher Scientific) was used. In brief, Caco-2 cells were grown on 96-well plates until fully confluent and differentiated. The cells were loaded with the nonfluorescent, thallium-specific FluxOR dye and treated apically with increasing concentrations of OEA or PEA. When potassium channels are stimulated, thallium flows into the cell and binds the FluxOR dye, generating a fluorescent signal, proportional to channel activity, which was compared to the effects of a high-potassium solution.

Measurement of endocannabinoid levels
A quantitative liquid chromatography-tandem mass spectrometry (LC-MS/MS) method was used for analysis of OEA and PEA in cell samples, based on a previously reported procedure (51). For these experiments, Caco-2 cells were grown in T75 flasks and subjected to inflammatory conditions (10 ng/ml IFNg for 8 h and 10 ng/ml TNFa for a further 16 h). Cell lysates and medium were stored at 280°C before analysis. Internal standard (0.42 nmol AEA-d8) was added to a 0.4 ml aliquot of each sample followed by solvent extraction (ethyl acetate:hexane; 9:1 v/v), centrifugation, and evaporation. Before analysis, each sample extract was reconstituted in acetonitrile. An MDS SCIEX 4000 Q-Trap hybrid triple-quadrupole-linear ion trap mass spectrometer (Thermo Fisher Scientific), operated in positive electrospray ionization mode, was used in conjunction with a series 10AD VP LC system (Shimadzu, Columbia, MD, USA), with an ACE 3 C8, 100 3 2.1 mm, 3 mm particle size column (Advanced Chromatography Technologies Ltd., Aberdeen, United Kingdom). Quantification was performed by measuring specific OEA and PEA precursor and product ions together with a calibrated internal standard method.

Chemicals and reagents
All chemicals and reagents used in these experiments were purchased from Sigma-Aldrich, unless otherwise stated. OEA and PEA and the receptor antagonists AM251, AM630, GW9662, GW6471, capsazepine, and O-1918 were purchased from Tocris (Bristol, United Kingdom). OEA and PEA were dissolved in ethanol to 10 mM, with further dilutions made in EMEM. All receptors antagonists were dissolved in DMSO to 10 mM, with further dilutions in EMEM. IFNg (100 mg) and TNFa (50 mg) were purchased from Thermo Fisher Scientific, and dilutions were made in FBS.

Statistical analysis
Results are expressed as means 6 SEM. Time-course data were compared by 2-way, repeated-measures (repeated by time factor) ANOVA, with Prism 6 (GraphPad Software, Inc., La Jolla, CA, USA). Statistical significance between manipulations and vehicle controls were determined by Dunnett's post hoc test. Results reaching P , 0.05 were statistically significant.

Permeability studies
Our initial experiments sought to explore whether the Nacylethanolamines were able to modulate the ionic conductance of the paracellular pathway, as a proxy for tight junction integrity. When applied to the apical membrane compartment, OEA increased Caco-2 cell monolayer transepithelial electrical resistance (TEER) (i.e., decreased permeability) in a concentration-dependent manner significantly different from control at 1, 3, and 10 mM (Fig. 1A). When applied to the basolateral membrane, OEA decreased TEER (i.e., increased permeability) in a concentrationdependent manner at 1, 3, and 10 mM (Fig. 1C). The log EC 50 of OEA at the apical membrane was 25.43 and at the basolateral membrane was 25.92 (Supplemental Fig. 1A).
PEA caused a large increase in TEER when applied to the apical membrane at 1, 3, and 10 mM (Fig. 1B). Although transient, the effects of 10 mM PEA remained significantly above the effect of vehicle until 48 h after administration. When applied to the basolateral membrane, PEA increased resistance from 30 min after application in a concentration-dependent manner from 1 mM (Fig. 1D). From 8 h on, after application, a significant effect of 300 nM PEA was observed. The log EC 50 of OEA at the apical membrane was 25.43; at the basolateral membrane, it was 25.92 (Supplemental Fig. 1B, C).
To ensure that these changes in permeability were not related to changes in the number of cells, we performed cell viability assays. Neither OEA nor PEA affected Caco-2 cell viability in fully confluent (Supplemental Fig. 2A) or proliferating (Supplemental Fig. 2B) cells. In addition, the expression of 2 AQPs found in mammalian intestines (AQP3 and -4) that transport water, glycerol, ammonia, and hydrogen peroxide (52) and could affect membrane permeability, was investigated. Apical treatment of Caco-2 cells with either OEA or PEA (10 mM, 1 h) led to a significant reduction in the membrane expression of AQP3 (Fig. 1E) and AQP4 (Fig. 1F). Furthermore, changes in transmembrane ion gradients generate osmotic alterations that can affect cell volume, and the involvement of potassium ion influx in cell volume regulation has only recently been recognized (53). Apical treatment of Caco-2 cells with either OEA or PEA led to a concentrationindependent increase in fluorescence indicative of activation of potassium channels (Fig. 1G).

Cytoskeletal changes
To clarify the impact of these lipid mediators on cytoskeletal changes, we treated and processed mature Caco-2 cells, to visualize F-actin (Fig. 2). At the apical focal plane, cell-to-cell adhesion is visible across all treatments with no gaps. OEA rendered the cortical F-actin to have an irregular morphology (top, middle) compared to vehicle control (top left), whereas PEA induced focal adhesion plaques at sites of cell-to-cell adhesion (top right). Cell adhesion to the slide can be seen at the basal focal plane in resting cells (bottom left), with OEA inducing a loss of both cellular tension through reduced actin filaments and focal adhesions (bottom middle). On the other hand, PEA caused an increase in polymerized F-actin filaments and focal adhesion plaques (bottom right). These occur throughout the cytoplasm of the cell, as well as some cortical accumulation. Images in Fig. 3 are representative fields of view from 4 separate experiments. Full z stacks projected into single images can be viewed in Supplemental Fig. 3.

Intracellular signaling
Because the action of the contractile cytoskeleton enables the cellular changes required to adjust permeability in response to its environment, we investigated the signaling events known to be important for cytoskeletal modifications-namely, FAK and the p42/44 MAPKs (54). OEA induced a transient increase in both FAK and Erk1/2 (Fig. 2, left panel, top and third blot down), peaking at 5 min and returning to basal levels by 30 min (Fig. 3B, C). PEA induced phosphorylation of Erk1/2 to significantly higher levels than OEA (Fig 3A, right panel, third blot down), but FAK activation by PEA continued to increase up to 1 h after application (Fig 3B, C, right panel, top blot).
We performed further experiments using Luminex technology and commercially available panels for multiple pathways and the SRC pathway. As seen with Western blot analysis (Fig. 1A), OEA significantly increased phosphorylated ERK1/2 and also p70s6K, CREB and NF-kB, and significantly decreased phosphorylated p38 and JNK (Fig. 3D). PEA significantly increased phosphorylated ERK1/2, p70s6K, and CREB, and significantly decreased phosphorylated p38 (Fig. 3E). In this panel, significant differences between OEA and PEA were observed in the ERK1/2 and Akt response (Supplemental Fig. 4D, F). In the Src family panel of signaling proteins, OEA and PEA significantly reduced phosphorylated Src, Yes, Lck, Lyn, Fgr, and Blk (significance is not shown in Fig. 3F and G; for clarity, please refer to Supplemental Fig. 5). OEA also significantly reduced phosphorylated Fyn and Hck. The reduction was more pronounced at 10 min for OEA (Fig.  3F) and at 2 min for PEA (Fig. 3G).

Receptor mechanism of action
The ability of a submaximal concentration of OEA (3 mM, apical application) to increase TEER was inhibited by capsazepine (a TRPV1 antagonist) only (Fig. 4A). The ability of OEA (3 mM, basolateral) to decrease TEER was inhibited by the TRPV1 antagonist capsazepine and the PPARa receptor antagonist GW6471 (Fig. 4C). The effect of PEA at the apical membrane was inhibited by the PPARa antagonist GW6471 (Fig. 4B). The effect of PEA at the basolateral membrane was inhibited by a PPARa antagonist (Fig. 4D).
OEA and PEA are degraded by FAAH. When OEA or PEA were applied in combination with the FAAH inhibitor URB597, their effects were amplified. OEA (3 mM, apically) caused further increases in TEER when coapplied with an FAAH inhibitor (URB597; Fig. 5A) to the apical membrane, and this effect was inhibited by TRPV1 antagonism. OEA (3 mM) also caused a further decrease in resistance when coapplied with URB597 to the basolateral membrane, via TRPV1 and PPARa (Fig. 5C). PEA (3 mM) caused further increases in resistance when coapplied with URB597 (at either the apical or basolateral membrane), and this resistance was inhibited by the PPARa antagonist GW6471 (Fig. 5B, D).

Effects of OEA and PEA on cytokineinduced hyperpermeability
When applied to the apical membrane concurrently with cytokines, OEA (3 mM) prevented the fall in TEER (Fig.  6A). Apically, OEA also recovered the increased permeability when applied 24 h after cytokines (Fig. 5B). By contrast, application of OEA to the basolateral membrane (at either time 0 or 24 h) caused further decreases in TEER than was caused by cytokines alone, indicating further increased permeability (Fig. 6A, B). PEA (3 mM) prevented the decline in TEER caused by cytokines when applied at the same time to the basolateral membrane, evident as early as 8 h into the cytokine exposure (IFNg exposure only, Fig. 6A). This effect of PEA at the basolateral membrane was still observed when PEA was applied 24 h after exposure to cytokines (Fig. 6B). However, PEA has no effect on cytokine-increased permeability when applied to the apical membrane at either time point (Fig. 6A, B).
To establish whether OEA and PEA are produced endogenously in cells in response to simulated inflammatory conditions, we measured the cellular and secreted levels of these compounds by LC-MS/MS, using the inflammation protocol used to assess permeability changes. Cellular levels of OEA (P , 0.001; Fig. 6C) and PEA (P , 0.01, Fig.  5E) were significantly increased by the inflammatory protocol. Significantly raised levels of OEA (P , 0.0001, Fig. 6D) and PEA (P , 0.001, Fig. 6F) were also detectable in the medium in response to simulated inflammation.

Mechanisms of action of OEA and PEA on cytokine-induced hyperpermeability
When applied to the apical membrane concurrently with cytokines, as before (Fig. 6), OEA (3 mM) prevented the decrease in TEER, and this effect was inhibited by the TRPV1 antagonist capsazepine (Fig. 7A). Apically, OEA also recovered the increased permeability when applied 24 h after cytokines, also inhibited by capsazepine (Fig. 7C). As before, application of OEA to the basolateral membrane caused further decrease in TEER (when added at time 0 or 24 h), which was inhibited by the PPARa antagonist GW6471 but not by capsazepine (Fig. 7A, C).
As before, PEA (at the basolateral membrane) prevented the decline in TEER caused by cytokines when applied at the same time or 24 h later, and this PEA effect was inhibited by GW6471 (Fig. 7B, D).
The effects of OEA and PEA on prolonged cytokine exposure Last, we examined whether OEA and PEA can alter the permeability response to prolonged cytokine exposure. At 48 h after application of cytokines, apical application of OEA restored permeability to baseline (Fig. 8A). However, after 72 h of inflammation, this ability of OEA was lost (Fig.  8C). At the basolateral membrane, PEA restored permeability to baseline when applied 48 h after cytokine exposure (Fig. 8B) and even at 72 h after cytokine exposure (Fig.  8D), and this effect of PEA was inhibited by the PPARa antagonist GW6471 (Fig. 8D).

DISCUSSION
This study has shown the effects of the endocannabinoidlike compounds OEA and PEA on the function and permeability of intestinal epithelial cells in control conditions and in inflammation. Both compounds reversed the hyperpermeability associated with inflammatory conditions through different mechanisms: OEA through TRPV1 on the apical membrane and PEA at the basolateral membrane through PPARa. Increased cellular and secreted OEA and PEA levels were observed in response to inflammation, suggesting that their local release plays a role in intestinal permeability. Inhibition of the degradation of these compounds augmented their responses, indicating that their effects are via the compounds themselves and not by their metabolites. It also suggests that the beneficial effects of these compounds could be augmented by coadministration of inhibitors of their degradation.

OEA
OEA production in the gut is stimulated by food intake (38) or reduced by food deprivation (39). It suppresses food intake, induces satiety, and decreases body weight gain (43) via PPARa activation (44). In intestinal epithelial cells, under control conditions, we found that apical administration of OEA increased Caco-2 monolayer resistance (i.e., decreased permeability) in a concentrationdependent manner via TRPV1. In contrast, OEA increased permeability when applied to the basolateral membrane by activation of TRPV1 and PPARa receptors. Although it is not known whether OEA stimulation by food intake would occur at the basolateral or apical membrane, based on the findings of the present study, alterations in permeability are likely to be associated.
Contractile filamentous actin networks regulate cellular shape change, which can be spatially and temporally modulated during physiologic processes such as cell adhesion, where cytoskeletal mechanics facilitate cell spreading and stiffening in response to environmental cues. F-actin structures, such as lamella and stress fibers, can facilitate adhesion, whereas cortical F-actin influences shape. FAK is a nonreceptor protein kinase that can modulate barrier function (54), and our data confirm that OEA transiently activates FAK. However, the F-actin changes that ensue are 2-fold. Apically, the cortical arrangement indicates shape change, whereas basally, the filamentous structure is reduced. Reduced adhesion at the base of the cells could explain why OEA has a differential effect on TEER, depending on where it is acting. It is unclear how the change in apical morphology connects to a change in cell-cell junctional complexes such that the interactions are tighter, but certainly reduced cellular adhesion to the extracellular matrix (or glass slide in this instance) could account for the OEA effect on permeability when applied basally. It is important to note that cells remain attached to each other with no gaps, implying that the changes in permeability are not related to pore formation or destruction of the epithelial monolayer (also indicated by the lack of effect of OEA on cell viability).
There are many molecular markers that are associated with barrier integrity and membrane permeability. Apical junctional complex structure can be dynamic and the precise location of some of the component parts can influence the final outcome. The contribution and mechanisms of AQPs in regulation of membrane permeability in the gut is unclear. In our study, the reduction in AQP4 membrane protein expression by OEA was unlikely to affect water transport, because knockdown of AQP4 has no impact on the colonic osmotic water permeability coefficient (55,56), but could be related to other functions, such as the intestinal inflammatory response (57). With regard to AQP3, apical expression in the ileum is reduced in early IBD (58), which may limit excessive water loss or alleviate oxidative stress. However, Zhang and colleagues (59) showed that intestinal barrier integrity is impaired by the knockdown of AQP3 by enhancement of paracellular permeability. OEA led to a modest reduction in expression of AQP3 in our study that would be unlikely to affect TEER through water transport. The role of these AQPs in glycerol and lipid metabolism is beyond the scope of this study, although it is tempting to speculate that the accepted contribution of OEA in fat sensing and transport of dietary lipids (60) could be mediated through AQP expression. We also showed that OEA activated potassium channels in Caco-2 cells-an effect that has also been observed in arteries (61,62). Potassium channel activation in the intestine is associated with many aspects of colonic epithelial function, including regulating electrogenic transport, regulating cell volume, and cellular migration (63,64), suggesting OEA modulates epithelial cell functions in the intestine at many levels, which requires further investigation.
Regulation of the intercellular junctional interactions that maintain barrier function is highly complex. However, FAK activity through phosphorylation has been well correlated with TEER and Src dependency may be crucial to this function, particularly in Caco-2 cells (54). In our study, OEA transiently increased the autophosphorylation of FAK, but reduced Src phosphorylation in the same time frame. Reduced phosphorylation of Src and JNK have been shown to attenuate stretch-induced reorganization of the actin cytoskeleton (65), and increased Src is associated with tight junction disruption in the intestinal epithelium (66). The increase in p70S6K and CREB phosphorylation is likely to relate to downstream gene transcription and protein translation, which is similar to PEA. However, the increase in NF-kB activity, which is unique to OEA in this system, requires further investigation. NF-kB has pleiotropic roles in cell survival and the immune response. The precise role of NF-kB in TEER in this context is unclear, but may explain the basolateral reduction in TEER, reminiscent of TNFa-induced barrier disruption (67).
In our model of inflammation, IFNg and TNFa applied to the basolateral membrane of confluent Caco-2 cells increased permeability, similar to that previously reported by our group (35). We found that application of OEA apically, concurrently with the cytokines, or even 24 or 48 h later, reversed the increased permeability via TRPV1. This is the first study to investigate the effects of OEA on intestinal permeability in vitro, but OEA has been found to decrease blood-brain barrier permeability in ischemia in vivo and in vitro similarly by PPARa activation (68). Pharmacological activation of TRPV1 may contribute to colonic inflammation (30); thus, the anti-inflammatory actions of OEA through TRPV1, may be brought about by desensitization of the TRPV1 receptor. We also showed that inflammation significantly increased OEA levels in Caco-2 cells, suggesting that these observations of the pharmacological effects of OEA have a physiologic relevance. Others have similarly shown that OEA is upregulated in response to inflammation (69) or by feeding (70), and this may be as a result of increased OEA synthesis or reduced degradation.
To summarize the effects of OEA, at the apical membrane OEA decreases permeability and inhibits increased permeability when applied before or after the induction of increased permeability associated with inflammation via TRPV1 activation, and inflammation increases cellular levels of OEA. By contrast, at the basolateral membrane, OEA causes increased permeability through both TRPV1 and PPARa. Activation of FAK, inactivation of Src, changes in F-actin, activation of K + channels, and downregulation of AQPs may underlie these cellular responses to OEA.

PEA
PEA is currently available as a nutraceutical food for medical purposes under the brand names Normast, Pelvilen, and PeaPure; has been studied in humans, mostly within trials on pain management; and is well tolerated (71). Several preclinical animal studies have shown that in vivo treatment with PEA reduces intestinal injury and inflammation via PPARa (47,48) and also CB 2 and GPR55 (49). In support of this, we showed that PEA decreases Caco-2 cell permeability when applied to either the apical or basolateral membrane in a time-and concentrationdependent manner, also via activation of the PPARa receptor. Furthermore, basolateral application of PEA, as might occur with systemic administration, also reversed the hyperpermeability associated with inflammation via PPARa. Unlike with OEA, there was no negative (i.e., increased permeability) response to PEA application at either membrane. In inflammation, this beneficial effect of PEA was observed when PEA was added before the insult, or even after 24, 48, and 72 h after induction of inflammation. This ability of PEA to prevent increased permeability at the intestine barrier, via PPARa, is likely to underpin some of the beneficial effects seen in vivo. The increase in cellular PEA levels in response to our inflammatory protocol is in keeping with the proposed protective effects of endogenously produced PEA in the gut (49,72,73).
Like OEA, PEA also induced FAK activity, but the timing was more extended and the F-actin lamella structure seen with PEA was pronounced. These differences imply different cellular outcomes. The effect of PEA on the filament formation appears more typical, in that FAK phosphorylation resulted in F-actin polymerization and the filaments formed with focal adhesion plaques, both at the apical cell-to-cell contacts and the basement membrane (glass slide in our case), meaning that, functionally, these increases in turn increase cellular tension and adhesion to each other as well as to the "matrix"/adhesive surface, resulting in increased TEER, whether applied apically or basally. Like OEA, PEA also led to a modest but significant reduction in both AQP3 and -4 in the membrane fraction and activated potassium channels (see earlier paragraph). The role of FAK activity in terms of transient and prolonged phosphorylation could have an impact on the transient vs. sustained cytoskeletal changes in this study. However, the rather blunt tool of immunoblot analysis may distort the more subtle contribution of location and binding partners.
To summarize the effects of PEA at both the apical and basolateral membrane, PEA decreases permeability and inhibits increased permeability when applied before or up to 72 h after the induction of inflammation via PPARa, and inflammation increases cellular levels of PEA. Activation of FAK, inactivation of Src, changes in F-actin, activation of K + channels, and down-regulation of AQPs may underlie these cellular responses. PEA treatment is feasible and tolerated in humans and the present studies provide a potential rationale to justify controlled clinical trials of PEA in GI disorders.

CONCLUSIONS
OEA and PEA modulate intestinal permeability in normal and inflammatory conditions; OEA can both increase and decrease permeability ( via TRPV1) when applied to the apical or basolateral membrane, respectively, whereas PEA always decreases permeability i.e., increases resistance (via PPARa). Cellular levels of OEA and PEA are increased in intestinal epithelial cells in response to inflammation, which may limit the increased permeability associated with inflammation. The beneficial effects on intestinal permeability may at least partly underlie the protective effects of PEA on intestinal damage recently observed in preclinical studies. PPARa agonism, PEA administration or inhibiting PEA enzymatic degradation represent a novel range of therapeutic approaches against several intestinal disorders associated with increased intestinal permeability, including inflammatory bowel disease and the acute intestinal ischemia associated with circulatory shock.