Single cell profiling of rectal organoids from patients with perianal CD
To begin the assessment on the ability of patient derived rectal organoids to harbor disease signatures, they were first established from crypts dissected from freshly obtained mucosal biopsies. Patient organoids expanded in Intesticult™ media typically ranged from 250 to 550 µm at the time of experimentation, demonstrating a rounded morphology across the samples when viewed microscopically (Fig. 1A). The organoids displayed a mix of spherical and oblong structures, but no associations were discernable between disease phenotype and organoid structures, regardless of whether the corresponding patient had CD, severe disease including perianal CD with inflamed mucosa, or a non-IBD control. After 3–4 weeks of growth in culture, gene expression in the organoids was analyzed by scRNA-seq and used to assess the subtypes and abundance of epithelial cells present in the culture (Fig. 1B, C). The expression levels of genes from single organoid cells were normalized, scaled, clustered, and shown as UMAP color coded to show epithelial cell subtypes (left) and lineage groups (right) in Fig. 1C. At the single cell level, 10 epithelial subtypes were identified in organoids, corresponding to four basic epithelial cell lineages: undifferentiated/stem, early absorptive, early secretory and a fourth group comprising lineages including a microfold-like cluster (Fig. 1C). Expression levels of marker genes used to identify the clusters are visualized in the dot plot in Fig. 1D. Thus, as expected the rectal organoids expressed features typical for the base of an intestinal crypt9.
Single-cell transcriptomics of organoids derived from patients with and without perianal fistulizing Crohn’s disease. (A,B) Patient biopsies were used (n = 13 patients; 8 CD, 5 controls) to establish rectal organoids that were processed for single-cell RNA sequencing using 10 × Genomics Technology at the 3rd passage and sequenced using Illumina sequencing at 50,000 reads per cell. (C) Organoid epithelial cells were normalized, scaled, clustered, and shown as UMAP colored by epithelial cell subtypes (left) and lineage groups (right) (D) Expression of key epithelial cell type markers used to define the subtypes in C visualized as a dot plot. Shown are the mean expression level as dot color and fraction of expressing cells as dot size. Bar represents 150 μm (schematics in A and B were created with BioRender).
Perianal fistulizing CD patients have mucosal epithelial cells with changes in WNT signaling and disruptions to their metabolic, epigenetic, and proliferative pathways
Next, we examined the abundance of epithelial cell subtypes amongst the phenotypic and genotypic groups. Cells from each patients’ organoid sample are shown in Fig. 2A color coded by their epithelial subtypes. The relative abundance of these epithelial subtypes for each patient organoid sample are quantified in Fig. 2B along with a comparison across the phenotypes is shown in Fig. 2C. The proportions of absorptive progenitors were significantly different across the ancestral and gender groups (Fig. 2C) but had similar levels of abundance between disease phenotypes. The proportions of the proliferating G2-M/TA subtypes were significantly different between genders (P < 0.05*, 0.01**, Wilcoxon and Anova). Thus, while rectal organoids did not show significant epithelial compositional changes amongst the CD groups, surprisingly they did reflect differences between the individuals’ gender and ancestry.
Patient and disease-based phenotypic differences using single-cell profiles of patient-derived organoids. (A) Patient (n = 13 patients; 8 CD, 5 controls) derived organoid cells were graphed by UMAP and color coded based on patient phenotype. (B) Stacked bar plot visualization of organoid cells showing relative abundance of subtypes for each patient or (C) grouped as disease phenotype, ancestry, and gender. Asterisk next to bar plot indicates the P < 0.05*, 0.01** significance using Wilcoxon and Anova. (D) Pathway enrichment analysis comparing disease phenotypes from mucosal samples and significant pathways were determined by FDR significance with P adjust < 0.05.
Pathway analysis of biological activity using transcriptional data from the organoids of different CD phenotypes indicated significant changes across multiple reactome pathways shown in Fig. 2D and Supplemental Table 1 (FDR significance with P adjust < 0.05). For example, metabolic pathways in organoids from the inflamed group revealed decreased fatty acid and cholesterol metabolic activity, among other metabolic changes. Organoid samples established from inflamed patient mucosa had increased expression of epigenetic pathway genes involving chromatin modification, and an enrichment for Rho GTPase. They also revealed a reduced enrichment of genes linked to pathways involved in proliferation as compared to organoids derived from patients with inactive perianal fistulizing CD or to non-IBD controls. No remarkable signals were noted for inflammatory pathways in the organoids although the inflamed group did show enrichment for NLRP3 related activity.
To further investigate properties of the rectal organoids, we compared them with sequenced mucosal cells from the base of the crypt. To this end, most of the organoids in this study were established from a set of biopsies that were collected at the same mucosal site during a patient’s endoscopy procedure and then processed for scRNA-seq, thus allowing for the patient’s mucosal epithelial cells to be compared to their cultured organoids. The epithelial compartment from the mucosal samples of these patients (141,367 mucosal epithelial cells; n = 29 patients; 18 CD, 11 controls) comprised 18 epithelial cell subtypes (Fig. 3A), ranging from stem cells to fully differentiated goblet, colonocytes, and tuft cells. As expected, integration of mucosal epithelial and organoid cells by rPCA and shown by UMAP conveyed that organoids recapitulated most of the cell types in the lower crypt region (Fig. 3B) but did not produce cells that clustered with fully differentiated cell subtypes including goblet and tuft cells. Consistent with this, crypt axis scoring that used marker genes to define regions of the crypt axis, showed organoids aligned mostly with transcriptional profiles corresponding to the base region of the crypt (Fig. 3C). Mapping the mucosal epithelial and organoid cells to their corresponding epithelial lineages as shown in Fig. 3D revealed that rectal organoids grown with Intesticult™ media produced a high abundance of undifferentiated (stem-like) cells, whereas mucosal epithelial cells consisted of higher proportions of differentiated absorptive cell subtypes.
Mucosal epithelial and organoid cell type differentiation and transcription factor signatures. (A) Mucosal epithelial cells derived from rectal biopsies (n = 29 patients) from the same group of patients from which organoids were established were normalized, scaled, and clustered and then visualized as UMAP. (B) UMAP of rPCA integrated organoid and mucosal epithelial cells at the top, with split mucosal and organoid at bottom with cell lineage color coded. (C) Crypt axis scoring of organoids and mucosal epithelial cells was generated using AddModuleScore(). (D) Relative proportions of epithelial lineage groups in organoids and mucosal epithelial cells is color coded based on lineages. (E) Transcription factor profiles for epithelial lineages measured in mucosa and organoid cells and visualized by violin plot. (F) Real-time PCR analysis of epithelial lineage gene expression in mucosa and organoid from total RNA extracts. (G) Ridge plot visualization of feature numbers per cell for mucosal and organoid samples. Overall samples are denoted by disease phenotype and sample type, “O” for organoids and “M” for mucosa. Control (CTRL), Perianal CD inflamed (PIF), Perianal CD non-inflamed (PNI). On the left shows overall mucosal and organoid data combined. The right top shows the mucosal epithelial cell subtype and right bottom is organoids.
To look at transcriptomic differences under these conditions that might account for the lack of differentiated cell types in rectal organoids, we compared the gene expression levels of several cell-fate regulating transcription factors expressed in organoids and mucosal epithelial cells (Fig. 3E). Organoid transcriptomes lacked signatures for mature secretory (e.g. co-expression of ATOH1 and SPINK4) or absorptive lineages but were like the undifferentiated cell types of the mucosal epithelium showing robust expression for HES1 and the stem marker OLFM4, along with the proliferative marker MKI67. Real time PCR confirmed that rectal organoids expressed comparable levels of HES1 to that of rectal mucosa, but unlike secretory cells of the mucosa, the organoids expressed only minute levels of ATOH1 and SPINK4 (Fig. 3F). We also noticed remarkable differences in the transcription density of unique genes expressed by cells in vivo versus those grown in vitro, with cells in vitro expressing twice the number of unique genes per cell (Fig. 3G). The distinguishing bimodal pattern in unique gene expression that is shown by all but one mucosal sample, is absent from about half of the inflamed mucosal samples (Fig. 3G; IF, inflamed; NI, non-inflamed). However, when analyzed similarly by cell subtypes, the stem cells from the inflamed mucosa show a bimodal pattern, whereas the non-inflamed and control cells are unimodal. Thus, with more unique genes expressed per cell than mucosal epithelial cells, the organoids appeared to have ectopic gene expression that might be associated with imbalances in the WNT/NOTCH pathways and the constitutive HES1 expression keeping organoids in an undifferentiated state.
We continued with a supervised assessment of WNT and NOTCH signaling activity based on known pathway genes that are differentially expressed across organoid phenotypic groups, and between organoid and mucosa cells (Fig. 4, P < 0.05). Inflamed rectal mucosa from perianal CD patients had higher levels of WNT and NOTCH signaling compared to the non-inflamed perianal CD and controls. Higher levels of expressed WNT and NOTCH genes were also associated with organoids obtained from inflamed perianal CD patients, showing increased expression of FBXW11, CTNNB1, CHD8 and PPP2R5A. Inflamed mucosal samples had increased levels of RAC1 expression, however RAC1 expression was highest in the organoids from non-inflamed controls.
Wnt and Notch signaling in mucosal epithelial and organoid cells. (A) Clustered dot plot visualization of supervised comparative analysis based on alignment of WNT and (B) NOTCH pathway genes with differentially expressed genes from mucosal epithelial and organoid cells grouped based on phenotype (P < 0.05).
Rectal organoids replicate metabolic and epigenetic transcriptional signatures associated with undifferentiated epithelial cells from the base of the crypt
To confirm that the signatures detected in organoids were representative of mucosal signatures of disease, we performed pathway analysis on mucosal single cell RNA sequencing samples of undifferentiated epithelial cell populations (base of crypt) that is shown in Fig. 5A and Supplemental Table 2 (P < 0.05 and Log2FC > 0.25 using MAST algorithm). Consistent with our previous study8, active inflammatory disease in the mucosa had a clear impact on the epithelium, affecting transcript levels of genes related to pathways across the spectrum of cellular behavior (Fig. 5A). For instance, in the mucosal epithelial cells, inflammation was associated with changes in transcriptional signatures associated with metabolism and chromatin modification pathways. There were also notable differences in WNT and NOTCH pathways across the organoid disease groups.
Pathway enrichment in undifferentiated mucosal epithelium relative to patient-derived organoids. (A) Comparative analysis between enriched biological pathways generated from significantly differentially expressed genes from mucosal epithelial cells (P < 0.05 and Log2FC > 0.25 using MAST algorithm). Different phenotypic groups are denoted on the top. (B) Comparison of enriched pathways in the undifferentiated mucosal epithelial cells versus organoids. Statistical cutoffs were defined at FDR P adjust < 0.05, with enrichment direction is color coded (blue down regulated, red up regulated). Bidirectional enrichment in A and B represent changes in expression of unique genes belonging to the same pathway.
To assess differences in pathway activity based on transcription profiles between the mucosa epithelial cells and the corresponding patient derived organoids, we calculated pathway enrichment scores shown in Fig. 5B and Supplemental Table 3. The mucosal epithelial cells were enriched for genes associated with TLR and chromatin modification pathways, while the organoids were enriched for IL-1, antigen processing, and TCR signaling. In some instances, there was partial enrichment of a transcript profile associated with pathways when comparing profiles between organoids and mucosa, as in the case for antigen presentation, C-lectin, and cell proliferating pathways scores. This phenomenon is reflected in the bar graph as enrichment in both organoid and mucosa (Fig. 5, red and blue) for the same pathway because different genes in the pathway were significantly enriched in organoids but other genes in the pathway were enriched for mucosa.
Organoid transcription patterns are strongly associated with patient genetics, but not inflammation
We also analyzed psuedobulked organoid single cell data by PCA without integrated mucosal epithelial cells to look for additional similarities and differences across patient organoid samples that we did not detect in our supervised analysis. Results from the unsupervised approach shown in Fig. 6A and Supplemental Table 5 revealed that ancestry and gender contributed the most to variance amongst the patient organoid samples, while patient age and disease status were both significantly correlated with patient organoid gene expression (Spearmen P < 0.05*, 0.01**, 0.001***). There were no associations found with inflammation in the gene expression profiles of the patient organoid samples.
Ancestry and gender related gene expression are primary components of epithelial cell transcriptomes. (A) PCA with psuedobulked organoid single-cell RNA seq data denoted by shape and color of disease phenotype, inflammation, ancestry, and gender. corrplot() visualization of the first 10 principal components are shown on the right with P < 0.05*, 0.01**, 0.001***. (B) Network analysis performed using STRING database on the group of genes driving the variance in each of the first six principal components. (C) PCA with psuedobulked organoid single-cell RNA seq data of subset epithelial cell absorptive progenitor subtype (left) and Microfold-like (right).
To find additional associations and biological implications of the gene expression levels driving variance in each principal component, we entered the top genes from each component (listed in Supplemental Table 5) into the STRING network10 and the analytical outputs making connections within the gene group are shown in Fig. 6B. Considering each PC individually, PC1 is defined by gene expression related to patient genetics that spans several STRING networks involving genes KRT20, DUOX2 and DDX3Y, for example. In PC2, the expression levels of TMSB4Y, PRKY and KDM5D, each related to gender-specific biology, are the primary drivers of the component’s variance. In PC3, the age-related variance correlated with the expression patterns of several homeobox genes, including HOXD9-11 and -13 (Fig. 6B). In PC5, the disease-related gene expression driving variance are MMP7, CXCL5, and CXCL14 levels, along with associated STRING networks involving HLA-DQB1, HLA-DRB1, and HLA-DRB5 (Fig. 6B).
Lastly, we also performed PCA on individual epithelial cell subtypes in the organoid cultures for transcriptional signals that might give further mechanistic insight. Examining the 10 epithelial subtypes found in organoids revealed that gene expression in the absorptive progenitor cells drove most of the observed variance for gender and ancestry, whereas no other significant variance patterns were present in the other subtypes, shown for example are Microfold-like cells (Fig. 6 C and Supplemental Table 4).
- SEO Powered Content & PR Distribution. Get Amplified Today.
- PlatoData.Network Vertical Generative Ai. Empower Yourself. Access Here.
- PlatoAiStream. Web3 Intelligence. Knowledge Amplified. Access Here.
- PlatoESG. Carbon, CleanTech, Energy, Environment, Solar, Waste Management. Access Here.
- PlatoHealth. Biotech and Clinical Trials Intelligence. Access Here.
- Source: https://www.nature.com/articles/s41598-024-75947-4