Main<\/h2>\n\nThe intestine constitutes the largest compartment of the immune system, organized in the lamina propria, epithelium and lymphoid follicles defined by the gut-associated lymphoid tissue (GALT). Immune cell types and GALT are regionally distributed across the mucosal layers and along the gut tube1<\/a><\/sup>. Immune\u2013epithelial crosstalk is essential to the maintenance of intestinal homeostasis, defense against pathogens and immunologic tolerance to dietary components and commensal bacteria. For instance, specialized epithelial microfold (M) cells, located in the follicle-associated epithelium, play a central role in gut immune sensing by transporting luminal antigens in the lamina propria to activate immune cells2<\/a><\/sup>. Immune\u2013epithelial crosstalk mediated by cytokines is involved in gastrointestinal tissue development, homeostasis and disease states3<\/a>,4<\/a>,5<\/a><\/sup>. Although experimental mouse models have been widely used to dissect the biological mechanisms of intestinal immune tissue development and disease, there are still substantial differences that prevent direct extrapolation to human biology6<\/a><\/sup>. Similarly, coculture of HIOs and immune cells has not fully recapitulated the complexity of human immune intestinal tissue7<\/a>,8<\/a>,9<\/a><\/sup>. Deeper understanding of the perturbations of the gastrointestinal immune system that commonly result in chronic human diseases requires new human models that represent patient-specific gastrointestinal immune tissue.<\/p>\nDuring the past decade, pluripotent stem cell (PSC)-derived HIOs have advanced our knowledge of human intestinal development, physiology and diseases10<\/a>,11<\/a><\/sup>. A key feature of the PSC-derived HIO model is the ability to generate complex human intestinal tissue. We have previously developed and validated an in vivo model that uses transplantation of in vitro PSC-derived HIOs into immunocompromised mice. We demonstrated the induction of intestinal maturation in transplanted HIOs, characterized by the presence of stem cells and functionally differentiated epithelial cells, as well as a crypt\/villus architecture with mesenchymal submucosal layers including lamina propria12<\/a>,13<\/a>,14<\/a><\/sup>. However, this system lacked the developing immune cells that would normally be found in the developing fetal intestine. To overcome this limitation, here we report a next-generation HIO system containing human immune cells, obtained by growing HIOs in mice with a humanized immune system.<\/p>\nHumanized immune system mouse models are commonly used to investigate human hematopoiesis or inflammatory diseases and can be generated by the engraftment of human peripheral blood leukocytes, hematopoietic stem cells or fetal tissues (bone marrow, thymus and\/or liver) in lymphopenic mice such as the NOD\/SCID\/Il2rg\u2212\/\u2212<\/sup> (NSG) mouse strain15<\/a>,16<\/a><\/sup>. Recently, Wunderlich et al. reported that transgenic expression of human SCF, GM-CSF and IL-3 in NSG mice, named NSGS mice, improves hematopoietic engraftment, reconstitution and function17<\/a>,18<\/a><\/sup>. However, because these lymphopenic mice do not express Il2rg<\/i>, resulting in impaired signaling in lymphoid tissue inducer (LTi) cells, they do not have Peyer\u2019s patches and lymphoid follicles in their intestine19<\/a>,20<\/a><\/sup>. The lack of GALT in these mouse models represents a major limitation to understanding numerous human gastrointestinal diseases. Here, we aimed to bridge the gap by integrating human immune cells in a human intestinal tissue.<\/p>\nIn our current model, we performed a serial time course following transplantation of HIOs under the kidney capsule to characterize the human immune cells that infiltrate the HIO lamina propria and epithelium. Mass cytometry and immunostaining confirmed the presence of GALT-associated B cells in cellular immune aggregates present in transplanted HIOs, resembling lymphoid follicles developing in human fetal intestine. Finally, we demonstrated that microbial exposure in the HIO lumen is required to induce the epithelial expression of glycoprotein 2 (GP2) at the cell surface of M cells, which consequently activate immune cells in the lamina propria. Altogether, these results demonstrate that crosstalk between the HIO epithelium and the immune cells induces the formation of lymphoid-like structures and M cell differentiation and function.<\/p>\n<\/div>\n<\/div>\n
\nResults<\/h2>\n\nHuman immune cells integrate into transplanted HIOs<\/h3>\n
To bioengineer an HIO with immune cells, we transplanted PSC-derived HIOs under the kidney capsule of humanized NSGS mice (see workflow in Extended Data Fig. 1a<\/a>). We collected and analyzed HIOs at 12, 16 and 20\u2009weeks post-transplant. We confirmed the presence of human immune cells in the blood of humanized mice at each time point (Extended Data Fig. 1b\u2013d<\/a>) and found that HIOs successfully engrafted and grew to a similar size compared with those grown in nonhumanized control mice (Fig. 1a,b<\/a>). However, we noticed that HIOs transplanted for periods of 16 and 20\u2009weeks were more heterogeneous in size and this phenomenon may be due to an accumulation of mucus within the HIO lumen which cannot be drained out. Regardless, the effect is independent of the presence of immune cells as we observed the same heterogeneity in size at 16 and 20\u2009weeks in control mice. In addition, there was no correlation between the size of HIO and the percentage of human immune cells in the peripheral blood of humanized mice (Extended Data Fig. 1e<\/a>). By immunohistochemistry (IHC) staining, we demonstrated that human CD45+<\/sup> cells had migrated to the mucosal layer and populated the lamina propria as well as the epithelium and formed cellular aggregates comparable to a human fetal or adult gut immune landscape (Fig. 1c\u2013e<\/a>). Notably, this did not occur within the humanized mouse small intestine (SI) (Extended Data Fig. 1f<\/a>). As demonstrated in our previous study in immunodeficient mice12<\/a><\/sup>, HIOs engrafted and differentiated in humanized mice expressed all major intestinal cell lineages (Extended Data Fig. 2<\/a>).<\/p>\n\nFig. 1: Integrating immune cells in HIO using immune system-humanized mouse model.<\/b><\/figcaption>\n![\"Science](\"http:\/\/media.springernature.com\/lw685\/springer-static\/image\/art%3A10.1038%2Fs41587-022-01558-x\/MediaObjects\/41587_2022_1558_Fig1_HTML.png\")
<\/picture><\/a><\/div>\na<\/b>, Transplanted HIOs at 12, 16 and 20\u2009weeks with mouse kidney seen underneath from control or humanized mice. b<\/b>, Graph represents the length of HIOs from control (black circle) or humanized (red square) group at 12\u2009weeks (n<\/i>\u2009=\u20096 control and 7 humanized mice), 16\u2009weeks (n<\/i>\u2009=\u20099 control and 9 humanized mice) and 20 weeks (n<\/i> = 4 control and 3 humanized mice) post-transplantation. Mean\u2009\u00b1\u2009s.d. Multiple Mann\u2013Whitney tests (two-sided); P<\/i>\u2009=\u20090.1666 for 12\u2009weeks, P<\/i>\u2009=\u20090.7120 for 16\u2009weeks and P<\/i> = 0.714 for 20 weeks. NS, not significant. Graph representative of at least three independent experiments. c<\/b>, Formalin-fixed paraffin-embedded sections of transplanted HIO at 12 (n<\/i> = 5), 16\u2009(n<\/i> = 7) and 20 (n<\/i> = 3) weeks stained by IHC with anti-human CD45 antibody. Scale bars, 100\u2009\u03bcm. Representative of at least three independent experiments. d<\/b>, Human fetal intestine at 14.7 and 20.7\u2009PCW stained, by immunofluorescence, with anti-human CDH1 (E-cadherin) (blue), anti-human CD45 (green) antibodies and DAPI (white). Scale bars, 100\u2009\u03bcm. Representative of two samples. e<\/b>, Human adult jejunum stained by IHC with anti-human CD45 antibody. Scale bar, \u2009100 \u03bcm. Representative of three samples.<\/p>\n<\/div>\n
Full size image<\/span><\/a><\/p>\n<\/figure>\n<\/div>\nMass cytometry analysis reveals a GALT-like immune signature profile in HIO<\/h3>\n
By mass cytometry, we next determined the immune signature in transplanted HIOs, and this was compared with the humanized mouse SI as an internal control. Using unsupervised clustering analysis, human CD45+<\/sup> immune cells were classified and organized into 13 clusters as visualized in a heatmap and uniform manifold approximation and projection (UMAP) graph (Fig. 2a,b<\/a>). Based on the combination and level of expression of markers on the heatmap and the expression patterns on individual UMAP graphs (Fig. 2a<\/a> and Extended Data Figs. 3<\/a> and 4<\/a>), we were able to define 12 immune cell types corresponding to the major immune cell lineages found in the intestine, for instance, T cells, B cells and innate lymphoid cells, as well as tissue-specific immune cells such as intraepithelial lymphocytes, mucosal-associated invariant T-like cells and LTi-like cells (Fig. 2a,b<\/a>). To compare the immune profile between HIO and humanized mouse SI post HIO transplantation, we determined the proportion of each immune cell type in each group over time (Fig. 2c<\/a>). Our results revealed a comparable immune cell composition between the groups, except that B cells and CD4+<\/sup> T cells were present in relatively higher numbers in HIOs compared with humanized mouse SI with increasing abundance from 12 to 16\u2009weeks. However, further increases were not seen in the immune profile at 20\u2009weeks post HIO transplantation compared with the profile at 16\u2009weeks. There was no inflammation associated with the increase in CD4+<\/sup> T cells, as demonstrated by a similar cytokine profile in cells isolated from HIOs or humanized mouse SI (Extended Data Fig. 5a,b<\/a>). Notably, in the SI, B cells are mostly found in GALT structures such as lymphoid follicles or Peyer\u2019s patches which also contained CD4+<\/sup> T cells. Our humanized mice lack IL-2R\u03b3, indicating that they lack intestinal lymphoid follicles and Peyer\u2019s patches, resulting in fewer intestinal T and B cells which correlates with our mass cytometry results. Taken together, our findings suggest that the augmentation of B and T cell frequencies observed in HIOs could come from GALT structures.<\/p>\n\nFig. 2: Immune profiling in HIOs and humanized mouse SI by mass cytometry.<\/b><\/figcaption>\n![\"Science](\"http:\/\/media.springernature.com\/lw685\/springer-static\/image\/art%3A10.1038%2Fs41587-022-01558-x\/MediaObjects\/41587_2022_1558_Fig2_HTML.png\")
<\/picture><\/a><\/div>\na<\/b>, Heatmap illustrating the level of expression of each marker (x<\/i> axis) for each cluster corresponding to their identified cell type (y<\/i> axis). b<\/b>, Visualization of high-dimensional data with UMAP overlaid with identified cell types. Heatmap (a<\/b>) and UMAP (b<\/b>) were generated with combined CyTOF dataset from transplanted HIO and humanized mouse SI at 12, 16 and 20 weeks post-transplantation. c<\/b>, Stacked bar graph representing the frequencies of each cell type per tissue and time point. IEL, intraepithelial lymphocyte; LTi, lymphoid tissue inducer cell; ILC, innate immune cell; MAIT, mucosal-associated invariant T cell; NK, natural killer cell; NKT, natural killer T cell; w, weeks.<\/p>\n<\/div>\n
Full size image<\/span><\/a><\/p>\n<\/figure>\n<\/div>\nImmune aggregates in the developing HIO are lymphoid follicle-like structures<\/h3>\n
Lymphoid follicles are well-organized structures formed by aggregation of B cells surrounded by T cells. IHC staining with anti-human CD3 and CD20 antibodies revealed that aggregates found in HIOs contained T and B cells, respectively, at all time points (Fig. 3a,b<\/a>). Surprisingly, in contrast to 12-week-old transplanted HIOs, a distinct cellular zonation of T and B cell populations appeared in 16- as well as 20-week-old HIOs resembling a lymphoid follicle-like structure (Fig. 3a,b<\/a>). However, we did not observe any aggregates of T and B cells in the humanized mouse SI at any time point (Extended Data Fig. 6a,b<\/a>). This result suggests that HIOs promote and influence the formation and maturation of lymphoid follicular structures over the course of their development.<\/p>\n\nFig. 3: Immune cellular aggregates in transplanted HIO contain T and B cells.<\/b><\/figcaption>\n![\"Science](\"http:\/\/media.springernature.com\/lw685\/springer-static\/image\/art%3A10.1038%2Fs41587-022-01558-x\/MediaObjects\/41587_2022_1558_Fig3_HTML.png\")
<\/picture><\/a><\/div>\na<\/b>,b<\/b>, Presence of human T cells (anti-hCD3) (a<\/b>) and human B cells (anti-CD20) (b<\/b>) in HIO at 12 (n<\/i>\u2009=\u20095), 16 (n<\/i>\u2009=\u20097) and 20 (n<\/i>\u2009=3) weeks post-transplantation. Scale bar, 100 \u03bcm. Representative of three independent experiments.<\/p>\n<\/div>\n
Full size image<\/span><\/a><\/p>\n<\/figure>\n<\/div>\nTemporal and spatial development of lymphoid-like structures in transplanted HIOs correlate to human lymphoid follicle development<\/h3>\n
Because the developing HIO is fetal in nature11<\/a><\/sup> and the lumen lacks exposure to antigens and microbiota, we compared the immune aggregates developing in HIO with what has been observed previously in the developing human fetal gut. Lymphoid follicles have been reported to develop in the human fetal gut starting at the second trimester of gestation21<\/a>,22<\/a><\/sup>. In Extended Data Fig. 7<\/a>, we summarized these findings on the developing human fetal gut immune tissue and used them as a reference to compare with the features observed in transplanted HIOs. Starting at 12\u2009weeks post HIO transplant, we found in the lamina propria as well as the epithelium the presence of T cells (mostly CD4+<\/sup> cells) with few scattered CD20+<\/sup> B cells, pointing out a similarity with the cellular composition of a fetal intestine at 11 post conceptual weeks (PCW)21<\/a>,22<\/a><\/sup> (Fig. 4a,b<\/a>). Additionally, in some areas, we observed the colocalization of T and B cells which resemble the aggregates described in the fetal intestine at 14\u201316\u2009PCW (refs. 21<\/a>,22<\/a><\/sup>) (Fig. 4a,c<\/a>). Notably, the cellular zonation of T and B cells observed in the fetal intestine at 19\u2009PCW (refs. 21<\/a>,22<\/a><\/sup>) was found in 16- and 20-week HIOs, where T cell zones are represented with high proportions of CD4+<\/sup> T cells and low CD8+<\/sup> T cells (Fig. 4a,d<\/a>). As described in human fetal intestine21<\/a>,22<\/a><\/sup>, we also observed the presence of plasma cells and few neutrophils in both 16- and 20-week HIOs, demonstrating a more mature and complex cellular composition over the course of HIO differentiation (Fig. 4a,e,f<\/a>). These data confirm that the HIO environment expresses unique signals to influence the maturation of the immune system as well as the formation of a lymphoid tissue.<\/p>\n\nFig. 4: Lymphoid-like structures in transplanted HIOs correlate to lymphoid development in fetal gut.<\/b><\/figcaption>\n![\"Science](\"http:\/\/media.springernature.com\/lw685\/springer-static\/image\/art%3A10.1038%2Fs41587-022-01558-x\/MediaObjects\/41587_2022_1558_Fig4_HTML.png\")
<\/picture><\/a><\/div>\na<\/b>, Prevalence of immune developmental features observed in HIO. Representative of two independent experiments21<\/a>,22<\/a><\/sup>. b<\/b>\u2013f<\/b>, Images illustrating features observed in HIO that resemble immune cell development described in fetal gut. b<\/b>\u2013d<\/b>, HIO sections stained with anti-human CD3 (top) to demonstrate the presence of T cells, anti-human CD4\/CD8 (middle) to distinguish T helper cells versus T cytotoxic cells and anti-human CD20 (bottom) to highlight the presence of B cells. e<\/b>, Hematoxylin and eosin (H&E) staining indicating the presence of neutrophils (black arrowheads) observed in late developing HIOs only. f<\/b>, H&E staining highlight the presence of plasma cells (black arrowheads, top) then confirmed with anti-human MUM1 IHC staining (bottom). Scale bars, 50\u2009\u03bcm. Representative of two independent experiments.<\/p>\n<\/div>\nFull size image<\/span><\/a><\/p>\n<\/figure>\n<\/div>\nM cells are induced in transplanted HIOs after microbial exposure<\/h3>\n
M cells are expressed in the epithelium overlying follicles and play a key role in immune responses by transporting antigens from the lumen to the lamina propria2<\/a><\/sup>. Crosstalk between lymphocytes and epithelial cells is known to promote M cell differentiation23<\/a><\/sup>. By histology, we were not able to detect GP2, a marker for M cells, at baseline in HIOs post-transplantation. However, in organoids derived from transplanted HIO epithelium, called enteroids, we found that GP2 was induced in vitro and expressed at a high level in enteroids derived from HIOs transplanted in humanized mice compared with control mice (Extended Data Fig. 8<\/a>).<\/p>\nGP2 is a transcytotic receptor and binds FimH, an adhesin molecule expressed by Escherichia coli<\/i>. It was reported that the number of M cells increased after bacterial challenge24<\/a><\/sup>. Therefore, we aimed to determine if bacterial exposure could induce M cells in HIOs at 16\u2009weeks post-transplantation. Indeed, we found that 72\u2009h after administration of E. coli<\/i> lysate in HIO lumen, GP2 was detected on the cell surface of epithelial cells overlying immune cells, indicating the presence of M cells (Fig. 5a,b<\/a>). In contrast, transplanted HIOs injected with saline only demonstrated background expression, probably induced by antigens present in the environment at the time of injection. In line with our histology results, we confirmed by quantitative PCR (qPCR) the induction of GP2<\/i> expression in transplanted HIOs after E. coli<\/i> exposure compared with the saline group (Fig. 5d<\/a>). M cell-mediated translocation of luminal antigens is required to initiate the production of IgA antibodies2<\/a><\/sup>. To determine whether M cells were functional and able to translocate luminal antigens to activate immune cells, we measured the level of IgA secreted in the mucus of transplanted HIOs in response to E. coli<\/i> lysate injection. Even though plasma cells were present in both groups, unlike in the saline-treated group, we found that IgA antibodies were present at a high level in the mucus of HIOs exposed to E. coli<\/i> lysate (Fig. 5c,e<\/a>). This result indicates that M cells are functional in translocating antigens to activate immune cells, which subsequently respond to the microbial exposure by activating plasma cells to produce IgA antibodies. We then confirmed in the E. coli<\/i>-treated group a colocalization of M cells and B cells by immunofluorescence, indicating a crosstalk between epithelial and immune cells (Fig. 5f<\/a>).<\/p>\n\nFig. 5: M cells are induced after microbial exposure in transplanted HIOs.<\/b><\/figcaption>\n![\"Science](\"http:\/\/media.springernature.com\/lw685\/springer-static\/image\/art%3A10.1038%2Fs41587-022-01558-x\/MediaObjects\/41587_2022_1558_Fig5_HTML.png\")
<\/picture><\/a><\/div>\na\u2013c<\/b>, 16-week HIO sections at 72 h post injection with saline (n<\/i> = 3) (left) or E. coli<\/i> lysate (n<\/i> = 4) (right) stained with anti-human CD45 (a<\/b>), anti-human GP2 (b<\/b>) and anti-MUM1 (c<\/b>). Arrows indicate M cells positive for human GP2. Scale bars represent 100 \u00b5m, except b<\/b> (bottom) human GP2 scale bars represent 50 \u00b5m. Representative of two independent experiments. d<\/b>, Graph represents level of human GP2<\/i> gene expression evaluated by qPCR in 16-week HIO at 72 h post injection with saline (n<\/i> = 3) or E. coli<\/i> lysate (n<\/i> = 6). Human GP2<\/i> gene expression is normalized to human GAPDH<\/i> gene. Mean \u00b1 s.d. Mann\u2013Whitney test (two-sided); P<\/i> = 0.1667. NS, not significant. e<\/b>, Level of human IgA measured by ELISA in mucus from 16-week HIO 72 h after being injected with saline (n<\/i> = 2) or E. coli<\/i> lysate (n<\/i> = 4). Mean \u00b1 s.d. Mann\u2013Whitney tests (two-sided); P<\/i> = 0.533. NS, not significant. f<\/b>, Immunofluorescence staining with anti-human CDH1 (white), anti-human GP2 (red) and anti-human CD20 (blue) in 16-week HIOs injected with E. coli<\/i> lysate (n<\/i> = 4) for 72 h. Scale bar, 50 \u00b5m. Representative of two independent experiments.<\/p>\n<\/div>\n
Full size image<\/span><\/a><\/p>\n<\/figure>\n<\/div>\n<\/div>\n<\/div>\n\nDiscussion<\/h2>\n\nHere, we report a next-generation in vivo model of HIOs with functional human immune tissue. Using this model, we demonstrate that immune cells temporally infiltrate and populate the HIO in the mucosa similarly to the immune landscape in the developing human gut. Mass cytometry data and immunostaining demonstrated the formation of GALT-like structures during HIO development, and their formation correlated temporally and spatially with human intestinal immune tissue development. Because M cells are present in the epithelium overlaying the GALT, we confirmed their presence and validated their function by exposing the HIO lumen to microbial components.<\/p>\n
Intestinal immune tissue contains diverse immune cell types to protect the mucosal barrier against invaders25<\/a><\/sup>. The human fetal gut has a diverse, functional immune system even without being exposed to antigens26<\/a>,27<\/a>,28<\/a>,29<\/a>,30<\/a><\/sup>. Our mass cytometry analysis indicated that the immune cell composition was diverse and increased in cell number as the transplanted HIOs developed, similar to the human fetal gut. Our results correlate with recent mass cytometry and transcriptomic data showing an increase in immune cell number and cell subsets, in particular CD4+<\/sup> T cells, during human fetal intestinal development29<\/a>,30<\/a><\/sup>. During HIO development, we found by mass cytometry an increase in CD4+<\/sup> T and B cell frequencies and we confirmed by histology that these cell types together formed aggregates that later developed into lymphoid follicles. These features were not identified in the intestine of humanized mice, confirming reports describing the absence of lymphoid follicles in this model19<\/a>,20<\/a><\/sup>.<\/p>\nThe timing of humanization and transplantation did not affect the timing of GALT structure formation. In a preliminary study in which HIO transplantation was delayed until the immune system was fully reconstituted, histology and flow cytometry studies indicated that the development of immune cellular aggregates composed of T and B cells was similar to that of early HIO transplantation (Extended Data Fig. 9<\/a>). These observations indicate that immune precursors were present even after the immune system was reconstituted. Taken together, our results suggest that the developing HIO and cells seeding the lamina propria influence and frame the formation of immune aggregates by expressing specific signals to attract and activate immune precursors as well as serving as a foundation for subsequent lymphoid tissue development.<\/p>\nRecently, transcriptomic analysis of human fetal intestine has identified subsets of fibroblasts in GALT formation that correspond to lymphoid tissue organizer cells30<\/a><\/sup>. These cells increased in number with fetal development and interacted with diverse immune cell populations, including LTi cells via the expression of CCL19, CCL21 and CXCL13 (ref. 30<\/a><\/sup>). In our model, preliminary data indicated that CCL19<\/i>, CCL21<\/i> and CXCL13<\/i> were expressed in HIOs at 16\u2009weeks post-transplantation, whereas their mouse gene orthologs were decreased by at least tenfold (Extended Data Fig. 10a,b<\/a>). These results indicate that, in contrast to the humanized mouse SI, the HIOs express unique microenvironment signals that support the formation of GALT-like structures. In line with this, our model could be used to dissect the mechanism of interaction between stromal lymphoid tissue organizer cells and LTi cells and to identify key pathways in the development of GALT in humans.<\/p>\nGALT is a site of adaptive immune responses and influences the epithelial cells overlaying the follicles to promote the generation of M cells that transport luminal antigens to activate immune cells2<\/a><\/sup>. GP2 is a glycoprotein expressed on the apical side of M cells and translocates luminal antigens to the immune cells in the lamina propria. Even though transplanted HIOs expressed the GP2<\/i> gene, we were not able to detect the protein by IHC likely due to the lack of microbial antigens in the lumen, similar to a fetal intestine11<\/a><\/sup>. Of note, nonhumanized HIOs exposed to microbial antigens in other models being developed in the laboratory also do not express GP2 protein (data not shown). Using our current model, we demonstrated that exposing the HIO lumen to E. coli<\/i> lysate induced M cells, characterized by GP2 expression as well as activation of an immune response indicated by the presence of IgA antibodies. After being produced by plasma cells in the lamina propria, IgA antibodies bind to polymeric immunoglobulin receptors (pIgR) expressed on the basal side of epithelial cells and, by transcytosis, epithelial cells then secrete IgA antibodies in the intestinal lumen31<\/a><\/sup>. Because we detected the presence IgA antibodies in the mucus, our findings suggest that epithelial cells in HIOs express pIgR and transport antibodies to the lumen.<\/p>\nHere, we have demonstrated that our in vivo HIO model with GALT-like structures and M cells provides opportunities for studies on food allergies, intestinal infectious diseases and the development of mucosal vaccines.<\/p>\n<\/div>\n<\/div>\n
\nMethods<\/h2>\n\nAnimals<\/h3>\n
Immunodeficient NOD\/Scid Il2rg null Tg (hIL3, hGM-CSF and hSCF) (NSGS; The Jackson Laboratory no. 013062 (ref. 17<\/a><\/sup>)) mice were bred and housed in the animal facility at the Cincinnati Children\u2019s Hospital Medical Center (CCHMC). All experiments were performed with the approval of the Institutional Animal Care and Use Committee of CCHMC (IACUC protocol no. 2018-0092).<\/p>\nHuman intestinal tissue<\/h3>\n
Normal, de-identified developing human fetal tissues from elective abortions were obtained from the University of Washington, Laboratory of Developmental Biology, and all work was approved by the University of Washington and the University of Michigan Institutional Review Boards (IRBs).<\/p>\n
Normal adult human jejunum was obtained from patients undergoing bariatric procedures between the ages of 16 and 25\u2009yr. Informed consent or assent was obtained from all patients and\/or parents\/legal guardians as appropriate. Human tissue collection was performed with the previous approval of CCHMC\u2019s IRB.<\/p>\n
HIOs<\/h3>\n
HIOs were generated and maintained as previously described12<\/a>,32<\/a>,33<\/a><\/sup>.<\/p>\nBriefly, human H1 embryonic stem cells (WA-01; WiCell) (passage number 40\u201355), obtained from the Pluripotent Stem Cell Facility in our institute, were grown in feeder-free conditions in mTESR1 medium (Stem Cell Technologies). For induction of definitive endoderm, cells were split with Accutase (Invitrogen) and plated at a density between 70,000 and 100,000 cells per well in a Matrigel-coated 24-well plate. Once the cells reached 80\u201395% confluency, they were treated with 100\u2009ng\u2009ml\u22121<\/sup> Activin A for 3\u2009d as previously described34<\/a><\/sup>. Definitive endoderm was then treated for 4\u2009d with hindgut induction medium containing 500\u2009ng\u2009ml\u22121<\/sup> FGF4 (R&D) and 3\u2009\u03bcM Chiron 99021 (Tocris) to induce formation of mid-hindgut spheroids. Spheroids were then plated in Matrigel (Corning) and maintained in intestinal growth medium supplemented with 100\u2009ng\u2009ml\u22121<\/sup> EGF (R&D) to generate HIOs. Medium was changed twice weekly and HIOs were replated in fresh Matrigel at 14\u2009d.<\/p>\nHuman umbilical cord blood cell engraftment<\/h3>\n
The Translational Trials Development Support Laboratory of CCHMC collected and distributed the umbilical cord blood units according to an IRB-approved protocol (protocol no. 02-3-4x). Mice were humanized as described previously35<\/a><\/sup>. Briefly, whole cord blood was subjected to hetastarch-induced aggregation of red blood cells. Cord blood cells (CBCs) were isolated, washed and viably frozen for future use. Thawed CBCs were resuspended in IMDM medium with 3% FBS and antibiotics and diluted to 23.3\u201328.3\u2009\u00d7\u2009106<\/sup> CBCs per ml. OKT3 antibody was spiked into the cell solution at a concentration of 1\u2009\u00b5g per 1\u2009million cells to prevent graft-versus-host disease. Immunodeficient NSGS mice at 6\u20138\u2009weeks old were conditioned by receiving a dose of busulfan (30\u2009mg\u2009kg\u22121<\/sup> by intraperitoneal injection) 24\u2009h before intravenous injection of 7\u20138.5\u2009\u00d7\u2009106<\/sup> CBCs in 300\u2009\u00b5l. Busulfan is a chemotherapy drug that partially depletes cells from the mouse bone marrow and allows the human hematopoietic stem cells to engraft. As a preventative measure against possible systemic infections, mice were fed with doxycycline chow (0.0625%, Purina) for 2\u2009weeks before and after human cell engraftment.<\/p>\nTransplantation of HIOs<\/h3>\n
As previously described, a single HIO, matured in vitro for 28\u2009d, was removed from Matrigel and then transplanted under the kidney capsule12<\/a><\/sup>. Briefly, the mice were anesthetized with 2% inhaled isoflurane (Butler Schein) and 2.5\u20133\u2009l\u2009min\u22121<\/sup> oxygen. The left side of the mouse was then prepped in sterile fashion with isopropyl alcohol and providine-iodine. A small left-posterior subcostal incision was made to expose the kidney. A subcapsular pocket was created in the kidney capsule and the HIO was then placed into the pocket. The kidney was then returned to the peritoneal cavity and the mice were given an intraperitoneal flush with 2\u20133\u2009ml of piperacillin\/tazobactam (100\u2009mg\u2009kg\u22121<\/sup>; Pfizer) to help prevent bacterial infection. The skin was closed in a double layer. For pain control, mice were then given a subcutaneous injection with buprenorphine (0.05\u2009mg\u2009kg\u22121<\/sup>; Midwest Veterinary Supply) or carprofen (4\u2009mg\u2009kg\u22121<\/sup>; Midwest Veterinary Supply) and were monitored for the next 48\u2009h following surgery. Additional injections of pain medication were given if needed. At 12 and 16\u2009weeks following engraftment, the mice were then euthanized and the tissues were collected and analyzed.<\/p>\n\n E. coli<\/i> lysate preparation and injection<\/h4>\n
Grown in LB medium overnight, E. coli<\/i> suspension was then washed three times with PBS and centrifuged at 800 g<\/i> for 5\u2009min. Bacteria were resuspended in saline solution at ~107<\/sup> colony-forming units per ml and centrifuged at 10,000 g<\/i> for 5\u2009min at 4\u2009\u00b0C. Bacteria were then lysed by freeze\u2013thaw cycles repeated four times.<\/p>\nAt 16\u2009weeks post HIO transplantation, the mice were anesthetized and a small left-posterior subcostal incision was made to expose the transplanted HIO. Then, 50\u2013100\u2009\u00b5l of E. coli<\/i> lysate was injected with a 0.5-ml insulin syringe in the HIO lumen.<\/p>\n
The HIO was then returned to the peritoneal cavity and the skin was closed in a double layer. For pain control, mice were then given a subcutaneous injection with carprofen (4\u2009mg\u2009kg\u22121<\/sup>; Midwest Veterinary Supply) and were monitored for the next 48\u2009h following surgery. Additional injections of pain medication were given if needed. At 72\u2009h post injection, the mice were euthanized and the tissues were collected and analyzed.<\/p>\nFlow cytometry<\/h3>\n
To confirm the expression of human immune cells from CD34+<\/sup> engraftment in mice, retro-orbital bleeding was performed 8\u201310\u2009weeks after the engraftment and a day before each tissue collection. Approximately 50\u2009\u03bcl of whole blood was collected in a BD Microtainer Dipotassium\/EDTA-coated tube and then lysed in 5\u2009ml of red blood cell lysis buffer (155\u2009mM NH4<\/sub>Cl, 12\u2009mM NaHCO3<\/sub> and 0.1\u2009mM EDTA pH 8.0; diluted in dH2<\/sub>O), for 5\u2009min at room temperature. An isovolume of PBS was added and tubes were centrifuged at 300 g<\/i> for 5\u2009min. The pellet was washed in PBS, centrifuged and resuspended in FACS buffer (0.5% BSA, 2\u2009mM EDTA in PBS 1\u00d7). Cells were stained for 30\u2009min on ice with the following combination of antibodies: FITC-conjugated anti-human CD45, PE-Cy5-conjugated anti-mouse CD45, BV650-conjugated anti-human CD19, PE-Cy7-conjugated anti-human CD3, BV421-conjugated anti-human CD56, PE-conjugated anti-human CD13, PE-conjugated anti-human CD33 and Zombie NIR fixable viability kit to exclude dead cells (Supplementary Table<\/a>). All antibodies were used at 1:200 dilution except Zombie NIR which was used at 1:2,000 dilution. Samples were washed twice and resuspended in FACS buffer. The samples were then recorded on an LSR Fortessa instrument (BD Biosciences) and the data were analyzed with FlowJo software (TreeStar).<\/p>\nHIO and SI dissociation and cell preparation<\/h3>\n
One-third of the HIO was used for cell dissociation. The cell dissociation protocol for the HIOs was modified from Weigmann\u2019s protocol for mouse colon cell dissociation36<\/a><\/sup>. Briefly, HIOs were cut into small pieces and incubated under slow rotation for 20\u2009min at 37\u2009\u00b0C in 5\u2009ml of predigestion solution containing EDTA and dithiothreitol in HBSS. The epithelial cell suspension was filtered through a 100-\u03bcm cell strainer, washed with cold PBS and kept on ice until pooled with the lamina propria-isolated cells. The remaining pieces of HIO were minced and placed in a new tube with 5\u2009ml of digestion solution containing collagenase D (Roche), DNAse I (Roche) and dispase II (Roche) in PBS and incubated for 15\u2009min at 37\u2009\u00b0C under slow agitation. The lamina propria cell suspension was passed through a 70-\u03bcm cell strainer, washed with cold PBS and re-incubated in 5\u2009ml of digestion solution for 15\u2009min at 37\u2009\u00b0C under slow agitation. This step was repeated one more time. Epithelial and lamina propria cells were combined in one tube and spun down at 450 g<\/i> for 5\u2009min.<\/p>\nThe method to isolate mononuclear cells from the humanized mouse gut was adapted from the protocol reported by Lee et al.37<\/a><\/sup>. Briefly, 10\u201312\u2009cm of proximal SI was collected, longitudinally cut open and washed in HBSS to remove any debris. The SI was incubated in HBSS containing 5\u2009mM EDTA on ice for 5\u2009min and vortexed at medium intensity, for a total of four incubations. The epithelium was then collected in a separate tube. After digestion of the tissue at 37\u2009\u00b0C for 30\u2009min with DNAse I (Roche) and Collagenase A (Roche), the homogenate was filtered and centrifuged at 450 g<\/i> for 5\u2009min.<\/p>\nBoth HIO and SI cell suspensions were individually resuspended in 44% Percoll (GE), loaded on a 67% Percoll layer and centrifuged at 650 g<\/i> without brake for 30\u2009min at room temperature. The cell layer at the interface of the two gradients was collected, washed, counted and used for in vitro assays, or stained for immune markers and analyzed by mass cytometry.<\/p>\nMass cytometry staining and data analysis<\/h3>\nMass cytometry staining<\/h4>\n
Antibodies used in this panel were purchased from Fluidigm except anti-human CD45RO antibody which was purchased from BioLegend and was labeled with Maxpar X8 Antibody Labeling Kit (Fluidigm) according to the manufacturer\u2019s instructions. All the reagents used in the following protocol were purchased from Fluidigm and all incubations were done at room temperature. Samples were first stained for 5\u2009min with Cell-ID Cisplatin at a final concentration of 5\u2009\u03bcM in Maxpar PBS and then washed in 5 volume of Maxpar Cell Staining Buffer. Before being stained for cell surface markers, cells were fixed in Fix I Buffer for 10\u2009min, washed twice with Maxpar Barcode Perm Buffer and barcoded with Cell-ID 20-Plex Pd Barcoding Kit for 30\u2009min. Samples were pooled and then stained for 30\u2009min with a cocktail of 25 antibodies in a final volume of 100\u2009\u03bcl of Maxpar Cell Staining Buffer (antibody dilution 1:200) (Supplementary Table<\/a>). After cell surface staining, samples were washed, fixed in 1.6% paraformaldehyde (PFA) (EMS) for 10\u2009min and, finally, incubated in 1\u2009ml of intercalator solution at a final concentration of 125\u2009nM. The samples were kept cold and shipped in intercalator solution to the Cytometry Research Core Facility at the University of Rochester, NY. The samples were washed and resuspended in Maxpar water before being acquired using a cytometry by time of flight (CyTOF)\/mass cytometry Helios instrument (Fluidigm). Data were then debarcoded using Fluidigm debarcoder software (available online).<\/p>\nBioinformatics analysis<\/h4>\n
CyTOF datasets from transplanted HIO- and humanized mouse SI-isolated immune cells were generated using mass cytometry (Fluidigm) variation at three time points, 12\u2009weeks, 16\u2009weeks and 20\u2009weeks. Datasets at aforementioned time points were generated in two batches for each tissue at 12 and 16\u2009weeks post-transplantation with a total of n<\/i>\u2009=\u20097 samples and n<\/i>\u2009=\u200910 samples, respectively, and in one batch for each tissue at 20\u2009weeks post-transplantation. Combining two batches, ~1.1\u2009million cells were analyzed, amounting to 63,076 cells in 12-week HIO, 350,132 cells in 16-week HIO, 148,274 cells in 20-week HIO, 137,887 cells in 12-week SI, 368,893 cells in 16-week SI and 71,569 cells in 20-week SI.<\/p>\n
Raw .fcs files for each sample were read and analyzed using read.flowSet<\/i> function from the flowCore R (v.3.6.3) package. The .fcs files stored in flow.set<\/i> objects were than normalized using inverse hyperbolic sine (Arcsinh<\/i>) transformation (https:\/\/support.cytobank.org\/hc\/en-us\/articles\/206148057-About-the-Arcsinh-transform<\/a>) using a cofactor value of 5. Normalized reads from the .fcs files were then used to create a Seurat object using CreateSeuratObject<\/i> function from Seurat (v.3.0.2), a single-cell analysis package in R.<\/p>\nFor each Seurat object, the following steps were carried out: variable features set to surface markers, scaling, dimension reduction and clustering. Dimension reduction was applied in both linear (principal components) and nonlinear (UMAP) approaches to obtain respective components. Clustering was performed using a k<\/i>-nearest neighbor graph of cells based on marker abundance similarity. This k<\/i>-nearest neighbor graph was finally partitioned into clusters based on highly interconnected colonies of cells.<\/p>\n
In the end, all the processed Seurat objects for each sample from two batches were integrated together using Seurat\u2019s reciprocal principal component analysis as the numbers of cells were large (~1.1\u2009million). Briefly, each dataset was projected into principal component space of other datasets to learn the anchors (cell pairs) based on the mutual nearest neighbor graph requisite.<\/p>\n
Dimension reduction plots were generated using DimPlot<\/i> function and heatmaps were generated using the DoHeatmap<\/i> function in Seurat.<\/p>\n
Example datasets and analysis code have been deposited to GitHub: https:\/\/github.com\/praneet1988\/Analyze_CyTOF_Using_Seurat.<\/a><\/p>\nMeasurement of cytokines from HIO- and mouse SI-derived immune cells<\/h3>\nDetection of cytokines by flow cytometry<\/h4>\n
Immune cells isolated from HIO or mouse SI were stimulated for 4\u2009h with 1:500 Cell Activation Cocktail (BioLegend) containing, according to the manufacturer, an optimized concentration of phorbol 12-myristate-13-acetate (PMA) and ionomycin. After an hour of stimulation, cells were then incubated with 1:1,000 Brefeldin A Solution (Biolegend) to block the secretion of cytokines. Finally, the cells were stained with anti-human CD3 and anti-human CD4 antibodies (Supplementary Table<\/a>), fixed with Cytofix (BD Biosciences) and then permeabilized with CytoPerm (BD Biosciences) overnight at 4\u2009\u00b0C. The following day anti-IFN\u03b3, anti-TNF\u03b1 and anti-IL-2 antibodies (Supplementary Table<\/a>) were added to the cells for at least 1\u2009h at room temperature. Samples were washed twice and resuspended in FACS buffer. The samples were then recorded on an Aurora instrument (Cytek) and the data were analyzed with FlowJo software (TreeStar).<\/p>\nDetection of cytokines by Milliplex assay<\/h4>\n
Indicated cytokines were measured in supernatants from immune cells (cell density 106<\/sup> per ml) isolated from HIO or mouse SI stimulated for 3\u2009d with a cocktail of anti-human CD3\/anti-human CD28 antibodies (STEMCELL Technologies). The presence of cytokines in supernatants was measured using Milliplex kits (Millipore), following the manufacturer\u2019s instructions.<\/p>\nIn vitro HIO-derived enteroid culture (M cell induction)<\/h3>\nHIO-derived enteroid preparation and in vitro expansion<\/h4>\n
At 20\u2009weeks post-transplantation, a section of transplanted HIOs was used to isolate the crypts following our protocol for human intestinal tissue38<\/a><\/sup>. Briefly, the mucosal layer from portions of transplanted HIOs was dissected under a microscope and scraped to remove the villi and debris. The mucosa was then incubated with freshly prepared 2\u2009mM EDTA solution and gently shaken for 30\u2009min. After several washes with ice-cold chelation buffer, the intestinal crypts were collected by gently scraping the mucosa with curved forceps and filtered twice through a 150-\u03bcm nylon mesh to remove any debris. Due to limited amounts of tissues, the crypts collected from each group were pooled. The crypts were then washed in ice-cold chelation buffer and 50\u2009\u03bcl of crypts, resuspended in Matrigel (Corning), was added per well in a 24-well plate. After polymerization of the Matrigel, 500\u2009\u03bcl of human IntestiCult Organoid Growth medium (STEMCELL Technologies) was added to each well. The crypts were cultured and expanded for 10\u201314\u2009d before being frozen down for later use.<\/p>\nM cell induction in vitro<\/h4>\n
Enteroids were plated on Transwells as described previously7<\/a>,39<\/a><\/sup>. Briefly, after being washed from Matrigel, enteroids were fragmented and plated on human collagen IV-precoated 24-well Transwells (0.4-\u03bcm pore size) and incubated in IntestiCult Organoid Growth medium (STEMCELL Technology) at 37\u2009\u00b0C until monolayer confluence was reached. To induce M cell differentiation, enteroid monolayers were cultured in differentiation medium supplemented with 50\u2009ng\u2009ml\u22121<\/sup> TNF-\u03b1 and 100\u2009ng\u2009ml\u22121<\/sup> RANK-L for a period of 5\u2009d, as previously described39<\/a><\/sup>.<\/p>\n
The intestine constitutes the largest compartment of the immune system, organized in the lamina propria, epithelium and lymphoid follicles defined by the gut-associated lymphoid tissue (GALT). Immune cell types and GALT are regionally distributed across the mucosal layers and along the gut tube1<\/a><\/sup>. Immune\u2013epithelial crosstalk is essential to the maintenance of intestinal homeostasis, defense against pathogens and immunologic tolerance to dietary components and commensal bacteria. For instance, specialized epithelial microfold (M) cells, located in the follicle-associated epithelium, play a central role in gut immune sensing by transporting luminal antigens in the lamina propria to activate immune cells2<\/a><\/sup>. Immune\u2013epithelial crosstalk mediated by cytokines is involved in gastrointestinal tissue development, homeostasis and disease states3<\/a>,4<\/a>,5<\/a><\/sup>. Although experimental mouse models have been widely used to dissect the biological mechanisms of intestinal immune tissue development and disease, there are still substantial differences that prevent direct extrapolation to human biology6<\/a><\/sup>. Similarly, coculture of HIOs and immune cells has not fully recapitulated the complexity of human immune intestinal tissue7<\/a>,8<\/a>,9<\/a><\/sup>. Deeper understanding of the perturbations of the gastrointestinal immune system that commonly result in chronic human diseases requires new human models that represent patient-specific gastrointestinal immune tissue.<\/p>\n During the past decade, pluripotent stem cell (PSC)-derived HIOs have advanced our knowledge of human intestinal development, physiology and diseases10<\/a>,11<\/a><\/sup>. A key feature of the PSC-derived HIO model is the ability to generate complex human intestinal tissue. We have previously developed and validated an in vivo model that uses transplantation of in vitro PSC-derived HIOs into immunocompromised mice. We demonstrated the induction of intestinal maturation in transplanted HIOs, characterized by the presence of stem cells and functionally differentiated epithelial cells, as well as a crypt\/villus architecture with mesenchymal submucosal layers including lamina propria12<\/a>,13<\/a>,14<\/a><\/sup>. However, this system lacked the developing immune cells that would normally be found in the developing fetal intestine. To overcome this limitation, here we report a next-generation HIO system containing human immune cells, obtained by growing HIOs in mice with a humanized immune system.<\/p>\n Humanized immune system mouse models are commonly used to investigate human hematopoiesis or inflammatory diseases and can be generated by the engraftment of human peripheral blood leukocytes, hematopoietic stem cells or fetal tissues (bone marrow, thymus and\/or liver) in lymphopenic mice such as the NOD\/SCID\/Il2rg\u2212\/\u2212<\/sup> (NSG) mouse strain15<\/a>,16<\/a><\/sup>. Recently, Wunderlich et al. reported that transgenic expression of human SCF, GM-CSF and IL-3 in NSG mice, named NSGS mice, improves hematopoietic engraftment, reconstitution and function17<\/a>,18<\/a><\/sup>. However, because these lymphopenic mice do not express Il2rg<\/i>, resulting in impaired signaling in lymphoid tissue inducer (LTi) cells, they do not have Peyer\u2019s patches and lymphoid follicles in their intestine19<\/a>,20<\/a><\/sup>. The lack of GALT in these mouse models represents a major limitation to understanding numerous human gastrointestinal diseases. Here, we aimed to bridge the gap by integrating human immune cells in a human intestinal tissue.<\/p>\n In our current model, we performed a serial time course following transplantation of HIOs under the kidney capsule to characterize the human immune cells that infiltrate the HIO lamina propria and epithelium. Mass cytometry and immunostaining confirmed the presence of GALT-associated B cells in cellular immune aggregates present in transplanted HIOs, resembling lymphoid follicles developing in human fetal intestine. Finally, we demonstrated that microbial exposure in the HIO lumen is required to induce the epithelial expression of glycoprotein 2 (GP2) at the cell surface of M cells, which consequently activate immune cells in the lamina propria. Altogether, these results demonstrate that crosstalk between the HIO epithelium and the immune cells induces the formation of lymphoid-like structures and M cell differentiation and function.<\/p>\n<\/div>\n<\/div>\n To bioengineer an HIO with immune cells, we transplanted PSC-derived HIOs under the kidney capsule of humanized NSGS mice (see workflow in Extended Data Fig. 1a<\/a>). We collected and analyzed HIOs at 12, 16 and 20\u2009weeks post-transplant. We confirmed the presence of human immune cells in the blood of humanized mice at each time point (Extended Data Fig. 1b\u2013d<\/a>) and found that HIOs successfully engrafted and grew to a similar size compared with those grown in nonhumanized control mice (Fig. 1a,b<\/a>). However, we noticed that HIOs transplanted for periods of 16 and 20\u2009weeks were more heterogeneous in size and this phenomenon may be due to an accumulation of mucus within the HIO lumen which cannot be drained out. Regardless, the effect is independent of the presence of immune cells as we observed the same heterogeneity in size at 16 and 20\u2009weeks in control mice. In addition, there was no correlation between the size of HIO and the percentage of human immune cells in the peripheral blood of humanized mice (Extended Data Fig. 1e<\/a>). By immunohistochemistry (IHC) staining, we demonstrated that human CD45+<\/sup> cells had migrated to the mucosal layer and populated the lamina propria as well as the epithelium and formed cellular aggregates comparable to a human fetal or adult gut immune landscape (Fig. 1c\u2013e<\/a>). Notably, this did not occur within the humanized mouse small intestine (SI) (Extended Data Fig. 1f<\/a>). As demonstrated in our previous study in immunodeficient mice12<\/a><\/sup>, HIOs engrafted and differentiated in humanized mice expressed all major intestinal cell lineages (Extended Data Fig. 2<\/a>).<\/p>\n a<\/b>, Transplanted HIOs at 12, 16 and 20\u2009weeks with mouse kidney seen underneath from control or humanized mice. b<\/b>, Graph represents the length of HIOs from control (black circle) or humanized (red square) group at 12\u2009weeks (n<\/i>\u2009=\u20096 control and 7 humanized mice), 16\u2009weeks (n<\/i>\u2009=\u20099 control and 9 humanized mice) and 20 weeks (n<\/i> = 4 control and 3 humanized mice) post-transplantation. Mean\u2009\u00b1\u2009s.d. Multiple Mann\u2013Whitney tests (two-sided); P<\/i>\u2009=\u20090.1666 for 12\u2009weeks, P<\/i>\u2009=\u20090.7120 for 16\u2009weeks and P<\/i> = 0.714 for 20 weeks. NS, not significant. Graph representative of at least three independent experiments. c<\/b>, Formalin-fixed paraffin-embedded sections of transplanted HIO at 12 (n<\/i> = 5), 16\u2009(n<\/i> = 7) and 20 (n<\/i> = 3) weeks stained by IHC with anti-human CD45 antibody. Scale bars, 100\u2009\u03bcm. Representative of at least three independent experiments. d<\/b>, Human fetal intestine at 14.7 and 20.7\u2009PCW stained, by immunofluorescence, with anti-human CDH1 (E-cadherin) (blue), anti-human CD45 (green) antibodies and DAPI (white). Scale bars, 100\u2009\u03bcm. Representative of two samples. e<\/b>, Human adult jejunum stained by IHC with anti-human CD45 antibody. Scale bar, \u2009100 \u03bcm. Representative of three samples.<\/p>\n<\/div>\n Full size image<\/span><\/a><\/p>\n<\/figure>\n<\/div>\n By mass cytometry, we next determined the immune signature in transplanted HIOs, and this was compared with the humanized mouse SI as an internal control. Using unsupervised clustering analysis, human CD45+<\/sup> immune cells were classified and organized into 13 clusters as visualized in a heatmap and uniform manifold approximation and projection (UMAP) graph (Fig. 2a,b<\/a>). Based on the combination and level of expression of markers on the heatmap and the expression patterns on individual UMAP graphs (Fig. 2a<\/a> and Extended Data Figs. 3<\/a> and 4<\/a>), we were able to define 12 immune cell types corresponding to the major immune cell lineages found in the intestine, for instance, T cells, B cells and innate lymphoid cells, as well as tissue-specific immune cells such as intraepithelial lymphocytes, mucosal-associated invariant T-like cells and LTi-like cells (Fig. 2a,b<\/a>). To compare the immune profile between HIO and humanized mouse SI post HIO transplantation, we determined the proportion of each immune cell type in each group over time (Fig. 2c<\/a>). Our results revealed a comparable immune cell composition between the groups, except that B cells and CD4+<\/sup> T cells were present in relatively higher numbers in HIOs compared with humanized mouse SI with increasing abundance from 12 to 16\u2009weeks. However, further increases were not seen in the immune profile at 20\u2009weeks post HIO transplantation compared with the profile at 16\u2009weeks. There was no inflammation associated with the increase in CD4+<\/sup> T cells, as demonstrated by a similar cytokine profile in cells isolated from HIOs or humanized mouse SI (Extended Data Fig. 5a,b<\/a>). Notably, in the SI, B cells are mostly found in GALT structures such as lymphoid follicles or Peyer\u2019s patches which also contained CD4+<\/sup> T cells. Our humanized mice lack IL-2R\u03b3, indicating that they lack intestinal lymphoid follicles and Peyer\u2019s patches, resulting in fewer intestinal T and B cells which correlates with our mass cytometry results. Taken together, our findings suggest that the augmentation of B and T cell frequencies observed in HIOs could come from GALT structures.<\/p>\n a<\/b>, Heatmap illustrating the level of expression of each marker (x<\/i> axis) for each cluster corresponding to their identified cell type (y<\/i> axis). b<\/b>, Visualization of high-dimensional data with UMAP overlaid with identified cell types. Heatmap (a<\/b>) and UMAP (b<\/b>) were generated with combined CyTOF dataset from transplanted HIO and humanized mouse SI at 12, 16 and 20 weeks post-transplantation. c<\/b>, Stacked bar graph representing the frequencies of each cell type per tissue and time point. IEL, intraepithelial lymphocyte; LTi, lymphoid tissue inducer cell; ILC, innate immune cell; MAIT, mucosal-associated invariant T cell; NK, natural killer cell; NKT, natural killer T cell; w, weeks.<\/p>\n<\/div>\n Full size image<\/span><\/a><\/p>\n<\/figure>\n<\/div>\n Lymphoid follicles are well-organized structures formed by aggregation of B cells surrounded by T cells. IHC staining with anti-human CD3 and CD20 antibodies revealed that aggregates found in HIOs contained T and B cells, respectively, at all time points (Fig. 3a,b<\/a>). Surprisingly, in contrast to 12-week-old transplanted HIOs, a distinct cellular zonation of T and B cell populations appeared in 16- as well as 20-week-old HIOs resembling a lymphoid follicle-like structure (Fig. 3a,b<\/a>). However, we did not observe any aggregates of T and B cells in the humanized mouse SI at any time point (Extended Data Fig. 6a,b<\/a>). This result suggests that HIOs promote and influence the formation and maturation of lymphoid follicular structures over the course of their development.<\/p>\n a<\/b>,b<\/b>, Presence of human T cells (anti-hCD3) (a<\/b>) and human B cells (anti-CD20) (b<\/b>) in HIO at 12 (n<\/i>\u2009=\u20095), 16 (n<\/i>\u2009=\u20097) and 20 (n<\/i>\u2009=3) weeks post-transplantation. Scale bar, 100 \u03bcm. Representative of three independent experiments.<\/p>\n<\/div>\n Full size image<\/span><\/a><\/p>\n<\/figure>\n<\/div>\n Because the developing HIO is fetal in nature11<\/a><\/sup> and the lumen lacks exposure to antigens and microbiota, we compared the immune aggregates developing in HIO with what has been observed previously in the developing human fetal gut. Lymphoid follicles have been reported to develop in the human fetal gut starting at the second trimester of gestation21<\/a>,22<\/a><\/sup>. In Extended Data Fig. 7<\/a>, we summarized these findings on the developing human fetal gut immune tissue and used them as a reference to compare with the features observed in transplanted HIOs. Starting at 12\u2009weeks post HIO transplant, we found in the lamina propria as well as the epithelium the presence of T cells (mostly CD4+<\/sup> cells) with few scattered CD20+<\/sup> B cells, pointing out a similarity with the cellular composition of a fetal intestine at 11 post conceptual weeks (PCW)21<\/a>,22<\/a><\/sup> (Fig. 4a,b<\/a>). Additionally, in some areas, we observed the colocalization of T and B cells which resemble the aggregates described in the fetal intestine at 14\u201316\u2009PCW (refs. 21<\/a>,22<\/a><\/sup>) (Fig. 4a,c<\/a>). Notably, the cellular zonation of T and B cells observed in the fetal intestine at 19\u2009PCW (refs. 21<\/a>,22<\/a><\/sup>) was found in 16- and 20-week HIOs, where T cell zones are represented with high proportions of CD4+<\/sup> T cells and low CD8+<\/sup> T cells (Fig. 4a,d<\/a>). As described in human fetal intestine21<\/a>,22<\/a><\/sup>, we also observed the presence of plasma cells and few neutrophils in both 16- and 20-week HIOs, demonstrating a more mature and complex cellular composition over the course of HIO differentiation (Fig. 4a,e,f<\/a>). These data confirm that the HIO environment expresses unique signals to influence the maturation of the immune system as well as the formation of a lymphoid tissue.<\/p>\n a<\/b>, Prevalence of immune developmental features observed in HIO. Representative of two independent experiments21<\/a>,22<\/a><\/sup>. b<\/b>\u2013f<\/b>, Images illustrating features observed in HIO that resemble immune cell development described in fetal gut. b<\/b>\u2013d<\/b>, HIO sections stained with anti-human CD3 (top) to demonstrate the presence of T cells, anti-human CD4\/CD8 (middle) to distinguish T helper cells versus T cytotoxic cells and anti-human CD20 (bottom) to highlight the presence of B cells. e<\/b>, Hematoxylin and eosin (H&E) staining indicating the presence of neutrophils (black arrowheads) observed in late developing HIOs only. f<\/b>, H&E staining highlight the presence of plasma cells (black arrowheads, top) then confirmed with anti-human MUM1 IHC staining (bottom). Scale bars, 50\u2009\u03bcm. Representative of two independent experiments.<\/p>\n<\/div>\n Full size image<\/span><\/a><\/p>\n<\/figure>\n<\/div>\n M cells are expressed in the epithelium overlying follicles and play a key role in immune responses by transporting antigens from the lumen to the lamina propria2<\/a><\/sup>. Crosstalk between lymphocytes and epithelial cells is known to promote M cell differentiation23<\/a><\/sup>. By histology, we were not able to detect GP2, a marker for M cells, at baseline in HIOs post-transplantation. However, in organoids derived from transplanted HIO epithelium, called enteroids, we found that GP2 was induced in vitro and expressed at a high level in enteroids derived from HIOs transplanted in humanized mice compared with control mice (Extended Data Fig. 8<\/a>).<\/p>\n GP2 is a transcytotic receptor and binds FimH, an adhesin molecule expressed by Escherichia coli<\/i>. It was reported that the number of M cells increased after bacterial challenge24<\/a><\/sup>. Therefore, we aimed to determine if bacterial exposure could induce M cells in HIOs at 16\u2009weeks post-transplantation. Indeed, we found that 72\u2009h after administration of E. coli<\/i> lysate in HIO lumen, GP2 was detected on the cell surface of epithelial cells overlying immune cells, indicating the presence of M cells (Fig. 5a,b<\/a>). In contrast, transplanted HIOs injected with saline only demonstrated background expression, probably induced by antigens present in the environment at the time of injection. In line with our histology results, we confirmed by quantitative PCR (qPCR) the induction of GP2<\/i> expression in transplanted HIOs after E. coli<\/i> exposure compared with the saline group (Fig. 5d<\/a>). M cell-mediated translocation of luminal antigens is required to initiate the production of IgA antibodies2<\/a><\/sup>. To determine whether M cells were functional and able to translocate luminal antigens to activate immune cells, we measured the level of IgA secreted in the mucus of transplanted HIOs in response to E. coli<\/i> lysate injection. Even though plasma cells were present in both groups, unlike in the saline-treated group, we found that IgA antibodies were present at a high level in the mucus of HIOs exposed to E. coli<\/i> lysate (Fig. 5c,e<\/a>). This result indicates that M cells are functional in translocating antigens to activate immune cells, which subsequently respond to the microbial exposure by activating plasma cells to produce IgA antibodies. We then confirmed in the E. coli<\/i>-treated group a colocalization of M cells and B cells by immunofluorescence, indicating a crosstalk between epithelial and immune cells (Fig. 5f<\/a>).<\/p>\n a\u2013c<\/b>, 16-week HIO sections at 72 h post injection with saline (n<\/i> = 3) (left) or E. coli<\/i> lysate (n<\/i> = 4) (right) stained with anti-human CD45 (a<\/b>), anti-human GP2 (b<\/b>) and anti-MUM1 (c<\/b>). Arrows indicate M cells positive for human GP2. Scale bars represent 100 \u00b5m, except b<\/b> (bottom) human GP2 scale bars represent 50 \u00b5m. Representative of two independent experiments. d<\/b>, Graph represents level of human GP2<\/i> gene expression evaluated by qPCR in 16-week HIO at 72 h post injection with saline (n<\/i> = 3) or E. coli<\/i> lysate (n<\/i> = 6). Human GP2<\/i> gene expression is normalized to human GAPDH<\/i> gene. Mean \u00b1 s.d. Mann\u2013Whitney test (two-sided); P<\/i> = 0.1667. NS, not significant. e<\/b>, Level of human IgA measured by ELISA in mucus from 16-week HIO 72 h after being injected with saline (n<\/i> = 2) or E. coli<\/i> lysate (n<\/i> = 4). Mean \u00b1 s.d. Mann\u2013Whitney tests (two-sided); P<\/i> = 0.533. NS, not significant. f<\/b>, Immunofluorescence staining with anti-human CDH1 (white), anti-human GP2 (red) and anti-human CD20 (blue) in 16-week HIOs injected with E. coli<\/i> lysate (n<\/i> = 4) for 72 h. Scale bar, 50 \u00b5m. Representative of two independent experiments.<\/p>\n<\/div>\n Full size image<\/span><\/a><\/p>\n<\/figure>\n<\/div>\n<\/div>\n<\/div>\n Here, we report a next-generation in vivo model of HIOs with functional human immune tissue. Using this model, we demonstrate that immune cells temporally infiltrate and populate the HIO in the mucosa similarly to the immune landscape in the developing human gut. Mass cytometry data and immunostaining demonstrated the formation of GALT-like structures during HIO development, and their formation correlated temporally and spatially with human intestinal immune tissue development. Because M cells are present in the epithelium overlaying the GALT, we confirmed their presence and validated their function by exposing the HIO lumen to microbial components.<\/p>\n Intestinal immune tissue contains diverse immune cell types to protect the mucosal barrier against invaders25<\/a><\/sup>. The human fetal gut has a diverse, functional immune system even without being exposed to antigens26<\/a>,27<\/a>,28<\/a>,29<\/a>,30<\/a><\/sup>. Our mass cytometry analysis indicated that the immune cell composition was diverse and increased in cell number as the transplanted HIOs developed, similar to the human fetal gut. Our results correlate with recent mass cytometry and transcriptomic data showing an increase in immune cell number and cell subsets, in particular CD4+<\/sup> T cells, during human fetal intestinal development29<\/a>,30<\/a><\/sup>. During HIO development, we found by mass cytometry an increase in CD4+<\/sup> T and B cell frequencies and we confirmed by histology that these cell types together formed aggregates that later developed into lymphoid follicles. These features were not identified in the intestine of humanized mice, confirming reports describing the absence of lymphoid follicles in this model19<\/a>,20<\/a><\/sup>.<\/p>\n The timing of humanization and transplantation did not affect the timing of GALT structure formation. In a preliminary study in which HIO transplantation was delayed until the immune system was fully reconstituted, histology and flow cytometry studies indicated that the development of immune cellular aggregates composed of T and B cells was similar to that of early HIO transplantation (Extended Data Fig. 9<\/a>). These observations indicate that immune precursors were present even after the immune system was reconstituted. Taken together, our results suggest that the developing HIO and cells seeding the lamina propria influence and frame the formation of immune aggregates by expressing specific signals to attract and activate immune precursors as well as serving as a foundation for subsequent lymphoid tissue development.<\/p>\n Recently, transcriptomic analysis of human fetal intestine has identified subsets of fibroblasts in GALT formation that correspond to lymphoid tissue organizer cells30<\/a><\/sup>. These cells increased in number with fetal development and interacted with diverse immune cell populations, including LTi cells via the expression of CCL19, CCL21 and CXCL13 (ref. 30<\/a><\/sup>). In our model, preliminary data indicated that CCL19<\/i>, CCL21<\/i> and CXCL13<\/i> were expressed in HIOs at 16\u2009weeks post-transplantation, whereas their mouse gene orthologs were decreased by at least tenfold (Extended Data Fig. 10a,b<\/a>). These results indicate that, in contrast to the humanized mouse SI, the HIOs express unique microenvironment signals that support the formation of GALT-like structures. In line with this, our model could be used to dissect the mechanism of interaction between stromal lymphoid tissue organizer cells and LTi cells and to identify key pathways in the development of GALT in humans.<\/p>\n GALT is a site of adaptive immune responses and influences the epithelial cells overlaying the follicles to promote the generation of M cells that transport luminal antigens to activate immune cells2<\/a><\/sup>. GP2 is a glycoprotein expressed on the apical side of M cells and translocates luminal antigens to the immune cells in the lamina propria. Even though transplanted HIOs expressed the GP2<\/i> gene, we were not able to detect the protein by IHC likely due to the lack of microbial antigens in the lumen, similar to a fetal intestine11<\/a><\/sup>. Of note, nonhumanized HIOs exposed to microbial antigens in other models being developed in the laboratory also do not express GP2 protein (data not shown). Using our current model, we demonstrated that exposing the HIO lumen to E. coli<\/i> lysate induced M cells, characterized by GP2 expression as well as activation of an immune response indicated by the presence of IgA antibodies. After being produced by plasma cells in the lamina propria, IgA antibodies bind to polymeric immunoglobulin receptors (pIgR) expressed on the basal side of epithelial cells and, by transcytosis, epithelial cells then secrete IgA antibodies in the intestinal lumen31<\/a><\/sup>. Because we detected the presence IgA antibodies in the mucus, our findings suggest that epithelial cells in HIOs express pIgR and transport antibodies to the lumen.<\/p>\n Here, we have demonstrated that our in vivo HIO model with GALT-like structures and M cells provides opportunities for studies on food allergies, intestinal infectious diseases and the development of mucosal vaccines.<\/p>\n<\/div>\n<\/div>\n Immunodeficient NOD\/Scid Il2rg null Tg (hIL3, hGM-CSF and hSCF) (NSGS; The Jackson Laboratory no. 013062 (ref. 17<\/a><\/sup>)) mice were bred and housed in the animal facility at the Cincinnati Children\u2019s Hospital Medical Center (CCHMC). All experiments were performed with the approval of the Institutional Animal Care and Use Committee of CCHMC (IACUC protocol no. 2018-0092).<\/p>\n Normal, de-identified developing human fetal tissues from elective abortions were obtained from the University of Washington, Laboratory of Developmental Biology, and all work was approved by the University of Washington and the University of Michigan Institutional Review Boards (IRBs).<\/p>\n Normal adult human jejunum was obtained from patients undergoing bariatric procedures between the ages of 16 and 25\u2009yr. Informed consent or assent was obtained from all patients and\/or parents\/legal guardians as appropriate. Human tissue collection was performed with the previous approval of CCHMC\u2019s IRB.<\/p>\n HIOs were generated and maintained as previously described12<\/a>,32<\/a>,33<\/a><\/sup>.<\/p>\n Briefly, human H1 embryonic stem cells (WA-01; WiCell) (passage number 40\u201355), obtained from the Pluripotent Stem Cell Facility in our institute, were grown in feeder-free conditions in mTESR1 medium (Stem Cell Technologies). For induction of definitive endoderm, cells were split with Accutase (Invitrogen) and plated at a density between 70,000 and 100,000 cells per well in a Matrigel-coated 24-well plate. Once the cells reached 80\u201395% confluency, they were treated with 100\u2009ng\u2009ml\u22121<\/sup> Activin A for 3\u2009d as previously described34<\/a><\/sup>. Definitive endoderm was then treated for 4\u2009d with hindgut induction medium containing 500\u2009ng\u2009ml\u22121<\/sup> FGF4 (R&D) and 3\u2009\u03bcM Chiron 99021 (Tocris) to induce formation of mid-hindgut spheroids. Spheroids were then plated in Matrigel (Corning) and maintained in intestinal growth medium supplemented with 100\u2009ng\u2009ml\u22121<\/sup> EGF (R&D) to generate HIOs. Medium was changed twice weekly and HIOs were replated in fresh Matrigel at 14\u2009d.<\/p>\n The Translational Trials Development Support Laboratory of CCHMC collected and distributed the umbilical cord blood units according to an IRB-approved protocol (protocol no. 02-3-4x). Mice were humanized as described previously35<\/a><\/sup>. Briefly, whole cord blood was subjected to hetastarch-induced aggregation of red blood cells. Cord blood cells (CBCs) were isolated, washed and viably frozen for future use. Thawed CBCs were resuspended in IMDM medium with 3% FBS and antibiotics and diluted to 23.3\u201328.3\u2009\u00d7\u2009106<\/sup> CBCs per ml. OKT3 antibody was spiked into the cell solution at a concentration of 1\u2009\u00b5g per 1\u2009million cells to prevent graft-versus-host disease. Immunodeficient NSGS mice at 6\u20138\u2009weeks old were conditioned by receiving a dose of busulfan (30\u2009mg\u2009kg\u22121<\/sup> by intraperitoneal injection) 24\u2009h before intravenous injection of 7\u20138.5\u2009\u00d7\u2009106<\/sup> CBCs in 300\u2009\u00b5l. Busulfan is a chemotherapy drug that partially depletes cells from the mouse bone marrow and allows the human hematopoietic stem cells to engraft. As a preventative measure against possible systemic infections, mice were fed with doxycycline chow (0.0625%, Purina) for 2\u2009weeks before and after human cell engraftment.<\/p>\n As previously described, a single HIO, matured in vitro for 28\u2009d, was removed from Matrigel and then transplanted under the kidney capsule12<\/a><\/sup>. Briefly, the mice were anesthetized with 2% inhaled isoflurane (Butler Schein) and 2.5\u20133\u2009l\u2009min\u22121<\/sup> oxygen. The left side of the mouse was then prepped in sterile fashion with isopropyl alcohol and providine-iodine. A small left-posterior subcostal incision was made to expose the kidney. A subcapsular pocket was created in the kidney capsule and the HIO was then placed into the pocket. The kidney was then returned to the peritoneal cavity and the mice were given an intraperitoneal flush with 2\u20133\u2009ml of piperacillin\/tazobactam (100\u2009mg\u2009kg\u22121<\/sup>; Pfizer) to help prevent bacterial infection. The skin was closed in a double layer. For pain control, mice were then given a subcutaneous injection with buprenorphine (0.05\u2009mg\u2009kg\u22121<\/sup>; Midwest Veterinary Supply) or carprofen (4\u2009mg\u2009kg\u22121<\/sup>; Midwest Veterinary Supply) and were monitored for the next 48\u2009h following surgery. Additional injections of pain medication were given if needed. At 12 and 16\u2009weeks following engraftment, the mice were then euthanized and the tissues were collected and analyzed.<\/p>\n Grown in LB medium overnight, E. coli<\/i> suspension was then washed three times with PBS and centrifuged at 800 g<\/i> for 5\u2009min. Bacteria were resuspended in saline solution at ~107<\/sup> colony-forming units per ml and centrifuged at 10,000 g<\/i> for 5\u2009min at 4\u2009\u00b0C. Bacteria were then lysed by freeze\u2013thaw cycles repeated four times.<\/p>\n At 16\u2009weeks post HIO transplantation, the mice were anesthetized and a small left-posterior subcostal incision was made to expose the transplanted HIO. Then, 50\u2013100\u2009\u00b5l of E. coli<\/i> lysate was injected with a 0.5-ml insulin syringe in the HIO lumen.<\/p>\n The HIO was then returned to the peritoneal cavity and the skin was closed in a double layer. For pain control, mice were then given a subcutaneous injection with carprofen (4\u2009mg\u2009kg\u22121<\/sup>; Midwest Veterinary Supply) and were monitored for the next 48\u2009h following surgery. Additional injections of pain medication were given if needed. At 72\u2009h post injection, the mice were euthanized and the tissues were collected and analyzed.<\/p>\n To confirm the expression of human immune cells from CD34+<\/sup> engraftment in mice, retro-orbital bleeding was performed 8\u201310\u2009weeks after the engraftment and a day before each tissue collection. Approximately 50\u2009\u03bcl of whole blood was collected in a BD Microtainer Dipotassium\/EDTA-coated tube and then lysed in 5\u2009ml of red blood cell lysis buffer (155\u2009mM NH4<\/sub>Cl, 12\u2009mM NaHCO3<\/sub> and 0.1\u2009mM EDTA pH 8.0; diluted in dH2<\/sub>O), for 5\u2009min at room temperature. An isovolume of PBS was added and tubes were centrifuged at 300 g<\/i> for 5\u2009min. The pellet was washed in PBS, centrifuged and resuspended in FACS buffer (0.5% BSA, 2\u2009mM EDTA in PBS 1\u00d7). Cells were stained for 30\u2009min on ice with the following combination of antibodies: FITC-conjugated anti-human CD45, PE-Cy5-conjugated anti-mouse CD45, BV650-conjugated anti-human CD19, PE-Cy7-conjugated anti-human CD3, BV421-conjugated anti-human CD56, PE-conjugated anti-human CD13, PE-conjugated anti-human CD33 and Zombie NIR fixable viability kit to exclude dead cells (Supplementary Table<\/a>). All antibodies were used at 1:200 dilution except Zombie NIR which was used at 1:2,000 dilution. Samples were washed twice and resuspended in FACS buffer. The samples were then recorded on an LSR Fortessa instrument (BD Biosciences) and the data were analyzed with FlowJo software (TreeStar).<\/p>\n One-third of the HIO was used for cell dissociation. The cell dissociation protocol for the HIOs was modified from Weigmann\u2019s protocol for mouse colon cell dissociation36<\/a><\/sup>. Briefly, HIOs were cut into small pieces and incubated under slow rotation for 20\u2009min at 37\u2009\u00b0C in 5\u2009ml of predigestion solution containing EDTA and dithiothreitol in HBSS. The epithelial cell suspension was filtered through a 100-\u03bcm cell strainer, washed with cold PBS and kept on ice until pooled with the lamina propria-isolated cells. The remaining pieces of HIO were minced and placed in a new tube with 5\u2009ml of digestion solution containing collagenase D (Roche), DNAse I (Roche) and dispase II (Roche) in PBS and incubated for 15\u2009min at 37\u2009\u00b0C under slow agitation. The lamina propria cell suspension was passed through a 70-\u03bcm cell strainer, washed with cold PBS and re-incubated in 5\u2009ml of digestion solution for 15\u2009min at 37\u2009\u00b0C under slow agitation. This step was repeated one more time. Epithelial and lamina propria cells were combined in one tube and spun down at 450 g<\/i> for 5\u2009min.<\/p>\n The method to isolate mononuclear cells from the humanized mouse gut was adapted from the protocol reported by Lee et al.37<\/a><\/sup>. Briefly, 10\u201312\u2009cm of proximal SI was collected, longitudinally cut open and washed in HBSS to remove any debris. The SI was incubated in HBSS containing 5\u2009mM EDTA on ice for 5\u2009min and vortexed at medium intensity, for a total of four incubations. The epithelium was then collected in a separate tube. After digestion of the tissue at 37\u2009\u00b0C for 30\u2009min with DNAse I (Roche) and Collagenase A (Roche), the homogenate was filtered and centrifuged at 450 g<\/i> for 5\u2009min.<\/p>\n Both HIO and SI cell suspensions were individually resuspended in 44% Percoll (GE), loaded on a 67% Percoll layer and centrifuged at 650 g<\/i> without brake for 30\u2009min at room temperature. The cell layer at the interface of the two gradients was collected, washed, counted and used for in vitro assays, or stained for immune markers and analyzed by mass cytometry.<\/p>\n Antibodies used in this panel were purchased from Fluidigm except anti-human CD45RO antibody which was purchased from BioLegend and was labeled with Maxpar X8 Antibody Labeling Kit (Fluidigm) according to the manufacturer\u2019s instructions. All the reagents used in the following protocol were purchased from Fluidigm and all incubations were done at room temperature. Samples were first stained for 5\u2009min with Cell-ID Cisplatin at a final concentration of 5\u2009\u03bcM in Maxpar PBS and then washed in 5 volume of Maxpar Cell Staining Buffer. Before being stained for cell surface markers, cells were fixed in Fix I Buffer for 10\u2009min, washed twice with Maxpar Barcode Perm Buffer and barcoded with Cell-ID 20-Plex Pd Barcoding Kit for 30\u2009min. Samples were pooled and then stained for 30\u2009min with a cocktail of 25 antibodies in a final volume of 100\u2009\u03bcl of Maxpar Cell Staining Buffer (antibody dilution 1:200) (Supplementary Table<\/a>). After cell surface staining, samples were washed, fixed in 1.6% paraformaldehyde (PFA) (EMS) for 10\u2009min and, finally, incubated in 1\u2009ml of intercalator solution at a final concentration of 125\u2009nM. The samples were kept cold and shipped in intercalator solution to the Cytometry Research Core Facility at the University of Rochester, NY. The samples were washed and resuspended in Maxpar water before being acquired using a cytometry by time of flight (CyTOF)\/mass cytometry Helios instrument (Fluidigm). Data were then debarcoded using Fluidigm debarcoder software (available online).<\/p>\n CyTOF datasets from transplanted HIO- and humanized mouse SI-isolated immune cells were generated using mass cytometry (Fluidigm) variation at three time points, 12\u2009weeks, 16\u2009weeks and 20\u2009weeks. Datasets at aforementioned time points were generated in two batches for each tissue at 12 and 16\u2009weeks post-transplantation with a total of n<\/i>\u2009=\u20097 samples and n<\/i>\u2009=\u200910 samples, respectively, and in one batch for each tissue at 20\u2009weeks post-transplantation. Combining two batches, ~1.1\u2009million cells were analyzed, amounting to 63,076 cells in 12-week HIO, 350,132 cells in 16-week HIO, 148,274 cells in 20-week HIO, 137,887 cells in 12-week SI, 368,893 cells in 16-week SI and 71,569 cells in 20-week SI.<\/p>\n Raw .fcs files for each sample were read and analyzed using read.flowSet<\/i> function from the flowCore R (v.3.6.3) package. The .fcs files stored in flow.set<\/i> objects were than normalized using inverse hyperbolic sine (Arcsinh<\/i>) transformation (https:\/\/support.cytobank.org\/hc\/en-us\/articles\/206148057-About-the-Arcsinh-transform<\/a>) using a cofactor value of 5. Normalized reads from the .fcs files were then used to create a Seurat object using CreateSeuratObject<\/i> function from Seurat (v.3.0.2), a single-cell analysis package in R.<\/p>\n For each Seurat object, the following steps were carried out: variable features set to surface markers, scaling, dimension reduction and clustering. Dimension reduction was applied in both linear (principal components) and nonlinear (UMAP) approaches to obtain respective components. Clustering was performed using a k<\/i>-nearest neighbor graph of cells based on marker abundance similarity. This k<\/i>-nearest neighbor graph was finally partitioned into clusters based on highly interconnected colonies of cells.<\/p>\n In the end, all the processed Seurat objects for each sample from two batches were integrated together using Seurat\u2019s reciprocal principal component analysis as the numbers of cells were large (~1.1\u2009million). Briefly, each dataset was projected into principal component space of other datasets to learn the anchors (cell pairs) based on the mutual nearest neighbor graph requisite.<\/p>\n Dimension reduction plots were generated using DimPlot<\/i> function and heatmaps were generated using the DoHeatmap<\/i> function in Seurat.<\/p>\n Example datasets and analysis code have been deposited to GitHub: https:\/\/github.com\/praneet1988\/Analyze_CyTOF_Using_Seurat.<\/a><\/p>\n Immune cells isolated from HIO or mouse SI were stimulated for 4\u2009h with 1:500 Cell Activation Cocktail (BioLegend) containing, according to the manufacturer, an optimized concentration of phorbol 12-myristate-13-acetate (PMA) and ionomycin. After an hour of stimulation, cells were then incubated with 1:1,000 Brefeldin A Solution (Biolegend) to block the secretion of cytokines. Finally, the cells were stained with anti-human CD3 and anti-human CD4 antibodies (Supplementary Table<\/a>), fixed with Cytofix (BD Biosciences) and then permeabilized with CytoPerm (BD Biosciences) overnight at 4\u2009\u00b0C. The following day anti-IFN\u03b3, anti-TNF\u03b1 and anti-IL-2 antibodies (Supplementary Table<\/a>) were added to the cells for at least 1\u2009h at room temperature. Samples were washed twice and resuspended in FACS buffer. The samples were then recorded on an Aurora instrument (Cytek) and the data were analyzed with FlowJo software (TreeStar).<\/p>\n Indicated cytokines were measured in supernatants from immune cells (cell density 106<\/sup> per ml) isolated from HIO or mouse SI stimulated for 3\u2009d with a cocktail of anti-human CD3\/anti-human CD28 antibodies (STEMCELL Technologies). The presence of cytokines in supernatants was measured using Milliplex kits (Millipore), following the manufacturer\u2019s instructions.<\/p>\n At 20\u2009weeks post-transplantation, a section of transplanted HIOs was used to isolate the crypts following our protocol for human intestinal tissue38<\/a><\/sup>. Briefly, the mucosal layer from portions of transplanted HIOs was dissected under a microscope and scraped to remove the villi and debris. The mucosa was then incubated with freshly prepared 2\u2009mM EDTA solution and gently shaken for 30\u2009min. After several washes with ice-cold chelation buffer, the intestinal crypts were collected by gently scraping the mucosa with curved forceps and filtered twice through a 150-\u03bcm nylon mesh to remove any debris. Due to limited amounts of tissues, the crypts collected from each group were pooled. The crypts were then washed in ice-cold chelation buffer and 50\u2009\u03bcl of crypts, resuspended in Matrigel (Corning), was added per well in a 24-well plate. After polymerization of the Matrigel, 500\u2009\u03bcl of human IntestiCult Organoid Growth medium (STEMCELL Technologies) was added to each well. The crypts were cultured and expanded for 10\u201314\u2009d before being frozen down for later use.<\/p>\n Enteroids were plated on Transwells as described previously7<\/a>,39<\/a><\/sup>. Briefly, after being washed from Matrigel, enteroids were fragmented and plated on human collagen IV-precoated 24-well Transwells (0.4-\u03bcm pore size) and incubated in IntestiCult Organoid Growth medium (STEMCELL Technology) at 37\u2009\u00b0C until monolayer confluence was reached. To induce M cell differentiation, enteroid monolayers were cultured in differentiation medium supplemented with 50\u2009ng\u2009ml\u22121<\/sup> TNF-\u03b1 and 100\u2009ng\u2009ml\u22121<\/sup> RANK-L for a period of 5\u2009d, as previously described39<\/a><\/sup>.<\/p>\nResults<\/h2>\n
Human immune cells integrate into transplanted HIOs<\/h3>\n
<\/picture><\/a><\/div>\nMass cytometry analysis reveals a GALT-like immune signature profile in HIO<\/h3>\n
<\/picture><\/a><\/div>\nImmune aggregates in the developing HIO are lymphoid follicle-like structures<\/h3>\n
<\/picture><\/a><\/div>\nTemporal and spatial development of lymphoid-like structures in transplanted HIOs correlate to human lymphoid follicle development<\/h3>\n
<\/picture><\/a><\/div>\nM cells are induced in transplanted HIOs after microbial exposure<\/h3>\n
<\/picture><\/a><\/div>\nDiscussion<\/h2>\n
Methods<\/h2>\n
Animals<\/h3>\n
Human intestinal tissue<\/h3>\n
HIOs<\/h3>\n
Human umbilical cord blood cell engraftment<\/h3>\n
Transplantation of HIOs<\/h3>\n
\n E. coli<\/i> lysate preparation and injection<\/h4>\n
Flow cytometry<\/h3>\n
HIO and SI dissociation and cell preparation<\/h3>\n
Mass cytometry staining and data analysis<\/h3>\n
Mass cytometry staining<\/h4>\n
Bioinformatics analysis<\/h4>\n
Measurement of cytokines from HIO- and mouse SI-derived immune cells<\/h3>\n
Detection of cytokines by flow cytometry<\/h4>\n
Detection of cytokines by Milliplex assay<\/h4>\n
In vitro HIO-derived enteroid culture (M cell induction)<\/h3>\n
HIO-derived enteroid preparation and in vitro expansion<\/h4>\n
M cell induction in vitro<\/h4>\n