Human intestinal organoids can be cultured with CM from L-WRNH cells
In a previous study, we successfully established L cells that stably overexpressed Wnt3a, R-spondin1, and Noggin at each required concentration (8). The culture medium for human intestinal organoids was routinely prepared by adding 50 ng/mL human recombinant HGF to 4-fold diluted CM harvested from L-WRN cells, which enabled a proliferation rate similar to that of organoids cultured with WRN recombinant proteins. Although HGF is seemingly not essential for human intestinal organoid culture (6, 18), we previously found that it significantly increased organoid growth, as observed in primary human colon organoids (18), which would be beneficial for assays that require large numbers of organoids. Additionally, we showed that L-WRNH cells, whose culture supernatant contained approximately 200 ng/mL HGF, were established by infecting L-WRN cells with a human HGF lentivirus (8). However, we did not examine whether CM prepared from L-WRNH cells could be applied to human intestinal organoid culture for further cost reduction by replacing recombinant HGF in the culture medium.
We cultured hiPSOs with either 25% WRN or 25% WRNH CM and found that hiPSOs cultured with WRNH CM proliferated more prominently than those cultured with WRN CM, as determined by time-lapse microscopic observation (Fig. 1A) and viable cell numbers using CellTiter-Glo 3D reagents (Fig. 1B). Therefore, we compared the 25% WRNH CM with the conventional culture medium (25% WRN CM with 50 ng/mL recombinant HGF) from the perspective of hiPSO proliferation capacities. Time-lapse microscopy revealed that hiPSOs proliferated at similar rates in both media (Fig. 1C). Furthermore, we quantified the number of viable cells using CellTiter-Glo 3D reagents and found no difference between the two groups after 7 days of passage (Fig. 1D). Notably, CM prepared from intact L cells scarcely proliferated hiPSOs (Fig. 1A), showing that both FBS supplemented in the medium and endogenous secretions from intact L cells had little effect on organoid growth. Importantly, WRNH CM stored in a deep freezer for more than two years did not change the organoid proliferation capacity determined by time-lapse microscopic images (Fig. 1E) and viable cell numbers (Fig. 1F), indicating that WRNH CM can be stably stored for a long time. From these data, we conclude that WRNH CM is useful for expanding human intestinal organoids and contributes to their culture cost reduction.
Matrigel Can Be Replaced With Type I Collagen Gel For Human Intestinal Organoid Culture
In addition to changing the culture medium from WRN to WRNH CM, Matrigel was replaced with another ECM to embed organoids for 3D culture. Given that Matrigel is expensive and is derived from mouse sarcoma, which causes lot-to-lot variations and severely limits its use in clinical studies (19), an alternative candidate should be pure, xeno-free, and inexpensive. We selected type I collagen (18) and alginate hydrogels (17) as potential ECM for human intestinal organoid culture. Although the alginate hydrogel was much cheaper than the collagen gel, hiPSOs barely proliferated when the gel was used (Figure S1). Type I collagen gel for 3D culture is commercially available, and we chose three more inexpensive products than Matrigel, namely porcine tendon collagen (Nitta Gelatin), bovine dermis collagen (Koken), and bovine dermis atelocollagen (Koken), to compare their suitability for organoid culture. As observed in full-focused z-stack images, hiPSOs embedded in the porcine tendon collagen gel proliferated better than those embedded in the bovine dermis collagen gel (Figure S2). Atelocollagen is a low-immunogenic collagen derivative prepared by removing non-helical telopeptide regions of collagen by protease digestion (20). Although atelocollagen is somewhat cheaper than collagen, we found that the gel could not maintain stable dome structures during the culture and collapsed as organoids expanded (Figure S2), resulting in the loss of organoids before harvest. Therefore, we chose the type I collagen derived from the porcine tendon as a potentially suitable matrix for human intestinal organoid culture. As embedded in Matrigel, hiPSOs in the collagen gel proliferated more prominently with WRNH CM than with WRN CM (Figs. 2A and 2B), emphasizing the effectiveness of WRNH CM in the collagen scaffold. More importantly, time-lapse microscopy revealed that hiPSOs proliferated in the collagen gel at rates similar to those in Matrigel (Fig. 2C). Viable cell numbers determined by CellTiter-Glo 3D reagents were also found to be comparable between the two types of gel after 7 days of culture (Fig. 2D). Subsequently, we examined the mRNA levels of various intestinal epithelial genes using qRT-PCR. We found that the expression of the intestinal epithelial stem cell marker gene LGR5 (21), Paneth cell marker gene LYZ (6), goblet cell marker gene MUC2 (22), mature intestinal epithelial cell marker gene VIL1 (23), and intestinal gene that maintains IEC integrity HNF4A (24) was not significantly different between the organoids cultured with Matrigel and the collagen gel (Fig. 2E). Additionally, whole-mount immunostaining revealed that hiPSOs cultured with both Matrigel and the collagen gel contained proliferative cells (Ki67+), Paneth cells (lysozyme +), and goblet cells (mucin 2+) (Fig. 2F). Furthermore, we also confirmed the continuous growth of hiPSOs in the collagen gel over 10 passages with constant passage dilution rates. These results clearly indicate that the replacement of Matrigel with type I collagen gel did not significantly affect the organoid growth rate or fundamental intestinal marker gene expression, thus ensuring further cost reduction of organoid culture.
IECs can be developed from iPSOs and PIOs cultured with WRNH CM and collagen gel and can be maintained with WRNH CM
Despite their usefulness for examining physiological intestinal epithelium function and response, intestinal organoids may not be suitable for studies to evaluate the absorption and/or permeability of nutrients or drug candidates, especially those with high hydrophilicity, since the inside of the organoids corresponds to the intestinal lumen where digested food or xenobiotics physically interact. To overcome this issue, we previously established monolayer IECs from disrupted organoids by trypsinization, which allows exogenous factors to have direct access to the apical (luminal) side of the IECs (10, 25). Since the generation of monolayer IECs at low cost is considered to expand their applications in various assays, we attempted to develop IECs from hiPSOs cultured with WRNH CM and the type I collagen gel before propagating the IECs with WRNH CM. As a result, we confirmed the proliferation of hiPSO-derived IECs (Fig. 3A). Moreover, we previously showed that IECs can also be developed from hPIOs cultured with WRN CM (plus recombinant HGF) and Matrigel (11). We also confirmed the proliferation of hPIOs with WRNH CM and the collagen gel, as well as the proliferation of hPIO-derived IECs with WRNH CM (Fig. 3B). Furthermore, we examined whether WRNH CM-cultured IECs developed from hiPSOs propagated with collagen gel and WRNH CM exhibited characteristics similar to those of WRN CM plus recombinant HGF-cultured IECs developed from hiPSOs propagated with Matrigel, WRNH CM, and recombinant HGF. Gene expression analysis revealed that the mRNA levels of intestinal epithelial marker genes (LGR5, LYZ, MUC2, VIL1, and CDX2), as well as intestinal functional genes (HNF4A, MTTP, and CD36), were not statistically different between the IECs (Fig. 3C). These results indicate that physiologically relevant monolayer IECs can be developed from human intestinal organoids using WRNH CM and the type I collagen gel.
Organoid-derived Iec-selective Cytotoxic Compounds Were Identified By High-throughput Screening
We recently performed compound screening and discovered that compounds with selective cytotoxicity against Caco-2 cells exclusively include anticancer drugs, highlighting that the origin of Caco-2 cells is cancer cells (11). To further investigate the different biological responses between intestinal organoids and Caco-2 cells, we performed a large-scale culture of hiPSOs with WRNH CM and the type I collagen gel and conducted another screening of compounds that selectively induce cytotoxicity against hiPSOs from a library of approximately 3500 chemical compounds consisting of drugs and pharmacologically active substances which were provided by Drug Discovery Initiative in the University of Tokyo. Consistent with a previous procedure, the dispersed IECs of organoids were used instead of organoids or monolayered IECs to promote cell homogeneity, which is important for cell-based drug screening to minimize measurement variation (26). Furthermore, SN-38, a topoisomerase I inhibitor and active metabolite of irinotecan, was chosen as a positive control compound that causes cytotoxicity against IECs because it is known to exhibit intestinal toxicity in vivo (27). The cytotoxic effect of SN-38 was confirmed by cell viability assay using CellTiter-Glo 3D Reagent, and 1 µM of SN-38 was used to guarantee the assay quality of each plate (Fig. 4A).
The cytotoxic compounds against organoid-derived IECs were first screened at every single concentration of 2 µM in 0.2% dimethyl sulfoxide. The average Z′-factor of the screening exceeded 0.6, indicating that the assay was sufficiently robust to filter the compounds (Fig. 4B). One hundred sixty-seven compounds with > 40% inhibitory activity, which was a statistical cutoff of 4×standard deviation (SD) of control groups without compound treatment (only with 0.2% dimethyl sulfoxide), were chosen as initial hits (Fig. 4C). Next, the hit compounds were evaluated by a counter-assay using Caco-2 cells, and 62 compounds with < 40% inhibitory activity were selected as potentially selective cytotoxic compounds against organoid-derived IECs (Figure S3). Subsequently, a dose-dependent analysis of these compounds was performed using Caco-2 cells and IECs derived from hiPSOs and hPIOs, and compounds with IEC-selective dose-dependent cytotoxicity were identified.
The identified compounds comprised ABT-737, N-(4-hydroxyphenyl)retinamide (4-HPR), ivermectin, N-oleoyldopamine, and 3-(5′-hydroxymethyl-2′-furyl)-1-benzyl indazole (YC-1) (Fig. 4D). Although these compounds include antiproliferative agents, such as 4-HPR for breast cancer cells (28), ABT-737 for senescent cells (29), their effects on normal IECs have not been thoroughly investigated. These results revealed that drug responsiveness is significantly different between normal IECs and Caco-2 cells, which emphasizes the different physiological responses of intestinal organoids and Caco-2 cells.
YC-1 induces apoptosis through MEK (mitogen-activated protein kinase) /ERK (extracellular signal-regulated kinase) signaling in organoid-derived IEC
Among compounds with selective cytotoxicity against organoid-derived IEC, this study focused on YC-1, also called lificiguat, as this compound displayed least cytotoxicity against Caco-2 cells. YC-1 is known as an activator of soluble guanylyl cyclase (sGC) while, at the same time, an inhibitor of hypoxia-inducible factor-1α (HIF-1α) (30, 31). Previously, sGC activation was reported to inhibit human pulmonary arterial smooth muscle cell proliferation (32) and HIF-1α enhanced the proliferation of certain cells, including mesangial and renal carcinoma cells (33, 34). To elucidate whether the cytotoxic effect of YC-1 on IECs was mediated by sGC activation or HIF-1α inhibition, other chemotypes possessing the same actions as YC-1 namely, vericiguat as an sGC activator and LW6, PX-478, and PT-2385 as HIF-1α inhibitors, were used. Unexpectedly, these compounds failed to cause little cytotoxicity against IECs, suggesting that their cytotoxic effect on IECs was mediated by neither sGC nor HIF-1α (Fig. 5A and 5B). This conclusion is also supported by data showing that the expression of HIF1A and its target gene, VEGF, was comparable between monolayered IECs from hiPSOs and Caco-2 cells, regardless of their distinct sensitivity to YC-1 (Fig. 5C). Furthermore, we investigated whether YC-1 cytotoxicity was associated with the activation of apoptotic signals. YC-1 dose-dependently induced the cleavage of caspase-8 and its downstream pro-apoptotic proteins, RIP1, caspase-3, and PARP, together with decreased the anti-apoptotic protein, XIAP (Fig. 5D), indicating the induction of apoptosis. Consistent with a previous finding that YC-1-induced apoptosis accompanies the reduction of CCND1 expression (33), the mRNA levels of CCND1, as well as the levels of intestinal epithelial cell marker, VIL1, were found to be downregulated by the YC-1 treatment (Fig. 5E).
In previous studies, YC-1-induced apoptosis was mediated by the activation of the MEK/ERK (35), p38 MAPK (33), and Jun N-terminal kinase (JNK) (34) pathways in various cells. To further elucidate the pathway important for apoptosis induction in organoid-derived IECs, inhibitors of each kinase were used. Interestingly, while a p38 MAPK inhibitor (SB203580) and a JNK inhibitor (SP600125) had almost no effect, a MEK/ERK inhibitor (PD98059) substantially prevented the dose-dependent cytotoxicity induced by YC-1 (Fig. 6A, 6B, and 6C). Alternatively, PD98059 inhibited the cytotoxicity of 1 µM YC-1 in a dose-dependent manner (Fig. 6D). Finally, inhibition of YC-1 induced cytotoxicity by PD98059 was also confirmed in IECs from hiPSOs that were expanded with Matrigel (Fig. 6E), indicating that the difference of ECM composition had little effect on the cellular response to YC-1. Taken together, these results indicate that YC-1 caused cytotoxicity in organoid-derived IECs through the activation of MEK/ERK signaling.