Graphical abstract

   graphic file with name fx1.jpg
   [31]Open in a new tab

Highlights

     * •
       We successfully induce hGOs with MM by culturing for 8 weeks
     * •
       The emergence of MM occurs after epithelial maturation
     * •
       HH, TGF-β, and the mechanical environment are suggested to be
       involved in MM formation
     __________________________________________________________________

   In this article, Aoi and colleagues successfully induce hiPSC-derived
   human gastric organoids with muscularis mucosa (MM), which allow us to
   investigate human gastric MM development. A time course analysis
   revealed that MM differentiates after epithelial maturation. Hedgehog
   and TGF-β signaling from the epithelium as well as the mechanical
   environment may be involved in the MM formation of human stomach tissue

Introduction

   The gastric muscularis mucosa (MM), which is contained within the
   mucosa, is a thin layer of smooth muscle (SM) that separates the mucosa
   from the submucosa ([32]Ross and Pawlina, 2020). The function of the
   gastric MM is to help expel the contents of the deeper gastric glands,
   prevent clogging, and enhance contact between the gastric epithelium
   and the contents of the gastric lumen for absorption ([33]Young et al.,
   2014). In gastric cancer, the rates of lymph node metastasis are higher
   in submucosal cases than in intramucosal cases; therefore, in cancer
   staging, the T category is changed if gastric cancers invade beyond the
   MM ([34]Gotoda et al., 2000). In gastric ulcer, if the damage is
   confined to the mucosa, gastric mucosa can be completely restored
   without scarring. However, damage that reaches a deeper part of the
   stomach wall beyond the MM can cause fibrosis, and a depressed pit can
   be formed ([35]Rosai et al., 2011). Thus, the gastric MM plays
   important roles in both normal and pathological states.

   Several signaling pathways regulating endoderm patterning, gastric
   specification, stomach regionalization, and morphogenesis have been
   reported to be involved in gastric development ([36]McCracken and
   Wells, 2017); however, the details are not fully understood, especially
   concerning the gastrointestinal mesenchymal development. There are some
   reports on the formation of the muscularis propria in model animals
   that have demonstrated the involvement of Hedgehog (HH), Transforming
   growth factor beta (TGF-β), and Barx1 in mesenchymal and SM
   differentiation of the gut in chicks, mice, and zebrafish ([37]Aubin
   et al., 2002; [38]Fukuda et al., 1998; [39]Gays et al., 2017;
   [40]Jayewickreme and Shivdasani, 2015; [41]Mao et al., 2010;
   [42]Ramalho-Santos et al., 2000). However, how the human stomach
   develops in the embryo stage and how the human gastric MM is induced
   remain to be clarified.

   The scarcity of specimens of human embryo and fetal organs has hindered
   studies on human developmental biology. In previous decades, various
   organs and tissues derived from the endoderm, mesoderm, and ectoderm
   have become able to be induced from human pluripotent stem cells, such
   as human embryonic stem cells (hESCs) and human induced pluripotent
   stem cells (hiPSCs) ([43]Koterazawa et al., 2020; [44]Shi et al., 2017;
   [45]Suzuki et al., 2019), which has allowed us to investigate normal
   and pathological human developmental processes and their underlying
   mechanisms. McCracken et al. reported the generation of human
   pluripotent stem cell-derived gastric organoids (hGOs); however, the
   hGOs lacked subepithelial SM tissue ([46]McCracken et al., 2014).

   In the present study, we successfully generated GOs with MM from hiPSCs
   by culturing for a longer duration than in previous reports.
   Furthermore, using this induction method, we revealed that
   epithelial-derived HH signaling, TGF-β signaling, and the mechanical
   properties of the substrate worked together to induce MM formation in
   hiPSC-derived hGOs.

Results

Induction of hGOs accompanied by the subepithelial SM layer

   First, we performed differentiation of hiPSCs into gastric lineage
   according to previous reports ([47]McCracken et al., 2011, [48]2014,
   [49]2017). In brief, hiPSCs were differentiated into definitive
   endoderm (DE) for the first three days using Activin A. Subsequently,
   DE was differentiated to posterior foregut endoderm (FG) by inhibitor
   of glycogen synthase kinase 3 (CHIR99021), FGF4, and Noggin. Posterior
   foregut spheroids were transferred to a Matrigel-based
   three-dimensional (3D) culture. We then maintained the 3D culture for
   7 weeks, which was 3 weeks longer than in the previously reported study
   ([50]McCracken et al., 2014) ([51]Figure 1A).

Figure 1.

   [52]Figure 1
   [53]Open in a new tab

   Induction of human gastric organoids (hGOs) with muscularis mucosa (MM)
   from human induced pluripotent stem cells (hiPSCs)

   (A) A schematic diagram of hGOs with MM generated from hiPSCs using
   three-dimensional culture with Matrigel.

   (B) Morphological changes during the differentiation of hGOs.
   Stereomicrographs showed that the epithelium of hGOs had a folded and
   glandular architecture at week 8. Scale bars, 500 μm (hiPSCs; weeks 1
   and 3) and 2 mm (weeks 6 and 8).

   (C) Hematoxylin and eosin (H&E) staining and αSMA immunohistochemical
   (IHC) staining of hGO with MM at week 8 revealed folded epithelium that
   resembled gastric foveolae or glands. Subepithelial spindle cells were
   arranged like a bundle on αSMA IHC staining. Scale bars, 500 μm (upper
   panel) and 50 μm (lower panel).

   (D) Immunohistochemistry (IHC) of hGOs. After week 3, hGOs were
   accompanied by subepithelial αSMA-positive spindle cells (MM), and the
   epithelium of hGOs was positive for MUC5AC. Scale bars, 50 μm.

   (E) A principal component analysis of the RNA-seq data of hiPSCs,
   differentiated derivatives into stomach organoids (weeks 1, 3, 6, and
   8) that were obtained in two independent differentiated experiments,
   and two stomach samples. The horizontal axis represents PC1 scores,
   while the vertical axis represents PC2 scores.

   (F) Heatmap showing changes in the expression of gastric epithelial
   development, maturation, mesenchymal development, SM, and gastric
   epithelial markers of RNA-seq data.

   On stereomicroscope observation, the epithelium of the derivatives of
   hiPSCs showed increasingly complex and folded structures from 6 to
   8 weeks ([54]Figure 1B). Histological examinations by hematoxylin and
   eosin (H&E) staining at week 8 also revealed that the epithelium formed
   glandular or foveolar structures ([55]Figure 1C), as in the previous
   report ([56]McCracken et al., 2014). From weeks 6 to 8, the epithelium
   of the differentiated structures became positive for the gastric
   epithelial markers MUC5AC ([57]Figure 1D, lower panel) and TFF-1
   ([58]Rio et al., 1988) ([59]Figure S1A, lower panel). Notably,
   subepithelial αSMA-positive spindle cells began to appear at week 6 and
   formed a band-like structure that resembled MM at week 8 ([60]Figures
   1C and 1D, upper panel). Subepithelial spindle cells were found to be
   positive for Desmin, another muscle marker, at week 8 ([61]Figure S1A,
   upper panel). Some of the subepithelial spindle cells were positive for
   histamine receptor H1 (HRH1), which mediates the contraction of SMs
   ([62]Figure S1B). Based on these results, we considered the structure
   to correspond to gastric MM and thus successfully generated hGOs with
   MM from hiPSCs.

   To clarify the gene expression dynamics during differentiation into hGO
   with MM, an RNA-seq analysis was performed at different time-lapse
   sampling points (hiPSC, week 1, week 3, week 6, and week 8). According
   to a principal component analysis (PCA), which enables the visual
   assessment of the similarities and differences between samples to
   determine whether or not samples can be grouped ([63]Ringnér, 2008), we
   noted that the stomach samples showed variability, and the hGOs with MM
   at week 8 were similar to one of the two stomach samples
   ([64]Figure 1E), with hiPSCs gradually differentiated toward the
   stomach. Consistent with the immunohistochemical findings that
   subepithelial αSMA-positive cells appeared following CDH1-positive
   epithelial cells ([65]Figure 1D), the heatmap showed that gastric
   epithelial and mesenchymal development, maturation markers, and SM
   markers were gradually increased up to week 8, and their expression
   became similar to that in the stomach. The expression of TFF-1, a
   gastric surface mucous cell marker ([66]Aihara et al., 2017), preceded
   the expression of SM markers after week 3 ([67]Figure 1F). These
   findings indicated that subepithelial SM cells begin to differentiate
   after gastric epithelial maturation, and some epithelial-derived
   factors might have been involved in the differentiation of
   subepithelial SM cells (MM) after week 3.

   Human GOs with MM enabled us to examine the factors involved in MM
   differentiation, and we focused on three candidates: HH signaling,
   TGF-β signaling, and the mechanical properties of the substrate,
   according to previous reports on SM differentiation.

Cyclopamine inhibited MM formation of hGOs

   To examine whether or not the HH signaling, which has been reported to
   be involved in gut mesenchymal SM formation of chicks and mice
   ([68]Kolterud et al., 2009; [69]Mao et al., 2010; [70]Ramalho-Santos
   et al., 2000), was involved in MM formation, we assessed the changes in
   the Sonic hedgehog (SHH) expression during the induction of hGO with MM
   by immunohistochemistry (IHC) and in situ hybridization (ISH). We first
   confirmed that IHC appropriately worked using negative and positive
   controls ([71]Figure S2). As shown in [72]Figure 2A, positivity of IHC
   for SHH gradually increased from weeks 3 to 8 and were found in the
   epithelium, indicated by CDH1-positivity. Furthermore, ISH revealed SHH
   mRNA in the epithelium at week 6 ([73]Figures S3A and S3B). These data
   suggested that SHH was mainly synthesized in and secreted by the
   epithelium. We also performed immunostaining of Indian hedgehog (IHH),
   as shown in [74]Figure S3C. As with SHH, IHH expression was mainly
   observed in the epithelium at week 3, and the IHC positivity for IHH
   increased in the epithelial cells at weeks 6 and 8.

Figure 2.

   [75]Figure 2
   [76]Open in a new tab

   Muscularis mucosa formation of human gastric organoids (hGOs) was
   reduced by cyclopamine (Hedgehog signal inhibitor)

   (A) Immunohistochemistry of Sonic hedgehog (SHH) during induction of
   hGOs with MM. After week 3, the SHH expression was found mainly in the
   organoid epithelium, and IHC positivity for SHH in epithelial cells
   gradually increased toward week 8. Scale bars, 100 μm.

   (B) H&E staining of hGOs with and without cyclopamine treatment. The
   hGOs treated with cyclopamine were smaller and showed fewer
   subepithelial cells than those without cyclopamine treatment. Scale
   bars 250 μm (upper panels) and 50 μm (lower panels).

   (C) IHC of hGOs with and without cyclopamine treatment. In hGOs with
   cyclopamine treatment, subepithelial αSMA-positive cells were not
   found, and MM formation was reduced. TFF-1 expression was found in the
   epithelium of both hGOs, and gastric differentiation was preserved.
   Scale bars, 100 μm.

   (D) qRT-PCR of hGOs with and without cyclopamine treatment. The
   expression of αSMA (MM marker) and GLI1 (downstream molecule of HH
   signaling) was reduced by cyclopamine. TFF-1 (gastric epithelial
   marker) was not significantly changed. αSMA, number of independent
   differentiation experiments: n = 6; GLI1, n = 5; TFF-1, n = 6. Mean
   values ± SE, unpaired t test: ^∗p < 0.01.

   To examine the role of HH during the induction of hGOs with MM, we
   added cyclopamine, an HH signaling inhibitor, to the medium from week
   3. Although the gastric foveolar-like architecture was preserved under
   the administration of cyclopamine, hGOs treated with cyclopamine tended
   to be smaller, and there were fewer subepithelial cells constituting
   the hGOs in comparison with those that were not treated with
   cyclopamine, according to a histological examination ([77]Figure 2B).

   Immunohistochemistry revealed that αSMA-positive subepithelial spindle
   cells were decreased by inhibiting HH signaling. An MM-like structure,
   with an αSMA-positive spindle cell bundle, was not found in hGOs
   treated with cyclopamine ([78]Figure 2C upper panels). Hoechst staining
   also showed that cyclopamine treatment resulted in the defect of the
   subepithelial cells. Epithelial TFF-1-positive cells were observed
   regardless of cyclopamine, indicating that gastric epithelial
   differentiation was preserved even with HH signaling inhibition
   ([79]Figure 2C lower panels). A quantitative RT (qRT)-PCR analysis also
   showed that cyclopamine reduced the αSMA expression, whereas the
   expression of TFF-1 and TFF-2, another gastric epithelial marker
   ([80]Uhlen et al., 2015), was not significantly changed ([81]Figures 2D
   and [82]S3D). The expression of GLI1, a known downstream molecule of HH
   signaling, was significantly decreased, indicating that HH signaling
   was successfully inhibited by cyclopamine ([83]Figure 2D).

   In addition, the expression of Myocardin (MYOCD), which is a master
   transcriptional regulator of SM differentiation and has been reported
   to be a direct target of HH signaling in intestinal mesenchymal cells
   ([84]Cotton et al., 2017; [85]Wang et al., 2003; [86]Zacharias et al.,
   2011), was also reduced in the hGOs by cyclopamine ([87]Figure S3E).
   These results suggested that HH secreted from the epithelium may act on
   the subepithelial cells to activate HH signaling and increase the
   expression of αSMA via MYOCD.

SB431542, an inhibitor of TGF-βR1, reduced MM of hGOs

   We next evaluated whether or not TGF-β signaling was involved in MM
   formation. Although TGF-β signaling has been reported to be involved in
   gastrointestinal SM differentiation in zebrafish and vascular SM
   differentiation in humans ([88]Gays et al., 2017; [89]Guo and Chen,
   2012), the relationship between TGF-β signaling and human gastric
   mesenchymal differentiation has not been clarified. To assess the
   expression of TGF-β1 and TGF-βR1 during gastric differentiation in our
   hGO development, immunohistochemistry was performed. Regarding TGF-β1
   IHC findings, we confirmed that IHC appropriately detected TGF-β1 using
   negative and positive controls ([90]Figure S4). TGF-β1 was largely
   detected in the epithelium of hGOs, and IHC positivity for TGF-β1 in
   the epithelium increased from weeks 3 to 8 ([91]Figure 3A upper
   panels). Furthermore, ISH at weeks 6 and 8 revealed that TGF-β1 was
   synthesized de novo predominantly in the epithelium rather than in
   subepithelial cells ([92]Figures S5A and S5B). The IHC positivity for
   TGF-βR1 in hGOs was also gradually increasing, and subepithelial cells
   were weakly positive after week 3 ([93]Figure 3A lower panels).

Figure 3.

   [94]Figure 3
   [95]Open in a new tab

   MM differentiation of hGOs was reduced by SB431542 (TGF-βR1 inhibitor)

   (A) IHC of TGF-β1 and TGF-βR1 of hGOs with MM during induction. The
   TGF-β1 expression was mainly found in the epithelia of hGOs. TGF-βR1
   was weakly positive in not only the epithelia but also subepithelial
   cells. Scale bars, 50 μm.

   (B) H&E staining and IHC of αSMA and TFF-1 of hGOs with the treatment
   of SB431542. αSMA expression was not found in subepithelial cells, and
   MM differentiation was inhibited by SB431542. TFF-1 expression was
   positive in the epithelia, and gastric differentiation was preserved
   under SB431542. Scale bars, 50 μm.

   (C) Results of a qRT-PCR analysis of αSMA, MUC5AC, and TFF-1 of hGOs
   treated with SB431542. αSMA expression was reduced by SB431542, whereas
   the MUC5AC and TFF-1 expression was not significantly changed. αSMA,
   number of independent differentiation experiments: n = 6; MUC5AC, n =
   6; TFF-1, n = 5. Mean values ± SE, unpaired t test: ^∗p < 0.01.

   We next administered SB431542, an inhibitor of TGF-βR1, from week 3 and
   examined the induced hGOs at week 8, since the MM of hGOs appeared
   after week 3. Immunohistochemical staining revealed that the epithelium
   of the hGOs was positive for the gastric epithelial marker TFF-1, and
   gastric differentiation was conserved under the addition of SB431542.
   However, αSMA-positive subepithelial cells were not detected
   ([96]Figure 3B), indicating that the MM formation of hGOs was
   suppressed by inhibition of TGF-β signaling. A qRT-PCR analysis showed
   that the expression of αSMA was significantly decreased by the addition
   of SB431542, whereas the gastric epithelial differentiation was
   preserved, as the MUC5AC and TFF-1 expression did not significantly
   change ([97]Figure 3C). The expression of αSMA was also reduced by
   SB431542 at the protein level ([98]Figure S6).

   These data suggest that SB431542 reduced the MM induction of hGOs after
   week 3, and epithelial-derived TGF-β signaling might be involved in
   human gastric MM differentiation.

Mechanical environment promotes subepithelial αSMA expression

   During their development, our hGOs were maintained carefully within the
   Matrigel in order to prevent contact with the surface of the dishes.
   However, they occasionally unintentionally penetrated the Matrigel and
   contacted the surface of the dishes. On such occasions, we noticed that
   spindle cells spread around the hGOs (data not shown), prompting the
   hypothesis that attachment to the surface of dishes induces SM
   differentiation. To test this hypothesis, we transferred the hGOs that
   developed in the Matrigel at week 3 to an ultra-low-attachment plate or
   attachment plate without re-embedding in the Matrigel and maintained
   them for 2 weeks ([99]Figure 4A) before examining the SM
   differentiation by immunohistochemistry, qRT-PCR, and western blotting.

Figure 4.

   [100]Figure 4
   [101]Open in a new tab

   Attaching to the surface of the dishes affects MM differentiation of
   hGOs

   (A) A schematic diagram of the generated hGOs to investigate the effect
   of attaching to the surface of the dishes.

   (B) Stereomicrographs showed spreading cells around the hGO with
   attaching to the surface of the dishes. Scale bars, 1 mm.

   (C) Immunohistochemical staining showed that the spreading cells were
   positive for αSMA in the hGOs with attachment. αSMA-positive cells were
   not found around the hGOs without attachment. Scale bars, 200 μm.

   (D) A qRT-PCR analysis of hGOs with and without attachment. Attaching
   (substrate property) promoted αSMA expression. The TFF-1 expression was
   preserved. n = 10 independent experiments, mean values ± SE, unpaired t
   test: ^∗p < 0.001

   While the hGOs on the ultra-low-attachment plate were not surrounded by
   spreading cells, the hGOs on the attachment plates were surrounded by
   proliferating spindle cells right after the transfer ([102]Figure 4B).
   As we hypothesized, immunohistochemistry revealed that only the spindle
   cells around the hGOs that had attached to the surface of the dishes
   were positive for αSMA ([103]Figure 4C). Consistent with
   immunohistochemistry, western blotting and qRT-PCR showed that
   attachment of the hGOs to the surface of the dishes resulted in the
   upregulated expression of αSMA ([104]Figures 4D and [105]S7A).

   The next issue to be addressed was which cells in hGOs were affected by
   the mechanical environment. Mesenchymal stem cells (MSCs) on stiff
   substrate reportedly differentiate into SM cells in response to TGF-β
   ([106]Park et al., 2011). Our hGOs at week 3 had some subepithelial
   cells that were positive for the MSC markers CD146 or CD44 ([107]Crisan
   et al., 2008) ([108]Figure S7B), suggesting that the subepithelial MSCs
   in the hGOs were affected by the mechanical environment under the
   epithelial-derived TGF-β and thereby differentiated into SM cells.

Effects of TGF-β signaling on HH signaling and mechanical properties

   We next examined the relationship between HH and TGF-β signaling in
   hGOs using the TGF-β inhibitor SB431542. TGF-β reportedly induces the
   expression of the HH signaling molecule GLI2 in various human cell
   types and cancer cell lines ([109]Dennler et al., 2007). In addition,
   GLI1 is a downstream target of HH signaling.

   The GLI2 mRNA expression was significantly reduced by SB431542
   treatment according to qRT-PCR of six independent experiments
   ([110]Figure 5A), indicating the successful suppression of TGF-β
   signaling in the hGOs. Although the GLI1 mRNA expression was not
   significantly changed (data not shown), GLI1-positive subepithelial
   spindle cells of hGOs treated with SB431542 were reduced compared with
   the subepithelial cells of hGOs without SB431542 in
   immunohistochemistry ([111]Figure 5B). These results indicated that
   TGF-β and HH signaling act not only independently but also
   collaboratively in the induction of MM in hGOs.

Figure 5.

   [112]Figure 5
   [113]Open in a new tab

   Effects of TGF-β signaling on HH signaling and mechanical properties

   (A) A qRT-PCR analysis of GLI2 in hGOs with or without the treatment of
   SB431542. The GLI2 expression was reduced by SB431542. Normalized to
   control. n = 6 independent experiments, mean values ± SE, unpaired t
   test, ^∗p < 0.01.

   (B) According to IHC of hGOs, the subepithelial GLI1 expression was
   reduced compared with hGOs without SB431542 treatment. Scale bars,
   20 μm.

   (C) A pathway analysis (WikiPathways and Reactome) for genes of RNA-seq
   data downregulated more than 10-fold by treatment of SB431542. Pathways
   with -log10 p value >5 were listed.

   (D) A gene set enrichment analysis (GSEA) of RNA-seq data showed that
   gene sets associated with collagen fibril organization (in the
   biological process [BP]) and extracellular matrix structure constituent
   (in the molecular function [MF]) were negatively regulated by SB431542.

   To broadly investigate the effects of TGF-β signaling inhibition on hGO
   induction, we analyzed the global gene expression changes in hGOs after
   the treatment of SB431542 by RNA-seq. We confirmed the reduced
   expression of GLI2 (data not shown) and identified 70 upregulated genes
   and 66 downregulated genes (fold change >10) by the treatment of
   SB431542. For these differentially expressed genes, we performed a
   pathway enrichment analysis on WikiPathways and Reactome, and pathways
   with a p value <10^−5 were defined as significantly affected (listed in
   [114]Figure 5C). As a result, six categories (collagen degradation,
   assembly of collagen fibrils and other multimeric structures, integrin
   cell surface interactions, collagen biosynthesis and modifying enzymes,
   extracellular matrix organization, and collagen trimerization) were
   commonly selected as associated with the downregulated genes. In
   addition, a gene set enrichment analysis (GSEA) showed that gene sets
   concerning “Collagen Fibril Organization” and “Extracellular Matrix
   Structural Constituent” were regulated negatively by SB431542
   ([115]Figure 5D). These results suggested that TGF-β signaling may also
   contribute to extracellular matrix stiffness and thus promote MM of
   hGOs.

Regeneration of MM in the adult gastric tissues

   We demonstrated that epithelial maturation preceded MM induction in our
   hGOs, which mimicked the gastric development. To investigate whether or
   not this precedence was found in the regeneration process of adult
   gastric mucosa, we histologically examined post-endoscopic submucosal
   dissection (ESD) gastric mucosa. Although gastric epithelium completely
   regenerates after gastric mucosal injury, gastric MM has not been
   restored yet ([116]Figure 6A). Consistent with the epithelial
   development in our organoid model, epithelial regeneration precedes MM,
   and epithelial-derived factors that induce MM development may also be
   involved in regeneration.

Figure 6.

   [117]Figure 6
   [118]Open in a new tab

   Regeneration of MM in the adult gastric tissues

   (A) H&E staining of the post-ESD stomach. The black triangles indicate
   regenerating MM.

   (B) Representative images of H&E staining and TGF-β1 IHC in normal
   stomach mucosa and post-ESD ulcer. Scale bars, 200 μm (H&E) and 100 μm
   (TGF-β1 IHC).

   (C) Co-immunofluorescence of post-ESD ulcer for CDH1 and TGF-β1. Scale
   bar, 10 μm.

   We demonstrated that TGF-β signaling was involved in gastric MM
   induction. Generally, some factors that function during development are
   used for regeneration or disease, and the TGF-β1 expression has been
   reported to be increased in rats' gastric ulcers ([119]Tominaga et al.,
   1997). We therefore investigated whether or not TGF-β1 was involved in
   regeneration of MM. We performed TGF-β1 immunohistochemical staining on
   post-ESD ulcer specimens. As a control, we also performed
   immunohistochemistry on a tissue array of normal stomach mucosa that
   was commercially available. In the histological examination, gastric
   ulcers or ulcer scars were found in the specimens obtained from
   post-ESD gastrectomy. When evaluating TGF-β immunohistochemistry, the
   cases with ≥1% epithelial cells positive for TGF-β1 were considered
   immunohistochemically positive. Although some inflammatory cells are
   positive for TGF-β1, only epithelial positivity was evaluated
   ([120]Figures 6B and 6C) ([121]Li et al., 2007; [122]Tominaga et al.,
   1997). Four of 24 cases showed positivity in the normal gastric mucosa,
   and 13 of 18 cases showed positivity in the post-ESD ulcers or ulcer
   scars ([123]Table 1). The TGF-β1 expression in regenerative epithelium
   of ulcers or ulcer scars was significantly increased compared with that
   in the normal mucosa, indicating that epithelial TGF-β1 is involved in
   not only induction of the developmental process but also regeneration
   of gastric MM after adult tissue injury.

Table 1.

   Epithelial positivity of TGF-β1 IHC in normal stomach mucosa and
   post-ESD ulcer or ulcer scar. Fisher exact test
   TGF-β1
     __________________________________________________________________

                   p value
         +                       –
   Normal stomach 4 (16.7%)  20 (83.3%) 0.00043
   Post-ESD ulcer 13 (72.2%) 5 (27.8%)
   [124]Open in a new tab

Discussion

   In this study, we successfully generated for the first time to our
   knowledge hiPSC-derived GOs accompanied by MM. We defined the
   subepithelial bundle consisting of αSMA-positive spindle cells as MM in
   our hGOs. The human gastric lamina propria contains scattered SM cells
   and is continuous with the MM ([125]Ross and Pawlina, 2020). In
   addition, subepithelial myofibroblasts also express SM markers with
   variable intensity ([126]Powell et al., 2011; [127]Wu et al., 1999). It
   is thus difficult to clearly distinguish MM from the scattered SM cells
   and myofibroblasts by the cell marker expression alone without
   observing the histological structure. To investigate MM formation,
   immunohistochemical confirmation of the band-like structure of SM cells
   is essential.

   Human PSC-derived GOs reported by [128]McCracken et al. (2014)
   contained only scattered αSMA-positive cells and no αSMA-positive
   band-like structure, indicating that MM had not been formed in these
   GOs. In contrast, the mouse embryonic stem cell-derived GOs reported by
   [129]Noguchi et al. (2015) and human PSC-derived intestinal organoids
   reported by [130]Uchida et al. (2017) did show obvious SM
   marker-positive band-like structures. It took longer to generate GOs
   consisting of epithelium with MM than those without MM, suggesting that
   the MM develops later than the epithelium. Our present data indeed
   demonstrated that the epithelium preceded MM in the organoid
   development. These findings support the notion that epithelial-derived
   factors play important roles in MM formation.

   We demonstrated in the present study that epithelial-derived TGF-β
   signaling was involved in SM differentiation of subepithelial cells of
   hGOs. To our knowledge, TGF-β1 has not been reported to be secreted
   from gastric epithelium during development or involved in MM induction.
   However, our analysis of publicly available RNA-seq data ([131]Szabo
   et al., 2015) showed that TGF-β1 was expressed in human fetal stomach
   from 10 to 20 weeks' gestation (data not shown). Indeed, combined with
   substrate stiffness, TGF-β1 induces mouse mesenchymal cells to
   differentiate to SM cells through the phosphatidylinositol 3-kinase/Akt
   pathway and also induces MSCs to differentiate into SM cells ([132]Lien
   et al., 2006; [133]Park et al., 2011). The subepithelial mesenchymal
   cells in our hGOs that were positive for mesenchymal stem cell markers
   might have been induced to differentiate into SM cells by TGF-β and the
   substrate stiffness.

   We showed in the present study that TGF-β signaling inhibition resulted
   in a significantly suppressed expression of GLI2 but not GLI1. The
   relationship between HH and TGF-β signaling was previously indicated in
   mice ([134]Aubin et al., 2002) as well as in cancer and non-cancer cell
   lines ([135]Dennler et al., 2007). GLI2 was induced rapidly by TGF-β
   through SMAD in cooperation with β-catenin, and GLI1 was subsequently
   expressed ([136]Dennler et al., 2009), suggesting that GLI2 is a direct
   downstream target of TGF-β signaling, while GLI1 is indirectly
   regulated by TGF-β signaling via HH signaling activation. Although the
   relationship between TGF-β signaling and HH signaling may partially
   contribute to subepithelial SM differentiation in hGOs, with various
   kinds of cells interacting with each other via various signaling
   pathways, the effect of TGF-β signaling on GLI1 activation in hGOs may
   be smaller than that in the culture of single cell lines.

   The TGF-β1 expression in regenerating human epithelium of post-ESD
   ulcer was shown in this study. The expression of TGF-β1 as well as
   collagen, fibronectin, and laminin, was reportedly increased during
   gastric ulcer healing in rats ([137]Tominaga et al., 1997). These
   increased extracellular matrix proteins might contribute to the αSMA
   expression ([138]Tominaga et al., 1997). In mouse gastric ulcers,
   gastric epithelial HH signaling has also been investigated and shown to
   be involved in the differentiation of gastric progenitor cells
   ([139]Kang et al., 2009). However, the role of HH signaling in
   non-epithelial tissue regeneration in gastric ulcers has not been
   studied. In the present study, SHH decreased from week 6 to 8 in the
   ISH analyses ([140]Figure S3A), and the inhibition of HH signaling by
   cyclopamine resulted in the defect of the subepithelial cells
   ([141]Figures 2B and 2C). Given these findings, HH signaling may be
   involved in the maintenance of progenitor cells of the subepithelium,
   but not in the induction of αSMA-positive cells from subepithelial
   cells. Epithelial-derived factors and their downstream biological
   processes that play important roles in gastric MM development may help
   restore the damaged MM in the adult stomach.

   In conclusion, HH and TGF-β signaling from epithelium to stroma and the
   mechanical properties of the subepithelial environment were suggested
   to be involved in the emergence of MM of human stomach tissue.
   Furthermore, TGF-β signaling may function during not only the
   development but also regeneration of MM. In addition, mechanical strain
   reportedly organizes the SM layer ([142]Huycke et al., 2019). H&E
   staining and IHC analyses of the αSMA in our hGOs did not appear to
   demonstrate the formation of the subepithelial layered structure
   constituted by the lamina propria and MM with a distinct boundary, as
   seen in human stomach tissues ([143]Figure 1C). The contraction was
   unclear in our hGOs, probably due to the absence of the muscularis
   propria. A lack of contraction may result in the incomplete formation
   of the MM layer by αSMA-positive muscular cells. Further studies may
   reveal the relationship between these factors and defects in MM
   reformation in gastric cancer with submucosal invasion.

Experimental procedures

hiPSC culture

   Two hiPSC lines (201B7 and FF-PB-3AB4) were used in the experiments in
   this study. The validated hiPSC line 201B7 was purchased from Riken
   Cell Bank (Tsukuba, Japan) and transferred from on-feeder to
   feeder-free conditions in our laboratory. The hiPSC line FF-PB-3AB4 was
   established in our laboratory ([144]Suzuki et al., 2019). These hiPSC
   lines were cultured according to a previously described method
   ([145]Nakagawa et al., 2015) with slight modifications. In brief, the
   hiPSCs were maintained in StemFit AK02N medium (Ajinomoto, Tokyo,
   Japan) with penicillin (50 units/mL) and streptomycin (50 μg/mL)
   (#1514022; Gibco, Carlsbad, CA, USA) at 37°C with 5% CO[2]. The medium
   was changed every other day and passaged every 7 days using 0.5x TrypLE
   Select (1x TrypLE Select [#A12859-01; Gibco] diluted 1:1 with 0.5 mM
   EDTA [#06894-14; Nacalai Tesque, Kyoto, Japan]/PBS [-]) and
   Rho-associated kinase (Rock) inhibitor (Y-27632; WAKO, Osaka, Japan).
   iMatrix-511 silk (#892021; Nippi, Tokyo, Japan) was used for
   precoating.

Gastric differentiation from hiPSCs

   Differentiation into the hGOs derived from hiPSCs was performed
   according to previous reports ([146]McCracken et al., 2014, [147]2017).
   In brief, 2.0 × 10^5 hiPSCs were plated as single cells in a 24-well
   dish (#142475; Thermo Scientific, Waltham, MA, USA) precoated with
   iMatrix-511 silk (0.5 μg/cm^2) in StemFit medium with 10 μM of ROCK
   inhibitor Y-27632. To induce differentiation into DE, the medium was
   changed to RPMI-1640 medium (#30264-56; Nacalai Tesque) supplemented
   with Activin A (100 ng/mL, #120-14; PeproTech, Rocky Hill, NJ, USA)
   from the next day for 3 days. BMP4 (50 ng/mL, #314-BP; R&D,
   Minneapolis, MN, USA) was added on the first day, and fetal bovine
   serum (FBS; #F7524; Sigma Aldrich, St. Louis, MO, USA) was included for
   the first 3 days at increasing concentrations of 0%, 0.2%, and 2.0%.

   To generate foregut spheroids, RPMI-1640 medium supplemented with 2.0%
   FBS, 2 μM of CHIR99021 (#4423; TOCRIS, Bristol, UK), 500 ng/mL of FGF4
   (#GFH31; Cell Guidance, St. Louis, MO, USA), and 200 ng/mL of Noggin
   (#6057-NG; R&D Systems) was added for 3 days until day 6. Retinoic acid
   (2 μM, #186-01114; WAKO) was added on day 6.

   For the three-dimensional culture of gastric organoids using Matrigel
   (#354234; Corning, New York, NY, USA), spheroids were transferred to a
   three-dimensional in vitro culture system as previously described
   ([148]McCracken et al., 2011). Spheroids were collected, mixed with
   50 μL Matrigel, and plated into the middle of a well of a 4-well dish
   (#176740; Thermo Scientific). They were then cultured with advanced
   DMEM/F12 (#12634-010; Gibco) supplemented with N2 (1x) (#17502-001;
   Gibco), B27 (1x) (#17504-044; Gibco), 2 mM of L-glutamine (#25030-081;
   Gibco), 10 mM of HEPES (#15630-080; Gibco), penicillin/streptomycin,
   and 100 ng/mL of EGF (#236-EG; R&D). For the first 3 days of
   three-dimensional culture, 2 μM retinoic acid and 200 ng/mL Noggin were
   added to the media. The media was replaced every 2–4 days, as
   necessary. On day 20–21 (week 3), organoids were collected and
   re-plated with fresh Matrigel. Around days 41–43 (week 6), the
   organoids were transferred to fresh Matrigel again to prevent them from
   attaching to the bottom of the well of the plates. On day 56 (week 8),
   the organoids were analyzed by quantitative (q)RT-PCR, histological,
   immunohistological examination, and western blotting.

   To inhibit HH and TGF-β signaling during MM differentiation of hGOs,
   cyclopamine (20 μM, #C9710; LKT Laboratories, St. Paul, MN, USA) and
   SB431542 (10 μM, #033-24631; Wako) were added to the medium,
   respectively, after re-plating the organoids on day 20–21.

RNA isolation and PCR

   Total RNA was extracted using TRIzol (#15596026; Life Technologies,
   Grand Island, NY). Human GOs were removed from Matrigel and homogenized
   using a beads crusher, μT-12 (#0068700-000; TAITECH, Saitama, Japan).
   Genomic DNA was removed by ezDNase (#11766051; Life Technologies), and
   reverse transcription was performed using a PrimeScript II 1st strand
   cDNA Synthesis Kit (#2690A; Takara Bio, Shiga, Japan) according to the
   manufacturer's protocol.

   qRT-PCR was performed using TB Green Premix Ex Taq II (#RR820; Takara
   Bio) on a LightCycler 480 II (Roche, Basel, Switzerland). The PCR
   primers used for qRT-PCR were as follows: GAPDH, forward
   5′-AGCCACATCGCTCAGACAC-3′, reverse 5′-GCCCAATACGACCAAATCC-3′; αSMA,
   forward 5′-TTCAATGTCCCAGCCATGTA-3′, reverse 5′-GAAGGAATAGCCACGCTCAG-3′;
   MUC5AC, forward 5′-CTCAGCTGTTCTCTGGACGA-3′, reverse
   5′-GCTGGATGATCAGGCTCCTA-3′; TFF-1, forward 5′-TGTGCAAATAAGGGCTGCTG-3′,
   reverse 5′-GTCAGGATGCAGGCAGATCC-3′; TFF-2, forward
   5′-AGTGCTGCTTCTCCAACTTC-3′, reverse 5′-AGATGCATCCTCTGGAACCAG-3′; GLI1,
   forward 5′-ACATCAACTCCGGCCAATAG-3′, reverse 5′-GAGGATGCTCCATTCTCTGG-3′;
   GLI2, forward 5′-AACACATGACCACCATGCAC-3′, reverse
   5′-CTGCCACTGAAGTTTTCCAG-3′; MYOCD, forward
   5′-TCAGCAGATGGATGAACTCCTG-3′, reverse 5′-AGTTGGACTTCGGGAAGATCTG-3′.

RNA sequencing

   The total RNA was isolated as described above and sent to Macrogen
   (Seoul, South Korea) for library preparation using the SMARTer stranded
   total RNA-seq kit. Paired-end RNA sequencing datasets were produced
   using an Illumina NovaSeq6000 (Illumina, San Diego, CA, USA). Reads
   were aligned to the human transcriptome (hg38) reference sequences
   using the Strand NGS software program (Strand Life Sciences, Bangalore,
   India). The heatmap was generated based on transcripts per million
   (TPM) values. Total RNA derived from two normal stomach samples
   purchased from BioChain Institute (R1234248-50; Newark, CA, USA) were
   also analyzed as controls. They were from two different donors.
   Although from which part of the stomach these samples were derived is
   unclear, based on the RNA-seq data, they were considered to have been
   taken from the antrum and at least deeper than the MM, as some adipose
   markers and an antral marker were expressed. Using the R software
   program, a PCA (in [149]Figure 1E) was performed for the 8594 genes
   with TPM values exceeding 2 in all differentiated cells (weeks 1, 3, 6,
   and 8 and stomach). WikiPathways and Reactome were performed to
   identify pathways using the genes with a more than 10-fold change in
   their expression in the RNA-seq data according to the Strand NGS
   software program and R packages ReactomePA ([150]Yu and He, 2016). A
   GSEA was performed using the GSEA software program ([151]Subramanian
   et al., 2005).

Histological and immunohistochemical examinations and in situ hybridization

   The spheroids and organoids were fixed with 10% buffered neutral
   formalin solution for a period ranging from 3 h to overnight and
   embedded in paraffin blocks. They were sectioned at 4-μm thickness.
   After being deparaffinized and rehydrated, sections were stained with
   H&E. For immunostaining, antigens were retrieved by a pressure cooker
   with adequate retrieving buffer, and primary antibodies were incubated
   overnight at 4°C. The primary antibodies were as follows:
   anti-E-cadherin (CDH1) (goat polyclonal antibody, dilution 1:200; R&D
   AF648), anti-MUC5AC (mouse monoclonal antibody, CLH2, dilution 1:200;
   Abcam, Cambridge, MA, USA ab77576), and anti-αSMA (mouse monoclonal
   antibody, 1A4, dilution 1: 200; Dako, Carpinteria, CA, USA M0851),
   anti-Desmin (mouse monoclonal antibody, D33, dilution 1:200; Abcam
   ab8470), anti-SHH (rabbit monoclonal antibody, EP1190Y, dilution 1:200;
   Abcam ab53281), anti-TFF-1 (rabbit monoclonal antibody, EPR3972,
   dilution 1:200; Abcam ab92377), anti-TGF-β1 (rabbit polyclonal
   antibody, dilution 1:200; Abcam ab92486), anti-TGFβ Receptor I (rabbit
   polyclonal antibody, dilution 1:50; Abcam ab31013), anti-GLI1 (mouse
   monoclonal antibody, A-7, dilution 1:200, Santa Cruz, Dallas, TX, USA
   sc-515781), anti-IHH (rabbit polyclonal antibody, dilution 1:50; Abcam
   ab39634), and anti-HRH1 (rabbit polyclonal antibody, dilution 1:50;
   Proteintech 13413-1-AP).

   For immunofluorescent staining, following the primary antibody, slides
   were washed in phosphate-buffered saline (PBS) and incubated with
   secondary antibodies (at dilution of 1:500) for 1 to 2 h at room
   temperature. Secondary antibodies were as follows: Donkey anti-goat
   IgG, Donkey anti-mouse IgG and Donkey anti-rabbit IgG (Alexa Fluor 488
   and 594; Invitrogen, Carlsbad, CA, USA, A11055, A11058, A21202, A21203,
   A21206, A21207). Nuclei were stained with Hoechst 33342 (H3570; Thermo
   Scientific). Slides were washed in PBS, and coverslips were mounted
   using Fluoromount/Plus (K048; Diagnostic BioSystems, Pleasanton, CA,
   USA). The images were obtained with a fluorescence microscope (BZ-X700;
   Keyence, Osaka, Japan) and a confocal microscope (LSM700; Carl Zeiss,
   Jena, Germany).

   For immunohistochemistry, following primary antibody incubation,
   incubation with a secondary antibody (Histofine Simple Stain MAX-PO;
   Nichirei Bioscience, Tokyo, Japan) was performed for 1 h at room
   temperature, and signals were detected using a Histofine DAB substrate
   kit (Nichirei). A normal stomach tissue array (BN01011b; US Biomax,
   Rockville, MD, USA) was stained with TGF-β1 as control.

   For in situ hybridization, RNAscope technology (Advanced Cell
   Diagnostics, Newark, CA, USA) was used according to the manufacturer’s
   protocol. Slides were hybridized using the Hs_SHH (#600951; Advanced
   Cell Diagnostics) and the HS_TGFB1 probes (#400881; Advanced Cell
   Diagnostics). The signals were detected by RNAscope 2.5 HD Detection
   Kit RED (#322360; Advanced Cell Diagnostics). They were counterstained
   with hematoxylin or subjected to immunostaining with anti-CDH1 as a
   primary antibody and Donkey anti-goat IgG (Alexa Fluor 488) as a
   secondary antibody after appropriate antigen retrieval, as mentioned
   above.

Human tissues

   The IHC experiment for post-ESD gastric ulcer tissues was approved by
   the institutional review board of Kobe University Graduate School of
   Medicine and performed under protocol no. B200295. Eighteen post-ESD
   gastric ulcer cases were retrieved from the pathological archives of
   Kobe University Hospital between December 2017 and December 2019. All
   patients underwent gastric ESD for early gastric cancer, and
   gastrectomy was performed as an additional treatment within 3 months
   because of invasion at the deeper region of the submucosa,
   lymphovascular invasion, and poorly differentiated carcinoma
   components. Recurrence of gastric cancer was not found in any cases.

Whole-mount staining

   hGOs with and without attachment to the bottom of dishes were fixed
   with 4% paraformaldehyde (PFA) overnight at room temperature. Samples
   were permeabilized by PBS with 0.5% Triton X-(PBST) overnight at 4°C.
   Blocking was performed by PBST with 2% nonfat instant skim milk
   overnight at 4°C. Primary antibodies were diluted in PBST with 2%
   nonfat instant skim milk, and samples were incubated for 1 to 3 days at
   4°C. Before incubating with secondary antibodies, samples were washed
   five times with PBST. Secondary antibodies were also diluted in PBST
   with 2% nonfat instant skim milk, and samples were incubated for 1 day
   at 4°C. An LSM700 (Carl Zeiss) was used for observation and image
   capture. Confocal images were acquired using a z stack collection with
   a 1-μm step. The images were processed for Z-projection (SD) using
   ImageJ/Fiji ([152]Schindelin et al., 2012).

Statistical analyses

   All experiments were performed as at least three independent
   experiments. All data were analyzed using R ([153]R Core Team, 2021).
   Differences in the mean values between two groups were analyzed using
   the paired or unpaired t test. Values of p < 0.05 were considered
   significant.

Author contributions

   K.U., T.K., M.K-A., T.I., and T.A. conceived the project and designed
   the experiments. K.U. and T.K. performed the culture experiments. K.U.
   performed immunostaining, western blotting, and qRT-PCR. K.U., M.K-A.,
   and T.A. performed the bioinformatics analysis. T.A. supervised the
   project. K.U., M.K-A., and T.A. wrote the manuscript.

Conflicts of interest

   The authors declare no conflicts of interest in association with the
   present study.

Acknowledgments