Development-inspired approach for the derivation of functional and transplantable salivary gland organoids from human pluripotent stem cells Authors: Laura J. Sherwood, Ronel Z. Samuel, Sai Harsha Bhamidipati, Kihoon Nam, Frank Maslow, Travis Small, Yali Zhang, Jianmin Wang, Song Liu, Olga J. Baker, Stelios T. Andreadis Published: 2026 ABSTRACT Salivary gland (SG) hypofunction significantly affects the life quality of patients with hyposalivation. Current treatment options for these conditions are limited and often result in undesirable side effects. To address this issue, we’ve developed a developmentally inspired strategy to derive salivary gland epithelial progenitor (SGEP) cells from human pluripotent stem cells (PSC). Then we employ SGEP cells derived from human PSC with a SOX10::GFP reporter to engineer three-dimensional lumenized SG organoids, with well-defined lumens and secretory function, reminiscent of human salivary glands. Upon transplantation into the submandibular gland of athymic nude mice, mCherry+ salivary gland organoids (SGO) integrate into the gland and differentiated into salivary gland phenotypes, including acinar, ductal and myoepithelial structures. Our findings suggest that SG organoids may hold promise for studying SG development, disease progression and drug screening, as well as for development of cell therapies for salivary gland regeneration. Introduction Salivary gland (SG) hypofunction has a profound impact on the quality of life of affected individuals, leading to difficulties in speaking, swallowing, and eating1. The available treatment options for these disorders are limited, providing only temporary relief, and are often accompanied by undesirable side effects1,2. As a result, there is a pressing need for novel therapeutic approaches to promote salivary gland regeneration. Current therapies include the use of human salivary gland epithelial progenitor (SGEP) cells3–5 from minced glandular tissue or mesenchymal stem cells6–9 as a source of trophic factors to promote glandular regeneration. However, these approaches have limitations including scalability, standardization, reproducibility, and cell/tissue delivery strategies that may result in poor engraftment. A potential alternative cell source for glandular regeneration are salivary gland epithelial progenitor (SGEP) cells derived from human induced pluripotent stem cells (hiPSC). hiPSC possess the remarkable ability to differentiate into all human somatic cell types, including salivary gland cells and can be readily derived from easily accessible adult somatic cells, such as skin or blood cells, making them an attractive cell source for therapeutic applications. Recent studies have shed light on the use of iPSC-derived SGEP for gland regeneration10,11. However, the current methods used for differentiation rely on forced gene overexpression via viral transduction to induce maturation10; or technically challenging and inefficient techniques, such as microdissection of micrometre-sized epithelial buds from organoids10,11, hindering the widespread adoption of these protocols. In addition, while differentiation strategies that rely on embryoid bodies are relatively simple and scaffold-free, they often produce heterogeneous and disorganized structures containing multiple cell types, thereby complicating the isolation of specific cell populations or multicellular assemblies12. Moreover, the use of external forces, such as centrifugation, to aggregate cells can promote artificial cell–cell interactions that may not accurately reflect physiological conditions13. While these studies provided important proof-of-concept that SGEPs can be derived from hiPSCs, the methods are difficult to reproduce, underscoring the need for robust and scalable strategies to generate SGEPs for both basic research and the development of cell-based therapies. In this study, we present a step-by-step strategy that mimics SG development to generate multipotent SGEP cells that can self-assemble into lumenized acinar or ductal organoids, without the need for genetic modification or challenging microsurgical techniques for separating well-differentiated parts of microtissues. We utilize a scaffold in the form of a hydrogel for the growth of 3D organoids from the SGEP cells, directing differentiation toward the salivary gland lineage to obtain a more homogenous population. Furthermore, we demonstrate that hiPSC-SGEP-derived salivary gland organoids are functional in vitro and that they can be successfully engrafted into the submandibular gland of athymic nude mice. Notably, the transplanted organoids exhibited tri-potent differentiation potential, giving rise to acinar, ductal, and myoepithelial structures, providing compelling evidence for the potential of hiPSC-derived SGO as platforms enabling drug discovery and screening efforts or development of cell therapies for salivary gland regeneration. Results Establishment of the surface ectoderm lineage We hypothesize that salivary gland epithelial progenitor (SGEP) cells can be derived from human pluripotent stem cells by precise temporal control of pathway activation or inhibition that mimics embryonic salivary gland development. To address this hypothesis, we used two human pluripotent stem cell (PSC) lines, ATCC-BXS0114, an iPSC line sourced from bone marrow CD34+ cells and WA09 (H9), an ESC line modified with a SOX10::GFP BAC reporter14 for live monitoring of Sox10 expression and a constitutive promoter, EF1α driving expression of the fluorescent protein, mCherry. Principal component analysis of RNA-seq data revealed that the two cell lines clustered closely at every time point, indicating comparable transcriptional dynamics throughout the 46-day differentiation process (Fig. 1a).Fig. 1Derivation of surface ectoderm from pluripotent stem cells.a PCA analysis of two human iPSC cells lines, BXS0114 (B) and WA09 (H9), at the indicated times of differentiation towards the salivary gland lineage. b Heat map from differential gene expression analysis of RNA-seq data showing downregulation of genes associated with the embryonic, neuroectoderm, neural crest, neural and hematopoietic lineages boxed in blue. c Heat map of differentially expressed genes associated with neural crest, and upregulation of early and mature surface ectoderm and surface ectoderm derived lineages from D0 to D10 boxed in red. d–f RT-PCR for the indicated genes; gene expression levels were normalized to PSC D0. d Embryonic and ectodermal genes: NANOG, TUBB3 and CDH1; (e) Neural/neural crest genes: ASCL1, NGFR, FOXD3; and (f) Key surface ectodermal genes: MSX1, MSX2, TP63. g, h Western blots and quantifications for (g) the pluripotency factor, NANOG; (h) MSX1 and MSX2 over time. GAPDH served as housekeeping loading control. ** Denotes p < 0.01, *** denotes p < 0.001, and **** denotes p < 0.0001. (n = 3). i–k Immunostaining of D10 SE cells for (i) β3-Tubulin; (j) SOX2, (k) E-cadherin (green). All cells were counterstained for pan-Cytokeratin (red) and DAPI (blue, nuclei). * Denotes p < 0.05, n = 4, scale bar = 50 µm, magnification = 40x. All data is shown as mean +/− SD. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test for multiple comparisons. Statistical significance between two groups were calculated using student’s t-test. Cell lines used in the study are represented as: B for BXS0114 and H for WA09 (H9), while the numbers following them represent the timepoints in days. Source data are provided as a Source Data file. Derivation of surface ectoderm from pluripotent stem cells. a PCA analysis of two human iPSC cells lines, BXS0114 (B) and WA09 (H9), at the indicated times of differentiation towards the salivary gland lineage. b Heat map from differential gene expression analysis of RNA-seq data showing downregulation of genes associated with the embryonic, neuroectoderm, neural crest, neural and hematopoietic lineages boxed in blue. c Heat map of differentially expressed genes associated with neural crest, and upregulation of early and mature surface ectoderm and surface ectoderm derived lineages from D0 to D10 boxed in red. d–f RT-PCR for the indicated genes; gene expression levels were normalized to PSC D0. d Embryonic and ectodermal genes: NANOG, TUBB3 and CDH1; (e) Neural/neural crest genes: ASCL1, NGFR, FOXD3; and (f) Key surface ectodermal genes: MSX1, MSX2, TP63. g, h Western blots and quantifications for (g) the pluripotency factor, NANOG; (h) MSX1 and MSX2 over time. GAPDH served as housekeeping loading control. ** Denotes p < 0.01, *** denotes p < 0.001, and **** denotes p < 0.0001. (n = 3). i–k Immunostaining of D10 SE cells for (i) β3-Tubulin; (j) SOX2, (k) E-cadherin (green). All cells were counterstained for pan-Cytokeratin (red) and DAPI (blue, nuclei). * Denotes p < 0.05, n = 4, scale bar = 50 µm, magnification = 40x. All data is shown as mean +/− SD. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test for multiple comparisons. Statistical significance between two groups were calculated using student’s t-test. Cell lines used in the study are represented as: B for BXS0114 and H for WA09 (H9), while the numbers following them represent the timepoints in days. Source data are provided as a Source Data file. First, we treated the hPSC with an optimized differentiation factor cocktail comprising BMP-410,15, SB43154210,16 (a potent inhibitor of TGF-β signaling) and DAPT17,18 (an inhibitor of NOTCH signaling) (Fig. S1a–i) for three days to guide the differentiation towards a naïve surface ectoderm (SE) fate10,16. RNA-seq showed downregulation of mRNA levels of key pluripotency factors such as NANOG, MYC and OCT3/4 (POU5F1) (Fig. 1b, Fig. S2a, Fig. S3a)19–22 along with upregulation of early SE markers16,18,23–29 (Fig. 1c). RT-PCR and western blot (WB) verified downregulation of NANOG (Fig. 1d, g) at the mRNA and proteins levels, indicating departure from pluripotency. The preservation of E-Cadherin (CDH1) (Fig. 1d) and downregulation of classical epithelial-mesenchymal transition (EMT) markers such as CDH2 and SNAI1 (Fig. 1b) at this stage indicated maintenance of the epithelial phenotype16,30,31. Furthermore, we noted that the primary neuronal regulators ASCL1 and OTX2 were expressed at low levels, while key neural crest genes, FOXD3 and NGFR were downregulated significantly (Fig. 1b, e), indicating a strong SE phenotype32–35. A week after BMP-4 withdrawal (Fig. S1j–u), the cells matured into a multipotent SE as seen by transcriptional upregulation of key SE markers like MSX1/2 and TP63 (Fig. 1f, h)16–18,36. Overall, we observed a transition in gene expression from early-stage naïve (e.g. KRT7/8, ISL1 & TBX3)16,18,23–29 to late-stage mature (e.g. COL5A1, COL12A1, COL17A1, TFAP2A/B & KRT19) SE markers15,29,36–38 (Fig. 1c, Fig. S2b, Fig. S3b). Furthermore, genes associated with SE derivatives such oral epithelium (OE) (SLC4A4)39, presumptive lens ectoderm and cornea (MEIS1/2)40 and limbal development (IGFBP5)41 were expressed or even upregulated (Fig. 1c), indicating that these cells may have the potential to differentiate to more than one lineages. β3-tubulin (TUBB3), an early ectoderm marker42 that later recedes to the neuronal ectoderm and its derivatives, was shown to be downregulated by day 10 (Fig. 1b). At the protein level, a small fraction of these nascent cells expressed low levels of β3-tubulin (Fig. 1i & Fig. S1s), but high levels of SOX2, E-cadherin and Pan-cytokeratin (Fig. 1j, k & Fig. S1t, u), indicating that these cells maintained an epithelial phenotype. Therefore, at this stage we established a robust multipotent SE phenotype with the potential to differentiate into its daughter lineages. Specification to oral epithelium and SG epithelial progenitors The salivary gland originates from the placode of the oral epithelium where Activin/BMP/SMAD activity is suppressed43. This prompted us to treat the SE cells with an optimized differentiation factor cocktail (Fig. S4a–j) comprising of LDN19318910 and FGF-210,44 to inhibit the BMP-SMAD1/5/8 pathway and activate the FGFR pathway, respectively to coax the SE into the oral epithelium (OE). After three days of treatment, we observed upregulation of PITX1/2 proteins (Fig. 2a) as well the mRNA levels of SOX245, PITX245 and other genes associated with proper orofacial development37,39,45–51 (Fig. 2b, l, Fig. S3c). Concomitantly, mRNA and protein levels of NANOG and β3-tubulin were further reduced (Fig. 2c, f, i), while the mRNA levels of ASCL1, NGFR and FOXD3 remained low (Fig. 2d), indicating departure from the neural and neural crest lineages. Furthermore, there was downregulation in mRNA levels of MSX1/2 and TP63 (Fig. 2e), indicating a departure from surface ectoderm. These newly formed cells continued to maintain an epithelial phenotype, as evidenced by further reduced levels of β3-tubulin and high expression of SOX2, E-cadherin and Pan-cytokeratin (Fig. 2f–k, Fig. S4k–m). Genes associated with the dental lamina39,46,49,52 (SOX11, FOS, FST), and the stratum intermedium39,48 (NOTCH1), which are OE derivatives continued to be expressed (FST) or upregulated (SOX11, FOS, NOTCH1) on day 13 (Fig. 2l, Fig. S2c, Fig. S3c). However, only those specifically associated with the salivary gland lineage39 (SOX9, FOXC1, BARX2, COL14A1) were further upregulated by day 17, indicating salivary gland specification (Fig. 2l, Fig. S2d, Fig. S3d).Fig. 2Differentiation of surface ectoderm into oral epithelium.a Western blots and quantifications for PITX1, PITX2 from D10 to D17. GAPDH served as housekeeping gene control ** Denotes p < 0.01, *** denotes p < 0.001, and **** denotes p < 0.0001. (n = 3). b–e RT-PCR for the indicated genes from D3 to D17. Gene expression levels were normalized to PSC D0; (b) Oral ectodermal genes, SOX2, PITX2; (c) NANOG, β3-tubulin; (d) Neural/neural crest genes ASCL1, NGFR, FOXD3; and (e) Surface ectodermal genes MSX1/2 and TP63. f, k Immunostaining and quantification of D13 oral ectoderm (OE) cells for (f, i) β3-Tubulin; (g, j) SOX2, (h, k) E-cadherin (green). All cells were counterstained for pan-Cytokeratin (red) and DAPI (blue, nuclei). l Heat map of differentially expressed genes associated with orofacial development and various lineages associated with the oral ectoderm. *Denotes p < 0.05, n = 4, scale bar = 50 µm. All data is shown as mean +/- SD. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test for multiple comparisons. Statistical significance between two groups were calculated using student’s t-test. The two PSC lines that were used in this study are depicted as: B for BXS0114 and H for WA09 (H9) with numbers following the letters representing the timepoints in days. Source data are provided as a Source Data file. Differentiation of surface ectoderm into oral epithelium. a Western blots and quantifications for PITX1, PITX2 from D10 to D17. GAPDH served as housekeeping gene control ** Denotes p < 0.01, *** denotes p < 0.001, and **** denotes p < 0.0001. (n = 3). b–e RT-PCR for the indicated genes from D3 to D17. Gene expression levels were normalized to PSC D0; (b) Oral ectodermal genes, SOX2, PITX2; (c) NANOG, β3-tubulin; (d) Neural/neural crest genes ASCL1, NGFR, FOXD3; and (e) Surface ectodermal genes MSX1/2 and TP63. f, k Immunostaining and quantification of D13 oral ectoderm (OE) cells for (f, i) β3-Tubulin; (g, j) SOX2, (h, k) E-cadherin (green). All cells were counterstained for pan-Cytokeratin (red) and DAPI (blue, nuclei). l Heat map of differentially expressed genes associated with orofacial development and various lineages associated with the oral ectoderm. *Denotes p < 0.05, n = 4, scale bar = 50 µm. All data is shown as mean +/- SD. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test for multiple comparisons. Statistical significance between two groups were calculated using student’s t-test. The two PSC lines that were used in this study are depicted as: B for BXS0114 and H for WA09 (H9) with numbers following the letters representing the timepoints in days. Source data are provided as a Source Data file. Activation of the FGFR2(IIIb) and RARα pathways was shown to increase expression of SOX953,54, the primary regulator of salivary gland (SG) progenitors39,54. In addition, the developing gland exhibited active Wnt signaling solely in the mesenchyme, while inhibition of the Wnt pathway was necessary to generate the epithelial progenitors55. These findings prompted us to coax differentiation of OE cells into salivary gland epithelial progenitors (SGEP) with a combination of differentiation factors (Fig. S5a–j), including FGF-1010,54, the retinoic acid receptor alpha agonist Am58053,56 and Wnt inhibitor Wnt-C5955. The transition from the OE was characterized by decreased mRNA levels of SOX2 and PITX1/2 (Fig. 3a), and increased the expression of key salivary gland initiation genes such as SOX953,54,56, FOXC110, TP63, and KRT5/1415,18 (Fig. 3b). Moreover, this SOX9+ population displays two distinct lineage markers, indicative of proximal and distal progenitors, mirroring observations made during salivary gland development (Fig. 3c)57. Specifically, a gene marking the distal tip acinar progenitors, SOX1058 (Fig. 3d), as well as genes specifying ductal progenitors, like ASCL359 and KRT860 (Fig. 3e) were significantly upregulated, indicating generation of a mixed population of multipotent progenitor cells. Flow cytometry revealed that 98.22% of SGEP were positive for SOX9 (Fig. 3f) when gated to the isotype control. Immunofluorescence analysis showed nuclear localized SOX9, membrane bound E-cadherin and Pan-cytokeratin (Fig. 3g, Fig. S5k, l) in these cells. Some SGEP initiation genes like SOX9, FOXC1, COL14A139 were upregulated on day 17 and some continued to increase up to day 25 (IRF661) or up to day 32 (FOXC1); while others increased from day 17 to day 25 e.g., ETV4/562,63. In contrast, genes related to the dental lineage derivatives39 (e.g., SOX11, GAD1, LEF1, PRSS23, EDNRB, FOS & FST) were downregulated after day 13 or day 17, with major downregulation occurring after day 17 (Fig. 3h, Fig. S6). Therefore, through these sequential inductions with differentiating factors over time, we successfully established a robust SOX9 + SGEP phenotype via a transitional OE stage.Fig. 3Derivation of salivary gland epithelial progenitors from oral epithelium.a, b RT-PCR of the indicated genes; the data was normalized to gene expression levels of D0 PSC. a Oral Ectodermal genes, SOX2, PITX2; (b) Salivary gland progenitor cell initiation genes SOX9, FOXC1, TP63, KRT5, KRT14; (c) Schematic showing the developmental process of bud formation and the specification of the SOX9 positive proximal and the SOX10 positive distal progenitor populations. Created in BioRender.Andreadis, S. (2026) https://BioRender.com/c2v29xx (d, e) RT-PCR of the indicated genes; (d) Early acinar progenitor gene, SOX10; (e) Early ductal progenitor genes ASCL3, KRT8. f Flow cytometry of the D17 SGEP for SOX9 population; the data were gated to the isotype control. g Immunostaining of D17 SGEP cells for SOX9 and E-cadherin (green); the cells were counterstained for pan-Cytokeratin (red) and DAPI (blue, nuclei). h Heat map with differentially expressed genes associated with developing dental lamina and its derivatives and the developing salivary gland. Analysis of single cell RNA sequencing data of D17 SGEP cells: (i) UMAP plot; (j) Donut plot corresponding to the UMAP depicting the large bipotent population; (k) Ridge plots of different gene categories, and (l) Violin plots of the indicated genes. * Denotes p < 0.05, n = 4, scale bar = 50 µm. All data is shown as mean +/- SD. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test for multiple comparisons. Statistical significance between two groups were calculated using student’s t-test. B for BXS0114 and H for WA09 (H9) with numbers following the letters representing the timepoints in days. Source data are provided as a Source Data file. Derivation of salivary gland epithelial progenitors from oral epithelium. a, b RT-PCR of the indicated genes; the data was normalized to gene expression levels of D0 PSC. a Oral Ectodermal genes, SOX2, PITX2; (b) Salivary gland progenitor cell initiation genes SOX9, FOXC1, TP63, KRT5, KRT14; (c) Schematic showing the developmental process of bud formation and the specification of the SOX9 positive proximal and the SOX10 positive distal progenitor populations. Created in BioRender.Andreadis, S. (2026) https://BioRender.com/c2v29xx (d, e) RT-PCR of the indicated genes; (d) Early acinar progenitor gene, SOX10; (e) Early ductal progenitor genes ASCL3, KRT8. f Flow cytometry of the D17 SGEP for SOX9 population; the data were gated to the isotype control. g Immunostaining of D17 SGEP cells for SOX9 and E-cadherin (green); the cells were counterstained for pan-Cytokeratin (red) and DAPI (blue, nuclei). h Heat map with differentially expressed genes associated with developing dental lamina and its derivatives and the developing salivary gland. Analysis of single cell RNA sequencing data of D17 SGEP cells: (i) UMAP plot; (j) Donut plot corresponding to the UMAP depicting the large bipotent population; (k) Ridge plots of different gene categories, and (l) Violin plots of the indicated genes. * Denotes p < 0.05, n = 4, scale bar = 50 µm. All data is shown as mean +/- SD. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test for multiple comparisons. Statistical significance between two groups were calculated using student’s t-test. B for BXS0114 and H for WA09 (H9) with numbers following the letters representing the timepoints in days. Source data are provided as a Source Data file. Furthermore, single-cell RNA sequencing (scRNA-seq) revealed that the majority of SGEP cells were bipotent progenitors expressing both acinar and ductal genes. Smaller subsets of cells were identified as exclusively acinar or ductal, or exhibited low expression of either lineage (Fig. 3i, j). Overall, SGEP cells showed higher expression of ductal genes and lower expression of acinar genes. Consistent with bulk RNA-seq analysis, genes characteristic of oral ectoderm, surface epithelium, and myoepithelium were not detected at this stage (Fig. 3k, l). Generation of lumenized and SOX10+ organoids To form SG organoids (SGO), day 17 SGEP were embedded in Cultrex (an ECM rich in laminin, collagen IV, and nidogens, basement membrane components that play critical roles in salivary gland development and promote epithelial cell polarization64) in the presence of an optimized combination of differentiation factors (Fig. S7a–d), including FGF1065,66, EGF67,68 and dexamethasone69, for 8 days (day 17 to day 25) as these factors are well known to aid in organoid development (Fig. S7a–d). This was followed by a week-long treatment (day 25 to day 32) with another optimized mixture of differentiation factors (Fig. S7e–g) containing HGF, which is known to induce amylase promoter activity promoting acinar cell fate69; along with EGF67 and dexamethasone to induce SOX10 expression within the organoids (Fig. 4a). By day 25, the SGEP embedded in Cultrex formed organoids with well-defined lumens (Supplementary Movie 1), a phenotype that was maintained through day 32 (Fig. 4b & Suppl. Movie 2). At this stage, the organoids displayed significant upregulation of SOX10 as shown by RT-PCR (Fig. 4c, Fig. S7) and SOX10 promoter (pSOX10) activity as shown by pSOX10-driven GFP expression (Fig. 4d, Suppl. Movie 3). Flow cytometry revealed an increase in the percentage of cells that express the pSOX10-driven GFP as they differentiated from day 17 to day 25 and 32 (Fig. 4e). Salivary gland initiation genes that were upregulated on day 17, such as SOX9, HEY170, LAMA171, and COL14A1 were downregulated by day 25 and 32 (Fig. 4f, Fig. S2e). At the same time, two distinct gene sets associated with later stages of glandular development peaked on day 25 (IRF6, ITGA671, LAMA371 & MYC10) and day 32 (ETV4/5, ITGA371, LAMA571, MUC1/4/1672 & FOXC1, KRT17, KRT19, ST14) (Fig. 4f, g, Fig. S3e, f). These results demonstrate the formation of lumenized SOX10+ organoids expressing many key SG progenitor markers.Fig. 4Formation of SG organoids from SGEP cells.a Schematic showing the embedding of the D17 cells into Cultrex marking the transition into a 3D differentiation model, where the cells aggregate to form luminal organoids by D25 and later transition into a SOX10 expressing luminal acinar progenitor by D32. Created in BioRender. Andreadis, S. (2026) https://BioRender.com/7pu5lxv (b) 3D rendering of D32 organoids from a confocal image shows the organoids were completely lumenized. Hollow lumens shown by arrows as seen in the bottom sliced rendering of the organoids. Comparative gene expression levels normalized to PSC Day 0 showed: (c) RT-PCR for the early acinar progenitor gene, SOX10 shows increased expression from D17 to D32. The data was normalized to gene expression levels of D0 PSC * Denotes p < 0.05, n = 4. All data is shown as mean +/- SD. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test for multiple comparisons. Statistical significance between two groups were calculated using student’s t-test. d Fluorescence microscopy of organoids made from H9 pSOX10::GFP reporter cells that express GFP when the SOX10 promoter locus is activated. Scale bar = 50 µm. e Flow cytometry across shows increased percentage of cells that express pSOX10::GFP as the differentiation progresses from D17 to D32. f, g Heat maps of differentially expressed genes associated with developing salivary gland and its various compartments from D17 to D32. Downregulation of SG genes peaking at D25 (f) and upregulation of SG genes peaking at D32 (g) are boxed in blue and red respectively. B for BXS0114 and H for WA09 (H9) with numbers following the letters representing the timepoints in days. Source data are provided as a Source Data file. Formation of SG organoids from SGEP cells. a Schematic showing the embedding of the D17 cells into Cultrex marking the transition into a 3D differentiation model, where the cells aggregate to form luminal organoids by D25 and later transition into a SOX10 expressing luminal acinar progenitor by D32. Created in BioRender. Andreadis, S. (2026) https://BioRender.com/7pu5lxv (b) 3D rendering of D32 organoids from a confocal image shows the organoids were completely lumenized. Hollow lumens shown by arrows as seen in the bottom sliced rendering of the organoids. Comparative gene expression levels normalized to PSC Day 0 showed: (c) RT-PCR for the early acinar progenitor gene, SOX10 shows increased expression from D17 to D32. The data was normalized to gene expression levels of D0 PSC * Denotes p < 0.05, n = 4. All data is shown as mean +/- SD. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test for multiple comparisons. Statistical significance between two groups were calculated using student’s t-test. d Fluorescence microscopy of organoids made from H9 pSOX10::GFP reporter cells that express GFP when the SOX10 promoter locus is activated. Scale bar = 50 µm. e Flow cytometry across shows increased percentage of cells that express pSOX10::GFP as the differentiation progresses from D17 to D32. f, g Heat maps of differentially expressed genes associated with developing salivary gland and its various compartments from D17 to D32. Downregulation of SG genes peaking at D25 (f) and upregulation of SG genes peaking at D32 (g) are boxed in blue and red respectively. B for BXS0114 and H for WA09 (H9) with numbers following the letters representing the timepoints in days. Source data are provided as a Source Data file. Maturation of SGOs into acinar and ductal phenotypes Subsequently, the SOX10+ organoids were further differentiated by treatment with EGF73, retinoic acid66, hydrocortisone74, and triiodothyronine75. Interestingly, the pSOX10-GFP signal continued to increase during this differentiation period (Suppl. Movie 4). By day 46, the organoids reached an average diameter of 50-60 μm, with the largest organoids measuring 100–110 μm in diameter (Fig. S8). At this stage, the organoids exhibited positive immunofluorescence staining for mature acinar cell markers such as MIST176 (Fig. 5a, i), NKCC177 (Fig. 5b, i), and apically localized AQP578 (Fig. 5c). They also expressed the adherens junction protein, E-cadherin79 (Fig. 5d, i) and tight junction protein, ZO179 (Fig. 5e, i). Notably, the organoids expressed AMY1A80 and exhibited an apical F-actin ring (Fig. 5f & Suppl. Movie 5), indicative of acinar cell maturation and the establishment of epithelial polarity.Fig. 5Maturation of SG organoids.Confocal immunofluorescence images of mature D46 organoids displaying adult salivary gland acinar markers: (a) MIST1, (b) NKCC1 and (c) AQP5. The organoids also show formation of (d) E-cadherin adherens junctions; and (e) Z0-1 tight junctions, indicating polarity; along with (f) AMY1A, a secreted protein present in saliva produced by acinar cells. c, d, f Organoids display an inner F-actin ring, a hallmark of polarity. g, h Co-staining of ductal marker, K7 with acinar markers of (g) AMY1A and (h) NKCC1. Scale bar = 50 µm, images are 40X magnified. i Western blots and quantifications for MIST1, NKCC1, E-cadherin, ZO-1 over time. GAPDH served as housekeeping loading control. * Denotes p < 0.05, ** denotes p < 0.01, *** denotes p < 0.001, and **** denotes p < 0.0001. (n = 3). j, k Heat maps of differentially expressed genes associated with the developing salivary gland, the various progenitor gland compartments along with adult acinar, ductal, and myoepithelial lineages; as well as genes encoding for salivary protein components and proteins that aid in saliva secretion. Downregulation of primarily ductal markers (j) and upregulation of primarily acinar markers (k) are boxed in blue and red respectively. l Representative calcium flux in D46 (blue) and D32 (red) organoids in response to carbachol. (m) Maximum Fluorescence Intensity (mFI) associated with calcium flux responses for D32 (red) and D46 (blue) organoids (n = 12, p < 0.0001). All data is shown as mean +/- SD. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test for multiple comparisons. Statistical significance between two groups were calculated using student’s t-test. Dashed line indicates background fluorescence levels. Cell lines used in the study are represented as: B for BXS0114 and H for WA09 (H9), while the numbers following them represent the timepoints in days. Source data are provided as a Source Data file. Maturation of SG organoids. Confocal immunofluorescence images of mature D46 organoids displaying adult salivary gland acinar markers: (a) MIST1, (b) NKCC1 and (c) AQP5. The organoids also show formation of (d) E-cadherin adherens junctions; and (e) Z0-1 tight junctions, indicating polarity; along with (f) AMY1A, a secreted protein present in saliva produced by acinar cells. c, d, f Organoids display an inner F-actin ring, a hallmark of polarity. g, h Co-staining of ductal marker, K7 with acinar markers of (g) AMY1A and (h) NKCC1. Scale bar = 50 µm, images are 40X magnified. i Western blots and quantifications for MIST1, NKCC1, E-cadherin, ZO-1 over time. GAPDH served as housekeeping loading control. * Denotes p < 0.05, ** denotes p < 0.01, *** denotes p < 0.001, and **** denotes p < 0.0001. (n = 3). j, k Heat maps of differentially expressed genes associated with the developing salivary gland, the various progenitor gland compartments along with adult acinar, ductal, and myoepithelial lineages; as well as genes encoding for salivary protein components and proteins that aid in saliva secretion. Downregulation of primarily ductal markers (j) and upregulation of primarily acinar markers (k) are boxed in blue and red respectively. l Representative calcium flux in D46 (blue) and D32 (red) organoids in response to carbachol. (m) Maximum Fluorescence Intensity (mFI) associated with calcium flux responses for D32 (red) and D46 (blue) organoids (n = 12, p < 0.0001). All data is shown as mean +/- SD. Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test for multiple comparisons. Statistical significance between two groups were calculated using student’s t-test. Dashed line indicates background fluorescence levels. Cell lines used in the study are represented as: B for BXS0114 and H for WA09 (H9), while the numbers following them represent the timepoints in days. Source data are provided as a Source Data file. Genes associated with early developmental stages like ETV4/5, MYC, IRF6, ITGA3/6 and XBP163,76 reached maximum expression between days 25 and 32 (Fig. 5j, k, Fig. S2f); whereas genes associated with mature stages of gland development like SLC12A2 (NKCC1), CFTR81, PIP63 and ST1482 reached their highest expression on day 46 (Fig. 5k, Fig. S2g, Fig. S3g). Interestingly, while many keratin genes associated with ductal cell fate (KRT7, KRT8/18, KRT17/19) were downregulated on day 46, their absolute levels were similar or higher than those of acinar genes (Fig. S3h, Fig. S9). Indeed, day 46 SGO immunostained for ductal (K7) and acinar markers (AMY1A, NKCC1), suggesting bipotent phenotype (Fig. 5g, h). Mature SGOs develop salivary secretory function Saliva secretion is stimulated by the neurotransmitter acetylcholine, which is secreted by the parasympathetic nervous system and binds to muscarinic receptor 3 (M3R), prompting intracellular calcium release from the endoplasmic reticulum83. In turn, this leads to activation of NKCC1, which promotes recruitment of AQP5 to the apical cell membrane, enabling saliva secretion from the acinar cells into the lumen84. We examined whether organoids developed functionality at different times of maturation by measuring the calcium flux in D32 and D46 organoids in response to the acetyl choline analog, carbachol using Fluo4-AM. To this end, D32 or D46 organoids were released from Cultrex and placed on the surface of Matrigel coated wells. After settling onto the surface, the organoids were incubated with Fluo4-AM and stimulated with Carbachol to release intracellular calcium, cleaving the AM tag and generating a short fluorescence pulse at 525 nm upon excitation at 488 nm (Supplementary Movie 6). Representative fluorescence signatures of D46 (blue line) and D32 (red line) organoids in response carbachol are shown in Fig. 5l. The maximum fluorescence intensity (mFI) was used as a metric to compare D32 and D46 organoids. As shown in Fig. 5m, salivary organoids exhibited significantly higher mFI on D46 as compared to D32 (background level) (n = 12, p < 0.0001). These results indicate that our salivary organoids develop secretory function between D32 and D46, coinciding with expression of mature acinar cell markers MIST1, NKCC1 and apical AQP5 (Fig. 5a–c). SGO transplantation into an orthotopic SMG wound model At day 32, SGOs are Sox10⁺, show high cell viability ( > 98%) and express multiple key salivary gland progenitor markers, but are not yet fully mature. We therefore hypothesized that day 32 SGOs retain sufficient plasticity to generate all salivary gland lineages within the native gland microenvironment in vivo. To test this hypothesis, we examined whether the day 32 SGO could successfully engraft and differentiate into acinar and/or ductal tissue upon transplantation into an orthotopic submandibular gland wound model in vivo. The advantage of the wounded model over an irradiated model is the ability to control the size of the wound, thereby enabling direct monitoring of wound regeneration and the fate of the organoids over time. For this purpose, a surgical wound was created in the submandibular glands of athymic nude mice using a 3 mm biopsy punch and SGO were then introduced into the wound space within the Cultrex matrix (one dome per wound) (Fig. 6a). To track human cell fate in vivo and distinguish them from mouse cells, the H9 hESC cells were transduced with lentivirus encoding the mCherry reporter. The wound site was examined at 6-weeks post -transplantation.Fig. 6Generation of acinar and ductal structures upon orthotopic transplantation of human SGO into the SG of athymic mice.a Schematic of the orthotopic transplantation of D32 SGO embedded in Cultrex into the SMG of athymic nude mice. Created in BioRender. Missouri, B. L. (2026) https://BioRender.com/uzg9uvi (b) H&E staining of mouse submandibular gland. The wounded area is boxed in red. c Immunostaining of tissue section from the wounded area of the mouse SMG (red box) for PANCK, ZO1, αSMA, AQP5 (yellow arrows), K7, and K19 (green); mCherry (red) indicates the human cells of the transplanted human SGO. Scale bar = 25 µm. Number of animals per group, n = 4. Generation of acinar and ductal structures upon orthotopic transplantation of human SGO into the SG of athymic mice. a Schematic of the orthotopic transplantation of D32 SGO embedded in Cultrex into the SMG of athymic nude mice. Created in BioRender. Missouri, B. L. (2026) https://BioRender.com/uzg9uvi (b) H&E staining of mouse submandibular gland. The wounded area is boxed in red. c Immunostaining of tissue section from the wounded area of the mouse SMG (red box) for PANCK, ZO1, αSMA, AQP5 (yellow arrows), K7, and K19 (green); mCherry (red) indicates the human cells of the transplanted human SGO. Scale bar = 25 µm. Number of animals per group, n = 4. H&E staining of the submandibular gland at 6 weeks post transplantation shows the wound site that was filled with Cultrex embedded organoids (box, Fig. 6b). Immunofluorescence imaging shows that mCherry+ (human) SGO were able to engraft and form clear, organized structures with well-defined lumens which were not present in unwounded tissue (Fig. S10). Immunostaining demonstrated ubiquitous PANCK expression and apical ZO-1 localization, indicating epithelial identity and polarized structures, as well as expression of acinar (AQP5), ductal (K7 and K17/19) and myoepithelial (αSMA) markers indicating differentiation of human SGO into all salivary gland lineages (Fig. 6c). In conclusion, day 32 organoids could survive, engraft and differentiate into adult glandular lineages following implantation in vivo, indicating that day 32 SGO are multipotent. Discussion We present a strategy for deriving salivary gland organoids from human PSC. By providing stage-specific cues in a temporal sequence that mimics glandular development via the ectodermal germ layer and progenitor phenotypes, we generated salivary gland organoids exhibiting lumenized architecture and physiologically appropriate localization of proteins associated with distinct glandular compartments. To assess the efficacy of this strategy, we closely followed the expression of notable genes affected at each stage as they are discussion here (Suppl. Table 5). First, the SE phenotype was induced via stimulation of the BMP-SMAD 1/5/8 pathway85; while concomitant inhibition of the TGFβ and Notch pathways were employed to suppress mesendoderm differentiation86 and promote TP63 expression17, respectively. SE differentiation was accompanied by loss of embryonic factors like NANOG, MYC and OCT3/4. The loss in OCT3/4 expression was directly linked to down-regulation of ETS-related transcription factors ETV4 and ETV5 (ETV4/5) along with ES1 (LINC01108), a long noncoding RNA that is specifically expressed in undifferentiated embryonic stem cells and regulates pluripotency62,87. Other genes associated with the maintenance of pluripotency like MT1X21, PCSK988, TDGF189, POLR3G90, ZIC391 and LINC0067892 were also significantly downregulated. At the same time, genes that specify mature SE, including the Adaptor Protein-2 (AP-2α/β/γ) transcription factor family15,37 along with SCUBE338, GABRP36, TGM293, SFN15 and members of the collagen family36 (COL17A1, COL12A1, COL5A1), were significantly upregulated. Genes associated with SE derivative lineages such as COL3A194 and IGFBP541 (limbal development); IGFBP395 and MEIS1/240 (presumptive lens and corneal ectoderm); GUCY1A196 (olfactory epithelium); MMP297 (linked to teeth, lens, hair follicle, otic and mammary gland morphogenesis); SLIT2/398 (surface pharyngeal ectoderm); C5orf4699 (linked to the eccrine sweat and seborrheic glands of the skin); SLC4A439 (OE) were significantly upregulated, demonstrating the multipotency of the resulting SE. The OE lineage was derived by inhibiting the SMAD1/5/810 while activating the FGFR pathway10 to suppress the formation of other SE derivatives. This was evident by loss of MSX1/2, and downregulation of MEIS1, GUCY1A1 and SLIT3 with a concomitant increase of BNC1100 (a p63 transcriptional target), SOX1139,49, FOS39,49, FST46, CCN239,44, PROSER239 (dental lamina), SKAP239 (anterior epithelium), and ubiquitous expression of SOX2 throughout the cell monolayer. Later specification of OE to the SGEP compartment was coaxed by activating the FGFR2(IIIb)54 and RARα53 pathways, while inhibiting the Wnt55 pathway, as seen in the developing gland parenchyma. This led to ubiquitous albeit heterogenous expression of SOX9, which is known to specify distal acinar progenitors while simultaneously promoting ductal branching morphogenesis. Consistent with this role, loss of SOX9 results in arrested salivary gland development54,57, underscoring its central function in maintaining the bipotent salivary gland progenitor identity of SGEPs. At the same time, there was upregulation of SEMA3A101 (cell growth and morphogenesis), TIMP3102, CELSR103 (involved in branching morphogenesis), CRABP2104 (RAR pathway), NID2105 (basement membrane protein); and marked downregulation of odontogenic (dental) related genes FOS, FST, GAD1, SLC4A4. Interestingly, when embedded in Cultrex, SGEP cells self-assembled into 3-dimensional luminal organoids containing polarized salivary gland cells. Formation of lumenized organoids was accompanied by a shift in gene expression from SGEP initiation markers on day 17 to transition markers on day 25, including SERPINB266 (ductal cell development), CEBPB63 (pro-acinar cell development), TGM2106 (early myoepithelial cell marker), PTPRB107 (inhibition of cell proliferation), SPRY1108, DUSP6109 (branching morphogenesis) and DNER70 (non-secreted salivary gland protein). A week later (day 32), the day 25 genes were downregulated, while more acinar genes were upregulated including XBP176,110 (activator of MIST1 promoter), ZNF185111 (a TP63 target that maintains epithelial integrity by interaction with E-cadherin), NID1105 (basement membrane protein) and BARX2112 (involved in proliferation and differentiation of the gland epithelial cells). On day 46, XBP1 and ZNF185 were downregulated while late-stage acinar markers ZBTB2063 (pro-acinar cell development), SLPI113, PIP66 (salivary proteins), AREG114, EREG115 (plays roles in glandular patterning) and GATA3116 (maintains luminal epithelial differentiation) were upregulated. At the same time, genes associated with ductal network development and salivary gland maturation such as MUC1/MUC1672, ST14117 (involved in secretory epithelial cell function) were also upregulated significantly. While other ductal markers like CARHSP163, SERPINB2, TFCP2L1118, KRT8/17/18/1960 and myoepithelial marker TGM2 were downregulated. Nevertheless, the levels of some of those ductal genes, especially, KRT8/18/19 and myoepithelial gene, TGM2 continued to be high, suggesting development of a salivary gland phenotype with mixed acinar and ductal gene expression. Indeed, immunostaining of day 46 SGOs demonstrated differential expression of mature markers like MIST1, KNCC1 and K7, among cells comprising each organoid, possibly reflecting the differential response of individual cells to differentiation signals, or the heterogeneity of cell populations within organoids. Nevertheless, these organoids exhibited signs of maturation and function as shown by their response to the acetylcholine analogue, carbachol. Future studies are needed to define conditions that further enhance maturation and understand cell heterogeneity within organoids, which may further improve the secretory function and regenerative capacity of salivary gland organoids. Orthotopic transplantation of SGO into wounded mouse salivary glands demonstrated their multipotency and differentiation potential into mature salivary glands. The advantage of the wound model over the irradiated model is the ability to control the wound size and ensure the precise delivery of the hPSC-derived SGO at the desired location. As a result, it is relatively easy to assess cell viability and differentiation potential of human cells in the native microenvironment of the salivary gland. Notably, human iPSC-derived organoids were able to survive, engraft and differentiate into all salivary gland lineages including acinar (positive for AQP5), ductal (positive for KRT7,17/19), and myoepithelial cells (positive for αSMA) in vivo. This result demonstrates the multipotency of day 32 SGO, in agreement with the transcriptomic data. It would be interesting to examine how the stage of cell/organoid differentiation affects engraftment and differentiation in vivo; the long-term fate of human cells in the mouse glandular environment; and whether they contribute to saliva production. These studies may require longer-term in vivo evaluation, during which potential inflammatory responses and fibrotic encapsulation will also be assessed. Given that tumor-derived matrices, such as Matrigel and Cultrex, may elicit host responses including inflammation, foreign body reactions, and tissue remodeling, the development and evaluation of alternative biomaterials with enhanced biocompatibility and translational potential will be an important focus of future work119. Finally, the design and adaptation of biomaterial platforms and bioactive cues to promote the neovascularization and innervation of human organoids in vivo will also be explored, building on approaches that we have previously demonstrated to be effective120. In summary, our results demonstrate the successful derivation of SGEP cells and salivary gland organoids from hPSC. Human SGO exhibited discernible lumens and salivary function in vitro and could regenerate acinar, ductal and myoepithelial salivary gland cells in vivo upon transplantation into mouse SMG. Such hPSC-derived SGEP and SGOs may be a valuable model for studying gland development and disease and an effective platform for drug screening and testing to prevent xerostomia, as well as for cell therapy approaches for gland regeneration following injury and loss of function. Methods Human pluripotent stem cell culture and differentiation hiPSC (ATCC-BXS0114) from ATCC (ACS-1028) and the modified human embryonic stem cell (hESC) line (WA09) from WiCell (H9) were maintained in mTesR™ Plus (STEMCELL Technologies) on tissue culture-treated 6-well plates coated with growth factor reduced Matrigel® (GFR-MG, Corning) in a 5% O2 (Hypoxia) and 5% CO2 incubator. The medium was replenished every day and the cells from passage 15-40 (iPSC), 60-80 (ESC) were used in these experiments. Cells were gently passaged every 5 days by non-enzymatic dissociation of cell clusters using a 0.5 mM EDTA in PBS solution and reseeded in mTesR Plus medium containing 10 μM Y27632 (STEMCELL Technologies). For differentiation, hPSC were seeded at 400,000 cells per well of a 6-well plate (60% confluent) and medium was changed to 2 ml SE DM (Supplementary Table 1) 18 h later. Next, cells were differentiated under hypoxic conditions and SE DM (DMEM/F12 + 10% KOSR + 5 μM SB431542 + 5 μM DAPT + 35 ng/mL BMP-4) was changed daily from day 0 to day 2. The medium was changed to 2 ml SE MM (DMEM/F12 + 10% KOSR + 5 μM SB431542 + 5 μM DAPT) on day 3, (Supplementary Table 1). Cells were dissociated using TrypLE™ Express (ThermoFisher Scientific) and replated using a 1:2 dilution on tissue culture-treated 6-well plates coated with a growth factor-reduced Matrigel® in 2 ml of SE MM containing 10 μM Y27632 per well on day 4. Following this step, the medium was switched with 2 ml SE MM on day 5 which was then replenished on day 6. Cells were next dissociated using TrypLE™ Express (ThermoFisher Scientific) on day 8 and replated using a 2:3 dilution on tissue culture treated 6-well plates coated with a growth factor-reduced Matrigel® in 2 ml SE MM containing 10 μM Y27632 per well. The medium was switched back to 2 ml SE MM on day 9. The medium was next changed to 2 ml of OE DM (DMEM/F12 + 2% KOSR + 5 μM SB431542 + 150 nM LDN193189 + 5 μM DAPT + 25 ng/mL FGF-2) on day 10 (Supplementary Table 1), followed by medium change on day 12 with OE DM again. Finally, the medium was changed to 2 ml SGEP DM (DMEM/F12 + 2% KOSR + 5 μM SB431542 + 150 μM LDN193189 + 250 nM Am580 + 2uM Wnt-C59 + 100 ng/mL FGF-10 + 10 μg/mL Heparin) on day 13 (Suppl. Table 1) and replenished on day 15. A list of all media components is shown in Supplementary Table 2. The media was not degassed prior to application. Organoid formation and differentiation Cells were dissociated using TrypLE™ Express and reseeded at 8 × 106 cell/mL in Cultrex Ultimatrix RGF Basement Membrane Extract, Type 2 Select (R&D Systems) on day 17. Next, 20 µl of cell-containing Cultrex was deposited on the surface of non-tissue culture treated 6-well plates. A total of 10 domes were plated per well and allowed to solidify at 37 °C for 30 min prior to the addition of 2 ml SGO DM 1 (DMEM/F12 + 2% KOSR + 5μM SB431542 + 150 nM LDN193189 + 100 ng/mL FGF-10 + 80 ng/mL EGF + 100 nM Dexamethasone + 10 μg/mL Heparin) (Supplementary Table 1). The domes were cultured under normoxic conditions in a 10% CO2 incubator and medium replenished every two days till day 23. Next, the medium was replaced with 2 ml SGO DM 2 (DMEM/F12 + 2% KOSR + 5μM SB431542 + 150 nM LDN193189 + 100 ng/mL HGF + 80 ng/mL EGF + 100 nM Dexamethasone + 10 mM Nicotinamide) (Suppl. Table 1) on day 25 and replenished every two days till day 31. On day 32, the medium was changed to 2 ml SGO DM 3 (DMEM/F12 + 2% KOSR + 5μM SB431542 + 150 nM LDN193189 + 1X ITS + 100 nM Retinoic Acid + 2 nM Triiodothyronine + 1.5 μM Hydrocortisone + 80 ng/mL EGF + 100 nM Dexamethasone) (Suppl. Table 1) and replenished every two days till day 44. A list of all media components is shown in Suppl. Table 2. Media optimization Media optimization for the surface ectoderm (SE) stage was done by qPCR analysis for highest expression levels for SE related genes like MSX1/216 and p6317, while maintaining low expression of embryonic, neural and neural crest associated genes. Hence following this guideline, we optimized concentrations of the growth factor BMP-410,15 (3.5–35 ng/ml), a known SE inducer and inhibitors like SB43154210,16 (1–10 µM, TGF-β pathway inhibitor) and DAPT17,18 (1–10 µM, Notch pathway inhibitor) that are known to inhibit the mesendoderm development and upregulate the p63 pathway respectively (Fig. S1). Subsequently, the media optimization for the oral ectoderm (OE) was done similarly with target genes being SOX210,45 and PITX210,45, while downregulating embryonic, neural, neural crest and SE related genes. Hence, similarly we optimized concentrations of FGF-210,44 (5–50 ng/ml), a known inducer of FGFR signaling and LDN19318910 (50–450 nM, BMP pathway inhibitor) a known inhibitor of dental precursors (Fig. S2). Concomitantly to see the effect of the FGFR signaling on the OE development we inhibited it using SU5402121 (5 µM) (Fig. S2). Next, to optimize the differentiation towards salivary gland epithelial progenitors (SGEP), we targeted progenitor specific genes like SOX9, FOXC1, p63, Keratin 5 and 14. Here we optimized FGF-7/1010,54,65,66 (10–100 ng/ml, FGFR2b agonists) and Am58053 (25–250 nM, RARα agonist) concentrations as they are known salivary gland morphogens and inducers of SOX9, along with perturbation of the Wnt pathway with CHIR9902156 (0.2–2 µM, Wnt pathway agonist) or Wnt-C5955 (0.2–2 µM, Wnt pathway inhibitor), as Wnt activity is restricted to the mesenchyme in the developing gland (Fig. S3). For the next stage, organoid formation from the SGEP in Cultrex was key for the media component optimization, the presence of EGF66,122 being critical for organoid formation (Fig. S4). We further defined the media by using combinations of previously optimized concentrations of FGF-10, Am580 and Wnt-C59 to target an upregulation in SOX10 and FOXC1 while downregulating SOX9 (Fig. S4). Similarly, for the further differentiation of the organoids we targeted the same genes as above with the goal of further upregulating SOX10 and FOXC1 while maintaining low levels of SOX9 (Fig. S4). Here we optimized the concentration of HGF68,122,123 (10–100 ng/ml, Met receptor agonist) with or without FGF-10. HGF is a growth factor known to be involved in the regulation XBP1/MIST1 and hence aids the maturation of the salispheres. The resulting up- and down-regulation of several notable genes throughout the protocol are included here in the order in which they appear in the paper’s discussion (Fig. S5). Organoid extraction for downstream use Ten gel domes were dissolved in a 15 ml conical tube using 3 ml of Cultrex Organoid Harvesting Solution (R&D Systems) at 4 °C for 3 h with gentle stirring at 60 rpm to extract the organoids. Once the gel domes were dissolved, the organoids were sedimented under gravity for 10 min at 4 °C to collect them. For flow cytometry, the organoids were incubated with TrypLE™ for 15 min at 37 °C prior to fixation. RNA purification, reverse transcription, and qPCR Total RNA was extracted from the organoids/cells using the RNeasy® Plus Mini Kit (Qiagen). Reverse transcription was performed using the QuantiTect® Reverse transcription Kit (Qiagen). Quantitative PCR (qPCR) was performed using the generated complementary DNA samples with Power Up® SYBR® Green Master Mix (ThermoFisher Scientific) in the CFX96™ Real-Time System (BIO RAD). The primer list and sequences are shown in Suppl. Table 3. Quantification of the samples was performed by analyzing the threshold cycle values using the ΔΔCt method. The values presented in the graphs represent individual experimental values with means ± standard deviations internally normalized to its relevant levels of GAPDH. Western blots Total protein was extracted from the organoids/cells using a lysis buffer containing 62.5 mM Tris-HCl (pH 6.8 at 25 °C), 2% (w/v) SDS, 10% (v/v) glycerol, 0.1% (w/v) bromophenol blue, 41.67 mM dithiothreitol (DTT) (Cell Signaling, Danvers, MA), and protease inhibitor cocktail (Sigma-Aldrich). The protein concentration was determined using the Bradford assay. Lysates were denatured by incubation at 95 °C for 5 min and proteins were loaded at 45 µg per lane and were separated in 4-20% acrylamide gels (ThermoFisher Scientific) by SDS-polyacrylamide gel electrophoresis based on their molecular weight. After transferring proteins to 0.22 µM nitrocellulose membranes (Bio-Rad) using the Trans-Blot Turbo Transfer System (Bio-Rad), the membranes were blocked in 5% (w/v) non-fat dry milk in blocking buffer for 1 h at room temperature. Subsequently, membranes were incubated overnight at 4 °C with antibodies listed in Suppl. Table 4. Membranes were incubated for 1 h at room temperature with Anti-rabbit IgG HRP linked (Cell Signaling, Cat# 7074; RRID: AB_2099233) or Anti-mouse IgG HRP linked (Cell Signaling, Cat# 7076; RRID: AB_330924). Finally, the protein bands were visualized using horseradish peroxidase-conjugated secondary antibodies and a chemiluminescence kit (ThermoFisher Scientific) according to the manufacturer’s instructions. Luminescent blots were imaged using ChemiDoc™ Touch Imaging System (Bio-Rad). Immunofluorescence analysis Cells, gel domes or Tissue-Tek® O.C.T. Compound (Sakura) frozen sections of the gel dome were fixed in 4% paraformaldehyde (ThermoFisher Scientific) in PBS for 20 min and permeabilized with 0.1% Triton X-100 (Sigma Aldrich) in PBS for 15 min at room temperature. Next, samples were blocked for 3 h with 5% goat serum (ThermoFisher Scientific) with 0.01% Triton X-100 in PBS at room temperature. Gel domes were then labelled for 40 h at 4 °C whereas sections were labelled for 16 h at 4 °C with different primary antibodies (Supplementary Table 4). Following this step, gel domes were incubated with Alexa Fluor™ secondary antibodies (1:200 dilution; ThermoFisher Scientific) for 16 h at 4 °C whereas the sections were labeled for 2 h at room temperature. Nuclei were counterstained with Hoechst 33342 (Tocris) or TO-PRO™-3 Iodide (ThermoFisher Scientific). Images were acquired using an Axio Observer Z1 LSM 510 fluorescence microscope (Zeiss) equipped with a digital camera ORCA-ER C4742-80 (Hamamatsu) or a Stellaris 5 confocal microscope (Leica). Images were visualized with Zen Black Edition (Zeiss), or LAS X Life Science (Leica) Microscope Software Platformand then analyzed and quantified using FIJI (ImageJ) software. Finally, videos were compiled using either the LAS X Life Science Microscope Software Platform (Leica) or FIJI (ImageJ) software. Flow Cytometry Cells from day 17 or from organoids at D25/32 were fixed in cold 4% paraformaldehyde in PBS for 10 min and permeabilized with 0.1% Triton X-100 in PBS for 10 min at room temperature. Next, samples were blocked for 30 min with 5% goat serum with 0.01% Triton X-100 in PBS at room temperature. The sample was then labeled for 1 h at room temperature with different directly conjugated primary antibodies and isotype control IgG (Suppl. Table 4). The samples were washed with 500 µl cold PBS after each step, centrifuged at 300 × g at 4 °C for 5 min, and resuspended in 200 µl of cold PBS. Each sample (10,000 cells per sample) was analyzed by flow cytometry using a BD Fortessa X20, four-laser, 14-color analyzer (Franklin Lakes, NJ). The results were analyzed using FCS Express 6 Flow software. Live cell imaging Whole gel domes were visualized at a 4X magnification using the Incucyte® live cell imager (Sartorius) from day seventeen to day forty-six. Green Fluorescence Protein (GFP, green) and mCherry (red, pTRIP-EF1α-mCherry lentivirus (gift of Abdel Benraiss, University of Rochester)) channels were used to track Sox10 expression and organoid formation, respectively. Videos were compiled from images taken with the imager at 12 h intervals. Calcium flux assay Functionality of D32 and D46 acinar organoids was evaluated by measuring calcium flux in response to carbachol, as showed previously124–126. Cultrex domes were dissolved with Organoid Harvesting Solution (R&D Systems) and replated into wells of a 24-well plate coated with Growth Factor Reduced Matrigel (R&D Systems). The organoids were allowed to settle onto the surface in their respective media overnight. The next day, the organoids were incubated in 4 μM Fluo4-AM (Invitrogen) for 30 min at 37 °C, then washed twice with Calcium Flux Buffer (50 mL DMEM/F12 + 0.18% w/v glucose + 0.1% BSA + 1 mM CaCl2) and incubated in the same buffer for 30 min at 37 °C. After incubation, the plate was loaded into a temperature-controlled chamber and organoids were detected under phase contrast before addition of carbachol (100 μM). Images were acquired with the Zeiss AxioObserver.Z1 Fluorescence Microscope equipped with ORCA-ER CCD camera (Hamamatsu, Japan). Using the Timescale setting, images were captured starting 15–20 s before the addition of carbachol and continued until the fluorescence signal (Ex: 488 nm; Em: 525 nm) reached background levels. Single cell sequencing – Sample and Library Preparation Cells and embedded organoids were dissociated via TryplE™ Express. The cells isolated at D17 were processed at the University at Buffalo Genomics Core (UBGBC) for scRNA sequencing. After lysis treatment, the cells were counted on a Logos Biosystems LUNA II in bright field mode using 0.4% Trypan blue exclusion for viability. The cells were captured ( ~ 10,000 cells per capture) using the 10X Genomics Chromium Platform with reagents from the Single Cell GEMX v4 kit according to the manufacturer’s recommendations. After drop-seq capture, cDNA from captured transcripts was used as input to generate double-stranded Illumina sequencing libraries. Following final cleanup, qPCR (QuantaBio library quantification kit) was used to determine final concentration of the libraries. The single-cell RNA libraries were diluted and denatured to 250 pM for sequencing on the NovaSeq 6000 (28 × 91 paired end read length). Single cell sequencing – Data analysis Following library generation and sequencing on an illumina SP NovaSeq flowcell, reads were demultiplexed using the 10x Genomics Cell Ranger mkfastq command with bcl2fastq2 (v2.20.0.422). Once demultiplexed, individual samples were analyzed using Cell Ranger (v7.2.0) count against the human genome GRCh38. The resulting counts matrix served as input for subsequent downstream analysis. Further preprocessing and downstream analysis was performed using R (v4.5.1). First, ambient RNA removal was handled using the SoupX127 (v1.6.2) package and potential doublets were removed using scDblFinder128 (v1.16.0) package. The data was then processed using the standard Seurat129 (v5.4.0) workflow, i.e., SCTransform, RunPCA, FindNeighbors, FindClusters and RunUMAP. Appropriate per-cell cutoffs (feature_cutoff > 750, count_cutoff > 1000, Mitochondrial transcripts > 10%) were applied to the dataset pre normalization. Effect of Cell cycle was regressed out during SCTransform normalization. Cells were scored for various cell types using either AddModuleScore from the Seurat package or AUCell130 (v1.30.1). Cells that were assigned both Pro-Acinar and Pro-Ductal by AUCell were considered bipotent. Bulk RNA-sequencing – sample and library preparation The global gene expression profiles were characterized by next generation RNA sequencing using Illumina platform. To this end, total RNA was isolated for all conditions using RNeasy Mini Kit and quality control analysis was performed by RNA gel and Agilent Fragment Analyzer. Sequencing libraries were prepared as per standard Illumina protocols (Illumina Stranded Total RNA Prep with Ribo-Zero Plus), quality checked and quantified by Kapa Biosystems qPCR. The multiplexed libraries were sequenced in pair-end (2 × 50 bp) on the NovaSeq 6000 at 300 pM with 1% loading control. Bulk RNA sequencing – data analysis Sequenced reads were processed using nf-core/rnaseq (v3.14.0) pipeline131,132. Quality of the reads was assessed using FastQC (v0.12.1)133. Sequencing reads were trimmed using Trim Galore (v0.6.7)134 to eliminate the remains of Ilumina adaptors and filtering reads that were shorter than 20 bp. The filtered reads were then mapped to the mouse genome (GRCm39, Ensembl) using STAR (v2.7.9a)135 and quantified using Salmon (v1.10.1)136. The abundance estimates were imported into R using tximport (v1.26.1)137 package. Further preprocessing and downstream analysis was performed using R (v4.5.1). The resulting gene-level counts were used for differential gene expression and PCA analysis using the DESeq2 (v1.38.3)138 package. Genes that had at least 10 counts in minimum of 2 samples were considered for differential expression analysis and were considered to be differentially expressed if adjusted P < 0.05. Heatmaps were plotted using the pheatmap139 package and the volcano plots using the EnhancedVolcano140 package. Wounded submandibular gland mouse model The submandibular gland (SMG) wound model was created using a method previously reported141. Briefly, 12 female 8-week old NU/J mice were anesthetized with isoflurane (2.5%, 100 mL/min), and a 0.5 cm skin incision was made along the anterior surface of the neck. Mouse SMGs were exposed, and the surgical wounds were created using a 3 mm diameter biopsy punch. Next, a 20 μL dome of Cultrex containing D32 SGO as described above was placed into each wound with fresh Cultrex injected to fill the remaining empty space. An aluminum plate was placed underneath for 15 s to prevent hydrogel leakage. Finally, the skin incision was sutured, and the mice were housed in cages in a room with a controlled environment for a period of 1 month. All animal usage, anesthesia, and treatments were conducted in compliance with ARRIVE guidelines and approved by the University of Missouri-Columbia Animal Care and Use Committee (ACUC). SMG collection The mice were euthanized in a CO2 chamber at 2- and 4-weeks post-treatment, followed by exsanguination through incision of the abdominal aorta for tissue collection. Then, collected SMGs were visualized for red fluorescence using GelDoc Imaging System to briefly confirm the presence of injected cells. SMGs without cell injection were used to compare and subtract the endogenous fluorescence signal. SMGs were then fixed in 10% formalin at room temperature overnight, dehydrated in a 70% ethanol solution, embedded in paraffin wax, and cut into 5 μm sections. The survival and differentiation status of injected cells were assessed by mCherry fluorescence and visualized using a Stellaris 5 confocal microscope (Leica). Statistics and reproducibility Statistical significance was determined by one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc test for multiple comparisons. Statistical significance between two groups were calculated using student’s t-test. All experiments were repeated 3 times, yielding similar results. All the data were analyzed using GraphPad Prism 8 and Microsoft 365 Excel. All the immunofluorescence and PCR experiments were performed independently at least three times with reproducible results. No statistical method was used to pre-determine sample size. The experiments were not randomized. The investigators were not blinded to allocation during the experiments or outcome assessment. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Peer Review file Description of Additional Supplementary Files Supplementary Movie 1 Supplementary Movie 2 Supplementary Movie 3 Supplementary Movie 4 Supplementary Movie 5 Supplementary Movie 6 Reporting Summary Supplementary Information Peer Review file Description of Additional Supplementary Files Supplementary Movie 1 Supplementary Movie 2 Supplementary Movie 3 Supplementary Movie 4 Supplementary Movie 5 Supplementary Movie 6 Reporting Summary Source data Source Data 1 Source Data 2 Source Data 1 Source Data 2