The Punicalagin Metabolites Ellagic Acid and Urolithin A Exert Different Strengthening and Anti-Inflammatory Effects on Tight Junction-Mediated Intestinal Barrier Function In Vitro

Scope: Ellagitannins are polyphenols found in numerous fruits, nuts and seeds. The elagitannin punicalagin and its bioactive metabolites ellagic acid and urolithins are discussed to comprise a high potential for therapeutically or preventive medical application such as in intestinal diseases. The present study characterizes effects of punicalagin, ellagic acid and urolithin A on intestinal barrier function in the absence or presence of the proinflammatory cytokine tumor necrosis factor-α (TNFα). Methods and Results: Transepithelial resistance (TER), fluorescein and ion permeability, tight junction protein expression and signalling pathways were examined in Caco-2 and HT-29/B6 intestinal epithelial cell models. Punicalagin had less or no effects on barrier function in both cell models. Ellagic acid was most effective in ileum-like Caco-2 cells, where it increased TER and reduced fluorescein and sodium permeabilities. This was paralleled by myosin light chain kinase two mediated expression down-regulation of claudin-4, -7 and -15. Urolithin A impeded the TNFα-induced barrier loss by inhibition of claudin-1 and -2 protein expression upregulation and claudin-1 delocalization in HT-29/B6. Conclusion: Ellagic acid and urolithin A affect intestinal barrier function in distinct ways. Ellagic acid acts preventive by strengthening the barrier per se, while urolithin A protects against inflammation-induced barrier dysfunction.


INTRODUCTION
Health benefits of the ellagitannin punicalagin and its metabolites are being extensively discussed in recent years. Punicalagin can be found in pomegranate (Punica granatum), raspberries (Rubus idaeus), strawberries (Fragaria sp.) or walnuts (Juglans regia). Especially pomegranate has already been applied for thousands of years in the traditional medicine (Longtin, 2003) to benefit from its anti-diarrhea (Das et al., 1999;Zhao et al., 2018), anti-oxidant, anti-obesity, anti-cancerogenic, and anti-inflammatory properties (reviewed in Saeed et al., (2018). Ellagitannins are not directly absorbed into the blood stream (Seeram et al., 2004) but hydrolyzed. In the stomach and intestine, hydrolysis of punicalagin yields ellagic acid (EA), which in turn is metabolized to urolithin A and B by the intestinal microbiota (Cerdá et al., 2005;Espín et al., 2007). Several studies showed these metabolites exert anti-inflammatory effects in the intestine. In a mouse model of ulcerative colitis, ellagic acid was effective in reducing intestinal inflammation by inhibiting cyclooxygenase-2 and inducible nitric oxide synthases. It impeded proinflammatory signaling via nuclear factor 'kappa-light-chainenhancer' of activated B-cells (NFκB) and signal transducers and activator of transcription 3 (STAT3) (Marín et al., 2013). A very recent cell culture study reported ellagic acid to inhibit pro-inflammatory effects of tumor necrosis factor α (TNFα). This involved loss of epithelia barrier function, upregulation of interleukin-6 and -8 secretion and induction of oxidative stress by impeding TNFα-induced signaling via NF-κB, extracellular signal-regulated kinases 1/2 (ERK1/2) and myosin light chain kinase (MLCK) (Iglesias et al., 2020). Urolithin A (UroA) can be detected at relatively high amounts in the colon. Its antiinflammatory activities were reported from a rat model of colitis (Larrosa et al., 2010) and a colon fibroblast model (Giménez-Bastida et al., 2012).
Intestinal inflammation causes intestinal barrier dysfunction [(reviewed in Hering et al., (2012)). The intestinal barrier integrity plays a central role for gut health by preventing an abandoned passage of antigens, allergens, bacterial toxins or other noxious agents from the intestinal lumen into the mucosa and blood circulation. The intestinal epithelium is built up by a single row of epithelial cells, which are connected by tight junctions (TJs) at their most apical point. Dependent on the physiological condition, the epithelial TJs regulates the paracellular passive passage of water and nutrients. This is achieved by the specific interplay of different TJ proteins, including the large family of claudins, TJ-associated MARVEL proteins (Mineta et al., 2011), such as tricellulin, and junctional adhesion molecules (Raleigh et al., 2010). These transmembrane proteins are connected to intracellular scaffold proteins (e.g. zonula occludens proteins, ZO−1-3) and form a meshwork of numerous horizontally oriented strands surrounding the epithelial cells. Changes to this defined composition can result in altered barrier function. Pro-inflammatory cytokines, as e.g. tumor necrosis factor α (TNFα) are well-known to cause barrier dysfunction by inducing epithelial apoptosis and by affecting TJ architecture, including claudin protein expression and delocalization (Hering and Schulzke, 2009). So far, little is known about the impact of punicalagin, ellagic acid or urolithin A on intestinal barrier function. Objective of the present study was to elucidate their putative protective and barrier strengthening properties on epithelial TJ integrity per se or in state of inflammation. As their bioavailability differs along the gastrointestinal tract (Espín et al., 2013), we hypothesized that these bioactive compounds might act distinctly on barrier function in ileum or colon. Therefore, we investigated two different intestinal cell lines, ileum-like Caco-2 cells and HT-29/B6 colon cells.

Cell Culture and Dosage Information
Caco-2 cells are epithelial cells derived from a colorectal adenocarcinoma (ATCC ® HTB-37 ™ ). However, under specific culturing conditions Caco-2 cells differentiate and polarize such that they functionally and morphologically resemble the phenotype of distal ileum enterocytes. They are characterized by absorptive capabilities and active transport pathways, possess enzymatic activities and an apical brush border. When cultured on filter supports Caco-2 cells grow as polarized monolayers with epithelial TJs (Hidalgo et al., 1989).
Myosin light chain kinase inhibitor PIK (150 µM) was preincubated on Caco-2 monolayers 2 h before challenging with EA. Phosphorylation events were studied under serum free conditions.
The colon carcinoma cell line HT-29/B6 is a subclone of the human colon carcinoma cell line HT-29 (Kreusel et al., 1991) and was cultured on permeable filter supports (0.6 cm 2 effective area; 3.0 μm pores, Millicell PCF, Millipore) using RPMI medium (Sigma-Aldrich) containing 10% bovine serum and 1% penicillin/streptomycin. Monolayers grew confluent within one week, giving a TER of at least 350 Ω·cm 2 and were pre-incubated with 10 µM punicalagin, 150 µM ellagic acid and 150 µM or 250 µM urolithin A from both sides. Two hours later 500 U/ml TNFα (Pepro Tech, Hamburg, Germany) were added to the basal compartment. Changes in barrier integrity were assessed by measuring transepithelial resistance (TER) with a pair of chopstick electrodes at 37 C as described before (Heller et al., 2005).
For the measurement of Na + and Cl − permeability, voltage and TER were monitored while reducing NaCl concentration in one hemichamber. This was done by switching to a solution containing a reduced concentration of NaCl and mannitol for balancing osmolality. All other components were equivalent to standard solution. NaCl permeability was determined from dilution potentials and the Goldmann-Hodgkin-Katz equation as previously reported (Amasheh et al., 2002;Günzel et al., 2009;Yu et al., 2009).

Freeze Fracture Electron Microscopy
Freeze fracture electron microscopy analysis was performed and quantified as described elsewhere in detail ).

Stabilizing Effect of Punicalagin, Urolithin A and Ellagic Acid on Barrier Properties of Caco-2 Intestinal Cells
Punicalagin increased TER slightly from initial values in Caco-2 monolayers within 24h ( Figure 1A; p < 0.05, p < 0.001 vs. control). This TER increase could not be enhanced dosedependently ( Figure 1A; p < 0.05 Puni 50 vs. 150 µM). In contrast, the TER increase induced by urolithin A was more pronounced ( Figure 1B). While 25 µM was not effective, doses up to 100 µM UroA increased TER significantly from control ( Figure 1B; p < 0.001 vs. control). Higher doses than 100 µM of urolithin A had no further increasing effects ( Figure 1B). Ellagic acid caused the strongest TER increase within 24 h. 50 µM was as effective as 200 μM EA compared to control ( Figure 1C; p < 0.001 and p < 0.05 vs. control). Comparing the most effective dose of each compound in one experiment proved ellagic acid (150 µM) to induce the strongest TER increase in Caco-2 monolayers, followed by urolithin A (250 µM) and punicalagin (10 µM) ( Figure 1D; p < 0.001 vs. control and p < 0.001 vs. EA). The TER increase induced by 150 µM ellagic acid was paralleled by a permeability decrease of the 332Da marker molecule fluorescein in Caco-2 monolayers (Figure 2A; p < 0.001 vs. control). Neither 250 µM urolithin A nor 10 µM punicalagin reduced fluorescein permeability (Figure 2A). Measurements of dilution potentials for sodium and chloride showed that ellagic acid restricted sodium permeability, but not chloride permeability in Caco-2 monolayers ( Figure 2B; p < 0.001 vs. control). The permeability ratios of sodium and chloride (P Na /P Cl ) were reduced 3-fold from 28 ± 7 in control to 9 ± 1 in monolayers challenged with ellagic acid (p < 0.05 vs. control).
Signaling was studied by assessing different kinase inhibitors. Although ellagic acid induced phosphorylation of p38 and STAT3, specific phosphorylation inhibition of these kinases could not impede the ellagic acid-stimulated TER increase (data not shown). In contrast, inhibition of MLCK by PIK prevented Myosin Light Chain 2 (MLC2) phosphorylation and blocked the ellagic acid-induced TER increase ( Figure 4A; p < 0.001 vs. control). In parallel, PIK impeded the ellagic acid-depended expression down-regulation of claudin-4, -7 and -15 ( Figure 4B; p < 0.05 PIK + ellagic acid vs. ellagic acid alone). In PIK + ellagic acid co-treated monolayers, protein levels did not differ from control ( Figure 4B).

DISCUSSION
Epithelial barrier function is a key feature of intestinal health. The present study examined the molecular mechanisms behind the health claims of the bioactive polyphenol punicalagin and its metabolites ellagic acid and urolithin A on epithelial barrier function in vitro. Our data revealed that especially ellagic acid and urolithin A affect barrier function in different ways in our two cell models. Ellagic acid strengthened barrier function per se by reducing the expression of pore-forming claudin-4, -7 and -15 via MLC2 signaling in ileum-like Caco-2 cells. Urolithin A impeded the pro-inflammatory dysregulation and/or redistribution of claudin-2 and -1 in colonic HT-29/B6 cells.
Punicalagin alone had only little or no effects on barrier function in the two cell models.
The constitution of the epithelial TJ, particularly its composition of pore-forming and sealing TJ proteins, is crucial for intestinal barrier function. Barrier integrity can be estimated by measuring TER and paracellular permeability of differently sized molecules or ions. Especially in Caco-2 cells, ellagic acid turned out to strengthen epithelial barrier function per se, displayed by a strong increase in TER, reduction in fluorescein and sodium permeability and down regulation of claudin-4, -7 and -15 expression.
Tight junction ultra-structure was not changed as demonstrated by freeze fracture electron microscopy. Subsequently, barrier effects are not due to changes in overall protein content but to TJ protein composition. Especially, the downregulation of channel forming claudins seems rational. Claudin-15 is predominantly expressed in the small intestine (Inai et al., 2005;Fujita et al., 2006;Holmes et al., 2006) and forms a paracellular cation- (Colegio et al., 2002;Van Itallie et al., 2003;Samanta et al., 2018) and water-selective channel (Rosenthal et al., 2020). Overexpression of claudin-15 in Caco-2 cells caused a decrease in TER (Takehara et al., 2009). The physiological function of claudin-4 and -7 is less clear as it is not consistent and seems to depend on the interactions with other TJ proteins and differs in different cell types and conditions (Günzel and Fromm, 2012). Two studies on flavonoids in Caco-2 cells reported quercetin (Amasheh et al., 2008) or kaempferol (Suzuki et al., 2011) to enhance barrier function by up-regulating claudin-4 expression. In contrast, in a cell culture study on kidney cells, claudin-4 was suggested to act as a chloride channel. Knock down of claudin-4 resulted here in a decrease in P Na /P Cl , which FIGURE 6 | Effect of Urolithin A on TNFα-caused claudin-1 delocalization in HT-29/B6 monolayers. Localization of claudin-1 was studied by confocal laser scanning microscopy in (A) control (B) TNFα (C) urolithin A (UroA) +TNFα and (D) urolithin A-challenged monolayers. Merging of claudin-1 (green) with the TJ marker protein ZO-1 (red) was assessed by z-stack imaging, nuclei are DAPI stained (blue). TNFα caused delocalization of claudin-1 (indicated by white arrows in (B)), while parallel Urolithin A treatment enhanced claudin-1-ZO-1 merging (indicated by white arrows in (C)). Bars indicate 5 µm.
Frontiers in Pharmacology | www.frontiersin.org March 2021 | Volume 12 | Article 610164 8 was paralleled by an TER increase (Hou et al., 2010). In an epithelial co-culture model of Caco and HT29-MTX cells, a decrease in TER induced by oxidative stress was associated with an increase in claudin-7 (Bianchi et al., 2019). In kidney cells, claudin-7 overexpression was associated with a decrease in P Cl and an increase in P Na (Alexandre et al., 2005). Taken together, it seems rational that especially the reduction of claudin-15 and claudin-7 are responsible for the decrease in sodium permeability and the TER increase in ellagic acidchallenged Caco-2 monolayers.
Ellagic acid enhanced the phosphorylation of STAT3, p38 and MLC2 in Caco-2 cells. However, only inhibition of MLC2 phosphorylation by the MLCK-inhibitor PIK impeded the ellagic acid-stimulated TER increase, suggesting a central role of MLC2 in ellagic acid-dependent barrier regulation. Phosphorylation of the MLC2 by MLCK is linked to actomyosin contraction and TJ regulation (Turner et al., 1999;Zolotarevsky et al., 2002;Shen et al., 2006). We showed that phosphorylation inhibition of MLC2 impeded the ellagic acid-induced TER increase and down-regulation of claudin-4, -7 and -15. So far, the role of non-muscle MLCK and MLC2 for barrier function was predominantly discussed with respect to TNFα-caused barrier loss (Zolotarevsky et al., 2002;Wang et al., 2006;Ye and Ma, 2008). Iglesias et al., recently showed ellagic acid to inhibit TNFα-stimulated MLC2 phosphorylation (Iglesias et al., 2020). In contrast, our data show the activation of MLCK/MLC2 in a none-inflammatory state and suggest that MLC2-triggered TJ regulation is not necessarily related to barrier loss as it seems to depend on the specific type of TJ protein affected.
Within the present study, the impact of punicalagin, ellagic acid and urolithin A on inflammation-induced barrier dysfunction was examined in the HT-29/B6 cell culture model, which is a very well-studied inflammation model and more sensitive to TNFα than Caco-2 cells. TNFα is known to upregulate the expression of sealing claudin-1 and channel forming claudin-2, at which claudin-1 is additionally redistributed from the TJ in HT-29/B6. Together with enhanced epithelial apoptosis, these TJ changes are reported to critically contribute to the TNFαcaused barrier loss (Gitter et al., 2000;Mankertz et al., 2009;Amasheh et al., 2010). In contrast to urolithin A, neither punicalagin nor ellagic acid were effective to inhibit the TNFα-induced TER drop in HT-29/B6. Moreover, urolithin A impeded up-regulation of caudin-1 and -2 and seemed at least partially to prevent redistribution of claudin-1. These effects of urolithin A are very similar to effects we observed in a former study on the ginger-derived pungent component 6-shogaol that also impeded the TNFα-induced up-regulation of claudin-2 and disassembly of claudin-1 (Luettig et al., 2016). Furthermore, TNFα enhances epithelial apoptosis contributing to epithelial leakiness (Gitter et al., 2000). Urolithin A did not inhibit TNFαcaused epithelial apoptosis in HT-29/B6, but even seemed to stimulate it slightly. This is in concordance with other studies that showed ellagitannins from pomegranate and urolithin metabolites to inhibit proliferation and to induce apoptosis in HT-29 cells (Kasimsetty et al., 2010).
These very distinct effects of punicalagin, ellagic acid and urolithin A on intestinal barrier function in the two cell models seem rational, because the bioavailability of these compounds differs along the intestine. Punicalagin was reported to be hydrolyzed already during the stomach passage where it yields ellagic acid. Punicalagin itself probably does not reach the intestine in high amounts, while ellagic acid might predominantly interact with the enterocytes of the ileum. As reported from clinical trials, ellagic acid was not detected in high amounts in the colonic mucosa (Nuñez-Sánchez et al., 2014). In contrast, an increasing gradient of urolithins from the jejunum to the distal colon was described in an animal study (Espín et al., 2007). So far, only very few studies addressed the question, how much punicalagin has to be ingested to reach effective intestinal concentrations of ellagic acid or urolithin A. In the present study, the effects seemed to depend on the optimal dosage, which was figured out for each compound. González-Sarrías et al. simulated gastrointestinal digestion of pomegranate extracts yielding around 500 µM ellagic acid, while plasma concentrations remained low at 100 nM (Nuñez-Sánchez et al., 2014). Subsequently, there are more studies needed to elucidate dosage, intestinal side of conversion, and bioavailability of these components in vivo.

CONCLUSION
Our study reveals that the punicalagin metabolites ellagic acid and urolithin A have a protective impact on barrier function in vitro. These findings support the hypothesis that therapeutically application might act preventive by strengthening and protecting the epithelial barrier in case of diarrhea or inflammation. Moreover, the characterization of these compounds might be of interest for the development of multimodal functional food in the future.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.