*3.5. Prediction of F<sup>a</sup>* × *F<sup>g</sup> in Human using EpiIntestinal Microtissues*

Since most of the relevant drug transporters and DMEs are present in EpiIntestinal microtissues, we were interested in finding out whether this model could serve as an in vitro model for the prediction of GI firstpass availability of drugs in human (*F<sup>a</sup>* × *Fg*). For this purpose, we selected a panel of reference drugs with known human data and measured the recovery of these drugs in the basal compartment (equivalent to portal vene) after adding the drugs to the apical compartment (equivalent to GI lumen). The data are summarized in Table 6. We observed a good agreement between the recovery of the drugs in basal compartment of EpiIntestinal microtissues and the *F<sup>a</sup>* × *F<sup>g</sup>* in human.

**Table 6.** Comparison of GI firstpass availability measured in EpiIntestinal microtissues and *F<sup>a</sup>* × *F<sup>g</sup>* in human. GI firstpass availability in EpiIntestinal microtissues was determined as described in 2.7. *F<sup>a</sup>* × *F<sup>g</sup>* in human for the tested drugs was calculated from the clinical pharmacokinetic data, as described in Section 2.8.


#### **4. Discussion**

At present, the Caco-2 cell culture model is the most accepted in vitro model in the pharmaceutical industry for the estimation of drug absorption in human intestine. The lack of the major drug metabolizing enzyme CYP3A4, however, hampers the use of this model for the holistic understanding of drug absorption and metabolism during the first-pass GI transition. Recently, Ayehunie et al. described a new organotypic 3D human intestine model, the EpiIntestinal microtissues, with combined barrier/drug transproter functions and DME activities [11]. In this work, we could confirm the intact barrier function of the EpiIntestinal microtissues with transepithelial electrical resistance (TEER) measurement and with the permeability data of reference drugs and in-house compounds (data not shown). However, our data did not agree with the conclusion by Ayehunie et al. that rosuvastatin is a P-gp substrate. In our opinion, the discordance was mainly due to the different interpretation of the inhibition by Elacridar. Elacridar (GF120918) is a nonselective inhibitor for both BCRP [31] and P-gp [32]. At the concentration of 10 µM used by Ayehunie et al., a strong inhibition of both BCRP and P-gp can be expected. Thus, the inhibition of rosuvastatin efflux by Elacridar cannot be unequivocally attributed to P-gp inhibition. In contrast, the inhibitors we were using in this study are

selective [31,33]. Our in-house evaluation showed that, at the concentrations we were using here (5 µM zosuquidar and 3 µM Ko-143), differential inhibition of P-gp and BCRP can be achieved. As shown in Table 3, only Ko-143 reduced in Caco-2 and EpiIntestinal rosuvastatin efflux strongly. Involvement of multiple transporters has been reported in the hepatobiliary transport of rosuvastatin, including OATP1B1, OATP1B3, OATP2B1, MRP2 (ABCC2), MDR1 P-gp (ABCB1), and BCRP (ABCG2) [34]. Except OATP1B1 and OATB1B3, all other transporters are expressed also in human intestine and Caco-2 cells [35–37]. In the clinic, however, only BCRP interaction has been related to increased bioavailability of rosuvastatin [38]. Our results here are in agreement with the clinical observation.

EpiIntestinal microtissues are an improved in vitro model for the intestinal barrier function. A clear advantage of this model over the Caco-2 cellular model is the physiologically relevant activities of CYP3A4 (Figure 1 and Table 4), which accounts for about 80% of total CYP content in human small intestine [39]. We could also detect activities of CYP2B6, CYP2C8, CYP2C9, 2C19, 2D6 and 2J2 in EpiIntestinal microtissues (Table 4), as reported for human intestine [39]. CYP1A2 activity, which is very low in human intestine, were detected in both in vitro models, with Caco-2 showing seven-fold higher activity (Table 4), suggesting that EpiIntestinal microtissues are closer to human intestine. Moreover, enzymes involved in phase 2 biotransformation (UGTs and SULTs) and carboxylesterases are present in EpiIntestinal microtissues at substantial levels (Tables 4 and 5, Figure 3). It is important to note that we measured the respective metabolites of the tested reference drugs both in supernatant and in cell lysate. With the exception of amodiaquine and astemizole, the intracellular accumulation of metabolites was rather low. Since the intracellular accumulation of all measured metabolites is comparable in Caco-2 cells and in EpiIntestinal microtissues, we do not expect a bias in the relative comparison of enzyme activities between Caco-2 and EpiIntesitnal by measuring the metabolite in supernatant only (as shown in Table 4). In the case of carboxylesterase activities, we could demonstrate that EpiIntestinal microtissues are closer to human intestinal mucosa compared to Caco-2 cells, which resemble rather hepatocytes regarding relative CES1/CES2 activities. Because of the rather comprehensive expression of DMEs, EpiIntestinal microtissues can serve as a useful tool for the identification of intestine-specific metabolites, as we demonstrated with raloxifene and ezetimibe (Table 6). One surprising finding was the identification of sulfation of raloxifene not only in EpiIntestinal microtissue, but also in primary human intestinal mucosa. Mono- and diglucuronides were found in human plasma as major metabolites after oral administration; no other metabolites were identified (Prescription information for Evista, Eli Lilly). However, raloxifene was identified as a substrate for various SULTs and the sulfation of raloxifene occurs during incubation with cytosols from human liver and intestine [40,41] and in Caco-2 cells [42]. Moreover, raloxifene is reported to be a potent competitive inhibitor of sulfotransferase 2A1 (SULT2A1) with a K<sup>i</sup> value very similar to its K<sup>m</sup> value for SULT2A1 [41,43]. The sulfation of raloxifene we observed in EpiIntestinal microtissues was in line with the reported in vitro data in the literature and was obviously not due to a biased expression of SULTs in this model. A possible explanation for the missing raloxifene sulfate in human plasma could be a strong first-pass hepatic extraction and the subsequent excretion of the sulfate into bile.

Although cryopreserved primary human enterocytes and human intestinal mucosa are now available for the investigation of intestinal drug metabolism [44,45], the advantage of EpiIntestinal microtissues is the presence of both intact barrier function and comprehensive DME activities. The combined barrier function and DME activities in EpiIntestinal microtissues make it possible to evaluate intestinal first-pass availability (*F<sup>a</sup>* × *Fg*) in humans in a single experiment. Indeed, the in vitro intestinal availability of 12 marketed drugs in EpiIntestinal microtissues (% recovery in receiver compartment) is in good agreement with *F<sup>a</sup>* × *F<sup>g</sup>* calculated from the clinical pharmacokinetic data of these drugs (Table 6). It is important to note that the in vitro availability in our model was obtained after an incubation time of 24 h, while the drug absorption in human intestine is usually completed after a few hours. The longer incubation time in the in vitro model can be mainly attributed to the higher ratio of drug amounts applied to the microtissues (1 nmol) to the surface area of the microtissues (0.6 cm<sup>2</sup> ). The human small intestine mucosa, in contrast, has a surface area of 30 m<sup>2</sup> [46]. The ratio

of drug amounts to surface area is much lower. It would be interesting to compare human intestine tissues mounted in Ussing chambers with the EpiIntestinal microstissues in this regard. One would assume that the primary tissues would perform at least similarly to the EpiIntestinal model, and the low availability of suitable human tissues would limit the broader use of the primary materials in drug screening. There is however one caveat for using the EpiIntestinal model in this regard: the data are only meaningful if the quantities of DMEs and drug transporters in the model are comparable to the human intestine. Investigation into the expression of DMEs and drug transporters in EpiIntestinal microtissues is currently ongoing (transcriptomics) or planned (proteomics).

Although the EpiIntestinal microtissues provide a number of advantages compared to the currently available tools like Caco-2 cells, primary human enterocytes, or human intestinal mucosa, there are some limitations with regard to the use of the model in drug screening. One of the limitations is the unknown donor variability. According to the manufacturer, the microtissues we tested to date were derived from one single donor. For various reasons, we have not been able to get access to microtissues derived from other donors from the manufacturer to date. For drugs involving highly polymorphic metabolizing enzymes, data from a single donor are certainly not representative for the patient population. Therefore, it will be very important to investigate this model further in this regard in the future. Another limitation of this model is the static incubation conditions. Under physiological conditions, both the content in the intestine lumen and the blood at the basal side are under constant flow. The blood flow, for example, can reduce the diffusion of the drugs back into the enterocytes and thus limit the "recycling" of the drugs between blood and intestinal mucosa. Under static conditions, however, the recycling of the drugs might lead to an underestimation of the availability of the drugs, especially for those with extensive metabolic clearance in the intestinal mucosa. Due to the strong dilution effect in the apical-to-basal direction (100 vs. 5000 µL media volume in the apical and the basal compartment, respectively) we consider the effect of recycling, even under the static incubation, to be rather low. We tried to mimic the blood flow by replacing a large part of the media in the basal compartment with fresh media at the indicated sampling time points. The results were comparable to the static incubation (data not shown). Nevertheless, the integration of this model into a microfluidic system might still be interesting because this will make the combination with other organ models (e.g., liver-on-chip) possible.

In summary, our data here demonstrate that the EpiIntestinal microtissues are a useful tool for understanding drug absorption and metabolism in human intestine. The easy access of the model makes it very attractive for drug screening in the drug discovery process. It can also be used for the mechanistic understanding of intestinal drug–drug interaction or for the identification of intestine-specific metabolites.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1999-4923/12/5/405/s1, Figure S1: Concentration dependence of activities of drug-metabolising enzymes in EpiIntestinal microtissues.

**Author Contributions:** Conceptualization, Y.C., S.C., C.M.; methodology, Y.C., S.C., D.S., F.R., C.M.; investigation, Y.C., S.C., D.S., F.R.; writing—original draft preparation, Y.C., S.C.; writing—review and editing, Y.C., S.C., D.S., F.R., C.M.; visualization, Y.C., S.C., D.S.; supervision, Y.C., C.M.; project administration, Y.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors thanks Veronika Diesch, Samira Selman, Sarah Heine, Tilo Goletz, and Jesus Blanco-Santos for technical support and Thomas Ebner, Eva Ludwig-Schwellinger, Jens Borghardt, Achim Sauer, and Klaus Klinder for consulting and fruitful discussion.

**Conflicts of Interest:** The authors declare no conflict of interest. All authors are employees of Boehringer Ingelheim Pharma GmbH & Co. KG. The company had no role in the design, execution, interpretation, or writing of the study.
