This study was approved by the Institutional Review Board at the Baylor College of Medicine (under protocol H30712) and informed consent was obtained from all patients. Duodenal enteroids and proximal colon colonoids were established from fresh colonoscopy biopsies (ascending/proximal colon) or resected tissue from roux en y gastric by-pass surgeries (duodenum) obtained from healthy patients through the Texas Medical Center Digestive Diseases Center as previously described (12). Lines were generated from 9 individual subjects with 6 lines examined for duodenum and 6 lines examined for proximal colon (note: 3 subjects had lines developed from both duodenum and proximal colon, Supplementary Table S1 ). Briefly, intestinal tissue was collected and placed into PBS on ice and was washed 3 times. Samples were then placed in a single well of a 6-well plate with Ca⁺ and Mg⁺ free PBS containing 20 mM EDTA at 4°C and shaken for 30 minutes. The tissue was then collected and resuspended in HBSS (with Ca+ and Mg+) and shaken until crypts were released. Crypts were pelleted at 70 x g for 3 minutes at 4°C and the resultant pellet was resuspending in DMEM (containing penicillin-streptomycin and Fungizone) and re-pelleted (70 x g, 1 min). The final crypts were resuspended in Matrigel and plated in a 24 well plate. Following Matrigel solidification, complete medium with growth factors was added (CMGF+; Advanced Dulbecco’s modified Eagle medium (DMEM)/F-12 medium (Thermofisher), supplemented with 50% Wnt3A conditioned medium, (volume/volume), 20% R-Spondin conditioned medium (volume/volume), 10% Noggin-conditioned medium (volume/volume), 100 U/mL penicillin-streptomycin (Fisher Scientific), 10 mM HEPES buffer (Invitrogen), and 1X GlutaMAX (Invitrogen); 1X B-27 supplement (Life Technologies), 1X N-2 supplement (Life Technologies), 1 x B7 supplement, (Life Technologies), 10 mM nicotinamide (Sigma-Aldrich, St. Louis, MO), 1 mM N-acetylcysteine (Sigma-Aldrich), 10 nM Leu-Gastrin I (Sigma-Aldrich), 500 nM A-83 (Tocris Bioscience), 50 ng/mL human EGF (R&D Systems), and 10 nM SB 202190 (Sigma-Aldrich).

Enteroids and colonoids were maintained in CMGF+ with media changes (0.5 ml) every 2–3 days using standard cell culture conditions (37°C, 5% CO₂, relative humidity 85-95%). Cultures were passaged every week by disrupting the enteroids/organoids with an insulin syringe, followed by resuspension of enteroids/colonoid fragments in Matrigel and replating. Some enteroid and colonoid lines were frozen and stored in liquid nitrogen, and subsequently thawed and expanded again before use.

Six duodenal lines and 6 proximal colon lines were used for this experiment ( Supplementary Table S1 ). Enteroids and colonoids were passage into new Matrigel with the goal of having a density of 100 colonoids/enteroids per well. To create undifferentiated cultures, enteroids and colonoids were treated with CMGF+. To generate differentiated enteroids and colonoids, cultures were treated with differentiation medium CMGF- (CMGF+ without Wnt3A, R-Spondin, nicotinamide, or SB202). All cultures were grown for 3 days with one 0.5 ml media change. Subsequently, the cultures were treated with 0.5 ml media containing 100 nM calcitriol (Cayman) or vehicle (ethanol). Three wells were used for each sample and treatment condition and 24 hours following treatment, cultures were collected, pooled by treatment condition, and processed. Calcitriol was prepared from powder to a concentration of 10⁻⁴ M in ethanol then the calcitriol stock was serially diluted to 100 nM in culture medium. The experimental design is summarized in Figure 1 . Images of the four types of culture used for study are presented in Figure 2A .

Total RNA was isolated from differentiated or undifferentiated enteroids/colonoids using the E.Z.N.A.^® Total RNA Kit I (Omega BIO-TEK) according to the manufacturer’s protocol. RNA integrity determined using an Agilent 2100 Bioanalyzer. All samples submitted for bulk RNA sequencing had RIN scores greater than 7. Sequencing was performed using paired-end Illumina sequencing by Novogene (Novogene Corporation) following their standard procedures. Subsequently, sequences were aligned to reference genome hg38 using Kallisto (https://pachterlab.github.io/kallisto/) with output expressed as transcripts per million (TPM) (14). Sequencing data was submitted to GEO and is available as accession number GSE159811.

Data was analyzed in R using the DESeq2 package with output expressed as Fragments Per Kilobase of transcript per Million mapped reads (FPKMs). Differential expressed gene (DEG) analysis was conducted with DESeq2 in R using adjustments for the paired nature of our samples. False Discovery Rate (FDR)/adjusted p-values were generated Benjamin-Hochberg procedure that balanced type I and II statistical errors and differential gene expression was calculated at FDR (padj) < 0.05 or 0.1 both with and without a 1.5 fold-change cut off. Multivariate analysis of variance based on distances and permutations was performed using the PERMANOVA package in R using the adonis function in the vegan package with 999 permutations and pairwise distances were calculated using the Euclidean method. The model used included patient origin (Line), the intestinal segment (Segment), the culture condition (Culture), and vitamin D treatment status (Vitamin D).

Functional enrichment analysis was conducted using DAVID (15) and Gene set enrichment analysis (GSEA) (16). In DAVID the analysis was conducted using a background data set of 15,047 genes that were “present” in the human colon or duodenal cultures and for which a gene name/symbol was available. Analyzes were conducted by individual group and separately for up and down regulated transcripts using genes that met the 0.05 FDR cut-off. Analysis was conducted with medium stringency using three gene ontology categories (molecular function, biological process, cell compartment) and three pathway groups (Reactome, Biocarta, Kegg). Categories that were significantly enriched at FDR <0.05 were prioritized for interpretation. When no categories met the FDR cut-off, those with p-values < 0.01 were considered. In GSEA we examined enriched canonical pathways, hallmark pathways, and gene ontology terms using curated gene sets in the Molecular Signatures Database (MSigDB v.2022.1.Hs) (16, 17). The analysis was conducted on 16,770 transcripts identified is “present” in at least one sample set (50% of samples with FPKM values >0.25 for a gene in a group). The dataset was not collapsed, and default parameters were used.

We compared the DEG from our RNA-seq analysis (10% FDR cut off) to published RNA-seq data from several studies: (a) Fernandez-Barral et al. (9), 2469 DEGs (1097 up, 1372 down), 1,25(OH)₂D₃ (100 nM, 96 h) treatment of undifferentiated colonoid cultures derived from normal colon tissue adjacent to tumors of 6 colorectal cancer patients (GSE100785); (b) Li et al. (10), 1,25(OH)₂D₃ (100 nM, 6 or 24 h) treatment of undifferentiated colonoid cultures from a single patient but with 3 technical replicates (GSE100785); (c) Bustamante-Madrid et al. (18), 1,25(OH)₂D₃ (100 nM, 6 or 24 h) treatment of undifferentiated colonoids or differentiated colonoids from normal colon tissue adjacent to tumors of 6 colorectal cancer patients (10 ng/g BW, 48h; GSE248274), and (d) Aita et al. (3), 1,25(OH)₂D₃-regulated (100 nM, 4 h) expression of genes in small intestinal villus, small intestinal crypts, or colon mucosa isolated from mice (n=6–7 per group; GSE161038). Genomic VDR binding sites used were from vehicle and 1,25(OH)₂D₃ treated undifferentiated colonoids (n=182 peaks) (9), LS180 human colon cancer cells (n= 845 peaks), and RWPE1 human prostate epithelial cells (n=7032) (19).

Using Permutational multivariate analysis of variance (PERMANOVA) to identify the drivers of variance among samples ( Figure 2B ) we found that differentiation status (Culture, 37.9%), patient source (Line, 26.9%) and intestinal segment (Segment, 14.2%) accounted for the bulk of the variation in our model. After adjusting for these factors, 1,25(OH)₂D₃ treatment had a significant impact on the transcriptome (Vitamin D, 3.1% of the variance). Principal component analysis (PCA) on the entire data set revealed similar results to the PERMANOVA analysis ( Figure 2C ), with 37% of variance separating the duodenal and colonic derived organoids (PC1) and 29% of the variance separating differentiated and undifferentiated samples (PC2). In each culture group, patient line was a stronger driver of variance than 1,25(OH)₂D₃ treatment, even though consistent effects of treatment were seen for each line/patient ( Figure 2D , differentiated duodenal enteroids). Because of the strong impact of line and segment on the transcriptome, we used a paired DESeq2 procedure to control these effects and to identify genes that were differentially expressed by culture condition (i.e. differentiation) and 1,25(OH)₂D₃ treatment.

To demonstrate that the culture conditions achieved the expected goals of modeling both intestinal segment and differentiation state, we examined the RNA-seq data for known markers of various intestinal cell states. The active intestinal stem cell marker LGR5 was dramatically higher in undifferentiated cultures and two markers for reserve intestinal stem cells, LRIG1 and BMI1, were higher in undifferentiated cultures ( Figures 3A–C ). APO4 and GATA4 are markers of the duodenum and were elevated in the duodenal cultures ( Figures 3D, E ) while LCT is a marker of the differentiated small intestinal epithelium and was highest in the differentiated duodenal enteroids ( Figure 3F ). Similarly, SATB2 is a colon cell marker that was elevated in the both colon cultures while CA1 and SLC51B are markers of the differentiated colonic epithelium and were elevated only in the differentiated colonoids ( Figures 3G–I ).

When examining our RNA-seq data at the 5% FDR cut-off, we identified many differentially expressed genes ( Supplementary Table S2 ; see Supplementary Figure 1 for heat maps of the top 50 up and down regulated transcripts within each of the four culture conditions). A traditional Venn diagram is provided as Supplementary Figure 2 while Figure 4 shows pairs of comparisons to emphasize several important points. First, our data show that differentiated cultures have more DEGs in response to 1,25(OH)₂D₃ treatment. Undifferentiated enteroids had less than one-half as many differentially expressed genes (DEGs) as differentiated enteroids and undifferentiated colonoids were the least responsive cell type we studied (with only 174 DEGs vs 836 DEGs in the next lowest group). Second, we found that there was a significant overlap in the vitamin D-induced DEGs for all the planned comparisons (i.e. within duodenal enteroids, colonoids, undifferentiated cultures or differentiated cultures) and the overlapping genes were more likely to be upregulated (59.3-74.5%, depending upon the comparison group). Finally, there was a subset of 63 genes that were differentially regulated across all four culture conditions and 87% of these overlapping transcripts were upregulated ( Figure 4 , Supplementary Table S2 ). Collectively these findings confirm a core finding from our previous analysis of the mouse intestine (3), i.e. that there are both common and compartment/segment-specific vitamin D target genes.

It has been previously established that 1,25(OH)₂D₃ binds to the VDR to induce genes encoding proteins necessary for transcellular intestinal calcium transport (i.e. TRPV6, S100G, ATP2B1) and for 1,25(OH)₂D₃ catabolism (CYP24A1) (6, 7). Figure 5A shows that VDR levels were uniformly expressed across the four culture conditions and that 1,25(OH)₂D₃ did not alter VDR levels; this is consistent with our previous report showing that VDR was not induced in small intestinal crypts and villi or in colon mucosal scrapings (20). In contrast, CYP24A1 was absent in untreated cells but was uniformly induced by 1,25(OH)₂D₃ across the four culture conditions ( Figure 5B ). This is consistent with what we have reported in mouse intestine (3). As expected TRPV6, S100G, ATP2B1, and SLC30A10 were each strongly induced by 1,25(OH)₂D₃ in differentiated duodenal enteroids ( Figures 5C–E ). Like CYP24A1, TRPV6 was significantly induced in all four culture conditions. In contrast, ATP2B1 was not induced in undifferentiated colonoids, SLC30A10 was induced only in the duodenum, and S100G was only significantly induced in the differentiated enteroids. While S100G encodes a protein proposed to be involved in intestinal calcium absorption (calbindin D_9k), other S100 genes encode for calcium binding proteins that regulate cell proliferation/differentiation, apoptosis, metabolism, and inflammation (21). Several S100 mRNAs were also upregulated in differentiated duodenal enteroids (e.g. S100A2, S100A6, S100A10) while S100P, which is associated with tumor growth and colon cancer was downregulated by 1,25(OH)₂D₃ treatment in differentiated colonoids ( Supplementary Table S2 ).

Enrichment analysis for GO terms and pathway analysis revealed common, as well as segment and differentiation state-specific molecular functions regulated by 1,25(OH)₂D₃ ( Table 1 DAVID analysis summary; Table 2 GSEA analysis summary; Full analyses are available in Supplementary Tables S3 – S7 ). The diversity of these enriched functions suggests that 1,25(OH)₂D₃ regulated additional intestinal functions beyond its traditional role in calcium absorption. Using GSEA, we identified enrichment of cell signature gene sets that suggest vitamin D treatment was reducing cell proliferation and promoting differentiation (e.g. enrichment of gene sets for proliferating cells or stem cells in the Control group for the duodenal cultures; enrichment of mature enterocyte gene sets in the Vitamin D group for undifferentiated duodenum enteroids and colonoids). We also found that in the analysis of differentiated colonoids, the control group was enriched for cell signatures of goblet cells. Consistent with this, the DAVID analysis showed vitamin D-mediated down regulated genes were enriched for the pathways O-glycan processing and mucin biogenesis, two processes typical of goblet cells (e.g. B3GNT2, GALNT7, 10, 11, and 12, MUC2, MUC6). This suggests vitamin D treatment antagonizes goblet cell differentiation.

Our analysis revealed that multiple terms for the functional category of metabolic processes controlling xenobiotic, steroid, or drug metabolism using cytochrome P450s or UDP-glucuronosyltransferases were enriched in all four groups. The genes in these enriched categories include several regulators of 1,25(OH)₂D₃ degradation and excretion: CYP24A1, which was strongly upregulated in all four groups, UGT1A1, which had higher overall expression in duodenal cultures but was induced in all four groups, and CYP3A4, which had a larger fold induction in differentiated organoids compared to the undifferentiated organoids. We also found that while some CYP and UGT genes, like CYP3A5 were up-regulated by 1,25(OH)₂D₃ treatment in all groups, others had compartmental specificity (e.g. stronger up-regulation of CYP2B6 in colon cultures; upregulation of CYP2C19 or suppression of UGT2B17 in duodenal cultures; upregulation of UGT1A3 and UGT1A5 in differentiated colon and duodenal enteroids.). Collectively, this shows there are unique segmental and compartment responses of CYP and UGT genes in the intestinal epithelium upon 1,25(OH)₂D₃ treatment.

DAVID and GSEA analysis identified Rho kinase signaling pathways as being enriched in vitamin D treated differentiated colon and duodenal enteroids ( Tables 1 , 2 ). Rho proteins are important molecular regulators of the cytoskeleton, cell morphology (including tight junction dynamics), and cell migration in the intestine (22). The vitamin D upregulated genes within these gene sets/ontologies included regulators of Rho signaling (e.g. DOCK5, ARHGAP25, OPHN1), cytoskeleton (e.g. ARPC2, AKAP12, TUBA1A, FMNL2), and cell adhesion (e.g. ABL1, CDH1). Pathways for integrin signaling and cell adhesion were also enriched in vitamin D induced genes from differentiated duodenal enteroids (e.g. ICAM1, ITGA2 and A5, TIMP2, KLK8).

A pathway for response to wounding was also enriched in vitamin D-upregulated genes that were expressed across all groups, e.g. genes encoding a regulator of notch signaling, PLLP (Plasmollipin); a protease controlling cell growth and migration, KLK6 (Kallikrein-related peptidase 6); and a coagulation regulator that has been implicated as a modulator of intestinal inflammation and barrier function, THBD (Thromboplastin).

Differentiated colon and duodenal enteroids also had vitamin D-mediated enrichment of lipid metabolism genes that included a regulator of intestinal triglyceride synthesis (DGAT2), genes involved in fatty acid metabolism (e.g. ACAA2, ELOVL7, FADS1), and those controlling phospholipid metabolism (e.g. GDPD3. PNPLA8, GNPAT). Vitamin D-mediated gene expression in differentiated duodenal enteroids was also enriched for many genes whose proteins are involved in the metabolism of phosphorus (up, SLC34A2, SLC20A1), iron (up, FTH1; down, SLC40A1, SLC46A1), copper (down, ATP7B), and sodium-coupled transport (down, SLC9A3, SLC5A1, SLC6A19). Vitamin D treatment also reduced genes for water transport in undifferentiated colon and duodenal cultures (e.g. AQP1, 3, and 5).

Other groups have used RNA-seq to examine the impact of 1,25(OH)₂ D treatment on colon organoids growing under conditions that promote stem cell survival and proliferation (i.e. high Wnt signaling like our CMGF+ media for undifferentiated cultures) (9, 10). Although these studies reported more vitamin D-mediated DEGs than we found in undifferentiated colonoids (5% FDR =174; 10% FDR = 236), the concordance between the studies for these genes was high (124/236 for Li et al.; 168/236 for Fernandez-Barral) with 111/236 matching across all three studies (74 up, 37 down) and 57 of these genes containing a potential VDR binding site including classical (CYP24A1, TRPV6) and non-classical (CD1, CEBPD, CYP3A4, CYP3A5, FOS, JUNB, and TXNRD1) vitamin D target genes ( Supplementary Table S2 ). A report by Bustamante-Madrid et al. (18) recently reported that 103 genes were differentially expressed by 1,25(OH)₂D₃ (100 nM, 48h) in colonoids induced to differentiate in media lacking Wnt3a and supplemented with BMP4 and the Notch inhibitor dibenzazepine. We found that 69 of those genes were also in our DEG list for differentiated colon organoids (34 up, 35 down) and 22 of these matched genes had a VDR binding peak. Finally, we compared our human organoid data to our previously reported findings on vitamin D-regulated gene expression in mouse small intestine villus, small intestine crypt, and colon mucosal scrapings (3). Of the 1397 vitamin D-regulated DEGs from Aita et al., 438 genes were also identified as DEGs in the human organoids (122 with a VDR binding peak). Direct comparison between mouse intestinal compartments and the human duodenal or colonic cultures is imprecise because the mouse compartments have more cellular complexity than the human intestinal cultures (e.g. the colon mucosal scraping contains more goblet cells and some tissue resident immune cells). Comparisons with similar compartments showed some similarities: small intestine villus to differentiated duodenum, 72 genes matched (24 with VDR peaks); small intestine crypt to undifferentiated duodenal cultures, 116 genes (38 with VDR peaks); colon mucosal scraping to either differentiated or undifferentiated colon cultures, 59 genes (20 with VDR peaks).