Generating dorsal foregut, and anterior domain endoderm cells

Abstract

To provide compositions, systems, kits, and methods for generating dorsal foregut endoderm (DFE) cells by contacting pluripotent stem cells with a retinoic acid signaling pathway agonist and a bone morphogenetic (BMP) pathway inhibitor without exposing the stem cells to a transforming growth factor beta (TGFβ) pathway agonist.SOLUTION: In certain embodiments, the DFE cells are contacted with a retinoic acid signaling pathway agonist and a FGFR pathway inhibitor to generate pancreatic endoderm (PE) cells (e.g., with dorsal identity). In other embodiments, the PE cells are contacted with an ALK5 inhibitor and a Notch inhibitor to generate endocrine cells (e.g., insulin expressing endocrine cells). In certain embodiments, compositions, systems, kits, and methods are provided for generating anterior domain endoderm (ADE) cells.SELECTED DRAWING: None

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to U.S. Provisional Application No. 62 / 851,348, filed May 22, 2019, which is incorporated by reference in its entirety. [Technical field]

[0002] Provided herein are compositions, systems, kits, and methods for generating dorsal foregut endoderm (DFE) cells by contacting pluripotent stem cells with a retinoic acid signaling pathway agonist and a bone morphogenesis (BMP) pathway inhibitor without exposing the stem cells to a transforming growth factor beta (TGFβ) pathway agonist. In certain embodiments, the DFE cells are contacted with a retinoic acid signaling pathway agonist and an FGFR pathway inhibitor to generate pancreatic endoderm (PE) cells (e.g., with dorsal identity). In other embodiments, the PE cells are contacted with an ALK5 inhibitor and a Notch inhibitor to generate endocrine cells (e.g., insulin-expressing cells). In certain embodiments, compositions, systems, kits, and methods for generating anterior domain endoderm (ADE) cells are provided. [Background technology]

[0003] Endoderm is the developmental germ layer that gives rise to the majority of cells within most visceral systems, such as the lung, stomach, pancreas, liver, and intestine. Nearly all efforts to induce endoderm from pluripotent cells have relied on the use of TGFβ pathway agonists, most commonly activin A (AA), as Nodal mimics to push pluripotent cells through in vitro gastrulation events (D'Amour et al., 2005, Gadue et al., 2006). This results in an endoderm population that can be used to generate multiple progeny fates, including the intestine (Spence et al., 2011), pancreas (Kroon et al., 2008, Rezania et al., 2014, Pagliuca et al., 2014), and liver (Sampaziotis et al., 2015). Generation of more anterior endoderm fates, such as the lung, has been achieved by providing patterning input at subsequent stages (Green et al., 2011). However, recent studies argue that early patterning of the definitive endoderm can occur during development (Matsuno et al., 2016; Loh et al., 2014).

[0004] The pancreas is of particular interest for cell-based therapy in diabetes, which is characterized by a deficiency of insulin-producing cells. The pancreas forms from two spatially distinct primordia that arise on the dorsal and ventral sides of the primitive gut and then fuse. Although both pancreatic buds are capable of generating all lineages of the adult pancreas (Matsuura et al., 2009), the initial induction of the pancreatic domains on opposite sides of the gut is controlled by different transcriptional programs. In mice, the ventral pancreatic bud first forms around embryonic day 8.5 (E8.5) from a region of endoderm with bipotential competence for pancreas and liver (Angelo et al., 2012; Deutsch et al., 2001; Tremblay and Zaret, 2005). This early ventral endoderm is composed of a population of progenitor cells that transiently co-express Pdx1 / Sox17, which divide by E9.5 to form the ventral pancreas and extrahepatic biliary system, respectively (Spence et al., 2009). Specification of the ventral pancreas depends on HHex expression, whose genetic ablation model showed complete ventral aplasia without affecting dorsal pancreatic bud formation (Bort et al., 2004). In contrast, the dorsal pancreatic bud in mice appears at approximately embryonic day 9.0 and forms the lateral tail to gastric antrum region. Studies in mice have also identified factors involved in dorsal pancreas specification that do not affect ventral organogenesis. Mnx1 (Hlxb9) knockout models show dorsal aplasia occurs without a ventral phenotype (Li et al., 1999). Mnx1 expression is observed in the ventral region, but only following Pdx1 expression, whereas in the dorsal region, Mnx1 precedes Pdx1 expression. Raldh2 knockout models result in dorsal-specific agenesis due to loss of Pdx1 and Prox1 expression in the dorsal bud (Martin et al., 2005; Molotkov et al., 2005). Furthermore, studies in chick have shown that early dorsal pancreatic bud formation is dependent on selective inhibition of SHH in the dorsal midgut (Hebrok et al., 1998).Although it is unclear whether the mouse lineage is conserved between species, a recent study using laser capture and subsequent deep sequencing analysis described some fundamental differences between the ventral and dorsal pancreas during human development ( Jennings et al., 2017 ).

[0005] Despite differences in pathway utilization and distinct intracellular factors, the dorsal and ventral pancreatic programs share many commonalities. HNF1β (Tcf2) is required for pancreatic specification in both pancreatic buds and is important throughout pancreatic development. Tcf2 knockout mice fail to generate a ventral pancreas and have a severely reduced dorsal bud that cannot differentiate or proliferate (Haumaitre et al., 2005). HNF1β is expressed in the prepancreatic foregut, where the gene functions at the apex of a sequential transcriptional cascade leading to activation of Hnf6 (Oc1) and subsequently Pdx1 (Poll et al., 2006). Conditional inactivation of HNF1β results in loss of Glis3 and Ngn3 expression, resulting in loss of the pancreas and endocrine promoter characterized by the cystic duct (De Vas et al., 2015). In human development, the importance of HNF1β is highlighted by the occurrence of the "maturity-onset diabetes of the young type 5" (MODY5) syndrome, a condition resulting from mutations in the HNF1β gene. Heterozygous mutations in HNF1β show no phenotype in mouse studies, but in humans, heterozygous mutations in HNF1β have been shown to be associated with MODY5 or complete pancreatic agenesis, suggesting that the role of HNF1β in human pancreatic development is more important than in mice (Body-Bechou et al., 2014). Summary of the Invention

[0006] Provided herein are compositions, systems, kits, and methods for generating dorsal foregut endoderm (DFE) cells by contacting pluripotent stem cells with a retinoic acid signaling pathway agonist and a bone morphogenetic (BMP) pathway inhibitor without exposing the stem cells to a transforming growth factor beta (TGFβ) pathway agonist. In certain embodiments, the DFE cells are contacted with a retinoic acid signaling pathway agonist and an FGFR pathway inhibitor to generate pancreatic endoderm (PE) cells (e.g., with dorsal identity). In other embodiments, the PE cells are contacted with an ALK5 inhibitor and a Notch inhibitor to generate endocrine cells (e.g., insulin-expressing endocrine cells). In certain embodiments, compositions, systems, kits, and methods for generating anterior domain endoderm (ADE) cells are provided.

[0007] In some embodiments, provided herein are methods for generating dorsal foregut endoderm cells, comprising: a) contacting a population of pluripotent stem cells (e.g., iPSCs or embryonic stem cells) with a retinoic acid signaling pathway agonist and a bone morphogenetic (BMP) pathway inhibitor; and b) culturing at least a portion of the population of pluripotent stem cells to generate a population of dorsal foregut endoderm (DFE) cells, wherein the stem cells are not exposed to a transforming growth factor beta (TGFβ) pathway agonist during the culturing or contacting.

[0008] In certain embodiments, provided herein is a composition comprising a cell, wherein the cell comprises i) an exogenous retinoic acid signaling pathway agonist and ii) an exogenous bone morphogenetic (BMP) pathway inhibitor, and wherein the cell i) does not comprise any exogenous transforming growth factor beta (TGFβ) pathway agonist, and ii) is a pluripotent stem cell (e.g., an iPSC or an embryonic stem cell) or is a dorsal foregut endoderm cell.

[0009] In certain embodiments, provided herein is a composition comprising a cell culture medium, the cell culture medium comprising: i) a retinoic acid signaling pathway agonist compound; and ii) a bone morphogenetic (BMP) pathway inhibitor, the culture medium being free or detectably free of any exogenous transforming growth factor beta (TGFβ) pathway agonist. In certain embodiments, the cell culture medium comprises a suitable energy source and a compound that regulates the cell cycle. In certain embodiments, the cell culture medium comprises most or all of amino acids, vitamins, inorganic salts, glucose, and serum as sources of growth factors, hormones, and attachment factors. In some embodiments, the composition further comprises a cell population, the cells being pluripotent stem cells (e.g., iPSCs or embryonic stem cells) or dorsal foregut endoderm (DFE) cells. In other embodiments, the cells are pluripotent stem cells, and the retinoic acid signaling pathway agonist and the BMP pathway inhibitor are present in the culture medium at concentrations that cause at least a portion of the pluripotent stem cells to become DFE when cultured in the medium for at least 2 days.

[0010] In other embodiments, kits and systems are provided herein comprising, consisting essentially of, or consisting of a) a population of pluripotent stem cells (e.g., iPSCs or embryonic stem cells) present in a cell culture vessel, b) a retinoic acid signaling pathway agonist compound present in a first vessel, and c) a bone morphogenetic (BMP) pathway inhibitor present in a second vessel, wherein the cell culture vessel does not contain any, or any detectable exogenous transforming growth factor beta (TGFβ) pathway agonist, and the kits and systems do not contain any transforming growth factor beta (TGFβ) pathway agonist in the first vessel or the second vessel, or in a third vessel. In further embodiments, the kits and systems further comprise a culture medium in the culture vessel.

[0011] In other embodiments, the BMP pathway inhibitor is a BMP4 pathway inhibitor. In certain embodiments, the BMP pathway inhibitor comprises LDN193189. In further embodiments, the BMP pathway inhibitor is selected from the group consisting of DMH1, DMH2, dorsomorphin, K02288, LDN214117, LDN212854, follistatin, ML347, and noggin. In some embodiments, the retinoic acid signaling pathway agonist is selected from the group consisting of i) a retinoid compound, ii) a retinoid X receptor (RXR) agonist, and iii) a retinoic acid receptor (RAR) agonist. In certain embodiments, the retinoic acid signaling pathway agonist is selected from the group consisting of retinoic acid, Sr11237, adapalene, EC23, 9-cis retinoic acid, 13-cis retinoic acid, 4-oxoretinoic acid, and all-trans retinoic acid (ATRA). In certain embodiments, the culture is for 1 to 5 days (eg, 1, 2, 3, 4, or 5 days) or for 2 to 4 days.

[0012] In some embodiments, the method further comprises c) contacting at least a portion of the population of DFE cells with a retinoic acid signaling pathway agonist and an FGFR pathway inhibitor; and d) culturing at least a portion of the population of DFE cells to generate a population of pancreatic endoderm (PE) cells. In certain embodiments, the DFE cells are not exposed to a transforming growth factor beta (TGFβ) pathway agonist during the culturing in step d) or during the contacting in step c). In some embodiments, the culturing in step d) is performed for about 1-5 days (e.g., 1, 2, 3, 4, or 5 days) or 2-4 days. In certain embodiments, the PE cells have a dorsal pancreatic identity. In some embodiments, the FGFR pathway inhibitor is selected from the group consisting of PD0325901, arctigenin, PD184352, PD198306, PD334581, SL327, U0126, a MEK inhibitor, an FGFR inhibitor, and a MAPK inhibitor. In certain embodiments, the contacting in step c) further comprises contacting the DFE cell population with a sonic hedgehog (SHH) pathway inhibitor.

[0013] In certain embodiments, the method further comprises: e) contacting at least a portion of the PE cell population with a Notch pathway inhibitor and an ALK5 inhibitor; and f) culturing at least a portion of the PE cell population to generate a population of endocrine cells. In some embodiments, the endocrine cells are insulin-expressing cells. In other embodiments, the endocrine cells are fetal beta cells. In certain embodiments, the Notch pathway inhibitor comprises gamma secretase inhibitor XX, also known as DBZ. In other embodiments, the Notch pathway inhibitor is selected from the group consisting of DAPT, MRK-003, MRK-0752, z-Ile-leu-CHO, gamma secretase inhibitor, L-685,485, LY411575, compound E, F-03084014, RO4929097, and BMS-906024. In some embodiments, the ALK5 inhibitor comprises A8301. In a further embodiment, the ALK5 inhibitor is selected from the group consisting of A7701, A83-01, SB505124, SB431542, alantolactone, and LY2157299. In a further embodiment, the culturing in step f) is carried out for 5 to 15 days (e.g., 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, or 15 days).

[0014] In some embodiments, cells (e.g., endocrine cells) produced by the methods herein are transplanted into a subject to provide a therapeutic effect, in some embodiments, the transplanted cells are insulin-expressing cells, such as when the subject has diabetes.

[0015] In certain embodiments, the methods herein further include c) contacting at least a portion of the population of DFE cells with EGF; and d) culturing at least a portion of the population of DFE cells to generate a population of gastric endoderm (SE) cells.

[0016] In other embodiments, the methods herein further include c) contacting at least a portion of the population of DFE cells with BMP4 or a BMP pathway agonist (e.g., sb4); and d) culturing at least a portion of the population of DFE cells to generate a population of hepatic endoderm (LE) cells.

[0017] In other embodiments, the methods herein are used to generate anterior definitive endoderm (ADE), which is anterior to the DFE and DE and includes cells corresponding to pharyngeal endoderm. The ADE expresses specific marker genes, such as TBX1, PAX9, and OSR1, and represents progenitor cells of cell lineages of the pharyngeal region, including, but not limited to, lung, thymus, and thyroid.

[0018] In some embodiments, provided herein is a method for generating anterior domain endoderm cells, comprising: a) contacting a population of pluripotent stem cells with i) a Nodal protein and a BMP protein or a BMP pathway agonist, or a vector(s) encoding the Nodal protein and the BMP protein or a BMP pathway agonist (if it is a protein), and ii) an ALK5 inhibitor; and b) culturing at least a portion of the population of pluripotent stem cells to generate a population of anterior domain endoderm (ADE) cells. In certain embodiments, the pluripotent stem cells are not exposed to i) a BMP inhibitor, and / or ii) a transforming growth factor beta (TGFβ) pathway agonist during the culturing or contacting.

[0019] In certain embodiments, provided herein are compositions comprising a) a Nodal protein or a vector encoding a Nodal protein, b) a BMP protein or a BMP pathway agonist, or, in the case of a protein, a vector encoding a BMP protein or a BMP agonist, and c) an ALK5 inhibitor. In some embodiments, the composition does not comprise a BMP inhibitor, and / or ii) a transforming growth factor beta (TGFβ) pathway agonist. In certain embodiments, a single vector encodes both a Nodal protein and a BMP protein (or, in the case of a protein, a BMP pathway agonist), while in other embodiments, separate vectors are used. In other embodiments, the composition further comprises a population of cells, which are pluripotent stem cells or anterior domain endoderm (ADE) cells.

[0020] In some embodiments, provided herein are kits or systems that consist essentially of, or consist of, a) a cell population present in a cell culture container; b) a Nodal protein or a vector encoding a Nodal protein present in a first container; c) a BMP protein or a BMP pathway agonist or, if the protein is present in the second container, a vector encoding a BMP protein or a BMP agonist; and d) an ALK5 inhibitor present in a third container.

[0021] In certain embodiments, the cell culture container does not contain any or any detectable exogenous BMP inhibitors and / or exogenous transforming growth factor beta (TGFβ), and the kits and systems do not contain any detectable exogenous BMP inhibitors and / or exogenous transforming growth factor beta (TGFβ) in the first container, the second container, the third container, or the fourth container. In other embodiments, the cells are pluripotent stem cells or anterior domain endoderm (ADE) cells. In further embodiments, the ALK5 inhibitor is selected from the group consisting of A7701, A-83-01, SB505124, SB431542, alantolactone, and LY2157299. In further embodiments, the BMP protein is a BMP4 protein.

[0022] The patent file or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee. [Brief description of the drawings]

[0023] [Figure 1] 1 shows an exemplary method for generating dorsal foregut endoderm (DFE) cells. [Diagram 2] Results of effective endoderm induction in the absence of AA / WNT are shown. A: Schematic of endoderm genes through endoderm induction optimizing Activin A induction of the Anterior Primitive Streak (APS) versus HNF1β induction. B: Predicted conditions for conditions that meet APS gene induction (upper circle) versus HNF1β induction (lower circle). C: Validation of endoderm marker induction using two separate protocols. D: Coefficient plot of selected endoderm genes (primary effectors only). [Diagram 3] Non-APS derived endoderm is shown to be highly activated by retinoic acid and BMP inhibition, displaying dorsal foregut characteristics. A: Schematic of the experiment performed. B: Graph showing relative expression of several endoderm genes in response to retinoic acid and BMP inhibition. Genes were normalized to the average expression of endogenous levels of YWHAZ, GAPDH and TBP. DE-definitive endoderm, RA-retinoic acid, LDN-LDN3189, BMP inhibitor and WIN-Win XXX, ALDH inhibitor. C: Heatmap containing major endoderm genes from pluripotent cultures subjected to classical definitive endoderm differentiation conditions or differentiated using conditions predicted by the HNF1β optimizer. [Figure 4]We show that DFE is capable of generating pancreatic endoderm with highly dorsalized properties. A: Schematic showing modeling of optimization of gastric genes SOX2, OSR1 and pancreatic gene PDX1. B: Corresponding optimizers for predicted maximum induction of SOX2, OSR1 and PDX1, respectively. C: Dynamic profile of effectors most responsible for each gene activation. D: Representative IHC of PDX1Opt. E: KeyGenes prediction of pancreatic endoderm derived from DE and DFE, respectively. F: Heatmap assessing differential expression of several pancreatic genes and dorsal specific pancreatic genes between the two protocols. [Diagram 5] Results of effective and rapid induction of pancreatic endocrine fate from pluripotency via DFE endoderm using small molecules. A: Schematic of reactions performed to alter pancreatic induction. B: C-peptide levels detected across different experimental conditions as determined via ELISA. C: QuantStudio-based expression levels of several key pancreatic and endocrine specific genes normalized to the average expression of GAPDH, TBP and YWHAZ. [Figure 6] Results showing that DFE-derived endocrine cells have an immature phenotype. A: Schematic showing the directed differentiation protocol used to generate endocrine sections. B: IHC assessment of end-point cultures for various endocrine-specific factors. C: Endocrine induction over 14 days was assayed daily for c-peptide production via ELISA. D: Retention of dithizone by pluripotent derivatives grown as aggregates. E: Glucose-stimulated insulin release assay (GSIS) assessment of endocrine function of endocrine cultures. F: Microfluidic analysis of internal calcium release in response to glucose and potassium challenge. [Figure 7]Results of a bioequivalence study of a small molecule-based pancreatic endocrine induction protocol. A: Schematic of reactions performed in which HNF1βOpt and PDX1Opt were replaced with only the factors from the modeling experiment with the highest predicted factor contributions, retinoic acid and LDN3189, or retinoic acid and PD0325, respectively. B: QuantStudio-based transcript analysis of the four respective reaction conditions at either the PE stage (shown in light grey) or the endocrine stage (shown in dark grey) normalized to the average expression of GAPDH, TBP, and YWHAZ. [Figure 8] Kinetics of DFE induction are shown. A: Graph showing relative percentage of FOXA2, HNF1β, HLXB9 and OCT3 / 4 positive cells over a 4 day incubation period with HNF1βopt. B: Representative IHC for HNF1β, FOXA2 and HLXB9 on day 3 of the differentiation protocol. [Figure 9] Figure 1 shows that DFE protocol input does not alter DE patterning. A: Representative IHC showing differential expression between TGFβ or retinoic acid responsive genes. [Figure 10] We show that the patterning capacity of DFE is distinct from DE-derived endoderm. A: Schematic of how either HNF1β-optimized cultures or cultures differentiated into definitive endoderm respond to being challenged to differentiate towards gastric, pancreatic, or hepatic fates. B: Relative expression of selectable markers for the indicated genes across different protocols. Blue bars represent cultures provided with conditions designed to differentiate towards the stomach. Yellow bars represent cultures provided with conditions designed to generate pancreas. Red bars represent cultures provided with conditions designed to generate liver. C: Schematic of a follow-up experiment in which organoids were grown from HNF1βopt cultures challenged with gastric induction conditions. D: DFE-derived organoids differentiated and grown over a 23-day period. E: Representative organoid staining for PSC-derived putative antral PDX1+ / SOX2+ / OSR1+ gastric genetic markers. [Figure 11]Showing enhanced pancreatic competence in DFE versus DE-derived endoderm. A: Schematic showing four different early endoderm genes optimized by computer modeling and subsequently challenged for pancreatic competence. B: IHC validation of four respective cultures for the general endoderm gene FOXA2. C: IHC validation of four respective cultures for the pancreatic specific gene PDX1. [Figure 12] Results show robust and homogenous induction of pancreatic fate from DFE using iPSCs and female hESCs. A: IHC validation of the HNF1βopt protocol among three different pluripotent cell lines with upregulated expression levels of corresponding key factors. B: IHC validation of pancreatic induction from DFE among three different pluripotent cell lines with upregulated expression levels of corresponding key pancreatic factors. [Figure 13] Hierarchical clustering comparing pluripotent-derived cells with several primary tissues. Hierarchical clustering was performed on the KeyGenes dataset in R. All pluripotent-derived samples fell into the same cluster, including human fetal stomach, fetal pancreas, and adult islet samples, so only this region of the cluster is shown. [Figure 14] It shows that retinoic acid metabolites are effective in activating dorsal foregut endoderm. A: Schematic showing the experiment performed. B: HNF1βOpt for the experiment described in panel A. C: Graph showing the differential effect of different retinoids in activating selected endoderm genes. D: Corresponding IHC validation comparing protein levels between original HNF1βOpt to conditions predicted by the current experimental design. E: Added compounds for reaction conditions are referred to as old and new HNF1βOpt. [Figure 15] Figure 1 shows that all-trans retinoic acid activates Nodal. All-trans retinoic acid and retinoic acid metabolites have the differentiation potential to activate the primitive streak marker Nodal. A: Schematic of the experiments performed. B: Relative expression of selected endodermal genes compared to retinoid-based induction. C: Schematic showing the experiments performed to obtain retinoid-based NODALopt conditions. [Figure 16]Patterning of endoderm fate space by initiating gastrulation. A: PSC cultures were exposed to a defined matrix of DoEs of compounds known to either initiate or inhibit gastrulation. B: tSNE plot of all optimizers obtained for assayed endoderm genes. C: Corresponding heatmap showing which genes segregate into each of the four determined populations shown as defined in the tSNE plot. [Figure 17] 1 shows the contribution of Activin A, BMP4 and Retinoic Acid to endoderm patterning during gastrulation. The relative factor contributions obtained from DoE-based modeling experiments assaying the effects of Activin A, BMP4 and Retinoic Acid during in vitro gastrulation events are shown in the table. [Figure 18] Quantification of DFE protocols. A: Representative IHC from 4 different DFE protocols as detailed in the schematic diagram in the image above. A': Corresponding quantification of the different protocols from panel A. B: Representative IHC from 4 different DFE protocols with further differentiation towards pancreatic progenitors as detailed in the schematic diagram in the image above. B': Corresponding quantification of the different protocols from panel B. C: Representative IHC from 4 different DFE protocols with further differentiation towards pancreatic progenitors using only the key process parameters as defined through the differentiation model and detailed in the schematic diagram in the image above. C': Corresponding quantification of the different protocols from panel B. [Figure 19] Defining relative effector contributions to representative patterned endoderm populations. A: Ridge plot showing the relative levels of different effectors used throughout the experiment. The different plots refer to the optimizer's clustering and are color-matched to different clusters. B: The original clusters are again displayed for ease of reference with the corresponding cluster numbers (0-3) presented. [Figure 20]Evaluating expression patterns within endoderm clusters. A: The original clusters are relabeled for ease of reference with corresponding cluster numbers presented (0-3). B: Cluster 0 consists of dorsal foregut endoderm (DFE). C: Cluster 1 consists of anterior to midgut endoderm (ADE) populations. D: Expression patterns of primitive streak markers Nodal and ALDH1A2. [Figure 21] Anterior and midgut genes are induced in the absence of AA. Violin plots defining relative expression levels across different clusters of selected endodermal genes. [Figure 22] Anterior endoderm genes are BMP4 and Nodal responsive. Twenty independent optimizers for each gene are shown in individual graphs, focusing on the predicted concentrations of TGFβ family ligands AA, Nodal, and BMP4 involved in the activation of each gene. [Diagram 23] AA, BMP, Nodal, and retinoic acid all have their own responsive gene networks. The effector factor contributions from all respective optimizers to the measured endoderm genes were normalized and ranked from most to least responsive. [Figure 24] AA-stimulated pluripotent cells activate several TGFβ family ligands. A: RNA-seq data was used to determine the expression patterns of TGFβ family ligands that were activated in both DE and DFE conditions. Genes were ranked from most to least abundant transcripts in the DE population and undetected levels were removed from the heatmap. B: IHC confirming expression of Nodal in AA-stimulated pluripotent cells. [Diagram 25] Anterior and midgut patterning occurs in the absence of AA signaling. Representative optimizers for master regulators of anterior and midgut genes. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0024] Provided herein are compositions, systems, kits, and methods for generating dorsal foregut endoderm (DFE) cells by contacting pluripotent stem cells with a retinoic acid signaling pathway agonist and a bone morphogenetic (BMP) pathway inhibitor without exposing the stem cells to a transforming growth factor beta (TGFβ) pathway agonist. In certain embodiments, the DFE cells are contacted with a retinoic acid signaling pathway agonist and an FGFR pathway inhibitor to generate pancreatic endoderm (PE) cells (e.g., with dorsal identity). In other embodiments, the PE cells are contacted with an ALK5 inhibitor and a Notch inhibitor to generate endocrine cells (e.g., insulin-expressing endocrine cells). In certain embodiments, systems, kits, and methods are provided for generating anterior domain endoderm (ADE) cells.

[0025] Induction of endoderm and progeny organs such as pancreas, liver, and intestine has impacted disease modeling and regenerative medicine. Common methods for inducing definitive endoderm from pluripotency necessarily involve the use of TGFβ signaling agonism, most commonly activin A. Work conducted during the development of the embodiments described herein did not require activin A, but determined that BMP antagonism and retinoid inputs could be successfully used to lead to induction of dorsal foregut endoderm (DFE), and that activin-induced endoderm results in cells that exhibit ventral heterogeneous characteristics. Pancreatic identity could be rapidly and robustly derived from DFE, thereby demonstrating that such cells are of dorsal pancreatic identity. In such work, DFE populations were highly competent to differentiate into both gastric organoid and pancreatic tissue types, and could effectively generate fetal beta cells by subsequent differentiation of the two through a small molecule-based process. The generation of such pancreatic insulin-producing cells is useful, for example, for cell-based therapy of diabetic patients.

[0026] Work carried out during the development of the embodiments herein has demonstrated that efficient and regionalized patterned endoderm can be reliably differentiated directly from pluripotency without the use of TGFβ agonism, and that pancreatic and other endodermal derivatives can be generated from such cells. The embodiments herein provide a robust protocol for inducing specialized human endoderm representative of the dorsal foregut region of the intestinal tube. Such generated cells can be effectively converted into dorsal pancreatic progenitor cells, which can subsequently adopt endocrine fates, including the generation of fetal-like beta cells.

[0027] The present disclosure is not limited by the type of medium or culture system used. Any suitable medium and culture system may be used. For example, one skilled in the art will appreciate that multiple methods are available for the culture and growth of pluripotent stem cells, with or without mesenchymal feeder cells, and with or without culture in fully chemically defined conditions, such as serum, or other complex not fully defined additives. Also, multiple, flexible growth medium bases are available, exemplified by hybrid media (e.g., media mixtures such as CDM2, as used internally), as well as specialized media such as Essential-8, Essential-6, TesR media, and various supplements such as KOSR, used to support the growth of pluripotent stem cell states, regardless of cell origin, such as ESC or iPSC. In certain embodiments, a series of specialized media of a continuous, sequential nature may be used, all of which have an underlying basal medium that may be flexibly explored from a range of widely used cell culture media, such as RPMI1640, DMEM, F2, and others. Such media are then provided with a series of metabolic support molecules such as pyruvate, anaplerotic agents such as glutamine, chemically defined lipids, a fuel source such as glucose, and optionally a chemically defined growth support such as B27, as commonly practiced by those of skill in the art. In certain embodiments, a variety of basal media, as well as metabolic support molecules, are used.

[0028] In certain embodiments, the cells produced using the methods and compositions herein are endocrine pancreatic insulin-producing cells. In certain embodiments, such cells are macroencapsulated as known in the art and then implanted into a patient with diabetes (see, e.g., Hwa et al., Curr Diab Rep. 2018 Jun 16;18(8):50, incorporated herein by reference). In other embodiments, the insulin-producing cells are implanted into the liver of a subject (e.g., a subject with diabetes). In other embodiments, such cells are implanted subcutaneously in a protective device such as described in clinical trial NCT03163511. In yet other embodiments, such cells are implanted in a microencapsulated sphere such as described in clinical trial NCT01739829. In other embodiments, such cells are implanted in an oxygen delivery device such as described in clinical trial NCT02064309. In other embodiments, such cells can be implanted at a suitable site (e.g., subcutaneously, intraperitoneally, intravenously) without physical protection but in the presence of an immunomodulatory agent. In other embodiments, such cells are prepared from the method described in An et al., Proc Natl Acad Sci, incorporated herein by reference. The transplant may be performed using a retrievable polymeric nanoporous string coated with alginate and implanted in the peritoneal cavity, as described in USA.2018 Jan 9;115(2):E263-E272. In some embodiments, the diabetic patient may be a type II diabetic patient with no autoimmune symptoms. In some embodiments, the cells may be derived through a process of pluripotent stem cell reprogramming and may be of a similar genetic makeup to the recipient.

[0029] In certain embodiments, certain cells are substituted for pluripotent stem cells as the original set of cells treated. Such cells include EP (Cheng et al., Cell Stem Cell. 2012, incorporated herein by reference). pluripotent stem cells or other cells may be derived from, but are not limited to, endodermal progenitor cells as described in Tesar Proc Natl Acad Sci USA. 2005 Jun 7; 102(23): 8239-44, or embryonic germ cells as described by Kimura et al., Stem Cells. 2014 Oct; 32(10): 2668-78, which is incorporated herein by reference. The present disclosure is not limited by the species from which the above-mentioned pluripotent stem cells or other cells are derived. All vertebrate species produce similarly patterned endoderm that forms anteriorly located derivatives therein that generate anterior endoderm progeny tissues, and all vertebrate species produce cells of such endoderm that produce the insulin molecule that controls glucose homeostasis. In some embodiments, the generation of pancreatic endoderm, pancreatic insulin-producing cells from non-human species may be advantageously used to provide a means to induce cell-based therapy, such as when combined with either microencapsulation devices, macroencapsulation devices, or immunosuppressive drug regimens, or alternatively, when genetically modified to allow for loss of immune detection. Furthermore, such cells, regardless of species origin, can be used as the basis for drug discovery of in vitro derived insulin-producing cells, thus leading to the identification of novel drugs that affect the diabetic condition.

[0030] In certain embodiments, the insulin producing cells produced by the methods and compositions described herein are used with technologies from any of the following companies: 1) Seraxis (cell replacement therapy for type 1 diabetes), 2) Unicyte AG (diabetes cell therapy), 3) ViaCyte (insulin for diabetes), 4) Sernova (stem cells for diabetes treatment), 5) Betalin Therapeutics (insulin production), 6) AltuCell (stem cell therapy for diabetes type 2), 7) NextCell Pharma AB (stem cell therapy for diabetes type 1), 8) Osiris Therapeutics (control of type 1 diabetes), 9) Mesoblast (type 2 diabetes treatment), 10) Evotec and Sanofi (stem cells for diabetes), 11) Orgenesis (insulin producing cells), 12) Semma Therapeutics (type 1 diabetes therapy), 13) Novo Nordisk (type 1 diabetes treatment), 14) Beta-O2 (artificial pancreas for diabetes), 15) Eli Lilly and Sigilon (Stem Cells for Diabetes). EXAMPLES

[0031] Example 1 Robust and rapid induction of dorsal pancreatic endoderm from pluripotent cells This example describes the generation of dorsal foregut endoderm (DFE) cells from pluripotent stem cells. This example also describes the generation of dorsal pancreatic endoderm cells from DFE cells.

[0032] result Two separate and distinct pathways exist for endoderm induction We initially set out to predict the conditions necessary to define a primitive streak progeny definitive endoderm population as suggested by the literature. This was accomplished by defining an anterior primitive streak (APS) population by modeling for maximal expression of MESP1, EOMES, and BRACHYURY / T while minimizing EVX1 (a posterior primitive streak marker) (Loh et al., 2014). Through ISPA, the conditions predicted to generate this differentiation event consisted of low resistance to Wnt inhibition (FC=18.65) and high levels of Activin A (FC=18.85) (Figures 2A and 2B). These conditions are consistent with current protocols for generating definitive endoderm (DE) that rely on Nodal signaling (D'Amour et al., 2005). Indeed, when differentiating pluripotent cultures using this condition, the majority of the cultures converted to a FOXA2+ / SOX17+ population within 3 days (Figure 2C). Through ISPA, numerous other genes were predicted to be highly expressed under these conditions, including COL6A1, HHEX, MESP2, and SOX17 (Figure 2D, data not shown). Of note, POU5F1 (OCT4) remained high, consistent with a node-like state. All of these genes are known to be elevated in lineal progeny after gastrulation in mice. However, through ISPA, we noted that not all known early endoderm expressed genes were uniquely maximized through APS conditions. Inspecting the known space from the experiments described above, we were able to obtain a rather different set of solutions, focusing on HNF1β (TCF2) expression, which is also known to be expressed in definitive endoderm. Maximizing HNF1β expression resulted in the expression of concomitant genes such as FOXA2, HNF4a, MNX1, CXCR4, and MTF1, all known markers of endoderm, while HHEX and SOX17 expression remained low (ISPA results not shown). We decided to characterize these different states and the requirements for their induction.

[0033] Retinoic acid and Bmp inhibition synergistically induce a mutually exclusive endoderm program in activin-induced endoderm Using ISPA, we extracted the underlying logic governing endoderm gene induction. For APS-derived DE, it was evident that many early endoderm genes were under the direct control of TGFβ signaling and showed strong, positive coefficient terms from AA in the complex regulatory model. These genes included, but were not limited to, SOX17, CXCR4, LEFTY1, MIXL1, and HHEX (Figure 2D). This was not the case for multiple other known endoderm markers. A subgroup of endoderm genes did not respond to activin A stimulation but was directly dependent on retinoic acid signaling and, to a lesser extent, required inhibition of the BMP pathway. These genes included FOXA2, EPCAM, ONECUT1, CDX2, and MNX1 (Figure 2D), and were predicted to be directly regulated through the synergistic effect of retinoic acid and BMP inhibition with high factor contributions including FOXA2 (RA=24.8, FC at BMPi=25.3), HNF1β (RA=30.1, FC at BMPi=31.7), and MNX1 (RA=22.7, FC at BMPi=30.5) (Figure 2B, data not shown). Notably, activin A was predicted not to contribute to the activation of these genes, rather inhibition of the TGFβ pathway was predicted to benefit the expression of these genes, with factor contributions of Alk5i of 7.59 for HNF1β, 8.88 for FOXA2, and 18.98 for MNX1 (Figure 2B, data not shown). Predicted conditions for optimizing HNF1β for differentiation of pluripotent cells for validation based on protein expression that can all activate HNF1β, MNX1, and FOXA2 and effectively transform cells (HNF1β Opt ) was tested. Opt Application of the conditions resulted in the majority of cells being HNF1β+ (Fig. 8), whereas APS-differentiated cells showed only weak HNF1β expression (Fig. 2C).

[0034] The two pathways to endoderm activation were fundamentally different and based on contradictory input logics suggesting that these pathways are mutually exclusive. This was tested by creating a hybrid protocol assaying the effects of RA and BMPi in the presence of activin A (Figure 3A). The inclusion of retinoic acid in the APS-based DE generation protocol (D'Amour et al., 2005) proved to only modestly increase gene expression of retinoic acid-responsive genes, with only CDX2 (data not shown) and OSR1 (Figure 3B) being significantly upregulated in this method. Conversely, known TGFβ-responsive genes were shown to be significantly downregulated when retinoic acid was included in the APS-type DE generation reaction, including HHEX, SOX17, and GSC (Figure 3B and Supplementary Figure 2). This indicates that the presence of either of the key protocol drivers (AA vs. RA) suppresses the other protocol drivers. Moreover, as predicted by ISPA, genes upregulated in the presence of retinoic acid and LDN3189 were more efficiently activated when activin A was omitted from these responses. Importantly, HNF1β OPT The two main protocol inputs for conditioning, RA and LDN3189, were sufficient to effectively initiate differentiation (Figure 3B).

[0035] Retinoic acid / BMP inhibition-induced endoderm is of dorsal foregut phenotype To better understand the distinct nature of the endoderm populations, cultures were subjected to RNA sequencing. Common endoderm genes expressed in both populations included CXCR4, FOXA2, EPCAM, GATA4, and GATA6 (Figure 3C). However, striking differences were observed for genes related to patterning, including HNF1β. OptWe found that the inductive endoderm was enriched for genes characteristic of the known dorsal (MNX1 and Pax6) and foregut endoderm HOXA1, HOXA3, HNF4A, and HNF1β (Figure 3C). In contrast, the APS-type DE showed enrichment for genes representative of the ventral endoderm NR5A2, HHEX, and SOX17, as well as more posterior endoderm (SOX17 and AFP, but not CDX1) (Figure 3C). HNF1β Opt Because the cultures appeared to have more of a dorsal foregut endoderm (DFE) phenotype, differentiation competence towards stomach, pancreas and liver (tissues derived from the posterior foregut) was addressed using conditions previously shown to induce these fates from APS-derived DE (Figure 3A, Supplementary Figure 3B) compared to APS-derived DE. A striking difference in competence was observed when DFE was compared to APS-DE populations. In all cases, APS-DE cultures activated liver genes (APOB, HHEX, and EVX1 (Supplementary Figure 3B)) more efficiently than DFE. However, stomach (OSR1) and pancreas (PDX1) were more effectively activated in DFE cultures. When working with DFE cultures to generate gastric organoids (Figures 10C and 10D, (McCracken et al., 2014)), co-expression of OSR1 and PDX1 was observed, as was co-expression of SOX2 and PDX1, suggesting that gastric organoids preferentially transformed into the antral-type, most posterior stomach (Figure 10E).

[0036] Mechanism of organ specification from dorsal foregut endoderm When addressing cultures for pancreatic induction, DFE-patterned cultures had greater competence for PDX1 activation compared to traditional DE induction methods (Figure 10A). Therefore, we returned to ISPA to examine pancreatic induction by optimizing different early endoderm markers at stage 1 to determine which markers would best enable pancreatic conversion. Using ISPA, we identified and compared independent conditions that maximized expression of HNF1β or OC2 as representative genes under the control of retinoic acid, or optimized expression of NR5A2 or SOX17 as representative genes under the control of TGFβ signaling (Figure 11A). We first exposed pluripotent cultures to induction conditions for each of the four genes (for 3 days), and then induced pancreas formation using previously defined conditions (Mfopou et al., 2010). Cultures optimized for HNF1β, or OC2, showed much more homogenous induction for the pancreatic markers PDX1, as well as FOXA2 (Figure 11B and C). Having observed that DFE cultures were competent for both gastric and pancreatic induction, we next assessed the nature of this bipotentiality through sequential DoE modeling experiments performed on DFE cultures.

[0037] Results from ISPA maximization of SOX2, OSR1 or PDX1 expression were compared (Figure 4A and B). Dynamic profiling analysis (Figure 4C) revealed that both OSR1 (stomach) and PDX1 (pancreas) are under similar regulatory mechanisms. Both genes responded strongly to RA with FC=30.68 and FC=31.38, respectively (Figure 4B and C). However, differential responsiveness to FGF signaling was predicted, with FGF4 predicted to be important for OSR1 (stomach) activation with an FC of 13.45, while MEK pathway inhibition was shown to strongly contribute to PDX1 activation with a substantial FC of 29.62 for the small molecule inhibitor PD0325901 (Figure 4B and C). Thus, the bipotentiality of gastric / pancreatic fates is resolved by FGF / FGF inputs, respectively. Notably, SOX2 expression was strongly reduced by RA and highly regulated by SHH, with FCs of -28.59 and 29.58, respectively (Figure 4B and C). SOX2 is a continuous marker of the esophagus and fundus, and thus immediately anterior to the PDX1-expressing gastric antrum in rodents (McCracken et al., 2014). Our data argue that retinoic acid anchors the posterior antrum and that active inhibition of SHH contributes to the downregulation of SOX2 expression, thereby enabling the switch from the gastric to the pancreatic portion. This observation is supported by developmental studies in chick (Hebrok et al., 1998).

[0038] Pancreases derived from DFE are of dorsal identity Using ISPA-defined PDX1 optimized conditions (shown in Figure 4B), it was demonstrated that PDX1-expressing DFE-derived PE cultures co-express several known pancreatic progenitor markers, including FOXA2, NR5A2, GATA4, and SOX9, indicating that a true PE state was rapidly induced (Figure 4D). This DFE-derived pancreatic induction was shown to be reproducible in H9 female embryonic stem cell lines, as well as in iPSC cultures (Figure 12). Using RNASeq-based KeyGenes analysis (Roost et al., 2015), we verified that this DFE-derived PDX1 induction was truly pancreatic destined (Figure 4E). We compared pancreatic endoderm (PE) derived from DFE to stage 4 PE derived from APS-type DE, as previously published (Xie et al., 2013). Interestingly, both initiating populations DE and DFE cultures initially showed a restricted signature to the brain, which was lost throughout successive stages of both protocols (Figure 4E). This is likely due to the emergent expression of neuronal fate suppressor genes activated during the successive stage(s) in both populations (Jennings et al., 2017) (Figure 4F). The published APS-derived DE population showed similarity to the lung, whereas the DFE population did not (Figure 4E). This is speculated to be because the lung bud is derived exclusively from the ventral endoderm. Hierarchical clustering comparison between the DFE, DE and their derived PE populations showed that the DFE and DE populations were highly similar to the DFE population clustered among the three sets of DE samples (Figure 13), proving that both the ventral and dorsal protocols achieve a basic endoderm program. However, when comparing PE derived from either DFE or DE to genes highly enriched in the developing dorsal pancreas in humans (Jennings et al., 2017), the PE populations were highly enriched in the genes highly enriched in the developing dorsal pancreas in humans (Jennings et al., 2017). (Esther et al., 2017), we found that the DFE population already expressed several of these genes, including DLL1, CNR1, FRZB, HOXA1, and ARMC3, whereas the APS-DE population showed only low expression of CNR1 and FRZB. Also, expression of the previously described dorsal marker MNX1 (Hlxb9) was expressed throughout the DFE cultures, whereas DE-derived PE only became detectable MNX1 expression at the PFG stage (Figure 4F). Thereafter, DFE-derived PE continued to express the majority of dorsal-specific genes (Figure 4F). In comparison to current pancreatic DD protocols (Xie et al., 2013), we conclude that DE-derived PE adopted a pancreatic phenotype at the published "PGT" stage, as evident through expression of HHEX, HNF1b, RFX6, HNF4A, PDX1, PROM1 and PROX1, but did not have a discernible dorsal identity until the cultures were exposed to retinoic acid at the "PFG" stage. As previously mentioned, in light of the ISPA results, although the present invention is not limited to any particular mechanism, and an understanding of the mechanism is not necessary to understand or practice the present invention, this implies that the dorsal phenotype previously attributed to these protocols is the result of redirecting the population to a dorsal fate later in the protocol.

[0039] Effective induction of endocrine pancreas from DFE-derived PE is time-dependent To further validate the identity of pancreatic precursors, DFE-derived PEs were characterized for their forward differentiation potential towards a terminal pancreatic fate using pancreatic lineage-specific markers. As the importance of NOTCH pathway inhibition for terminal differentiation of endocrine cells is well known (Jensen et al., 2000, Afelik et al., 2012), the temporal effect of NOTCH inhibition on endocrine commitment was assessed using three media inputs outlined in Figure 5A, all containing a NOTCH pathway inhibitor (g-secretase inhibitor XX), evaluating single SMAD (Alk5 inhibitor) and dual SMAD inhibition (Alk5i / LDN). Regardless of the conditions, these media were able to induce endocrine differentiation. Single SMAD inhibition appeared to be preferable to dual SMAD inhibition, with the absence of SMAD inhibition resulting in slightly increased acinar differentiation (Figure 5B-C). Multiple developmental systems (e.g., somitogenesis, neurodevelopment) depend on the total duration of NOTCH signaling, and therefore varied the residence time during pancreatic induction at which the NOTCH pathway becomes active. Notably, extending the duration of PE induction reduced endocrine competence. These data revealed that cultures progress through a competence window after PDX1 activation that, if extended, results in loss of endocrine fate even under permissive endocrine induction conditions (Figure 5C). Notably, increasing the duration at the PE stage increased competence toward acinar and, ultimately, toward ductal differentiation. Acinar competence, as assayed through expression of CPB1, MIST1 (BHLHA15), was observed when PE cultures remained capable of engaging in NOTCH signaling for 5 days before undergoing NOTCH inhibition (Figure 5C). This implies that endocrine lineage competence is temporally regulated first pre-acinar and then finally toward the ductal lineage (Figures 14A and 14C). Ductal differentiation, as observed through expression of F3, HNF1β, and PROM1, increased the longer the PE population was allowed to remain in the PE state with active NOTCH ( Figure 5C ), consistent with NOTCH being permissive for ductal fate ( Afelik et al., 2012 ).

[0040] Dorsal PE-derived β cells are fetal-like Acknowledging that conditions for forward differentiation towards an endocrine fate were primarily dependent on NOTCH signaling inhibition and not optimized using a QbD approach, we examined the resulting endocrine output and also determined the maturation state. Using a short duration PE state (Figure 6A), we found that PDX1+ / NKX6.1+ co-expression, a cell state representative of TrPC pro-endocrine cells and necessary to generate functional beta cells, occurred throughout the culture (Figure 6B). In humans, but not in mice, NKX2.2 is beta cell specific and is not expressed by MPCs. NKX2.2 was co-expressed with insulin C-peptide within patches of whole cultures. Furthermore, in cells co-expressing INS+ / GCG+ or INS+ / SST+, induction of multiple endocrine products was observed (Figure 6B). The longer the cultures were exposed to Notch inhibition, the greater the C-peptide production, with maximum levels occurring after 10 days and stabilizing thereafter (Figure 6C). Cell aggregates were maintained in a 10-day culture with dithizone retention, resulting in Zn 2+ We demonstrated a 4-fold increase in detectable C-peptide levels when challenged with insulin secretion (Figure 6D). When challenged with insulin secretion, KCl induced a 4-fold increase in detectable C-peptide levels, whereas glucose challenge (2 mM > 20 mM) did not induce C-peptide secretion (Figure 6E). Microfluidic evaluation of endocrine-derived cultures showed approximately 6% detectable calcium influx (Figure 6F), and again an absence of glucose response was observed. The results of classical GSIS and microfluidic assays were consistent and demonstrated immature β-cell physiology capable of synthesizing and storing insulin, but with a limited functional profile when responding to glucose excursions, a characteristic of the fetal-like β-cell state previously observed from APS-DE-derived β-cells (Hrvatin et al., 2014).

[0041] Small molecule methods for the induction of fetal-like β cells PDX1 OptExamination of conditions confirmed that retinoic acid and PD0325 had the highest factor contributions of 31.38 and 29.62, respectively (Figure 7A), with all other effectors tested having low factor contributions <10. To achieve a minimally complex method for pancreatic endocrine cell induction, such components for DFE induction (RA and LDN(R / L)) were combined with PDX1 Opt First, we combined HNF1β with key components (RA and PD0325(R / P)) and finally completed the conversion to an endocrine fate using NOTCH and ALK5 inhibition. opt Following DFE induction with HNF1β, a significant increase in the R / P ratio of the pancreatic markers assayed was observed, and this increase was amplified when endocrine induction continued (Figure 7B). Opt When the (R / L) step was replaced by the use of the (R / L) step, a further increase in pancreatic genes was observed. opt The conditions were used to produce the highest levels of the pancreatic genes INS, GCG, NKX6.1, CHGA, GLP1R, and PDX1 (Figure 7B). Opt and PDX1 Opt Comparable changes were observed when both were replaced together with R / L and R / P. A slight decrease in INS, GCG, NKX6.1, CHGA, GLP1R and PDX1 expression was observed under these conditions, whereas a moderate compensatory increase in the expression of NEUROD1, NGN3, PAX4, MAFA and SST was observed. We conclude that directed differentiation of pluripotent cells into fetal beta cells can be rapidly achieved through three successive steps using five small molecules: RA, LDN3189, PD0325901, gamma-XX and A8301.

[0042] We demonstrate a novel protocol that can rapidly convert pluripotent cells into a regionalized endoderm population (DFE) that has the competence to form dorsal pancreatic progenitors and undergo endocrine conversion through a three-step protocol that relies on the use of only five small molecules.

[0043] RNA-Seq data demonstrated that DFE is of dorsal phenotype based on multiple genes. Jennings et al. investigated the dorsal and ventral identity of the developing human pancreas and showed that a defined set of 13 genes can distinguish between dorsal and ventral pancreas. As a basis for our interpretation, the DFE population was consistent with a dorsal fate. Using the administration of retinoic acid at a later stage, this dorsal aspect was injected into the APS-DE. In the case of DFE, the dorsal identity is carried over into the pancreas and this state is permissive for induction of all pancreatic lineages. Although previous studies have demonstrated that DE undergoes a hepatic vagus fate (Mfopou et al., 2010), DFE-derived pancreas does not, we point out that DE is bipotential for antral gastric induction. This is expected because the ventral pancreas has bipotential competence for hepatic induction (Angelo et al., 2012; Deutsch et al., 2001; Tremblay and Zaret, 2005; Bort et al., 2004). Early lateral plate-derived BMPs directed this progenitor towards a hepatic fate (Chung et al., 2008), and PE treated with BMP4 also induced hepatic genes. In contrast, ISPA analysis suggested that SHH and retinoic acid are key process inputs influencing bipotential pancreatic / gastric progenitors. These findings are consistent with previous studies of dorsal pancreatic specification in chick (Hebrok et al., 1998). Regardless of the dorsal / ventral origin of the endoderm population, both DFE and APS-DE readily give rise to pancreatic endoderm capable of generating endocrine cells. As observed for APS-DE-derived cells, DFE-derived endocrine cells are more functionally similar to fetal β-cells than to fully mature glucose-responsive β-cells.

[0044] Materials and Methods cell culture Pluripotent cultures were adapted to Essential 8 medium (Gibco A15169-01) and grown on vitronectin (Gibco A14700) coated plates (Corning Inc. 3598). Differentiation experiments were performed on 1 cm 2APS cells were seeded at 75,000 cells per well and maintained in E8 medium for 48 h until cultures were approximately 90% confluent before initiating differentiation. All growth factors and small molecules were reconstituted according to the supplier's recommendations and aliquots were stored at -80°C for up to 1 year. Growth medium was changed daily and the basal medium used in all differentiation experiments was CDM2, except for the DE protocol (Loh et al., 2014). Opt 250 nM LDN3189 (Sellekchem The IFN-γ ... Opt The medium was supplemented with 250 nM LDN3189, 500 nM A8301 (Biogems 4463325), 12.6 ng / ml Wnt3a (R&D5036WN / CF), 25 ng / ml bFGF, 250 nM PD0325 (Selleckchme S1036), 40 ng / ml SHH (Peprotech 100-45) and 2 uM retinoic acid (Sigma Aldrich R2625). DE was generated using previously described conditions (Rezania et al., 2014; D'Amour et al., 2005). Opt The medium consisted of 500 nM A8301, 250 nM LDN3189, 100 nM Sant1, 250 nM PD0325, 2 uM retinoic acid and 1% B27 supplement (Gibco 12587) and was used for 3 consecutive days to differentiate the cells unless stated differently in the text. Endocrine Push Medium was supplemented with 100 nM gamma-secretase inhibitor XX (EMD Millipore 565789) and 500 nM A8301. Gastric organoids were generated using a modified protocol (McCracken et al., 2014). Briefly, HNF1β OptCultures were incubated for 3 days in stage 2 medium supplemented with 500ng / ml Wnt3a, 500ng / ml FGF4 (R&D 7460-F4), 250nM LDN3189 and 2uM retinoic acid (only on the first day of RA). Cultures were seeded in a 1:4 Matrigel (Fisher Scientific 354230) / CDM2 mixture, incubated at 37°C for 30 minutes, and then passaged 1:6 by applying stage 3 medium consisting of 2uM retinoic acid, 250nM LDN3189 and 100ng / ml EGF (R&D 236-EG) for 3 days. Cultures were then supplemented with stage 4 medium supplemented with 100ng / ml EGF for 23 days, after which organoids were harvested and fixed for sectioning.

[0045] Computer modelling All DoE experimental designs were computationally generated using D-optimal interaction design in MODDE software (Sartorius Stedim Data Analytical Solutions, SSDAS). Perturbation matrices were generated (96 independent experiments performed) on a Freedom Evo150 liquid handling robot (TECAN, CH). Each DoE differentiation experiment used a 3-day application of perturbation matrix with daily changes of medium. All cell cultures, including manual preliminary growth and seeding, as well as any robotic manipulations, were contained in a modular X-Vivo system (Biospherix, NY, USA) that provides process analytical technology (PAT) of cell culture conditions (providing %N2, %O2, %CO2, as well as temperature control and monitoring at any point in the experimental process). End-point data collection was performed using an Open Array RT-QPCR method (QuantStudio, Life Technologies, 53 genes / sample / card). QuantStudio gene expression data were normalized and standardized within Expression SuiteTM (Life Tech) and exported to MODDE. Mathematical modeling of the differentiation space was performed in MODDE using PLS to maximize the explained variance, and then gene-specific models were adjusted by Q2 maximization. Fitted model optimization terms for maximizing or minimizing the response were extracted using the software prediction tool "Optimizer" and "Dynamic Profile" analysis. Virtual experiments were performed in the known space covered by the experimental design, addressing the relative contribution of pathways to the desired parameters.

[0046] RNA and cDNA preparations RNA extraction for modeling analysis was performed using MagMax-96 Total RNA Isolation kit (Life Technologies AM1830) and was performed according to the manufacturer's protocol. Reverse transcription of RNA samples was performed using the reaction conditions provided by the High Capacity cDNA RT Kit (Life Technologies 4368814). RNA extraction and QRT-PCR validation for RNA Seq was performed in quadruplicate with RNA extraction generated using a Trizol (Life Technologies 15596018) based method. cDNA was then subjected to QuantStudio analysis for OpenArray validation (not modeling). The resulting data was then normalized to housekeeping genes and graphed using GraphPad Prism 5.02 software.

[0047] QRT-PCR for modeling analysis Samples were loaded onto custom-designed Quant Studio Cards using the OpenArray AccuFill System (Life Technologies 4471021) and run on a QuantStudio 12k Flex Real-Time PCR System (Life Technologies 4471090). QuantStudio open array analysis provided expression information for 53 genes selected for each open array configuration. The resulting data underwent QC assessment in ExpressionSuite to remove any genes that failed to amplify and were normalized to three internal standards (internal standards were changed between designs). Early designs included genes responsible for all major sub-lineage fates downstream of pluripotency, whereas later designs were more focused on the endodermal lineage with emphasis on pancreatic fate. For all open array designs used, the primer / TaqMan probe sets per platform provided typical confidence amplification (Cq Conf.) scores of 0.975 or higher. Each modeling experiment generated approximately 10,000 individual gene expression response data points on which the MODDE optimization was based.

[0048] Immunohistochemical analysis of cultures Histological characterization was performed by treating the cell cultures with 4% paraformaldehyde (EMD30525-89-4) solution for 15 min at ambient temperature. Samples were then blocked and permeabilized for 1 h with 0.1 M Tris / HCl (Promega H5123) pH 7.5 supplemented with 0.5% blocking reagent (Perkin Elmer FP1012) and 0.1% TrintonX-100 (Fisher Scientific BP151). All antibody solutions were diluted in 0.1 M Tris / HCl (Promega H5123) pH 7.5, primary antibodies were incubated overnight and secondary antibodies were incubated for 1 h. All incubations were performed at ambient temperature. Cultures were treated with a 1:2 dilution of Vectashield mounting medium (Vector Laboratories H-1200). The specific antibodies and dilutions used are shown in Table 1.

[0049] [Table 1-1]

[0050] [Table 1-2]

[0051] Cultures were generally stained within the TC culture vesicles in which they were differentiated, whereas aggregates or organoids were fixed and embedded in OCT (Sakura4583) followed by cryostat sectioning prior to staining.

[0052] Characterization of endocrine cells GSIS assays were performed by incubating cells in Krebs-Ringer buffer supplemented with 2 mM glucose for 30 min to determine basal C-peptide levels. Following this, the reaction buffer was exchanged for Krebs Ringer buffer supplemented with either 20 mM glucose or 30 mM KCl and incubated for an additional 30 min. Buffer samples were collected and C-peptide levels were quantified using ELISA (Mercodia 10-1141-01) according to the manufacturer's protocol. Microfluidic assays were performed as previously described (Adewola et al., 2010; Wang et al., 2012).

[0053] RNA Seq and key gene analysis RNA samples were sequenced by the University of Chicago Genomics Core. FASTQ files were aligned against reference hg19 using HISAT2 version 2.0.5 software. Gene counts of aligned reads were generated by HTSeq version 0.8.0 using RefSeq annotation files. Differential expression analysis was performed using edgeR version 3.18.1 with a cutoff of 4 counts per million. KeyGenes analysis was performed using the provided training set according to published protocols (Roost et al., 2015). Hierarchical clustering was performed using RStudio software with the R packages gplot and ggplot2 on the 87 KeyGenes classifier genes using only the part that contained all the hESC derivatives shown. Differential expression was determined using the EdgeR package and heat maps were generated using Microsoft Excel software.

[0054] Example 2 Generation of DFE cells using retinoid compounds This example describes the generation of dorsal foregut endoderm cells from pluripotent stem cells using retinoids with known biological activity. This example also describes how endoderm is patterned during gastrulation through the coordinated activity of TGFb agonism, TGFb inhibition, and retinoid agonism.

[0055] result Retinoic acid is produced in the body through a series of sequential enzymatic oxidation reactions that convert retinol to retinal and finally to retinoic acid. Although it is generally accepted that all-trans retinoic acid is the biologically relevant form of the molecule, retinoic acid exists in many isomeric forms due to its long conjugated backbone. Three common isoforms; 7-cis retinoic acid, 13-cis retinoic acid, and all-trans retinoic acid, have been observed in the circulation, and serum ratios of these different isomers have been shown to be consistent over time. Experiments using liver lysates suggest that the conversion between these different isomers is enzymatically regulated, although the enzymes involved in these conversions have not been clearly identified. Studies have shown that these isoforms have differential binding affinities for RAR and RXR. When retinoic acid is produced in or enters cells, it induces the production of Cyp26a, an enzyme that converts retinoic acid to 4-oxo-retinoic acid. Although it was initially believed that the production of 4-oxo-retinoic acids was inactive, several studies have demonstrated that these 4-oxo-retinoic acids have differential binding affinities for several retinoic acid targets, collectively suggesting that different retinoic acid isomers may have unique biological activities.

[0056] When retinoic acid is used in directed differentiation protocols, the chosen form used is all-trans retinoic acid, but this molecule is highly unstable and degrades rapidly over time or when exposed to light. Both time and light can create a mixture of different isoforms, resulting in epoxy and keto forms of retinoic acid. These epoxy and keto retinoic acid derivatives have been shown to have biological activity and affinity for the retinoic acid binding site. Taken all together, this creates several potential biologically active molecules with different differentiation potential when using retinoic acid in in vitro experiments. A DoE-based ISPA approach was used to determine which retinoic acid isoforms and derivatives can induce dorsal foregut endoderm.

[0057] BMP inhibition in the presence of 4-oxo and 9-cis retinoic acid induces DFE As shown in Figure 14A, a DoE design was created encompassing several known biologically active retinoids. When the ISPA-based optimization was performed (Figure 4B), it was predicted that several retinoids tested contributed to HNF1β induction. Both 13-cis retinoic acid (significant with FC=24.1609) and all-trans retinoic acid (less significant with FC=2.77281) indeed had a negative contribution to HNF1β induction, whereas 4-oxo-retinoic acid (FC=13.21), an enzymatically degraded form of retinoic acid, was predicted to have the highest factor contribution among all retinoids tested (Figure 4B). To validate ISPA, a transcript analysis of several key DFE genes was performed. A new optimizer consisting of the two retinoids predicted to have the highest contributing factors in the presence of LDN193189 was optimized to optimize the original HNF1β induction. Opt It was shown that this novel HNF1β could activate HNF1β to the same level as Opt The phenotype was further confirmed by the activation of several previously described DFE markers, including MNX1, SFRP5, OSR1, ONECUT1, MYT1, FOXA2 and FRZB (Figure 14C and D). By replacing 4-oxo-retinoic acid with the newly reconstituted retinoic acid, we were able to activate some of the DFE genes, but to significantly lower levels. Removal of LDN193189 from the differentiation medium formulation showed that most of the DFE gene markers were not activated, but several endoderm genes (OSR1, ONECUT1, MYT1, CDX2, data not shown) could be activated with retinoids, including the gastrulation gene NODAL (Figure 4C and D), suggesting that a distinct patterned endoderm was being induced.

[0058] All isomers of retinoic acid can induce NODAL signaling Most directed differentiation protocols generate endoderm using activin A as a Nodal mimic to induce in vitro gastrulation events, but during normal embryonic development, primitive streak formation is initiated with activin A, not Nodal. Previous studies have shown that Nodal and activin A signaling are not equivalent and that the definitive endoderm generated through Nodal signaling is distinct from that generated by activin A. Because our ISPA model for retinoid signaling predicted that retinoid acid can activate NODAL in the absence of BMP inhibition, we next assessed NODAL-induced endoderm formation against retinoic acid induction of NODAL. Responses using differentiation media supplemented with NODAL and CRIPTO (a binding partner of NODAL) in the presence of either BMP4, activin A or retinoic acid were compared to our ISPA-based NODAL induction model. Opt All reactions tested were able to induce the early and mid-gastrulation markers NODAL and ALDH1A2, respectively, with the lowest levels of NODAL. Opt It was noted that ventral induction of HHEX was observed in cultures exposed to retinoic acid (Figure 15B). Consistent with previous observations, the ventral marker HHEX was induced in all reactions lacking retinoic acid, however, the highest induction of the classical definitive endoderm marker SOX17 was observed in cultures exposed to retinoic acid-based NODAL. Opt This occurred in cultures exposed to retinoic acid (Figures B and C). All differentiation responses supplemented with retinoic acid showed the highest induction of DFE markers HNF1β, MNX1, SFRP5 and OSR1, although the levels obtained were orders of magnitude lower in the absence of BMP inhibition (Figure 15B).

[0059] Next, to better understand how the in vitro generated endoderm populations being studied are patterned, we performed a follow-up DoE-based gastrulation modeling experiment (Figure 16A). The ISPA optimizations for all individual endoderm genes measured were then subjected to cluster analysis using the Seurat R package (Figure 16B). This analysis split all optimizers into four distinct groups consisting of a single cluster containing DFE, patterned by retinoic acid and LDN193189 (Figure 16C and Figure 17). Another cluster was patterned by Nodal / Bmp4 and consisted of endoderm patterned primarily towards anterior and midgut fates, while the hindgut marker CDX2 was elevated within this cluster (Figure 16C and Figure 17). The cluster patterned by Nodal / Activin A consisted of endoderm patterned towards ventral and posterior fates (Figure 16C and Figure 17). Considering the factor contributions to the individual genes assayed throughout this experiment, it is noted that anterior and dorsal genes (except GSC) are generally not activated in the presence of activin A (Figure 17). It is also important to note that while some of these anterior or dorsal endoderm genes are predicted to be activated through Nodal / retinoic acid or Nodal / BMP synergistic effects, the following gene markers are optimally induced using retinoic acid in the absence of Nodal: DFE, HNF1b, MNX1, HNF6, OSR1, and SFRP5.

[0060] conclusion NODAL Opt HNF1β has strong contributions to all biologically relevant isomers of retinoic acid (9-cis, 13-cis, and all-trans), but also to the cyp26a1 inhibitor talarozole, ensuring that catalytic conversion to their respective 4-oxo forms does not occur (Figure 15A). This indicates that neither the 13-cis nor all-trans forms of retinoic acid are tolerated, but the 4-oxo metabolites and the 9-cis are required for HNF1β. OptIn contrast to the Nodal-dependent DFE phenotype (Figure S14B), retinoic acid can induce the DFE phenotype in a Nodal-dependent manner when isomers of retinoic acid are used, or in a Nodal-independent manner when 4-oxo-retinoic acid is used, suggesting that optimal induction of DFE occurs after retinoic acid is enzymatically degraded to 4-oxo-retinoic acid. Opt This means that DFE induction can occur via unwanted, transient NODAL induction, whereas retinoic acid induction of cyp26a1 abolishes any Nodal production. Taken together, these findings indicate that optimal DFE induction occurs in the absence of Nodal signaling.

[0061] Example 3 Validation of DFE optimizer conditions using immunohistochemistry.

[0062] To further validate the induction conditions for DFE markers, we performed immunostaining followed by high-content imaging to assess the conversion efficiency. By testing alternative agonists of the retinoic acid pathway, we found that utilizing a mixture of 4-oxo-retinoic acid and 9-cis retinoic acid enhanced the conversion efficiency of intact HNF1β. Opt The photostable RA agonist EC23 also behaves comparably to the RA and 4-oxo / 9-cis conditions.

[0063] Example 4 Assessment of key process parameters associated with endoderm induction and specification of multiple endoderm fates derived from pluripotency based on relative inputs from nodal, BMP, and retinoic acid.

[0064] A HD-DoE design incorporating Activin A (AA), BMP4, NODAL, retinoic acid and their respective inhibitors was used in pluripotent cultures to monitor key endoderm genes (Figure 16). The resulting data was imported into analysis software and clustered computational models predicting optimal culture conditions for activating all monitored genes (Figure 19B). Four distinct endoderm populations were predicted, all permissive for NODAL. The corresponding ridge plot (Figure 19A) demonstrates that retinoic acid signaling contributes to the induction of endoderm populations defined by Cluster 0 and Cluster 1 (red and green clusters, respectively). Cluster 0 is the dorsal foregut endoderm with previously characterized CPPs of retinoic acid and BMP inhibition, while Cluster 1 is the anterior endoderm population described below that is generated and patterned through the combinatorial effects of Nodal, BMP and retinoic acid. Only cluster 2 (blue cluster) was found to be dependent on AA, because the concentration of AA was highly variable in both clusters 1 and 3 (green and purple clusters), and the Alk4 / 5 / 7 inhibitor A8301, which specifically targets AA signaling, was present at high levels in clusters 0, 1, and 3. A8301 is essentially absent from the proposed AA-driven cluster 2 (blue cluster), which allows for uninhibited signaling. A similar pattern is observed with BMP4 signaling, with only cluster 1 (green cluster) having both high levels of BMP4 without providing a selective pathway inhibitor (in this case, the absence of the ALK2 / 3 inhibitor LDN193189) (Figure 19). The ALK1 / 2 / 6 inhibitor K02288 was tolerated within cluster 1, suggesting that BMP4 signaling occurs via the ALK3 receptor in this endodermal population. Nodal was shown to be present in all clusters with the two strongest contributions being cluster 2 (blue cluster) and cluster 1 (green cluster) in the proposed AA and BMP4-driven clusters, respectively.This suggests that Nodal can signal through either the Smad2 / 3-mediated AA pathway classically attributed to Nodal signaling, and may play a role in the Smad1 / 5 / 8 BMP-driven signaling transduction pathway that Nodal has more recently been shown to activate. Also, given that both clusters 0 and 3 were shown to have significant contributions from the effector AA, BMP4 and Nodal, and their respective inhibitors, this implies that endoderm patterning may have a strong contribution from non-canonical TGF-β signaling.

[0065] Example 5 Identification of anterior domain endoderm populations by analysis of region-specific genes.

[0066] We sought to further characterize the phenotypes of the specific clusters identified in Figures 16 and 19B. Figure 20 reveals the location of specific genes within the four major clusters. First, we show that cluster 0 represents the previously identified DFE. As can be seen in Figures 20B and 21, the ventral marker HHEX is largely absent, and the dorsal marker MNX1 is exclusively expressed in cluster 0. Both SFRP5 and HNF1β have broader expression beyond cluster 0 and are also present in cluster 1. This is because SFRP5 and HNF1β are not restricted to the dorsal part of the gut, but are expressed throughout the anterior region and throughout the midgut. Cluster 1 represents an endoderm population distinct from DFE and standard literature-derived DE (represented by cluster 2). Due to the specific marker expression, we determined that cluster 1 represents an endoderm population representative of the anterior / midgut region of the gut due to the expression patterns of the anterior genes TBX1, PAX9, and OSR1 (Figures 20C and 21). Here, we refer to this population as the anterior domain endoderm (ADE). During development, both TBX1 and PAX9 are expressed in the anterior pharyngeal region, and OSR1 is expressed throughout the pharynx and in the midgut (Figures 20C and 21).

[0067] The key process inputs of ADE are examined by ridge plot analysis (Figure 19A). Induction of ADE is most dependent on providing inhibition of Nodal, BMP and ALK5 in the absence of BMP inhibition. It is resistant to the input of RA signaling. This set of inputs reveals distinct BMP-dependent features not present in other types of endoderm. Ridge plots show distinct contributions of retinoic acid and Nodal for induction of ADE (Figure 19A), but RA, which produces the enzyme ALDH1A2, is not expressed and Nodal has altered expression throughout cluster 1. The absence of ALDH1A2 is consistent with this population representing the more anterior portion, since retinoic acid production begins more posteriorly in midgastrulation. Thus, any involvement in patterning of the anterior endoderm can be achieved through the diffusion of retinoids within the embryo. The lack of Nodal expression could be due to the fact that the anterior-most region of the primitive streak is thought to be the first to exit gastrulation, implying that this population has already exited the primitive streak phenotype but retains responsiveness to Nodal.In conclusion, we define a novel endoderm type, the induction of which types underlie its development.

[0068] Example 6 Comparison of key process parameters associated with specific endoderm expressed genes To better understand the importance of Nodal and BMP4 signaling in generating anterior / midgut endoderm populations, we compiled all predictive models for optimization of TBX1, PAX9, and OSR1 (Figure 22). As shown through the TBX1 and PAX9 optimizers, both Nodal and BMP4 showed equal average cumulative contributions to the activation of anterior endoderm. On the other hand, the more widely expressed OSR1 (anterior to the midgut) showed a much stronger response to Nodal signaling, although BMP4 still had a significant contribution. This means that the more distally gut patterning occurs, the less BMP signaling contribution there is. It is important to note that AA made little or no contribution to inducing TBX1, PAX9, or OSR1, as all previous efforts to generate pharyngeal or gastric progeny relied on the early use of AA (Figure 22). This time, it was noted that AA had increased effector contributions in selected genes representing midgut, posterior and ventral endoderm fates as shown through selected exemplary optimizers for APOB, CDX2 and SOX17, respectively (Figure 22), indicating that the AA input commonly encountered in the literature plays a role in generating the more posterior endoderm compartment.

[0069] Next, we performed ISPA on all major endoderm genes measured that could define distinct regions along the gut tube to directly define which genes responded to each TGFβ agonist. We then normalized the respective effector factor contributions and ranked the genes from most to least responsive for AA, Nodal, BMP4 and RA (Figure 23). Our previous findings that RA plays a central role in generating DFE were reaffirmed by showing that HNF6, HNF1β, SFRP5 and MNX1 are all highly regulated by RA and inhibited by AA (Figure 23). Consistent with the conditions for generating ADE, Nodal, BMP4 and RA, all have been shown to be involved in the activation of TBX1, PAX9 and OSR1, with highly negative contributions from AA, normalized factor contributions of -0.59, -0.85 and -1.0, respectively. Consistent with the current hypothesis that AA and Nodal have overlapping functions, several genes with overlapping input logic were observed between the two effectors. These included the primitive streak genes GSC and NODAL (NODAL is not shown in Figure 23 but is shown in Figure 21), the midgut genes NR5A2, CDX2, and PROX1, and the ventrally localized genes BMP4 and SOX17.

[0070] Since both BMP4 and NODAL have been shown to be activated by AA, we next used RNA-Seq data to assess which TGFβ family ligands were activated in both classical DE and novel DFE protocols. Incubation of pluripotent cultures with AA was shown to induce endogenous expression of several TGFβ family members, including TDGF1, LEFTY1 / 2, NODAL, BMP1 / 2 / 4 / 7, and TGFβ1, possibly explaining previous findings due to a broader effect for AA endoderm induction (Figure 24A). This induction of high levels of Nodal by generation of DE was confirmed by IHC-based protein detection (Figure 24B). On the other hand, RA / LDN treatment of pluripotent cells resulted in selective activation of BMP1 / 2 / 4 / 7, GDF11, AMH, and TGFβ2 (Figure 24). This retinoic acid-mediated BMP induction explains the importance of LDN in dorsalization of the DFE protocol. Overall, we demonstrate that NODAL and BMP pattern endoderm along the anterior to midgut tube and AA responsive genes are more representative of the primitive streak, ventral midgut and more posterior regions. Optimizers for selective activation of several Nodal / BMP / RA responsive genes that do not respond to AA are shown as examples in Figure 25.

[0071] References: ADEWOLA,et la.,Biomed Microdevices,2010,12,409-17. AFELIK et al.,Development,2012,139,1744-53. ANGELO et al.,2012,PLoS One,7,e40707. BODY-BECHOU,et al.,2014,Prenat Diagn,34,90-3. BORT,et al.,2004,Development,131,797-806. CARINHAS et al.,2012,Trends Biotechnol,30,377-85. CHAKRABARTY et al.,2013,Wiley Interdiscip Rev Syst Biol Med,5,181-203。 CHEN,et al.,2013,Development,140,675-86。CHUNG,et al.,2008,Dev Cell,15,738-48。 D'AMOUR, et al., 2005, Nat Biotechnol, 23, 1534-41. DAVENPORT et al.,2016,Stem Cells,34,2635-2647。 DE VAS, et al., 2015, Development, 142, 871-82. DEUTSCH,et al.,2001,Development,128,871-81。 GADUE,et al.,2006,Proc Natl Acad Sci US A,103,16806-11. GERIN,et al.,2014,BMC Syst Biol,8,96。 GREEN,et al.,2011,Nat Biotechnol,29,267-72。 HAUMAITRE, et al., 2005, Proc Natl Acad Sci USA, 102, 1490-5. HEBROK,et al.,1998,Genes Dev,12,1705-13。HRVATIN,et al.,2014,Proc Natl Acad Sci U SA,111,3038-43. JENNINGS,et al.,2017,Stem Cell Reports,9,1387-1394。 JENSEN,et al.,2000,Diabetes,49,163-76。 KITANO,H.2002.Computational systems biology.Nature,420,206-10。 KROON,et al.,2008,Nat Biotechnol,26,443-52。 KUMAR,et al.,2003,Dev Biol,259,109-22。 KUMAR,et al.,2014,Biotechnol Prog,30,86-99。 KWON,et al.,1999,Dev Cell,15,509-20。 LI,et al.,1999,Nat Genet,23,67-70。 LOH,et al.,2014,Cell Stem Cell,14,237-52。 MARTIN,et al.,2005,Dev Biol,284,399-411。MATSUNO,et al.,2016,Differentiation,92,281-290。 MATSUURA,et al.,2009,Mech Dev,126,817-27。 MCCRACKEN,et al.,2014,Nature,516,400-4。 MCDONALD,et al.,2014,Nat Cell Biol,16,1128-9。 MENDES,et al.,2016,Tissue Eng Part C Methods,22,473-86。 MERCIER,et al.,2013,J Biotechnol,167,262-70。 MFOPOU,et al.,Gastroenterology,138,2233-45,2245 e1-14。 MOLOTKOV,et al.,2005,Dev Dyn,232,950-7。 PAGLIUCA,et al.,2014,Cell,159,428-39。 POLL,et al.,2006,Diabetes,55,61-9。 RATHORE,et al.,2014,Biotechnol Prog,30,967-73。 REZANIA,et al.,2014,Nat Biotechnol,32,1121-33. ROOST,et al.,2015,Stem Cell Reports,4,1112-24. ROTTINGER,et al.,2015,Ptychodera flava.Biol Open,4,830-42. SADOWSKI et al.,2016,Trends Biotechnol,34,214-227. SAMPAZIOTIS,et al.,2015,Nat Biotechnol,33,845-852. SPENCE,et al.,2009,Dev Cell,17,62-74. SPENCE,et al.,2011,.Nature,470,105-9. SWAIN,et al.,2018,Curr Drug Discov Technol. TREMBLAY, et al., 2005, Dev Biol, 280, 87-99. ULVEN,et al.,2000,Dev Biol,220,379-91. WANG,et al.,2012,Biomed Microdevices,14,419-26. XIE,et al.,2013,Cell Stem Cell,12,224-37.

[0072] All publications and patents mentioned herein and / or listed below are incorporated herein by reference. Various modifications and variations of the described methods and systems of the invention will be apparent to those skilled in the art without departing from the scope and spirit of the invention. Although the invention has been described in connection with specific embodiments, it should be understood that the invention as claimed should not be unduly limited to such specific embodiments. Indeed, various modifications of the described modes for carrying out the invention that are obvious to those skilled in the relevant fields are intended to be within the scope of what is described herein.

Claims

1. 1. A method for producing endoderm cells, comprising: the endoderm cell is an anterior domain endoderm cell; a) contacting a population of pluripotent stem cells with i) a Nodal protein and a BMP protein or a BMP pathway agonist, or a vector(s) encoding said Nodal protein and said BMP protein, and ii) an ALK5 inhibitor; b) culturing at least a portion of the population of pluripotent stem cells to generate a population of anterior domain endoderm (ADE) cells; The method, wherein the pluripotent stem cells are not exposed to i) a BMP inhibitor, and / or ii) a transforming growth factor beta (TGFβ) pathway agonist during the culturing or during the contacting.

2. A composition comprising: a) a Nodal protein or a vector encoding said Nodal protein; b) a BMP protein or a BMP pathway agonist, or, if a protein, a vector encoding said BMP protein or BMP agonist; and c) ALK5 inhibitors.

3. A kit or system comprising, consisting essentially of, or consisting of: a) a population of cells present in a cell culture vessel, optionally the cells being pluripotent stem cells or anterior domain endoderm (ADE) cells; b) a Nodal protein or a vector encoding said Nodal protein present in a first container; c) a BMP protein or BMP pathway agonist present in a second container, or, if a protein, a vector encoding said BMP protein or BMP agonist; and d) an ALK5 inhibitor present in a third container.

4. The composition according to claim 2, which satisfies the following a) or b): a) the composition does not include a BMP inhibitor, and / or ii) a transforming growth factor beta (TGFβ) pathway agonist; b) further comprising a population of cells, said cells being pluripotent stem cells or anterior domain endoderm (ADE) cells.

5. 4. The kit or system of claim 3, the cell culture vessel does not contain any or any detectable exogenous BMP inhibitors, and / or exogenous transforming growth factor beta (TGFβ); and The kit or system does not contain any detectable exogenous BMP inhibitor and / or exogenous transforming growth factor beta (TGFβ) in the first container, the second container, the third container, or the fourth container.

6. The method according to claim 1, which satisfies the following α) or β): α) the ALK5 inhibitor is selected from the group consisting of A7701, A-83-01, SB505124, SB431542, alantolactone, and LY2157299; β) The BMP protein is a BMP4 protein.

7. The composition according to claim 2, which satisfies the following α) or β): α) the ALK5 inhibitor is selected from the group consisting of A7701, A-83-01, SB505124, SB431542, alantolactone, and LY2157299; β) The BMP protein is a BMP4 protein.

8. The kit or system according to claim 3, which satisfies the following α) or β): α) the ALK5 inhibitor is selected from the group consisting of A7701, A-83-01, SB505124, SB431542, alantolactone, and LY2157299; β) The BMP protein is a BMP4 protein.

Images