Together with our microbiome, more than a milliard of bacteria hosted by the colon, the colonic epithelium not only participates in the absorption/resorption of food constituents but also in their metabolism (1, 2). Indeed, a crucial role of the colonic epithelium is to serve as barrier against xenobiotics. Besides, it hosts immune responses to support the elimination of toxins and pathogens. High cell turnover and constant renewal of the epithelial structures are necessary to fulfill its physiological roles (3–6). This is of relevance also for pathological conditions, such as colon carcinoma, since chronic diseases are often the result of a disrupted epithelium and are accompanied by a disbalance in metabolism and immune signaling (7, 8). Besides, foodborne xenobiotics and microbial metabolites are frequently reported to impact the intestinal epithelial homeostasis and possibly contribute to the genesis of chronic gastrointestinal tract (GIT)-related diseases (9, 10). Hence, metabolic competence is crucial to support detoxification.

A highly conserved signaling pathway involved in this respect is triggered by the activation of the aryl hydrocarbon receptor (AhR), which supports intestinal metabolic capacities, epithelial barrier function (11, 12), and affects colonic inflammatory processes (13), all crucial for intestinal homeostasis (14). In detail, the AhR pathway target genes are involved in Phase I- and Phase II-metabolism and encode for CYP1A1, CYP1A2, CYP1B1, NQ01, ALDH3A1, and UGT1A, and GSTA1 enzymes (15–18). Moreover, activation of the AhR was reported to participate in the maintenance of the intestinal barrier structure by regulating the cellular distribution and expression of tight junction proteins (14). Furthermore, AhR signaling acts on the immune response via multiple interactions with transcription factors known to mediate pro-inflammatory processes, such as the IL-1β-induced NF-κB pathway (19).

A balanced diet offers a great variety of nurturing chemicals contributing to a healthy lifestyle and GIT system. Ellagitannins are commonly occurring constituents of berries, nuts, and fruits, and the extracts of ellagitannin sources are used as dietary supplements (20, 21). Throughout their way through our GIT, ellagitannins are transformed into ellagic acid and further to a group of dibenzo-α-pyrones (DAPs), called urolithins, by the gut microbiome. These microbial metabolites are claimed to be essential for some of the health promoting bioactivities of ellagitannins (22, 23). Protective effects against GIT related diseases, including colorectal cancer, and inflammatory bowel disease were postulated for both, a directly supplemented, as well as in vivo generated DAP: urolithin A (Uro A, Figure 1A; 24). Uro A was suggested to restore the colonic epithelial barrier integrity in chronic inflammatory diseases (25). It was reported to activate the Phase I- and Phase II-metabolism driving AhR pathway, acts as an antioxidant via the Nrf2-ARE pathway, and exert anti-inflammatory activities (26, 27). Bioavailability of urolithins formed from ellagitannins is frequently reported (28–30). Next to Uro A, human metabolites of Uro A recently recovered in plasma following 4-week oral supplementations included Uro A-glucuronide (Uro A-GlcA) and Uro A-sulfate (Uro A-S; 31).

In addition to food constituents, harmful compounds can also be present in the diet. Fungal secondary metabolites produced by the fungal species, Alternaria and Fusarium are regularly detected to co-contaminate food and feed products (32, 33). Alternariol (AOH, Figure 1B), a small molecule produced by Alternaria alternata, is considered an “emerging mycotoxin”, that requires toxicological risk assessment beyond its reported toxic effects (34, 35). AOH was previously described to hold, amongst others, genotoxic, endocrine disruptive, pro-oxidative, and immunomodulatory potential (36–39). Furthermore, AOH was suggested to impact cell membrane properties relevant for the innate immune response (40). Structurally, AOH belongs to the chemical class of DAPs. Sharing structural similarities with Uro A, it can be hypothesized for both, to exert beneficial as well as detrimental effects to human health, opening the research question concerning their singular and combined mechanisms of action (41). Biomonitoring studies recurringly detect AOH in human urine samples (42, 43). Data on human metabolites in vivo are still scarce; however, oxidative metabolites produced in human microsomes include 2-, 4-, 8-, and 10-hydroxy (OH)-AOH (44). A recent animal study also recovered 4-OH-AOH in rat feces after 24 h of oral administration of AOH (45).

Fusarium spp. infesting mostly grains, are regularly reported to produce a B-type trichothecene secondary metabolite, deoxynivalenol (DON, Figure 1C; 46). DON, also known as vomitoxin, can cause acute intoxication symptoms affecting the GIT (47). Moreover, gastrointestinal health can be compromised by low chronic exposure to DON as well. In relation to its proteostatic potential, DON was described to modulate the absorption of nutrients (48), and impact on intestinal barrier homeostasis, for instance, by regulating proteins responsible for epithelial structure, leading to disruptions in the epithelial barrier integrity in a time-, and dose-dependent manner (49). Furthermore, DON is known to exert pro-inflammatory as well as immunomodulatory effects in vitro, depending on the exposure scenario (48, 50, 51). According to literature, upon ingestion, DON is absorbed, metabolized, and excreted in urine within 24 h. Next to free DON, the recovered human metabolites include DON-glucuronides and DON-sulfates (52–54).

In light of the occurrence of the compounds and their mechanisms of action, it can be postulated that exposure to Uro A, AOH, and DON could affect intestinal functionality at several levels. Thus, this study aimed to decipher combinatory interactions of the foodborne substances, such as Uro A, AOH, and DON, taking multiple co-exposure scenarios into account. As food ingestion leads to exposure of the colonic epithelium, differentiated Caco-2 cells were chosen as a model for the evaluation of the repercussions of the compounds on intestinal metabolic capacity, and epithelial structural integrity. Finally, possible interactions of Uro A, AOH, and DON in the presence of a pro-inflammatory stimulus were also taken into account.

The present study focusses on the impact of foodborne xenobiotics on colonic cell metabolic competence, epithelial barrier functions, and absorption patterns. In this context, the role of the AhR pathway in maintaining intestinal barrier integrity, and subsequently Phase I- and Phase II-metabolism were investigated. For higher relevance of the potential exposure scenarios, the activities of the ellagic acid metabolite, Uro A and the two mycotoxins, AOH and DON were examined as single compounds along with binary combinations. For the first time, to the best of our knowledge, this work provides insights on DON acting differently on the colonic epithelial barrier integrity and its metabolic capacity compared to Uro A and AOH; furthermore, DON dominates the absorptive and metabolizing behavior of the cells to binary exposures in this context. In contrast, the two DAPs Uro A and AOH dominate the sustainment of the epithelial structural integrity in Caco-2 cell monolayers, in particular in combinatory exposure scenarios.

Exogenous and endogenous small molecules characterized by a planar scaffold are known to activate the AhR pathway. This reflects on cellular metabolism (CYP enzymes), immune response, and epithelial homeostasis (structural proteins turnover) in the gut (76, 77). Colonic cancer cells were found to express higher levels of CYP1A1/1B1/2E1 and GST isoforms in comparison to a healthy colonic epithelium, suggesting this signature to be a useful marker for the prognosis of cancer progression (78). Hence, we investigated the potential of the microbial metabolite, Uro A, and the mycotoxins, AOH and DON to induce CYP1A1/1A2/1B1 enzyme activation via EROD assays (Figures 2A–D). A recent study reported AhR activating Alternaria toxins within complex mixtures to interact synergistically upon CYP 1 family activation (79). Nonetheless, the effect of the binary mixture of Uro A and AOH did not exceed the efficacy of single toxins (Figures 2A,B) in our Caco-2 cell model. In line with our results, Uro A was recently described as AhR ligand; however, it could either support or antagonize 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCCD) induced activation of the AhR, depending on the in vitro model used (26). In turn, exposures to mixtures with DON diminished the potencies of single DAPs in this respect (Figures 2A,B). However, when exposed to the DON human colonic metabolite, DON-3-S (67) instead of DON, Uro A and AOH could exert their EROD activating capacities, respectively (Figures 2C,D). Hence, the suppressive effect of DON could be directly linked to its capability to inhibit protein synthesis. This interpretation finds confirmation in the cell viability tests (Figures 3A–D). DON reduced NR uptake alone, as well as in combination with AOH. However, in cells additionally stimulated with IL-1β, DON led to a significant reduction in the lysosomal activity as single compound, and in binary mixtures with AOH and Uro A (Figures 3A,B), possibly potentiating cellular stress due to the inflammatory stimulus. Intriguingly, its human metabolite, DON-3-S displayed no such effects, neither alone, nor in binary mixtures (Figures 3C,D). The NR uptake assay captures the ability of the cells to maintain pH gradients, membrane homeostasis, and functionality of lysosomes (80). As such, this assay could highlight DON to exert slight, yet significant, cytotoxicity, an effect even more pronounced in further stressed cells, due to inflammatory triggers (Figure 3B). This is in line with previous reports on DON exerting cytotoxic effects in a dose-dependent manner in differentiated Caco-2 cells (56), as well as epidermal carcinoma cells A431 (81), and targeting lysosomal homeostasis (82). Nonetheless, the cytotoxic effect was shown to comply even more distinctly in proliferating cells, compared to differentiated ones like the Caco-2 cells (83). Considering a more physio-pathological context, mediators of inflammatory responses, such as the pro-inflammatory cytokine IL-1β, are known to trigger signaling cascades like the NF-κB pathway, which was reported to cross-talk with AhR signaling (84). Indeed, also in our experimental setup, Caco-2 monolayers stimulated by 25 ng/mL of IL-1β showed reduced EROD enzyme activity levels compared to reference activities in solvent controls (Figures 2B,D). Moreover, Uro A and AOH, previously reported for their immunomodulatory potential, readily induced CYP1A1/1A2/1B1 enzyme activity to similar extents compared to the respective control (Figures 2B,D). DON affected EROD enzyme activities slightly different after inflammatory stimulus compared to without inflammatory triggers. This should be interpreted under the premise, that it was described to rather exacerbate intestinal inflammatory responses involving the NF-κB pathway in Caco-2 cells and other intestinal in vitro models at similar concentration levels as applied here (51, 56, 85).

In malignant intestinal tissues, the AhR expression is reportedly predominant within the epithelial cells, as well as immune cells located in the lamina propria (86, 87). Hence, its importance in several GIT- related pathological conditions by maintaining tissue homeostasis is regularly demonstrated (88–90). TEER measurements presented in this study (Figures 4A,B) revealed the compounds of interest to exhibit time-dependent effects toward Caco-2 monolayer integrity. Exposure to the AhR positive control B[a]P as well as AOH and Uro A resulted in elevated TEER values. The trend towards increasing TEER for the two DAPs was sustained in their binary exposure scenario (Uro A + AOH) (Figures 4A,B). In line with our findings, a recent study described a dose-dependent increase in Caco-2 TEER values for Uro A. Besides, Uro A was described to counteract a TNF-α induced loss in epithelial barrier integrity in a HT-29/B6 co-culture model, which was accompanied by maintenance of basal levels or reduction of protein expression levels of claudin 1 and claudin 2, respectively, which were both upregulated upon TNF-α stimulation (91). However, DON significantly reduced TEER values, and could reduce the effects exerted by both DAPs, when co-incubated (Figure 4). Slightly enhanced LY permeability was observed after exposure to AOH and Uro A + AOH (Figure 5). DON combined with Uro A and AOH increased the relative amount of permeating LY. In line, DON was previously reported to disrupt epithelial barrier integrity in vitro (Caco-2 cells, clone: HTB-37) and in vivo (male B6C3F1 mice) at similar concentration levels already after 8–12 h. DON exposure reduced tight junction protein levels and altered their localization. Concomitantly increasing mRNA transcription for tight junction and scaffold proteins, this effect was linked to the protein biosynthesis inhibiting capacity of DON (92). Of note, recurring exposure to high levels of DON was associated with a detrimental loss of barrier integrity, which was accompanied by substantial passage of bacteria to the basolateral compartment in vitro (49). In our experimental setup, we observed both AOH and DON to reduce ZO-1 immunofluorescence signals (Figure 6). Even if data on AOH are limited, a similar concentration of DON (2 μM) was suggested to induce epithelial barrier disruption by the degradation of tight junction proteins via p38 MAPK and JNK signaling (93), p38 MAPK being recognized as one pathway triggered by low doses of DON contributing to its pro-inflammatory effects (94). Combinatory exposure to Uro A and AOH led to enhanced ZO-1 immunolocalization (Figure 6). This finding is supported by a recent report of Uro A to induce tight junction protein expression in several colonic in vitro and in vivo models preserving the intestinal barrier function against inflammatory stress. At the molecular level, this was linked to its AhR and Nrf-2 pathway activating potential (25). Concerning their binary mixture effects, Uro A seems to dominate the effect toward ZO-1 protein expression. ZO-1 is a TJ protein considered as a parameter for Caco-2 (HTB-37) differentiation status (73, 95). Conversely, AOH possibly dictates the overall impact on TEER and LY, both serving as measures of barrier integrity per se (Figures 4, 5). Of note, changes in TEER during ongoing differentiation post-confluency were recently connected to alterations in cell circumferential junctional length (lining of junctional proteins along the cell–cell contacts) as well, while permeability was not further altered (96). However, other proteins of the apical junctional complex (97) could be involved in the effects seen toward the barrier integrity by AOH (Figures 4, 5) as well, as it was described for Uro A (25). Co-exposure of CH-22 with AOH re-established the ZO-1 immunofluorescence signal of the cells to the level of the solvent control (Figure 6). This is accompanied by a restoration of the EROD assay signal to values similar to those of controls (Figures 2A,C). This strongly infers for the involvement of the AhR activation in the effects of the DAPs at the intestinal level. Even if CH-22 showed a distinct efficacy toward the endpoints investigating the epithelial barrier structural integrity (Figures 4–6), this could be partly explained by a complementary mechanism. The molecule was previously reported to counteract ROS production and lipid peroxidation (LP) accompanied by AhR and CYP1A1 activation (98). Elevated ROS and LP, however, were both associated with epithelial barrier disruption proceeding as either reduced TEER or enhanced permeability of Caco-2 monolayers (99). From this perspective, it is worth mentioning that both Uro A and AOH were previously reported to hold the potential to modulate cellular redox status (27, 100–102).

Altered intestinal epithelial structure may impact the route and extent of absorption or passage through the intestinal lineage of exogenous as well as endogenous chemicals (74). A recent in vitro study also suggested basolateral exposure to low concentrations of DON to disrupt epithelial tight junction proteins and further barrier functionality of enteroids (103). This highlights the importance of investigations on cellular uptake and efflux of foodborne substances. Apical exposure to Uro A, AOH, and DON as single substances and in binary mixtures resulted in a significant passage of all three compounds (Figures 7A–C). The recovery of all three substances of interest or their metabolites within the cell lysate compartment, implies that the compounds can be taken up via the intracellular route in our experimental settings (104). Uro A and DON were mostly recovered unmodified (Uro A: 68 ± 9% – 129 ± 25%; DON: 93 ± 3% – 123 ± 1%), whereas AOH recovery rates were lower, ranging between 1.0 and 1.4% (Figure 7). Generally, the binary mixture exposure scenarios led to higher recovery rates (apical and basolateral compartment together) for all three substances. In relevance for the metabolism of Uro A, a recent Uro A intervention study described the bioavailability of DAP in vivo, with Uro A and its Phase II-metabolites reaching significant plasma levels within 24 h after oral supplementation (105). We found both, Uro A-S and Uro A-GlcA in all three compartments (Supplementary Figure 3). Nonetheless, high recoveries of the parent compound (Figure 7A) leave space for speculations concerning the kinetics of Uro A metabolism in our cell model. However, in vivo exposure to Uro A or structurally similar urolithins is considered to be substantially influenced by the individual metabolic competence, or “metabotype,” yielding different levels of formed urolithins depending on the persons’ age (106). In contrast, most recovered analytes concerning AOH, were observed as Phase II-metabolites, AOH-GlcA, AOH-3-S, and its monomethyl-ether AME (107). This is in line with the previous report of AOH to be rapidly absorbed, metabolized to AOH-GlcA and AOH-S, and transported to the basolateral side to a substantial degree within 3 h (108). Besides the distinct time frames investigated, a recent study suggests AOH to highly bind to human serum albumin (109): this aspect needs to be considered, when interpreting the overall low recoveries for all AOH analogs. Furthermore, AOH is also suggested to interact with cholesterol-rich cell membrane domains, possibly contributing to the reduced recoveries as well (40). As previously described, AOH is metabolized to AOH-S and AOH-GlcA in Caco-2 cells (110). Crudo et al. recovered Phase II-metabolites in a time-dependent manner, reaching the highest levels at 24 h of toxin exposure. Intriguingly, co-incubation with Uro C, a structural analog of Uro A, reduced the production of both AOH metabolites, and the distribution of AOH and the metabolites between the apical and basolateral compartment. Conversely, the exposure of Uro A to AOH led to no substantial alterations in metabolic or distribution pattern in our experimental setup (Figure 7 and Supplementary Figures 1–4). Nonetheless, the binary mixture of Uro A + AOH led to reduced recoveries of Uro A-GlcA, yet in turn, no significant change in AOH-GlcA recovery was detected (Supplementary Figures 3, 4). Overall, the co-exposure of DON to the two DAPs increased the recovery rates of not only the respective parent compounds, but also of Phase II-metabolites of Uro A (Uro-A-S in particular) and AOH (AOH-3-S, Supplementary Figure 2). By inhibiting the activity induction of Phase I-metabolism enzymes, CYP1A1/1A2/1B1 by Uro A and AOH (Figure 2), it is possible to hypothesize that more parent compounds are available as such for recovery (Figure 7) or for Phase II metabolism (Supplementary Figures 2, 3). Finally, DON was recovered mostly as the parent compound in our exposure scenario, showing almost equal distributions between the apical and the basolateral compartment. When DON was co-incubated with AOH, however, significantly more DON was found in the basolateral media compared to DON exposure alone (Figure 7). A recent study in Caco-2 cells reported DON to be metabolized to DON-3-S within 24 h; however, DON was applied in higher concentrations, and DON-3-S was recovered after 24 h in this case (67). Therefore, considering a possible production of DON-3-S in our cell model, concentrations would be too low to be considered for any impact on the endpoints measured in the study at hand.

Transcriptional activation of AhR target genes implies the regulation of several Phase I- and Phase II-metabolism enzymes, including CYP1A1, among others (18). After 6 h of substance incubation, 5 μM of B[a]P readily induced CYP1A1 relative gene transcription, while neither Uro A, AOH, nor the binary combination showed inductive potential (Supplementary Figure 18). On the contrary, the two DAPs significantly reduced mRNA transcription levels compared to the SC. The effect of Uro A and AOH toward the expression of CYP1A1 did not correlate with the EROD activities, suggesting that the expression of other CYP isoforms may be increased. Indeed, the outcome of the EROD assay measurements results from contributions of several CYP isoforms; thus, it cannot be excluded, that the individual isoforms might retain distinctive response profiles (111, 112). Accordingly, both DAPs could potentially also induce the enzyme activity of other CYP 1 family isoforms, such as CYP1A2/1B1, which are concomitantly addressed by EROD measurements as well (113, 114). Indeed, selectivity for the individual isoforms can occur in tissue- and substance-specific manners, as it was previously reported for B[a]P (115). In differentiated Caco-2, the effect of B[a]P on CYP1A1 (Figure 8 and Supplementary Figure 18) could be attributed to a higher sensitivity and/or different kinetic of the positive control when compared to Uro A and AOH. Of note, after 48 h of incubation, B[a]P induced EROD enzyme activity was of a much higher magnitude in comparison to that triggered by Uro A and AOH (Figures 2A–D). Other studies described enhanced mRNA transcription levels of CYP1A1 for both DAPs after 24 h of incubation, in different in vitro models (25, 116, 117). This contributes to the interpretation that Uro A and AOH could have different activity profiles in comparison to B[a]P in our cell model. The immunofluorescence data (Figure 8) could support this perception. The CYP1A1 signal in cells incubated with the binary mixture of Uro A + AOH was significantly higher in comparison to AOH alone, a response which infers for the capability of the substances to affect this pathway, albeit with a different threshold of activation.

In sum, the sub-cytotoxic concentrations (Figure 3) of Uro A and AOH applied to the differentiated Caco-2 monolayer, yielded additive efficacy toward EROD enzyme activity (Figure 2), and monolayer integrity measured as TEER enhancements (Figure 4). While AOH rather reduced both ZO-1 as well as CYP1A1 immunofluorescence signal, Uro A partially antagonized these effects (Figures 6, 8). The activity profile of DON seemed centered on other molecular levers and dominated by the proteostatic effect. Based on the structural similarity of AOH and Uro A, it seems legitimate to postulate alike mechanisms of action (41). However, AOH is classified as mycotoxin, while Uro A is primarily described for its health benefits (118), as it is considered safe for use as a dietary supplement (119). In any case, supplementation of ellagitannins containing fruit extracts, in combination with the respective “metabotype” account for considerable exposure of the intestinal epithelium. Of note, urolithin precursors were reported for their potential to inhibit topoisomerase II catalytically at low concentrations, and although low concentrated Uro A and Uro B could not sustain this effect, knowledge on the impact of higher concentrations in this respect is lacking (120). Moreover, a few studies suggested Uro A to interfere with hormone-dependent signaling (121, 122), which is plausible, since DAPs structurally resemble endogenous hormones, such as 17-β-oestradiol (E2) for example. On this note, the highly complex AhR signaling is regularly reported to interfere with physiological and pathological processes on multiple levels, including hormone metabolism and hormone-receptor-signaling (123). Cross-talks with other transcription factors and downstream signaling cascades can yield varying outcomes, necessitating further assessment of substances interacting with the signaling of AhR and its subsequent effects on human health. These are suspected to go way beyond the metabolism of environmental pollutants to potential carcinogens, and affect the immune responses, hormonal cascades, and the epithelial integrity of barrier tissues, such as the colon (41, 84, 124, 125).

Taken together, this study reports for the first time, single and combinatory effects of Uro A, AOH, and DON on intestinal AhR activation and implications toward the epithelial structural integrity and metabolic capacities of colon cancer. CYP1A1/1A2/1B1 enzyme activating potential of Uro A and AOH was diminished by co-exposure to DON, an effect reversed by exchanging DON for its human metabolite DON-3-S. Combination of the two food-derived DAPs showed to rather support the colonic barrier integrity. However, AOH did not pair the activity of Uro A in the modulation of CYP1A1 and ZO-1 localization, suggesting that the molecular pathways of the two compounds must separate downstream the AhR pathway activation. Conversely, DON altered the intestinal structure and efflux of Uro A, AOH, and their metabolites to the apical or basolateral compartment, respectively. Thus, considering the human colonic tissue being regularly exposed to chemical mixtures, more in-depth investigations on foodborne small molecules and their interactive impact on intestinal health and absorptive/metabolic capacities is required to support the risk assessment and ensure food safety.

The original contributions presented in this study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.

JG, GDF, and DM: conceptualization. JG and DM: data curation and project administration. JG and CS: formal analysis. JG: investigation, visualization, and writing—original draft. JG, GDF, EV, and DM: methodology. DM and GDF: resources. EV, GDF, and DM: supervision. JG, CS, EV, GDF, and DM: validation. EV, CS, GDF, and DM: writing, review and editing. All authors have read and agreed to the published version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.