Curcumin Mitigates Immune-Induced Epithelial Barrier Dysfunction by Campylobacter jejuni

Campylobacter jejuni (C. jejuni) is the most common cause of foodborne gastroenteritis worldwide. The bacteria induce diarrhea and inflammation by invading the intestinal epithelium. Curcumin is a natural polyphenol from turmeric rhizome of Curcuma longa, a medical plant, and is commonly used in curry powder. The aim of this study was the investigation of the protective effects of curcumin against immune-induced epithelial barrier dysfunction in C. jejuni infection. The indirect C. jejuni-induced barrier defects and its protection by curcumin were analyzed in co-cultures with HT-29/B6-GR/MR epithelial cells together with differentiated THP-1 immune cells. Electrophysiological measurements revealed a reduction in transepithelial electrical resistance (TER) in infected co-cultures. An increase in fluorescein (332 Da) permeability in co-cultures as well as in the germ-free IL-10−/− mouse model after C. jejuni infection was shown. Curcumin treatment attenuated the C. jejuni-induced increase in fluorescein permeability in both models. Moreover, apoptosis induction, tight junction redistribution, and an increased inflammatory response—represented by TNF-α, IL-1β, and IL-6 secretion—was observed in co-cultures after infection and reversed by curcumin. In conclusion, curcumin protects against indirect C. jejuni-triggered immune-induced barrier defects and might be a therapeutic and protective agent in patients.


Introduction
Campylobacter jejuni is the most prevalent pathogenic bacterium of zoonotic gastroenteritis [1]. Typical symptoms provoked by C. jejuni are watery to bloody diarrhea, abdominal pain, fever, and nausea [2]. This human pathogen is present in the intestinal microbiota of farm animals, especially poultry, which is the main source of infection for humans by ingestion of contaminated or undercooked food [1]. The bacteria adhere to the mucus and the surface of intestinal epithelial cells, invade the intestinal epithelium, while they pass the cells via the transcellular or paracellular route [3,4]. Consequently, direct epithelial barrier defects such as dysregulation of tight junction (TJ) proteins, induction of epithelial lesions or also indirect effects by inflammatory responses of epithelial or immune cells occur [5,6].
Epithelial lesions can go along with single-cell lesions by increased apoptosis, mid-sized leaks like focal leaks, as well as erosions or even ulcerations. These pathological findings could explain the type of diarrhea for the Campylobacter infection as leak flux pathomechanism, which is characterized by a loss of water and solutes from the organism into the intestinal lumen through a leaky epithelium [6,7]. For a physiological intestinal integrity, an intact epithelium is essential. The main components of intestinal epithelial barrier are the TJs. They are the apical components of the intercellular seal and divide the epithelial cell membrane into an apical and basolateral compartment (fence function) [8]. Tight junctions form a barrier for water, ions, and solutes (barrier function) [9] and consist of various molecular transmembrane proteins like claudins and occludin, junction adhesion molecules (JAM), tricellulin, and cytoplasmatic scaffolding proteins as zonula occludens protein-1 (ZO-1) [8,9].
Recently, the number of C. jejuni infections surpasses Salmonella infections as the leading cause of zoonotic bacterial gastroenteritis in the US and Europe. Moreover, high levels of fluoroquinolone resistance have been reported [10] and levels of multi drug resistances (MDR) are increasing [11]. In consequence, new therapeutic targets as well as innovative strategies to reduce the amount of C. jejuni in farm animals and environment are urgently required. A potent natural substance thought to interfere with Campylobacter infection is curcumin.
The aim of the presented study was to evaluate the barrier-protective and anti-inflammatory properties of curcumin on intestinal epithelial barrier function, which is compromised by the C. jejuni-mediated immune response. Therefore, a novel in vitro co-culture model consisting of intestinal epithelial HT-29/B6-GR/MR and immune THP-1 cells was established. This co-culture model enables the screening of potential barrier-protective and anti-inflammatory compounds in C. jejuni infection, as a result of which curcumin was identified as being effective.

Establishment of a Co-Culture with Colon Epithelial Cells HT-29/B6-GR/MR and Immune THP-1 Cells
To analyze the impact of the immune response during a Campylobacter infection on epithelial barrier function, human colon epithelial cells and immune cells were co-cultured. Colonic epithelial cells (HT-29/B6-GR/MR) were seeded on cell culture filter supports (0.4 µm pore size) and represented the epithelial compartment. Under the membrane of the epithelial cell filter, THP-1 immune cells were seeded on the bottom of the cell culture well and represented the subepithelial compartment. Basal bacterial infection with C. jejuni decreased transepithelial electrical resistance (TER) when colonic epithelial and immune cells (differentiated THP-1 cells) were co-cultured ( Figure 1). From the apical supernatant, no bacteria could be cultured after the experiment, indicating that the bacteria could not pass the filter membrane with 0.4 µm pore size. As a result of this, and in contrast to the results in the co-culture model, C. jejuni infection at the basal membrane had no effect on TER in HT-29/B6-GR/MR mono-culture ( Figure 1).

Curcumin Improves Disturbed Intestinal Barrier Function in the Co-Culture System
Using mRNA sequencing and subsequent ingenuity pathway analysis (IPA) of human mucosa biopsies from C. jejuni-infected patients, inhibition of curcumin-dependent pathways indicated that the presence of curcumin might prevent the C. jejuni-induced changes in gene expression. Based on this bioinformatic prediction of curcumin's effect on host gene expression, the addition of curcumin should activate downstream pathways and might reduce the effects of Campylobacter infections (Supplementary Table S1). In order to test the barrier protective and anti-inflammatory properties of curcumin in Campylobacter-induced barrier dysfunction, a curcumin solution was added to the infected co-culture model. As shown in Figure 2, C. jejuni reduced TER significantly 48 h after infection, whereas this change was inhibited by the presence of curcumin. Permeability studies were performed with the paracellular flux marker fluorescein (332 Da). Compared to controls, C. jejuni infection caused an increase in permeability to fluorescein. Application of curcumin not only ameliorated the permeability increase for fluorescein, but even resulted in an improved barrier function for paracellular passage of macromolecules when exceeded the control level ( Figure 3).

Curcumin Improves Disturbed Intestinal Barrier Function in the Co-Culture System
Using mRNA sequencing and subsequent ingenuity pathway analysis (IPA) of human mucosa biopsies from C. jejuni-infected patients, inhibition of curcumin-dependent pathways indicated that the presence of curcumin might prevent the C. jejuni-induced changes in gene expression. Based on this bioinformatic prediction of curcumin's effect on host gene expression, the addition of curcumin should activate downstream pathways and might reduce the effects of Campylobacter infections (Supplementary  Table S1). In order to test the barrier protective and anti-inflammatory properties of curcumin in Campylobacter-induced barrier dysfunction, a curcumin solution was added to the infected co-culture model. As shown in Figure 2, C. jejuni reduced TER significantly 48 h after infection, whereas this change was inhibited by the presence of curcumin.

Curcumin Improves Disturbed Intestinal Barrier Function in the Co-Culture System
Using mRNA sequencing and subsequent ingenuity pathway analysis (IPA) of human mucosa biopsies from C. jejuni-infected patients, inhibition of curcumin-dependent pathways indicated that the presence of curcumin might prevent the C. jejuni-induced changes in gene expression. Based on this bioinformatic prediction of curcumin's effect on host gene expression, the addition of curcumin should activate downstream pathways and might reduce the effects of Campylobacter infections (Supplementary Table S1). In order to test the barrier protective and anti-inflammatory properties of curcumin in Campylobacter-induced barrier dysfunction, a curcumin solution was added to the infected co-culture model. As shown in Figure 2, C. jejuni reduced TER significantly 48 h after infection, whereas this change was inhibited by the presence of curcumin. Permeability studies were performed with the paracellular flux marker fluorescein (332 Da). Compared to controls, C. jejuni infection caused an increase in permeability to fluorescein. Application of curcumin not only ameliorated the permeability increase for fluorescein, but even resulted in an improved barrier function for paracellular passage of macromolecules when exceeded the control level ( Figure 3). Permeability studies were performed with the paracellular flux marker fluorescein (332 Da). Compared to controls, C. jejuni infection caused an increase in permeability to fluorescein. Application of curcumin not only ameliorated the permeability increase for fluorescein, but even resulted in an improved barrier function for paracellular passage of macromolecules when exceeded the control level ( Figure 3).

Effect of Curcumin In Vivo on Intestinal Barrier Regulation in IL-10 −/− Mice
Secondary abiotic IL-10 −/− mice were infected with C. jejuni and treated with 0.5 mg/mL of curcumin, starting 4 days prior infection until the end of the observation period (i.e., day 6 post infection). Epithelial barrier function of distal colon specimens from IL-10 −/− mice was analyzed in Ussing chambers. Conductance (G, Figure 4A) and permeability to fluorescein ( Figure 4B) was significant higher in C. jejuni-infected mice than in untreated infected mice. Mice receiving curcumin showed a significantly lower conductance and permeability to fluorescein than untreated infected mice. The electrophysiological results obtained in vivo ( Figure 4) resemble the results from the in vitro co-culture system ( Figure 3).    Figure 4A) and permeability to fluorescein ( Figure 4B) was significant higher in C. jejuni-infected mice than in untreated infected mice. Mice receiving curcumin showed a significantly lower conductance and permeability to fluorescein than untreated infected mice. The electrophysiological results obtained in vivo ( Figure 4) resemble the results from the in vitro co-culture system ( Figure 3).

Effect of Curcumin In Vivo on Intestinal Barrier Regulation in IL-10 −/− Mice
Secondary abiotic IL-10 −/− mice were infected with C. jejuni and treated with 0.5 mg/mL of curcumin, starting 4 days prior infection until the end of the observation period (i.e., day 6 post infection). Epithelial barrier function of distal colon specimens from IL-10 −/− mice was analyzed in Ussing chambers. Conductance (G, Figure 4A) and permeability to fluorescein ( Figure 4B) was significant higher in C. jejuni-infected mice than in untreated infected mice. Mice receiving curcumin showed a significantly lower conductance and permeability to fluorescein than untreated infected mice. The electrophysiological results obtained in vivo ( Figure 4) resemble the results from the in vitro co-culture system ( Figure 3).

Effect of Curcumin on Bacterial Integrity
To determine whether curcumin had a direct effect on the viability of C. jejuni in the concentration used in our experiments, the minimal inhibitory concentration (MIC) for curcumin was determined. The MIC of curcumin in C. jejuni 81-176 amounted to 87 µM under physiological conditions at a pH of 7.4, and is thus above our experimental curcumin concentrations in the co-culture. Moreover, if curcumin is used as treatment, it is necessary to know whether the substance exerts any stress to the bacterium, leading to enhanced adaption by, e.g., stimulation of natural transformation capacity. Therefore, competence development and DNA uptake of C. jejuni was investigated under the treatment of curcumin using a single cell assay. After curcumin treatment, no change in the ratio of competent bacteria was registered (control 26% ± 2% versus treated 25.5% ± 0.5% with 50 µM curcumin, n = 3, n.s., unpaired Student's t-test). In addition, also the DNA uptake processes in competent C. jejuni was unaffected after curcumin treatment (control 100% versus treated 86.9% ± 6.7%, n = 3, n.s., unpaired Student´s t-test).

Cytokine Secretion in C. jejuni-Infected Co-Cultures
To identify which cytokines were released by the immune cells and affected epithelial cells function, cytometric bead arrays (CBA) from supernatants of treated co-cultures were performed. TNF-α, IL-1β, and IL-6 were increased in C. jejuni-infected co-cultures, while this changes were prevented by curcumin treatment ( Figure 5).

Effect of Curcumin on Bacterial Integrity
To determine whether curcumin had a direct effect on the viability of C. jejuni in the concentration used in our experiments, the minimal inhibitory concentration (MIC) for curcumin was determined. The MIC of curcumin in C. jejuni 81-176 amounted to 87 µ M under physiological conditions at a pH of 7.4, and is thus above our experimental curcumin concentrations in the coculture. Moreover, if curcumin is used as treatment, it is necessary to know whether the substance exerts any stress to the bacterium, leading to enhanced adaption by, e.g., stimulation of natural transformation capacity. Therefore, competence development and DNA uptake of C. jejuni was investigated under the treatment of curcumin using a single cell assay. After curcumin treatment, no change in the ratio of competent bacteria was registered (control 26% ± 2% versus treated 25.5% ± 0.5% with 50 µ M curcumin, n = 3, n.s., unpaired Student's t-test). In addition, also the DNA uptake processes in competent C. jejuni was unaffected after curcumin treatment (control 100% versus treated 86.9% ± 6.7%, n = 3, n.s., unpaired Student´s t-test).

NFκB Signaling Pathway is Involved in C. jejuni-Induced Barrier Dysfunction
Furthermore, we evaluated the possible relevance of the NFκB pathway for barrier function in our cell model using the NFκB inhibitor BAY 11-7082 (BAY). BAY inhibited the C. jejuni-induced reduction in TER, similar to the effect of curcumin ( Figure 6A). In addition, the increase in permeability to fluorescein could be reversed after BAY inhibition and almost reached control values ( Figure 6B).

NFκB Signaling Pathway is Involved in C. jejuni-Induced Barrier Dysfunction
Furthermore, we evaluated the possible relevance of the NFκB pathway for barrier function in our cell model using the NFκB inhibitor BAY 11-7082 (BAY). BAY inhibited the C. jejuni-induced reduction in TER, similar to the effect of curcumin ( Figure 6A). In addition, the increase in permeability to fluorescein could be reversed after BAY inhibition and almost reached control values ( Figure 6B).

Epithelial Apoptosis is Blocked by Curcumin in the Co-Culture Model
Given that the induction of epithelial apoptosis constitutes a barrier-relevant pathomechanism, and Campylobacter infections have been shown to induce epithelial apoptosis, the impact of curcumin on apoptosis was analyzed using cleaved caspase-3 staining and visualized by confocal laserscanning microscopy (CLSM) (Figure 7).

Epithelial Apoptosis is Blocked by Curcumin in the Co-Culture Model
Given that the induction of epithelial apoptosis constitutes a barrier-relevant pathomechanism, and Campylobacter infections have been shown to induce epithelial apoptosis, the impact of curcumin on apoptosis was analyzed using cleaved caspase-3 staining and visualized by confocal laser-scanning microscopy (CLSM) (Figure 7).

Influence of C. jejuni and Curcumin on Tight Junction Protein Expression in the Co-Culture Model
To define the C. jejuni-induced tight junction (TJ) expression changes, protein lysates were generated from colon epithelial cells of co-cultures after treatment with curcumin and infection with C. jejuni (Figure 8). The expression change was quantified by densitometric analysis with β-actin as loading control, which showed an increase in protein expression of claudin-1 protein expression after C. jejuni infection. Claudin-1 expression seemed to decrease again after curcumin treatment in the C. jejuni infected group. However, this did not reach statistical significance. No further expression changes of the tested TJ proteins could be observed. A reduction by trend after C. jejuni infection in occludin and claudin-8 was detectable, but without reaching statistical significance after multiple test correction (claudin-8 infected versus curcumin-treated p = 0.046, uncorrected Student´s t-test). After infection with C. jejuni, the apoptotic ratio was increased 3-fold in comparison to control. Incubation with curcumin during the infection with C. jejuni reduced the apoptotic ratio to that of controls ( Figure 7) (control 0.7% ± 0.4% versus infected 3.3% ± 0.4% versus treated 1.1% ± 0.2%, n = 6, ** p < 0.01, *** p < 0.001, one-way ANOVA with Bonferroni correction for multiple comparisons).

Influence of C. jejuni and Curcumin on Tight Junction Protein Expression in the Co-Culture Model
To define the C. jejuni-induced tight junction (TJ) expression changes, protein lysates were generated from colon epithelial cells of co-cultures after treatment with curcumin and infection with C. jejuni (Figure 8). The expression change was quantified by densitometric analysis with β-actin as loading control, which showed an increase in protein expression of claudin-1 protein expression after C. jejuni infection. Claudin-1 expression seemed to decrease again after curcumin treatment in the C. jejuni infected group. However, this did not reach statistical significance. No further expression changes of the tested TJ proteins could be observed. A reduction by trend after C. jejuni infection in occludin and claudin-8 was detectable, but without reaching statistical significance after multiple test correction (claudin-8 infected versus curcumin-treated p = 0.046, uncorrected Student´s t-test).

Influence of C. jejuni and Curcumin on Subcellular TJ Protein Distribution in Co-Cultures in Confocal Laser-Scanning Microscopy
Since the composition and presence of TJ proteins in the TJ strands influence barrier function, the subcellular distribution of the TJs was analyzed with confocal laser-scanning microscopy (CLSM). Z-stacks were generated from different TJ proteins of the immunostained epithelial cell monolayers, and intensity-distance plots were generated. The TJ strands after C. jejuni infection indicated a redistribution of claudin-4 ( Figure 9A) and claudin-8 ( Figure 9B). Claudin-4 and claudin-8 signals were retracted from the TJ strands. Moreover, claudin-4 and claudin-8 were no longer co-localized with zonula occludens protein-1 (ZO-1) after C. jejuni infection. Intensity-distance plots with high background intensities indicate re-localization into the intracellular regions. After treatment with curcumin, the appearance of claudin-4 and claudin-8 improved. As a consequence, the infected and curcumin treated group showed co-localization (merging), with high peak intensities in the areas of TJ contact, comparable to the control. Curcumin control groups showed no difference compared to the untreated controls (Supplementary Figure S2). Since the composition and presence of TJ proteins in the TJ strands influence barrier function, the subcellular distribution of the TJs was analyzed with confocal laser-scanning microscopy (CLSM). Z-stacks were generated from different TJ proteins of the immunostained epithelial cell monolayers, and intensity-distance plots were generated. The TJ strands after C. jejuni infection indicated a redistribution of claudin-4 ( Figure 9A) and claudin-8 ( Figure 9B). Claudin-4 and claudin-8 signals were retracted from the TJ strands. Moreover, claudin-4 and claudin-8 were no longer co-localized with zonula occludens protein-1 (ZO-1) after C. jejuni infection. Intensity-distance plots with high background intensities indicate re-localization into the intracellular regions. After treatment with curcumin, the appearance of claudin-4 and claudin-8 improved. As a consequence, the infected and curcumin treated group showed co-localization (merging), with high peak intensities in the areas of TJ contact, comparable to the control. Curcumin control groups showed no difference compared to the untreated controls (Supplementary Figure S2).

Discussion
It is important to contain the Campylobacter jejuni infection because of its impact on gastroenteritis worldwide and resulting sequelae, such as irritable bowel syndrome, reactive arthritis, or Guillian-Barré syndrome. Therefore, we aimed to study the underlying pathomechanisms for barrier dysfunction and the influence of the immune system in more detail. For this purpose, we used a new in vitro co-culture model and an in vivo mouse model to study the protective properties of the natural polyphenol curcumin.
Curcumin is a vitamin D receptor (VDR) ligand [24], and we found the downstream signaling pathway of curcumin to be inhibited in the mucosa of acutely infected C. jejuni patients by means of the bioinformatic prediction in Qiagen ingenuity pathway analysis (IPA) from RNA-sequencing (Supplementary Table S1). This prediction indicated that curcumin could be a potential therapeutic or preventive target against C. jejuni infections. Based on this bioinformatics prediction, we tested the effects of curcumin on C. jejuni infection in vitro and in vivo. The inhibitory effect of another top hit upstream regulator in IPA analysis on the mucosa of C. jejuni-infected patients was calcitriol (active form of vitamin D). Calcitriol exhibited a high significance value comparable to curcumin (Appendix A, Table S1). Annulment of C. jejuni-induced cytotoxicity after calcitriol treatment has been demonstrated in HT-29/B6 cells previously, supporting the view that the bacterium affects intestinal barrier function via a VDR-dependent pathway [6].
As predicted by the pathway analysis in acute human campylobacteriosis, curcumin was effective in our experiments. It could prevent the decrease in TER as well as the increase in fluorescein permeability in our co-culture model, indicating improved barrier function by curcumin in immune-mediated epithelial barrier dysfunction by C. jejuni.
Because we expected an essential contribution of the immune system, we established a co-culture model with basal infection. In this model, the C. jejuni-triggered immune response reduced the TER after 48 h. The apical bathing medium did not contain bacteria even after 48 h, which suggests that the effect on resistance was caused indirectly by the C. jejuni-induced immune response. The decrease in TER may therefore reflected an increase in paracellular permeability secondary to epithelial barrier defects, which would increase permeability to small molecules and toxins. Furthermore, this change in barrier function may increase the access of bacteria to the underlying tissue and potentiate the immune response, causing additional barrier disruption, with loss of cellular integrity [6,25]. Concomitantly, a leak-flux type of diarrhea is caused that represents a main diarrheal mechanism.
Depending on the cell type and the experimental infection conditions, C. jejuni was more or less able to modulate TER in previous studies. No effects on TER have been found in MKN-28 cells [26] or Caco-2 monolayers [4]. In contrast, exposure to C. jejuni has been reported to reduce TER in T84 and MDCK-I cells [27]. In our HT-29/B6-GR/MR cells, TER was also not different from control even after 48 h basal infection, although a decrease in TER after basolateral infection has been reported to be more pronounced than after apical infection in T84 cells [28]. That this was not seen in our HT-29/B6-GR/MR cells may reflect differences in cell type and/or the 0.4 µm filter pore size of our filters, limiting the bacterial access to the epithelial cells.
Within 6 days following peroral C. jejuni infection, secondary abiotic IL-10 −/− mice develop acute ulcerative enterocolitis mimicking key features of severe campylobacteriosis in humans, with a specific cytokine response [29]. We were able to show for the first time, using impedance spectroscopy in Ussing chambers, compromised barrier integrity with an increased epithelial conductance and permeability to macromolecules in this mouse model. These data reflect the results of our co-culture model, in which the barrier defects could be similarly prevented by curcumin. Bücker et al. [6] incubated colonic mucosal biopsies from C. jejuni-infected patients in culture medium and quantified cytokine release into the supernatant. Secretion of TNF-α, IFN-γ, IL-1β, and IL-13 was enhanced in comparison to controls [6]. It was concluded that this mucosal cytokine storm led to disruption of intestinal TJs and the induction of epithelial apoptosis. In our infected co-cultures, release of cytokines was also mainly responsible for the barrier defects, even though THP-1 cells produce and secrete a lower amount and spectrum of cytokines in comparison to fresh isolated peripheral blood mononuclear cells (PBMC). After C. jejuni infection, levels of the pro-inflammatory cytokine TNF-α, IL-1β, and IL-6 increased remarkably, and these changes were largely prevented by curcumin. Curcumin not only reduced local inflammation in the intestine, it also reduced systemic inflammation triggered by LPS [23]. In another study, curcumin nanoparticles suppressed the expression of mRNAs encoding pro-inflammatory mediators, including TNF-α, IL-1β, and IL-6 [14].
C. jejuni was found to activate the NFκB signaling pathway [28], which was suppressed by curcumin [12,17,19]. To determine the role of the NFκB pathway in maintaining barrier function, inhibition studies with the NFκB-inhibitor BAY 11-7082 were performed in C. jejuni-infected co-cultures. BAY 11-7082 treatment inhibited both the C. jejuni-induced decrease in TER and the increase in fluorescein permeability. This points to the involvement of NFκB in the regulation of epithelial barrier function. Similar results were obtained by our group previously with the bioactive ginger ingredient 6-shogaol after TNF-α stimulation in Caco-2 and HT-29/B6 cells [30].
In order to exclude antibacterial effects of curcumin in our study, the MIC was determined with 87 µM (at pH 7.4). This suggests that the protective effects of 50 µM curcumin were not based on its antibacterial properties. Thus, we conclude that curcumin had direct barrier-protective and anti-inflammatory properties in the host cells.
To further exclude interfering properties of curcumin with bacterial fitness and adaptive response, we investigated if curcumin changes natural transformation capacity. Therefore, we used a single cell DNA uptake assay and challenged the bacteria with 50 µM curcumin. C. jejuni is naturally competent for DNA uptake from the environment. Generally, this leads to genetic recombinations between bacterial strains, and as a consequence, to more diversity [37]. Curcumin did not influence the competence development and horizontal DNA uptake of C. jejuni in our experimental concentrations.
One of the major challenges to the development of therapeutic approaches is the remarkable diversity between different Campylobacter strains in animals, the environment, and food [2]. Consequently, the efficacy of curcumin should also be tested using other C. jejuni field strains and, subsequently, in other Campylobacter species such as C. coli or C. concisus. Curcumin is a natural agent, and should be more appropriate than synthetic drugs [12]. Curcumin has been shown to be beneficial in maintaining remission in ulcerative colitis patients when used as a complementary therapeutic substance together with the antiphlogistic compound mesalazine in several randomized clinical studies [38][39][40]. Since the immune response plays an important role in pathogenesis in IBD as well as in C. jejuni infection, curcumin may be equally effective in immune-induced barrier dysfunctions by C. jejuni as an add-on to other anti-inflammatory drugs. Indeed, multimodal therapies may have an even greater protective effect, and synergism has been reported between curcumin and quercetin [41], genistein [42], epigallocatechin-3-gallate [15,43], and antioxidants [12,44]. Thus, curcumin alone or in combination with other agents may be useful in preventing, treating, and combating Campylobacter infection in humans, C. jejuni-associated sequelae as well as colonization of farm animals and especially poultry.
Taken together, our results suggest that a significant part of the C. jejuni-induced barrier defect is mediated indirectly by the immune cells. Curcumin abolishes functional epithelial disorders as well as pro-inflammatory cytokine secretion. Further barrier-protective or anti-inflammatory compounds, as well as curcumin combinations with other therapies, should be tested in the co-culture model.

Epithelial Cell Culture and Differentiation of THP-1 Cells
The human colon carcinoma cell line HT-29/B6-GR/MR [45], a subclone of the HT-29/B6 line [46] was cultured in 25 cm 2 culture flasks in RPMI 1640 culture medium (Sigma Aldrich, St. Louis, MO, USA). HT-29/B6-GR/MR cells expressing the human glucocorticoid receptor α and the human mineralocorticoid receptor, were established earlier by stable transfection [45,47]. The culture medium was supplemented with 10% fetal calf serum (FCS; Gibco, Carlsbad, CA, USA), 1% penicillin/streptomycin (Corning, Wiesbaden, Germany), G418-BC (300 µg/mL; Invitrogen, Carlsbad, CA, USA) and hygromycin B (200 µg/mL; Biochrom GmbH, Berlin, Germany), and cells were passaged every week. Cells were seeded on Millicell PCF filters membranes (Merck Millipore, Billerica, MA, USA; 0.4 µm pore size and an effective growth area of 0.6 cm 2 ). The cell medium in the culture flask, as well as in the filters, were changed every other day. Experiments were performed between 7 and 9 days after seeding, when polarized cells formed confluent monolayers and the transepithelial electrical resistance was 600-900 Ω·cm 2 . These cells were chosen because of their stable growth behavior and the high resistance. The human monocyte leukemia cell line THP-1 (ATCC TIB-202) was provided by Verena Moos. THP-1 cells were cultured in RPMI 1640 supplemented with 10% heat-inactivated FCS and 1% penicillin/streptomycin. Cultures were maintained by the addition of fresh medium and were subcultured before reaching a density of 8 × 10 5 cells/mL once a week. THP-1 cells were re-suspended in antibiotic-free medium supplemented with phorbol 12-myristate 13-acetate (PMA; Sigma Aldrich, St. Louis, MO, USA; solved in DMSO) at a final concentration of 100 nM, seeded with a density of 1.8 × 10 5 in 12-well plates, and incubated for 24 h for differentiation to macrophages [48]. After incubation, PMA-containing medium was removed and the differentiated and adherent macrophages-like cells were cultured together with epithelial monolayers grown on filters in co-culture ( Figure 10). All cells were cultured at 37 • C in a humidified 5% CO 2 atmosphere. compounds, as well as curcumin combinations with other therapies, should be tested in the co-culture model.

Epithelial Cell Culture and Differentiation of THP-1 Cells
The human colon carcinoma cell line HT-29/B6-GR/MR [45], a subclone of the HT-29/B6 line [46] was cultured in 25 cm 2 culture flasks in RPMI 1640 culture medium (Sigma Aldrich, St. Louis, MO, USA). HT-29/B6-GR/MR cells expressing the human glucocorticoid receptor α and the human mineralocorticoid receptor, were established earlier by stable transfection [45,47]. The culture medium was supplemented with 10% fetal calf serum (FCS; Gibco, Carlsbad, CA, USA), 1% penicillin/streptomycin (Corning, Wiesbaden,Germany), G418-BC (300 µ g/mL; Invitrogen, Carlsbad, CA, USA) and hygromycin B (200 µ g/mL; Biochrom GmbH, Berlin, Germany), and cells were passaged every week. Cells were seeded on Millicell PCF filters membranes (Merck Millipore, Billerica, MA, USA; 0.4 µ m pore size and an effective growth area of 0.6 cm 2 ). The cell medium in the culture flask, as well as in the filters, were changed every other day. Experiments were performed between 7 and 9 days after seeding, when polarized cells formed confluent monolayers and the transepithelial electrical resistance was 600-900 Ω•cm 2 . These cells were chosen because of their stable growth behavior and the high resistance. The human monocyte leukemia cell line THP-1 (ATCC TIB-202) was provided by Verena Moos. THP-1 cells were cultured in RPMI 1640 supplemented with 10% heat-inactivated FCS and 1% penicillin/streptomycin. Cultures were maintained by the addition of fresh medium and were subcultured before reaching a density of 8 × 10 5 cells/mL once a week. THP-1 cells were re-suspended in antibiotic-free medium supplemented with phorbol 12-myristate 13-acetate (PMA; Sigma Aldrich, St. Louis, MO, USA; solved in DMSO) at a final concentration of 100 nM, seeded with a density of 1.8 × 10 5 in 12-well plates, and incubated for 24 h for differentiation to macrophages [48]. After incubation, PMA-containing medium was removed and the differentiated and adherent macrophages-like cells were cultured together with epithelial monolayers grown on filters in co-culture ( Figure 10). All cells were cultured at 37 °C in a humidified 5% CO2 atmosphere. Epithelial cells were then seeded on filter membranes and differentiated over 7-9 days to a polarized monolayer with tight junctions. 24 h before co-culture started, immune THP-1 cells were stimulated with phorbol 12-myristate 13-acetate (PMA) to allow a differentiation to adherent macrophage-like immune cells. After the differentiation of epithelial and immune cells, filters were placed into the 12-well plate with epithelial cells in the apical and immune cells in the basal compartment. Infection with C. jejuni was performed in the basal compartment.

Growth Conditions of C. jejuni, Treatment and Infection Procedure In Vitro
Campylobater jejuni wildtype (wt) 81-176 reference strain was pre-cultured on blood agar plates (Oxoid, Thermo Scientific, Waltham, MA, USA) under microaerobic conditions (5-10% O2, 10% CO2, Epithelial cells were then seeded on filter membranes and differentiated over 7-9 days to a polarized monolayer with tight junctions. 24 h before co-culture started, immune THP-1 cells were stimulated with phorbol 12-myristate 13-acetate (PMA) to allow a differentiation to adherent macrophage-like immune cells. After the differentiation of epithelial and immune cells, filters were placed into the 12-well plate with epithelial cells in the apical and immune cells in the basal compartment. Infection with C. jejuni was performed in the basal compartment.

Growth Conditions of C. jejuni, Treatment and Infection Procedure In Vitro
Campylobater jejuni wildtype (wt) 81-176 reference strain was pre-cultured on blood agar plates (Oxoid, Thermo Scientific, Waltham, MA, USA) under microaerobic conditions (5-10% O 2 , 10% CO 2 , 85% N 2 ) at 37 • C and re-cultured a second time before infection experiments. Microaerobic conditions were generated in a plastic jar with CampyGen gaspacks from Oxoid (Oxoid, Thermo Scientific, Waltham, MA, USA). Bacteria were then cultured in Mueller-Hinton broth for at least 2.5 h at 37 • C under microaerobic conditions, centrifuged (2 min, 5000 g, 10 • C), and re-suspended in the cell culture medium. The optical density at OD 600 was adjusted to 1 and the cells where infected from basal side with a multiplicity of infection (MOI) of 100. At least 1.5 h before the infection, cells were washed three times with antibiotic-free culture medium supplemented with 10% heat-inactivated FCS. To analyze the barrier-protective and anti-inflammatory properties of curcumin (Sigma Aldrich, St. Louis, MO, USA; final concentration 50 µM), different conditions were tested: control, C. jejuni infected, curcumin control (50 µM), and curcumin in combination with C. jejuni. Curcumin control showed no adverse effects on the cell viability of HT-29/B6-GR/MR monolayers at the concentration tested. Cells were pre-incubated with curcumin on both sides for 2 h, and infected with C. jejuni from the basal side for a direct immune cell infection and cytokine-induced barrier effect. Curcumin was added to the apical and basal side, since functional measurements depend on identical solutions on both the sides of the monolayer to avoid, e.g., potential osmotic effects. Moreover, curcumin readily accessed the basolateral compartment via C. jejuni-induced leaks. After infection, cells were incubated at 37 • C under microaerobic conditions favorable to the bacteria for 48 h.

Ethics Statement
The animal experiments were carried out in our animal facility according to the German animal protection law (LaGeSo Berlin; approval number G0172/16, 13th Oct. 2016).  (2.5), and the antibiotics piperacillin (50 mg/L) and imipenem (4 mg/L), and was equilibrated for 15 min with carbogen gas to a pH of 7.4. One-path impedance spectroscopy measurements were performed as described previously [51] to delineate between epithelial and subepithelial conductance (G).

Epithelial Permeability
For permeability measurements, unidirectional flux studies were conducted with the paracellular marker fluorescein (332 Da; 100 µM) from the mucosal to the serosal compartment, either directly in 12-well plates or under short circuit current (I SC ) conditions in Ussing chambers, as described earlier [7]. I SC was recorded under voltage clamp conditions by an automatic clamp device (CVC6, Fiebig Hard & Software, Berlin, Germany) at 37 • C over 1.5 h. Fluorescein was dissolved either in Ringer´s solution or in media. Samples were taken every 15 min for one hour from the basolateral side, and fluorescence was measured in a spectrophotometer (Tecan GmbH, Maennedorf, Switzerland). Permeability was calculated from flux over concentration difference.

Determination of Minimal Inhibitory Concentration Values
Determinations of minimal inhibitory concentration values (MICs) of Campylobacter strain 81-176 were performed in a broth microdilution assay. The wells of microtiter plates contained 2-fold serial dilutions of curcumin with a test range of 1-1024 µg/mL. For MIC determinations, inoculum level, growth medium, incubation time, and conditions were performed in accordance with the recommendations given in the Clinical and Laboratory Standards Institute (CLSI) document VET01-A4. In brief, Campylobacter colonies were taken from a blood agar plate and transferred into a tube containing 5 mL of 0.85% saline. The suspension was adjusted to the turbidity equivalent of 0.5 McFarland standard, and diluted 1:100 with growth medium. Subsequently, 50 µL of the diluted inoculum was added to each test well (50 µL) in the dilution series (containing substances at double the desired final concentration) and mixed. The microtiter plates were incubated for 24 to 48 h at 42 • C under microaerobic conditions to obtain sufficient growth. They were analyzed visually, and the lowest concentration preventing visible growth of bacteria was defined as the MIC. All susceptibility tests were performed in growth medium adjusted to pH of 7.4. Campylobacter jejuni reference strain DSM 4688 was used for quality control purposes. The MIC values of the quality control strain were determined in advance in three independent experiments using the broth microdilution and macrodilution method.
4.11. DNA-Uptake Assay C. jejuni strain 81-176 was streaked out from a −80 • C stock and subcultured on Columbia blood agar containing 5% sheep blood (Oxoid, Thermo Scientific, Waltham, MA, USA). Plates were incubated overnight under an atmosphere containing 3.5% H 2 , 6% O 2 , 7% CO 2 , rest N 2 at 37 • C. Cells were inoculated at OD ini~0 .3 in sterile-filtered brain heart infusion broth (BHI; Oxoid, Thermo Scientific, Waltham, MA, USA), and incubated under the same atmosphere at 37 • C for 6-9 h at 140 rpm. Liquid cultures were subcultured under the same conditions in 5 mL BHI with or without 50 µM curcumin. A suitable OD ini was chosen to reach optical densities of 0.1-0.6 after 17 h (± 3 h). 500 µL (± 300 µL) of the cell suspension was harvested by centrifugation (~16000× g for 5 min). The pellet was resuspended in 100 µL BHI with or without 50 µM curcumin. DNA labeling and uptake analysis were performed as previously described for H. pylori [52]. In short, C. jejuni BfR-CA-14430 genomic DNA was extracted using the PureLink Kit (Life Technologies, Thermo Scientific, Waltham, MA, USA), and labeled with fluorescein in a 1:1 (volume:weight) ratio of Label IT reagent to nucleic acid according to the manufacturer's protocol (Mirus Label IT Fluorescein, Mirus Bio LLC, Madison, WI, USA). One µL of labeled DNA (100 ng/µL) was added to the cell suspension and incubated at 5% O 2 , 10% CO 2 , rest N 2 for 30 min at 37 • C. Subsequently, cells were centrifuged (~16000× g for 5 min) and resuspended in 15 µL BHI supplemented with at least 3 U DNaseI (Roche AG, Basel, Switzerland). DNaseI digestion was performed at 37 • C for 5-10 min. The fluorescence microscope Axio Observer ZI (Zeiss, Jena, Germany) with a plan apochromatic 63x/1.4 objective and differential interference contrast (DIC) was used for the analysis of Campylobacter cells immobilized on 1.5% agarose pads. A metal halide light source (HXP120C) and a filter set with excitation at 470 ± 20 nm and emission at 525 ± 25 nm were used to visualize fluorescein. Exposing times varied between 250 ms and 750 ms. Images were taken by the 12-bit monochromatic AxioCam MRm camera. Cells with at least one fluorescent focus were considered active for DNA uptake, i.e., competent for natural transformation. For each condition, the fraction of competent cells was calculated and data are presented from three independent experiments.

Ingenuity Pathways Analysis
Ingenuity Pathways Analysis (IPA, Qiagen Silicon Valley, Reswood, CA, USA) was used to evaluate curcumin-dependent expression data from a dataset from human colon biopsies, generated, and analyzed previously [6].

Statistical Analysis
All data are expressed as mean values ± standard error of the mean (SEM). Statistical analyses were performed with GraphPad Prism (version 7.0, GraphPad Software, Inc., San Diego, CA, USA) using one-way ANOVA with Bonferroni adjustment for multiple comparison or unpaired Student´s t-test. For data that were not normally distributed, the nonparametric Mann-Whitney U-test was used. p < 0.05 was considered to be statistically significant. The funders had no influence in the study design, data collection, analysis, decision to publish, or preparation of the manuscript. We acknowledge support from the German Research Foundation (DFG) and the Open Access Publication Fund of the Charité-Universitätsmedizin Berlin.
Acknowledgments: In-Fah Maria Lee is gratefully acknowledged for her excellent technical support. We thank Geoffrey I. Sandle, (University of Leeds, Leeds, UK) for his critical discussion and support of the study.

Conflicts of Interest:
The authors declare no conflict of interest.