CDX2-induced intestinal metaplasia in human gastric organoids derived from induced pluripotent stem cells

Summary Intestinal metaplasia is related to gastric carcinogenesis. Previous studies have suggested the important role of CDX2 in intestinal metaplasia, and several reports have shown that the overexpression of CDX2 in mouse gastric mucosa caused intestinal metaplasia. However, no study has examined the induction of intestinal metaplasia using human gastric mucosa. In the present study, to produce an intestinal metaplasia model in human gastric mucosa in vitro, we differentiated human-induced pluripotent stem cells (hiPSC) to gastric organoids, followed by the overexpression of CDX2 using a tet-on system. The overexpression of CDX2 induced, although not completely, intestinal phenotypes and the enhanced expression of many, but not all, intestinal genes and previously reported intestinal metaplasia-related genes in the gastric organoids. This model can help clarify the mechanisms underlying intestinal metaplasia and carcinogenesis in human gastric mucosa and develop therapies to restitute precursor conditions of gastric cancer to normal mucosa.


INTRODUCTION
Gastric intestinal metaplasia frequently accompanies intestinal-type gastric cancer (Correa, 1992;Correa and Shiao, 1994;Plummer et al., 2015). Intestinal metaplasia has been regarded as a precancerous lesion in the Helicobacter pylori-induced metaplasia-dysplasia-carcinoma sequence (de Vries et al., 2008). However, some researchers have argued that intestinal metaplasia is a para-cancerous lesion, because more than two-thirds of microscopic gastric cancer do not have the intestinal phenotype (Kawachi et al., 2003) and because of inconsistencies in the phenotype expression of mucin between gastric cancers and the surrounding mucosa (Hattori, 1986). Regardless of whether gastric intestinal metaplasia is a precancerous or para-cancerous lesion, understanding the molecular mechanisms of intestinal metaplasia is important for clarifying the details of gastric carcinogenesis.
CDX2-an intestinal specific homeobox gene-has been suggested to play a crucial role in intestinal metaplasia as well as in the development and maintenance of the intestinal mucosa phenotype (Beck et al., 1999;Chen et al., 2021). The expression of CDX2 as well as CDX1 is confined to the posterior gut endoderm during later development and after birth (Guo et al., 2004). Cdx2 heterozygous knockout mice develop multiple intestinal polyp-like lesions that do not express Cdx2 and contain areas of squamous metaplasia (Grainger et al., 2013). Indeed, CDX2 and CDX1 are expressed in human gastric epithelial metaplasia (Almeida et al., 2003). Furthermore, several groups have shown that transgenic expression of CDX2 in mouse gastric mucosa results in intestinal metaplasia (Mutoh et al., 2002;Silberg et al., 2002). However, no intestinal metaplasia model of human gastric mucosa has been reported because transgenic experiments in humans are impossible. To resolve this issue, a range of human pluripotent stem cell differentiation technologies have been developed over the past decade, thereby enabling the creation of various types of human tissues in vitroalso referred to as ''organoids''-including gastric mucosa, which can be used in transgenic experiments (McCracken et al., , 2017.
In this study, we tried to establish a human gastric intestinal metaplasia model by overexpressing CDX2 in gastric organoids derived from pluripotent stem cells and achieved incomplete metaplasia. This model can

RESULTS
Validation of a human-induced pluripotent stem cell (hiPSC) line to differentiate into gastric organoids In this study, we used a hiPSC line FF-PB-3AB4 generated from a healthy donor's peripheral blood mononuclear cells (PBMCs) using episomal plasmid vectors under xeno-free and feeder-free conditions (Suzuki et al., 2019). FF-PB-3AB4 showed hES cell-like morphologies, just like the reference iPSC line 201B7 that was cultured under the same conditions ( Figure S1A), and expressed the pluripotent markers OCT3/4, SOX2, and NANOG at the mRNA ( Figure S1B) and protein levels ( Figure S1C). FF-PB-3AB4 successfully differentiated into three germ layers ( Figure S1D) in vitro and had a normal karyotype ( Figure S1E).
Next, we examined whether or not this iPSC line could differentiate into three-dimensional antral gastric organoids using a previously reported in vitro culture system (McCracken et al., , 2017. Phase contrast microscopy showed sequential morphological changes from undifferentiated iPSC to definitive endoderm (DE), foregut, and gastric organoids ( Figure S2A). A quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis showed that the endoderm markers SOX17 and FOXA2 were upregulated at the early stages; however, the foregut marker-SOX2-was continuously expressed but downregulated in later stages ( Figure S2B). The gastric pyloric epithelial cell marker-PDX1-and gastric surface mucous cell marker-MUC5AC-were upregulated at the late stages ( Figure S2B). The morphology of gastric organoids at day 61 is shown in Figure S3A. On HE staining, gastric epithelial cellular composition such as foveolar cells, parietal cells, chief cells, and neck mucous cells was unclear ( Figure S3B top panels). However, immunostaining clearly showed that most cells were positive for MUC5AC-a gastric foveolar cell marker ( Figures S3B and S4). Furthermore, we observed a few cells positive for H,K-ATPase (ATP4A) (Figure S5A)-a parietal cell marker-and for the endocrine cell markers-Somatostatin (SST), Synaptophysin (SYP), and Chromogranin A (CHGA) ( Figure S5B). We identified cells with obvious expression of SOX2 and PDX1 but no expression of the intestinal markers CDX2 or MUC2 on an immunohistological analysis ( Figures S3B and S4).

Generation of hiPSC with drug-inducible CDX2 expression
To induce the forced expression of CDX2 in the gastric organoids from hiPSC, we constructed the DOXinducible PB-transposon plasmid PB-TAC-CDX2-ERN ( Figure 1A) and introduced it with pCAG-PBase into the parental iPSC line FF-PB-3AB4. We then isolated the subclone that expressed mCherry after DOX treatment, subsequently referred to as ''CDX2-iPSC.'' Similar to the parental iPSC, CDX2-iPSC showed hES cell-like morphologies ( Figure 1B) and expressed the undifferentiated markers OCT3/4, SOX2, and NANOG at the mRNA ( Figure 1C) and protein levels ( Figure 1D).
We confirmed that almost all of the CDX2-iPSC expressed mCherry by adding 1 mM of Doxycycline (DOX) for 24 h (Figure 2A). The upregulation of CDX2 mRNA and protein was detected by RT-PCR ( Figure 2B) and Western blotting ( Figure 2C), respectively. Immunostaining showed that CDX2 was co-expressed with mCherry in CDX2-iPSC with DOX but not in parental iPSC or CDX2-iPSC without DOX ( Figure 2D).

Drug-inducible CDX2 expression in gastric organoids from CDX2-iPSC
Next, we established a system for drug-inducible forced expression of CDX2 in gastric epithelium derived from hiPSC ( Figure 3A). Consistent with gastric differentiation from the parental iPSC line, the differentiation progeny of CDX2-iPSC formed gastric organoids and expressed the gastric epithelial marker MUC5AC ( Figures 3B and S6A). Because the expression of MUC5AC was upregulated at approximately 30 days in the parental iPSC-derived organoids ( Figure S2B), we decided to start adding DOX after day 35.
We then tested whether or not DOX treatment upregulates the expression of CDX2 in the hiPSC-derived gastric organoids. The fluorescence of mCherry protein ( Figure 3C) and the enhanced expression of CDX2 mRNA ( Figure 3D) were detected in gastric organoids treated with 1 mM of DOX for 9 days but not in organoids without DOX. Furthermore, immunohistochemistry showed that CDX2 was co-expressed with mCherry protein by adding DOX (Figures 3E and S6B). Taken together, these findings indicate that we successfully induced the forced expression of CDX2 upon DOX addition in hiPSC-derived gastric organoids. When we treated the iPSC-derived organoids with DOX, phase contrast microscopy revealed the budding of crypt-like domains ( Figures 4A and S7A), which have been found in mouse and human intestinal organoids, but not gastric organoids derived from in vivo tissues (Barker et al., 2010;Fordham et al., 2013;Jung et al., 2011;Sato et al., 2009).
We evaluated the expression patterns of CK7 and CK20 in the epithelial cells in the organoids. The cells facing the lumen were considered to be epithelial cells, as all such cells were positive for E-cadherin ( Figures 4B and S8A lower panels). The epithelium in DOX(À) organoids contained CK7-positive and CK7-negative cells ( Figures 4B and S8A upper left panels), which was compatible with the fact that fetal (C) An RT-PCR analysis showed that CDX2-iPSC maintained the mRNA expression of the pluripotent markers OCT3/4, SOX2, and NANOG. GAPDH was used as an endogenous control. RT: reverse transcriptase. (D) Immunostaining showed that CDX2-iPSC (upper panels) as well as Parental-iPSC (lower panels) expressed the pluripotency markers NANOG, OCT3/4, and SOX2. The nuclei were stained blue with Hoechst33342. Scale bars, 50 mm.

OPEN ACCESS
iScience 25, 104314, May 20, 2022 3 iScience Article gastric epithelium at a certain point in time contains simultaneously CK7-positive and CK7-negative cells (Kirchner et al., 2001). Notably, in contrast to all epithelial cells in the DOX(À) organoids being negative for CK20, CK20-positive epithelial cells appeared in DOX(+) organoids, which contained CK7-positive and CK7-negative cells ( Figures 4B and S8A upper and middle right panels). A quantitative immunohistochemical analysis of three independent experiments revealed a statistically significant emergence of CK20positive cells in DOX (+) organoids ( Figure S8B). The immunophenotype of gastric intestinal metaplasia has been known to be CK7(À)/CK20(+) or CK7(+)/CK20(+) (Ormsby et al., 1999; Couvelard et al., 2001), and our iScience Article results suggested that intestinal metaplasia could be induced in the DOX(+) organoids ( Figures 4B, S6C, and S8A). The enhanced expression of CK20 in DOX(+) organoids was also confirmed with qRT-PCR (Figure 4C). An RNA-seq analysis showed that the forced expression of CDX2 resulted in the statistically significant upregulation of CK20 as well as other intestinal markers (CDH17, MUC13) and a tendency toward the upregulation of another intestinal marker, CDX1. In addition, the statistically significant downregulation of some stomach markers (CLDN18, TFF2, and MUC5AC) and a tendency toward the downregulation of other stomach markers (MUC1, GIF, and CK7) ( Figure 4D) were also noted. Furthermore, we detected cells expressing the intestinal marker MUC2 in DOX(+) organoids but not in DOX(À) organoids in immunofluorescence analyses in all four independent induction experiments ( Figures 4E, S6D, and S7B). MUC2positive cells were clustered in the crypt-like area of the DOX(+) organoid ( Figure S7B). In contrast, the DOX(À) organoids showed neither such a structure nor any MUC2-positive cells. These results revealed a significant emergence of MUC2-positive cells in DOX(+) organoids (chi-square test, p < 0.05). iScience Article Cells expressing E-cad, which is an indicator of epithelial cells, in DOX(À) organoids expressed MUC5AC, indicating a gastric epithelial phenotype ( Figure 4F upper panels and Figure S6E). In contrast, mCherry (+) epithelial cells did not express MUC5AC, although all mCherry (À) epithelial cells expressed MUC5AC in DOX(+) organoids ( Figure 4F lower panels). A quantitative immunofluorescence analysis of three independent experiments revealed that the expression of MUC5AC was significantly disappeared in mCherry-positive cells, i.e., CDX2-overexpressing cells of DOX(+) organoids ( Figure S8C).
Effect of CDX2 forced expression on the genome-wide gene expression profile of hiPSCderived gastric organoids We compared the genome-wide gene expression patterns between hiPSC-derived gastric organoids with and without the forced expression of the CDX2 gene using RNA-seq analysis. To assess the induction of the intestinal gene expression pattern, we employed two gene lists: a list consisting of genes highly expressed in the normal intestine compared to the normal stomach (''Intestinal Genes'') and a publicly available list of intestinal metaplasia marker genes (''IM Genes'') obtained by a microarray analysis of microdissected human intestinal metaplasia tissues (Lee et al., 2010). To create the first list, we analyzed the previously reported transcriptome data of human tissues (Fagerberg et al., 2014) (see the STAR Methods section).
To confirm the reliability of our experiments, we first performed a clustering analysis in ''All Genes,'' ''Intestinal Genes,'' and ''IM Genes.'' According to this analysis, the six samples of the organoids generated among the three independent experiments could be clearly divided into two groups: one consisting of only DOX(À) samples and the other consisting of only DOX(+) samples, indicating the reproducibility of our experimental system ( Figure S9).
In RNA-seq analysis of whole organoid samples, no obvious difference in the expression of SOX2, a gastric epithelial transcription factor found between the DOX(+) and DOX(À) organoids (fold change: 0.77-1.24) ( Figure 6A, right panel); however immunofluorescence demonstrated the disappearance of SOX2 expression on the CDX2-overexpressing epithelium, indicated by the expression of E-cadherin in DOX(+) organoids ( Figures 6B and S6F). This finding suggested that the expression of CDX2 suppresses the expression of SOX2 on the gastric epithelium.
GATA4 is a transcription factor expressed in both stomach and intestine tissues (Uhlen et al., 2015) and was reported to regulate proliferation in the early developing intestine (Kohlnhofer et al., 2016) and to control intestinal crypt cell replication in conjunction with CDX2 (San Roman et al., 2015). Consistent with these previous findings, GATA4 expression was widely observed in stomach tissues, the DOX (À) gastric organoids and the DOX (+) organoids ( Figure S11). Several animal models of gastric intestinal metaplasia (Honda et al., 1998;Mutoh et al., 2002;Silberg et al., 2002;Zheng et al., 2004) and the induction of the intestinal phenotype in human gastric cancer cell lines and an immortalized gastric epithelial cell line (Fujii et al., 2012) have been reported. However, an intestinal metaplasia model using human gastric epithelial tissues has yet to be established. Species differences in the pathogenesis of various diseases remain a subject of debate (Nature Medicine Editorial, 2013; Seok et al., 2013;Takao and Miyakawa, 2015). In addition, in H. pylori-related gastric carcinogenesis, not only the direct pathogenicity of H. pylori against gastric epithelial cells but also its indirect pathogenicity via immune-mediated inflammation is important (Chiba et al., 2006), and our in vitro system might be able to mimic various inflammatory microenvironments around gastric epithelial tissue by adding . Therefore, it is still important to study the pathogenesis of IM and ways to restore the gastric mucosa. The model established in this study can help clarify the molecular mechanisms underlying gastric carcinogenesis, leading to the development of novel therapies and/or prevention strategies for gastric cancer.
Our present model of human gastric intestinal metaplasia exhibited the downregulation of SOX2, a ''master transcription factor'' that determines the cell identity for the upper gastrointestinal tract (Que et al., 2007). Metaplasia is generally understood to be the consequence of a change in the expression of the master transcription factor for one tissue to that of another (Slack, 2009); in the case of gastric intestinal metaplasia, this involves changing from SOX2 to CDX2. Indeed, several reports have shown that the SOX2 expression decreased under conditions of intestinal metaplasia (Niu et al., 2017;Tsukamoto et al., 2005Tsukamoto et al., , 2006, although another report argued that the Sox2 expression increased in the intestinal metaplastic mucosa of Cdx2-transgenic mouse stomach (Mutoh et al., 2011). In the gastric epithelium, H. pylori infection was reported to cause upregulation of the CDX2 expression concomitantly with SOX2 downregulation (Chen et al., 2020), but the data in our present study suggest that the enhanced expression of CDX2 in the gastric epithelium may suppress the expression of SOX2. Consistent with our findings, an inverse correlation between SOX2 and CDX2 was recently reported in gastric cancer and colorectal cancer (Helal et al., iScience Article 2020; Lopes et al., 2020), suggesting the mutual suppression of the expression of SOX2 and CDX2. Of note, the endogenous CDX2 gene was not upregulated in our model or in the Cdx2-transgenic mouse model (Mutoh et al., 2009), indicating that neither model has the ability to recapitulate the changes from the gastric transcriptional network to the autonomous intestinal transcriptional network. Moreover, this RNA-seq analysis showed the significant upregulation of CDX1. A previous report argued that the expression of CDX1 plays an important role in intestinal metaplasia in the gastric epithelium (Mutoh et al., 2004). In the case of the generation of iPSC from gastric epithelial cells, the expression of transgenes are only required for induction but not for the maintenance of a pluripotent state (Aoi et al., 2008). This suggests that gastric epithelial cells can be changed into other types of cells, including intestinal cells, in which a master transcriptional network is established and thereafter maintained without the persistent expression of transgenes.
The genome-wide transcription analysis in the present study showed that >85% of ''Intestinal Genes'' were upregulated in the gastric organoids by the overexpression of CDX2.
Notably, the proportion of CDX2-upregulated genes among ''IM Genes'' was significantly less than that among ''Intestinal Genes.'' These results suggested two hypotheses: (1) IM, which is clinically observed and which is an important issue in the context of gastric carcinogenesis, is a different phenomenon from mere ''fate conversion of stomach into intestine,'' (2) the forced expression of CDX2 alone is not enough to induce bona-fide intestinal metaplasia in the gastric epithelium, and some other critical trigger is required. Based upon the system that we established in this study and hypothesis, we would be able to unravel the suppressive mechanisms of cell fate conversion and the reasons for the failure of the mechanisms. Furthermore, the overexpression of some additional factors or different culture conditions may enable us to establish a model that fully recapitulates the cell fate change from a gastric identity to an intestinal one.

Limitations of the study
In contrast to previously reported mouse intestinal metaplasia models (Silberg et al., 2002), our present model did not show the obvious upregulation of several conventional markers for intestinal metaplasia, such as MUC2 and villin (Park et al., 2010;Reis et al., 1999), in RNA-seq analysis, and only a few MUC2-positive cells were found in immunofluorescence. The phenotype of the stomach in previously reported Cdx2transgenic mouse models was analyzed at 1 to 15 weeks of age (Silberg et al., 2002), whereas we analyzed the phenotype 5 to 12 days after CDX2 overexpression in the present study. This difference in the period of assessing CDX2 overexpression may have resulted in the observed differences in the marker gene expression pattern. We applied gastric organoid culture medium , but not intestinal medium, to both DOX(À) and DOX(+) organoids in the present study. Our present results are consistent with those of a previous report on mouse tissue-derived gastric organoids artificially expressing CDX2, which were cultured for more than 35 days under gastric organoid culture conditions (Simmini et al., 2014). Crafting intestinal organoid culture conditions containing Noggin, R-Spondin, and Wnt3a (Fordham et al., 2013;Sato et al., 2009) might enable the maintenance of CDX2-overexpressing organoids for an extended period of time, thereby allowing us to obtain a complete intestinal phenotype.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

ACKNOWLEDGMENTS
We thank all members of our laboratory for their scientific comments and valuable discussion and all staff of Kobe University Hospital Advanced Tissue Staining Center for the tissue staining. We also thank Yukari Takatani and Yoko Matsuoka for administrative support. We thank Dr. Knut Woltjen at Kyoto University for providing the plasmids PB-TAC-ERN and pCAG-PBase.

AUTHOR CONTRIBUTIONS
T.K., M.K.A., Y.K., and T.A. designed the study, analyzed the data, prepared the figures, and wrote the manuscript. T.K. performed the vector construction and cell and organoid culture experiments. T.K. and K.U. performed the immunohistological analyses. All authors read, contributed to, and approved the final manuscript.

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Takashi Aoi (takaaoi@med.kobe-u.ac.jp).

Materials availability
This study did not generate new unique reagents.
Data and code availability d RNA-seq data have been deposited at GEO and are publicly available as of the date of publication.
d This paper does not report original code.
d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon reasonable request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS iPSC culture
The validated iPSC line 201B7 was purchased from Riken Cell bank (Tsukuba, Japan) and transferred from on-feeder to feeder-free conditions in our laboratory. We cultured the iPSC lines according to a previously described method (Nakagawa et al., 2014). In brief, the culture plates were precoated with iMatrix-511 (0.5 mg/cm 2 ), and the iPSC were maintained in Stem Fit medium (Ajinomoto) with penicillin (100 units/ mL) and streptomycin (100 mg/mL; Life Technologies, MA, USA) at 37 C with 5% CO 2 . The medium was changed every other day and passaged every 7-10 days using 0.53 TrypLE Select (13 TrypLE Select diluted 1:1 with 0.5 mM EDTA/PBS [-]; Life Technologies) and Rho-associated kinase (Rock) inhibitor (Y-27632; WAKO, Osaka, Japan).

Vector construction and generation of CDX2-iPSC
The cDNA encoding human CDX2 open reading frame (ORF) was amplified by polymerase chain reaction (PCR) using the forward primer 5 0 -CACCATGTACGTGAGCTACCTCCTGGACAAGGAC-3 0 and reverse primer 5 0 -TCACTGGGTGACGGTGGGGTTTAGCACCCCCCCAGTTG-3 0 , and the resulting PCR product was cloned into pENTR/D-TOPO (Life Technologies) to generate pENTR-CDX2 clone according to the manufacturer's protocol. LR clonase II (Life Technologies) recombination was then performed using a pENTR-CDX2 clone and the destination vector PB-TAC-ERN (Kim et al., 2016) to generate PB-TAC-CDX2-ERN.
To generate CDX2-iPSC, the dissociated single cells of FF-PB-3AB4 were seeded onto an iMatrix-511coated 6-well plate at a density of 5 3 10 5 cells/well. The next day, the cells were transfected with 1.5 mg of pCAG-PBase (Kim et al., 2016) and 1.5 mg of PB-TAC-CDX2-ERN using Fugene HD (Promega, WI, USA). Forty-eight hours after transfection, 100 mg/mL G418 (Nacalai Tesque, Kyoto, Japan) was added to select transduced cells for 4 days. After 8 hiPSC colonies were isolated and expanded, the subclone CDX2-iPSC that strongly expressed mCherry by DOX addition (1 mM for 3 days) was selected and used in this paper. G418 (100 mg/mL) was continuously added during the maintenance culture of CDX2-iPSC before differentiation into gastric organoids.  Figures 2D, 3B, 3E, 4E, 4F, 6B, S7B and S11, representative data of two or three independent experiments are shown.

In vitro spontaneous differentiation via embryoid body formation
For embryoid body (EB) formation, undifferentiated iPSC were dissociated into single cells, resuspended in Primate ES medium (Reprocell) containing 20 mM Rock inhibitor Y-27632 (WAKO) and seeded on low-celladhesion 96-well spindle-bottom plates (PrimeSurface, Sumitomo Bakelite, MS-9096M; Tokyo, Japan) at 1 3 10 4 cells per well. After 7 days of culture, the EBs were transferred to gelatin-coated 24-well plates and cultured in the same medium for another 7 days. The differentiated cells were immune-stained with the indicated antibodies.

Frozen section samples
The cultured organoids were fixed with 4% paraformaldehyde and then embedded in Tissue-Tek O.C.T Compound (Sakura Finetek Japan, Tokyo, Japan) and frozen at À80 C. The frozen samples were sectioned at 5-8 mm on a cryostat.

Histological and immunohistochemical analyses of the organoids
The organoids were embedded in paraffin blocks and sectioned at 4-mm thickness. The sections were deparaffinized and stained with Hematoxylin and Eosin (HE). Immunohistochemistry was performed using the Benchmark XT (Roche, Basel, Switzerland) autostainer with an XT ultraView Universal DAB Detection Kit (Ventana Medical Systems, Inc., AZ, USA).
For immunofluorescence, primary antibodies were incubated overnight at 4 C. Slides were washed in PBS and incubated with secondary antibody for 1 h at room temperature. For paraffin embedded sections, antigen retrieval was performed with 1 mM EDTA (pH8.0) for 3 min in pressure cooker before primary antibody incubation.

RNA sequencing
Total RNA was isolated using Trizol (Life Technologies) and treated with the TURBO DNA-free kit (Thermo Fisher Scientific), as described above. The RNA was sent to Macrogen (Seoul, South Korea, https://www. macrogen.com) for library preparation and paired-end RNA sequencing on the Illumina Novaseq6000 platform. Raw sequence files (fastq) were aligned to the human transcriptome (hg38) reference sequences using the Strand NGS software program (Strand Life Science, Karnataka, India) with default parameters. The aligned reads were normalized using Reads per kilobase of exon per million mapped reads (RPKM) again with the Strand NGS software program. For analysis, only the genes whose RPKM values are more than 0.1 (log2) in at least one sample of six samples were used to filter out noise from the expression data. The total ll OPEN ACCESS