Single-nucleus multi-omics analysis of mouse small-intestinal Lgr5+ cell populations reveals Foxa3-induced Paneth cell-lineage differentiation

Abstract

Lgr5+ stem cells play crucial roles in maintaining intestinal epithelial cell homeostasis. However, the cellular heterogeneity and underlying regulatory programs of Lgr5+ small intestinal stem cells (ISCs) remain elusive. In this study, we profiled gene expression and chromatin accessibility of Lgr5+ ISCs at single-cell resolution to gain a deeper understanding of the lineage specification and early fate determining mechanisms. Our analysis identified a total of 6 subsets of Lgr5+ cell populations, which exhibited heterogeneity in gene expression and chromatin structure. We found that early fate-determining processes diverged the absorptive and secretory lineages within Lgr5+ cells. We further constructed gene regulatory networks controlling lineage determination and identified Foxa3 as a key transcription factor that regulates the differentiation of the intestinal secretory precursor. In vitro knockdown of Foxa3 disrupted the differentiation of Paneth cells by modulating Peroxisome-Proliferator-Activated Receptors (PPARs). Further Foxa3-targeted CUT&Tag sequencing analysis also verified that Foxa3 predominantly drives Paneth cell differentiation in the small intestine by regulating the expression of core genes in the PPAR signaling pathway. These results provide a comprehensive reference map for advancing our understanding of intestinal epithelial development and related diseases.

Employing single-cell multi-omics and organoid technology, lineage specification and early fate determination of small intestinal stem cells is more deeply revealed, showing that Foxa3 influences Paneth cell differentiation by regulating Peroxisome-Proliferator-Activated Receptors (PPARs).

Introduction

The intestinal epithelium maintains homeostasis by rapidly repairing and regenerating in a complex environment. At the base of the crypt, Lgr5+ intestinal stem cells (ISCs) play a crucial role in the continuous renewal of intestinal tissue^1–3. These stem cells, known as crypt base columnar (CBC) cells, were proposed to generate all differentiated cell types in vivo and in vitro, and they are found intermingled with Paneth cells at the base of the crypt^4,5. Alternatively, the ‘+4 model’ was introduced, suggesting that stem cells are located in a ring of 16 cells positioned immediately above the Paneth cells^6–9. After decades of research, a unified theory incorporating both models’ aspects is emerging. The small intestinal crypt has a pool of distinct stem cells, and the crypt stem cell niche contains phenotypically distinct populations of active and reserve ISCs to maintain homeostasis and regeneration of the small intestinal epithelium. However, the functions and fates of different stem cells in maintaining homeostasis and responding to injury in the small intestine remain unclear^10–12.

The small intestine is composed of six distinct cell lineages that can be classified into absorptive cells (enterocytes and microfold cells) and secretory cells (goblet cells, enteroendocrine cells, tuft cells, and Paneth cells). At the base of the crypt, stem cells undergo rapid proliferation to generate transit-amplifying cells (TA cells). TA cells undergo two to five cycles of division before differentiating into enterocytes, while secretory cells can mature directly from stem cells¹³. Paneth cells and enteroendocrine cells both originate from a slowly dividing small intestinal label-retaining cell (LRC) progenitor^14–16. These LRCs express higher levels of the WNT target gene Lgr5 compared to CBCs¹⁷. The application of single-cell transcriptomics and organoid models has facilitated the identification of signals that regulate the differentiation of stem cells. The determination and differentiation of stem cell fate are driven by dynamic changes in signal pathways such as Notch, WNT, epidermal growth factor (EGF), BMPs, and MAPK^18–22. These signals can be manipulated in organoid cultures to enrich specific cell types, supporting a role for these signals in fate specification²³. However, many of these cell fate regulators, including WNT and Notch, also regulate stem cell proliferation^24,25. Targeting these pathways may therefore pose a risk of cancer transformation or stem cell depletion. Identifying specific regulators and mechanisms controlling intestinal cell fate would facilitate the generation of organ tissues enriched with specific cell types, accelerating the drug discovery process¹⁹.

Specific patterns of gene expression and chromatin state within cells control cellular differentiation. Advances in single-cell genome-wide molecular analysis techniques have facilitated the study of these processes. In recent years, there has been an increase in studies utilizing various single-cell analysis methods to investigate the gut. These studies have successfully characterized the complex epithelial subtypes of the human and mouse small intestine, as well as how cells are organized into different multicellular structural niches. These findings have provided crucial insights into the molecular identity and functional specialization of intestinal epithelial cells. However, most of these studies have focused on intestinal epithelial cells along the small intestinal crypt-villus axis, with limited attention given to heterogeneity within small intestinal stem cells at the single-cell level^26–29. To date, there is no specific profiling of the distinct functional subpopulations of enriched small intestinal Lgr5+ stem cells at the 10x genomics single-cell resolution. Therefore, the regulatory mechanisms that govern the early fate decisions of small intestinal stem cells with diverse differentiation potential remain poorly understood.

In this study, we focused on mouse Lgr5-expressing intestinal cells located at the small intestinal crypts. We enriched these cells and examined their different functional subpopulations, as well as the regulatory mechanisms that govern their behavior, by analyzing gene expression and chromatin accessibility. Our findings revealed that 6 distinct subpopulations of Lgr5⁺ ISCs exhibited significant heterogeneity in terms of cellular stemness and function. We also demonstrated distinct developmental differentiation trajectories for the progenitors of the absorptive and secretory lineages. Notably, we identified Foxa3 as a transcription factor that plays a crucial role in the differentiation of stem cells into secretory precursors, not only participating in the regulation of goblet cells differentiation²⁹ but also modulating the differentiation of Paneth cells. This study reveals a Foxa3-PPAR regulatory axis in Paneth cell differentiation. Our results provide a clear definition of the different functional subpopulations of ISCs and identify specific regulators that control early fate decisions of small intestinal Paneth cell lineage. This study underscores the importance of chromatin and transcriptional landscapes in the differentiation of small intestinal stem cells, while potentially offering insights into the interventions for gastrointestinal epithelial-related diseases.

Results

Cell population landscapes of the mouse small intestine

To elucidate the dynamic changes of cell composition and state in the small intestine during intestinal differentiation, we generated single-cell multi-omics data for a uniform stem cell population, which expressed the cell surface marker leucine-rich repeat-containing G protein-coupled receptor 5 (Lgr5), collected from the small intestinal jejunal region of an adult Lgr5-EGFP reporter mouse. These Lgr5+ stem cells were processed and sequenced using single-nucleus multi-omics (10x Genomics Multiomics), obtaining paired gene expression (messenger RNA) and chromatin accessibility (ATAC) information for the same cell (Fig. 1a, Supplementary Fig. 1a). As the cells are in the process of differentiation, we regressed out the difference between the G2M and S phase cells. This meant that signals separating non-cycling cells and cycling cells would be maintained, but differences in cell cycle phase among proliferating cells would be regressed out of the data. For ATAC data, we performed a second round of peak-calling on all cells together to identify a more accurate set of peaks. After quality filtering, 9189 cells were retained with 32,184 genes and 139,702 distinct peaks.

Fig. 1. Multi-omics profiling of mouse small intestinal epithelial cells.

a Schematic of Lgr5+ cell sorting and multi-omics sequencing. Feature plots showing the expression of Lgr5. b UMAPs showing identified cell types in mouse small intestine epithelial cells using combined gene (mRNA) and chromatin accessibility (ATAC) information. c Dot plot showing the expression levels and the percentages of gene expression of the canonical genes in each distinct cell type. d Trajectory and pseudotime analysis of the small intestinal crypt cells. The trajectories indicate that ISC differentiates in both directions, and the cells are colored by pseudotime. e CytoTrace2 prediction results for the differentiation potential of small intestinal epithelial cells.

We performed weighted nearest neighbor (WNN) analysis using methods implemented in the Seurat software suite, creating a multimodal Seurat object with paired transcriptome and ATAC-seq profiles and performing weighted neighbor clustering on RNA + ATAC data in single cells. The results were presented using uniform manifold approximation and projection (UMAP) (Supplementary Fig. 1b). Based on previously reported canonical cell markers, five main known cell types were identified (Fig. 1b, c): enterocytes (Slc13a1, Slc51a), Paneth cells (Lyz1, Defa26), goblet cells (Zg16, Clca1), enteroendocrine cells (Chgb, Neurod1) and tuft cells (Dclk1, Trpm5). Four of these functional epithelial cells were derived from Lgr5⁺ intestinal stem cells (ISCs) (Lgr5, Olfm4).

A trajectory analysis predicted a bidirectional differentiation from Lgr5+ stem cells into enterocytes and Paneth cells (Fig. 1d). This trajectory recapitulated cell fate determination and cell differentiation from Lgr5⁺ stem cells, as evidenced by lineage-tracing experiments performed in adult mice^4,19. The ISCs located at the base of the crypt reside within a permissive niche that fosters their identity. As these cells move upward and exit the niche, there is a gradual decrease in their stemness, ultimately leading to their differentiation into absorptive or secretory cell lineage cells (Fig. 1e). Trajectory analyses further indicated that ISCs underwent differentiation into progenitor cells via various pathways from a common origin, subsequently developing into functional cells with distinct characteristics.

Heterogeneity of Lgr5⁺ intestinal stem cells

Lgr5 has been identified as a highly specific ISC marker⁴. However, our data indicated that Lgr5 was expressed in a wide range of intestinal cell types (Fig. 1a, Supplementary Fig. 2a, c). The findings of recent studies suggest that Lgr5 is not a reliable indicator of stem versus differentiated cells and that additional stem cell populations exist beyond the crypt base³⁰. Moreover, diversities were identified between Lgr5-high and Lgr5-low stem cells. The cells with the highest levels of Lgr5 expression did not exhibit the highest stem cell potential, whereas the highest cellular potency was observed in some of the Lgr5-low cells^30,31. These findings collectively indicate cellular heterogeneity among small intestinal stem cells.

To fully elucidate the heterogeneity of small intestinal stem cells, we selected 7900 ISCs (86%) within the integrated dataset based on the selection of canonical marker Lgr5 for the subsequent definition of cell identity and study of lineage specification. The multi-omics datasets provided us the opportunity to delineate the identities of ISCs based on a more biologically regulatory status, integrating both transcriptomic and chromatin accessibility profiles. Analysis of single-omics data alone could not distinguish all stem cell subtypes (Supplementary Fig. 3), but the integrated analysis of mRNA and ATAC data allowed the clear identification of six subpopulations of ISCs. The annotation of each of the six subgroups was based on the gene expression levels of marker genes and the activity scores of marker gene sets, as well as the stemness differences of the cells. The six subgroups were thus named Lgr5-low stem cells (Lgr5-low SCs), absorptive progenitor cells (AP), Top2a+ transit-amplifying cells (Top2a+ TA cells), Hmgb2+ TA cells, crypt base columnar cells (CBCs), and Lgr5⁺ label-retaining cells (Lgr5⁺ LRCs) (Fig. 2 and Supplementary Fig. 4).

Fig. 2. Heterogeneity of cell stemness, gene expression, and their functions in Lgr5⁺ ISCs.

a UMAPs showing identified cell types in ISCs, sing combined gene (mRNA) and chromatin accessibility (ATAC) information (86% Lgr5⁺ small intestinal crypt cells). b CytoTrace2 prediction results for the differentiation potential of small intestinal stem cells. c Heat map showing the expression of the top 10 enriched genes in each cluster. d Gene Ontology Enrichment Analysis of genes characterizing AP cells was enriched in processes associated with the differentiation of columnar epithelial cells. The adjusted p-values were corrected with the Benjamini–Hochberg method. e The feature gene set of nonPaneth_LRCs activity AUC score cluster distributions.

The marked differences in function and stemness of these diverse cell subtypes indicate the presence of significant heterogeneity among ISCs (Fig. 2b, c, Supplementary Fig. 5). CytoTRACE2 was used to infer cell stemness among the six identified stem cell clusters, revealing that Lgr5-low-expressing cells may possess relatively higher developmental potential. Besides, secretory progenitor cells (Lgr5⁺ LRCs) and APs exhibited relatively lower stemness (Fig. 2b). Notably, Lgr5 was differentially expressed across distinct ISC subpopulations. These findings further support the lack of a strict correlation between Lgr5 expression levels and inferred developmental potential (Supplementary Fig. 2b, d). Meanwhile, this is consistent with recent studies that have detected the highest levels of cellular potency in some Lgr5-low cells^30,32. The other four types of stem cell subtypes can be divided into two broad categories: absorptive lineage progenitors and secretory lineage progenitors. AP cells mainly expressed genes related to small intestinal epithelial cell differentiation, including Gata4, Klf5, and Fgfr2 (Fig. 2c, d). In contrast, Lgr5⁺ LRCs have been observed to overexpress a number of markers associated with small intestinal quiescent cells located at the base of the crypt. These include markers of quiescent stem cells (Lrig1, Hopx), a stem cell marker (Lgr5), and a marker of the secretory lineage (Sox9) (Fig. 2e). Furthermore, Lgr5⁺ LRCs were predominantly in the G1/G0 phase (Supplementary Fig. 6), which is consistent with the previously described intestinal quiescent cells LRC. This is a secretory progenitor that co-expresses stem cell and secretory lineage markers^14,17,33. Additionally, the highest expression of Lgr5 was observed in Lgr5⁺ LRC, which exhibited a higher level of expression than TA cells (Supplementary Fig. 2d). This finding is consistent with previous research indicating that TA cells are Lgr5-negative³⁴ and that Lgr5^hi cells are predominantly found in clusters of cells associated with secretory progenitor cells³⁰. The heterogeneity that exists between these ISCs is consistent across the human small intestine. Independent validation was performed using single-cell transcriptome data from the human jejunum as published by Hickey, John W. et al.²⁹. In this data set, the marker for LRC (TERT) and the marker for AP (PCSK5) were found to be highly expressed in immature goblet cells and immature enterocytes, respectively. It was also observed that Lgr5 was co-expressed with ASCL2 (a marker for CBC), TERT (a marker for the secretory lineage precursor LRC), but was not expressed in the TA or immature enterocytes. Furthermore, within the stem cell populations, a distinct group of small intestinal stem cells exhibited low LGR5 expression and comparatively high expression of the Lgr5-low SC subpopulation markers (FOS, JUN, JUNB, IER2) (Supplementary Fig. 7).

We performed trajectory analyses to predict the differentiation between these ISCs (Fig. 3a). This trajectory was likely to recapitulate cell fate determination from ISCs when the cells left the stem cell niche, as reported in previous studies¹³. Trajectory analyses showed a continuous route between Top2a+ TA cells and AP cells, suggesting Lgr5-low SCs first differentiate into TA cells, which then divide and mature into absorptive lineage cells. Meanwhile, another route showed that Lgr5-low SCs differentiated directly into Lgr5⁺ LRCs and then matured into secretory cells (Fig. 3b).

Fig. 3. Transcriptional and chromatin accessibility profiles of key regulators affecting differentiation between Lgr5⁺ ISC subtypes.

a Trajectory and pseudotime analysis of the ISCs. b Models of differentiation between Lgr5⁺ ISC subtypes. c Enrichment of the peak signals of all ISCs was observed around the transcription start site (TSS). d Distribution of accessible ATAC peaks of ISCs in different regions of the genome. e Heat maps highlighting gene expression and ATAC motif accessibility of active transcription factors co-enriched with both features in Lgr5-low SC, AP, and Lgr5⁺ LRC. The colors indicate the level of TF expression (left) and the degree of opening of the chromatin region corresponding to the motif (right), respectively.

Key regulatory elements in the fate determination of small intestinal differentiation

To delve into the mechanisms underlying early fate determination in small intestinal differentiation, we conducted a comprehensive cross-referencing of gene expression data with information on chromatin accessibility associated with transcription factor binding motifs. This approach enabled us to identify active transcription factors during lineage differentiation that exhibit both high expression levels and the capacity to access their target binding motifs, thereby promoting the transcription of other genes.

At the transcriptome level, we analyzed genes that were altered as the cell progressed along the trajectory and organized these trajectory-variable genes into modules. It was observed that different cellular subtypes exhibited differential expression of distinct cellular modules (Supplementary Fig. 8). Subsequently, we constructed enhancer-driven gene regulatory networks (eGRNs) using the pySCENIC method to infer the transcription factors that may regulate these trajectory-variable genes. Here, we identified 113 active transcription factors that are specific to each stem cell subtype (Supplementary Fig. 9 and Supplementary Data 1 and 2). Among them, highly active transcription factors in small intestinal absorptive lineage precursor cells included Gata4³⁵, Onecut2³⁶, and Hnf4a³⁷, which have been previously demonstrated to play crucial roles in the differentiation of enterocytes. In contrast, Lgr5⁺ LRC have co-enrichment of Foxa3, Ets2, Mga, Ehf, Gata6. These results demonstrate the heterogeneity of the small intestinal stem-cell differentiation regulation.

For the ATAC data, all peaks are enriched near the TSS region (Fig. 3c), of which 80.71% were located within the annotated gene regions, including the proximal part of the promoter (−5 to −0.5 kb from the TSS), the UTR region, non-TSS exons and introns, and downstream (Fig. 3d). The peaks located in gene regulatory regions were then scored for transcription factor motifs using the HOCOMOCO database. Here, we highlight the top 10 identified active transcription factor motifs enriched in specific intestine stem cells (Supplementary Fig. 10 and Supplementary Data 3).

To ensure consistency between the RNA data and the ATAC data, we performed a correlation analysis of the transcription factors and motifs specific to each subtype and obtained seven transcription factors that were highly correlated with small intestinal differentiation (Fig. 3e). Among the TFs examined, Hnf4a was found to be up-regulated during the differentiation of small intestinal absorptive lineage cells (AP and Top2a+ TA cells), while Foxa3, Nfia, Sox9 and Tcf7l2 were highly expressed during the differentiation of small intestinal stem cells to secretory lineage cells (Lgr5⁺ LRC). Conversely, Klf3 and Klf4 were highly expressed in the Lgr5-low SCs of small intestinal stem cells, indicating that they may play a role in sustaining the stemness of small intestinal stem cells. The overall expression level of the Hnf4a regulon in each cell was found to be positively correlated with the degree of chromatin opening of the Hnf4a motif (Supplementary Fig. 11c) during the differentiation of small intestinal stem cells. The opening of Hnf4a motifs was markedly elevated in the AP (Supplementary Fig. 11b), concomitant with a notable enhancement in the specificity of the Hnf4a regulon in the AP (Supplementary Fig. 11a). This is consistent with previous studies indicating that inhibition of the HNF4-SMAD4 module leads to abnormal differentiation of enterocytes, favoring the development of secretory progenitor cells³⁷. Furthermore, Hnf4a’s target genes are involved in processes such as the positive regulation of gene expression and SH3 domain binding (Supplementary Fig. 11d). This suggests that Hnf4a may influence gene expression during absorptive lineage cell differentiation by recruiting transcription-regulatory proteins. This aligns with the conclusion that HNF4 drives intestinal cell differentiation by regulating gene expression through modulating RNA polymerase II dynamics, thereby affecting chromatin accessibility^38,39.

Integrative analysis identifies Foxa3 as a putative regulator of Paneth cell differentiation

Since the role of HNF4A in the differentiation of absorptive lineage cells has been demonstrated, we then investigated the secreted lineage-specific transcription factor Foxa3.

FOXAs are the subfamily of winged helix/forkhead box (Fox) transcription factors that regulate development and metabolism in various systems⁴⁰, including the embryo, liver⁴¹, lung⁴², and prostate⁴³. In particular, FOXAs interact with several important signaling pathways in regulating the differentiation fate of small intestinal stem cells, including WNT⁴⁴ and Notch.

In our experimental data, the transcriptional status and expression of Foxa3 were highly consistent across subpopulations. When pooling all ISC cells within a single cell type, the overall expression of the Foxa3 gene regulatory module and the degree of chromatin accessibility at the Foxa3 motif exhibit a high correlation (R = 0.99). In contrast, no such significant correlation was observed within every individual ISC cell-subtype, suggesting that dynamic variations in Foxa3 gene expression and chromatin accessibility between ISC subtypes are primarily driven by inter-lineage differences rather than intra-lineage differences. (Fig. 4a, Supplementary Fig. 12). The AP subpopulation barely expressed the Foxa3 gene compared to the Lgr5-low SCs, whereas the expression was increased in the Lgr5⁺ LRC subpopulation (Fig. 4c). At the ATAC level, the openness of the Foxa3 motif was decreased in AP and increased in Lgr5⁺ LRC (Fig. 4d). Meanwhile, the overall openness of Foxa3 and its regulatory genes was higher in Lgr5⁺ LRC and mature secretory lineage functional cells than in AP (Fig. 4b).

Fig. 4. Foxa3 and its regulators exhibit heightened activity during the differentiation of intestinal secretory lineage cells.

a Pearson’s correlation analysis of the total expression of the Foxa3(+) regulon and the mean open scoring of the Foxa3 motif across ISC subpopulations. The sequence of the Foxa3 motif is shown at the top left of the figure. b Enrichment of ATAC peaks of different cell subtypes around the TSS of Foxa3 and its regulatory genes. Secretory combined peak signals of pre-Paneth, Paneth, goblet, and enteroendocrine cells. c Violin plots of Foxa3 expression levels in ISC subpopulations. The adjusted p-values for the significance test of differences were analyzed using the t-test and corrected with the Benjamini–Hochberg method. d Violin plots of Foxa3 motif scoring across ISC subpopulations. The adjusted p-values for the significance test of differences were analyzed using the Wilcoxon test and corrected with the Benjamini–Hochberg method. e Results of multiple sequence alignment of Foxa3 proteins from five species. f Schematic representation of the domains contained in the Foxa3 proteins of the five species.

In addition, the conservation of Foxa3 proteins across species was determined through protein multiple sequence alignment. A highly conserved sequence of ∼100 amino acids is present in the Foxa3 proteins of five species (Fig. 4e). This conserved sequence is the forkhead structural domain (FH), which is the wing-helix DNA-binding domain in forkhead box protein A3 (FOXA3) and analogous proteins, including the Drosophila melanogaster protein forkhead (dFKH) (Fig. 4f). It has been demonstrated that the knockdown of FoxA/Fkh in the Drosophila intestine restores Notch signaling and inhibits the aberrant proliferation of stem cells by directing differentiation of stem/progenitor cells into the enterocyte lineage. This study also postulates a potential role for FOXAs in enteroendocrine cell differentiation⁴⁵. Another study on chronic lung disorders accompanied by goblet cell metaplasia demonstrated that FOXA3 was highly expressed in airway goblet cells of asthma and COPD patients. Meanwhile, FOXA3 induced goblet cell metaplasia and enhanced the expression of a network of genes mediating mucus production⁴².

In conjunction with previous studies, we propose that Foxa3 binds to specific sites within open chromatin regions, such as the promoters and other regulatory domains of target genes. In this location, it engages in a collaborative process with other proteins, thereby facilitating the transcription of genes that are instrumental in the differentiation of small intestinal secretory lineage cells^46–48. It is noteworthy that FOXAs exhibit highly similar DNA recognition motifs, and Foxa1/2 have been demonstrated to regulate the differentiation and maturation of small intestinal secretory lineage cells. The number of mature Goblet cells, enteroendocrine L cells, and D cells was diminished in Foxa1/2 conditional knockout mice⁴⁹. To further confirm the specific role of Foxa3 in the differentiation of secretory lineage cells, the expression levels of Foxa family genes in different stem cell subpopulations were examined. Gene expression and chromatin accessibility of the corresponding motifs of the Foxa family were higher in secretory lineage cells (Enteroendocrine, Goblet, Paneth cells). In particular, Lgr5⁺ LRC, which was previously suggested to be a Paneth cell precursor population¹⁴, also predominantly expressed high levels of Foxa3 (Supplementary Figs. 13, 14). Consequently, the hypothesis was formulated that Foxa3 is more likely to play a key role in regulating the differentiation of Paneth cells. In humans, the expression of the FOXA family is tissue-specific, predominantly observed in tissues associated with digestion, respiration, and the urinary system, which are derived from the endoderm. In gastrointestinal tissues, FOXA3 expression levels were found to be higher in comparison to FOXA1/2 (Supplementary Fig. 15a). At the single-cell level, FOXA1 is widely expressed in the jejunum region, FOXA2 is expressed in Goblet and Paneth cells, and FOXA3 is expressed in a subset of small intestinal stem cells as well as in Goblet and Paneth cells (Supplementary Fig. 15b)²⁹. These results are consistent with our inference. In conclusion, our findings suggest that Foxa3 may play a pivotal role in regulating the differentiation of Paneth cells.

Knockdown of Foxa3 disrupts the differentiation of Paneth cells

To delve into the role of Foxa3 in intestinal development, we employed the Wnt signaling agonist CHIR-99021 to establish intestinal organoids enriched with intestinal stem cells and implemented Foxa3 knockdown (Fig. 5a). Subsequently, we transferred intestinal organoids from both the control and Foxa3 knockdown groups into a differentiation culture system, to observe the specific effects of Foxa3 knockdown on the differentiation of intestinal stem cells. In order to identify transcriptional changes in Foxa3 knockdown groups, differential gene expression analysis was carried out using edgeR against the control group, with thresholds of |log2FC| > 0.5 and adjusted p-values < 0.05. The results obtained revealed a significant decrease in the expression of Paneth cell markers (i.e., Mmp7 and Ang4) and goblet cell markers (i.e., Zg16, Tff3, Ffar4, and Reg4) in Foxa3 knockdown groups (Fig. 5b). Furthermore, motif analysis of the 3 kb upstream region (candidate promoters) of the transcription start sites (TSS) of downregulated genes in Foxa3 knockdown groups revealed the enrichment of transcription factors such as Foxa3 and Sox9 (Fig. 5c), suggesting that these downregulated genes were direct targets of Foxa3. To further validate whether Foxa3 knockdown suppresses the differentiation of goblet cells and Paneth cells, we performed RT-qPCR assays and found that the mRNA expression levels of goblet cell markers (i.e., Reg4 and Tff3) and Paneth cell marker Mmp7 were reduced following Foxa3 knockdown (Fig. 5d). The results indicate that Foxa3 may regulate the expression of genes involved in Paneth cell and goblet cell differentiation by binding to regulatory regions, such as promoters of target genes.

Fig. 5. Knockdown of Foxa3 impairs the differentiation of intestinal Paneth cells.

a Schematic of experimental protocol for intestinal organoid treatment. b Volcano plot of differentially expressed genes in Foxa3 knockdown small intestinal organoids in comparison with normal small intestinal organoids. c Homer motif enrichment analysis identified a Foxa3 binding motif for downregulated genes in the Foxa3 knockdown group. d Quantative PCR analysis of indicated genes in control and shFoxa3 organoid. Data were presented as mean ± s.d (n = 3). *P < 0.05; ***P < 0.001. e Immunostaining of E-cadherin (ECAD) and KRT20 in control and shFoxa3 organoid. Scale bar, 50 μm. f Immunostaining of E-cadherin (ECAD) and MMP7 in control and shFoxa3 organoid. Scale bar, 50 μm. g Quantifcation of IF staining for KRT20 and MMP7 in control and shFoxa3 organoid (n = 6). Student’s t-test: *P < 0.05.

To further validate these findings, we collected organoid samples for sectioning and conducted immunofluorescent experiments. The results showed an increase in KRT20-positive intestinal epithelial absorptive cells and a decrease in MMP7-positive Paneth cells (Fig. 5e–g). Together, these results suggest that Foxa3 is an important regulator in controlling intestinal Paneth cell differentiation.

Foxa3 regulates Paneth cell differentiation by modulating PPARs

In order to investigate the specific regulatory mechanisms by which Foxa3 regulates Paneth cell differentiation, KEGG functional enrichment analysis was performed on genes that were significantly downregulated in Foxa3 knockdown groups. It was found that Foxa3 knockdown led to the suppression of genes associated with the PPAR signaling pathway. This included core transcription factors within the pathway (Ppara, Ppard, and Pparg), as well as their downstream target genes (Fig. 6a, b and Supplementary Data 4). Furthermore, the 3 kb upstream of the TSS (candidate promoters) of downregulated genes also contained motifs for the Ppara, Ppard and Pparg transcription factors (Fig. 6c). Protein–protein interaction (PPI) analysis further revealed regulatory relationships among Foxa3, the PPAR signaling pathway, and Paneth and goblet cell marker genes (Fig. 6d). The present findings suggest that Foxa3 may regulate Paneth cell differentiation by modulating the expression of PPARs and their target genes.

Fig. 6. The regulatory relationship between Foxa3 and the PPAR family genes at the transcriptomic and epigenomic levels.

a Heatmap of differentially expressed genes in Foxa3 knockdown small intestinal organoids in comparison with normal small intestinal organoids (n = 2). b KEGG functional enrichment analysis results for downregulated genes in the Foxa3 knockdown group. c Homer motif enrichment analysis identified PPARs binding motifs for downregulated genes in the Foxa3 knockdown group. d Protein interaction networks for Foxa3, PPAR signaling pathways, and Paneth and goblet cell markers from the STRING database. e Gene tracks showing CUT&Tag-seq data for Foxa3 (n = 2). S1: small intestinal organoid 1; S2: small intestinal organoid 2. f Homer motif enrichment analysis results for peaks in the CUT&Tag-seq data. * from NCBI Chip-seq experiments. g Model of Predicted Transcription Factor Binding Sites in the upstream regions of Foxa3. h Gene tracks showing CUT&Tag-seq data for genes in the PPAR signaling pathway. i Model of predicted transcription factor binding sites in the upstream regions of Ppara.

In order to strengthen this claim and identify the specific binding sites of Foxa3, Foxa3-targeted CUT&Tag sequencing was performed on two small intestinal organoids. Firstly, an analysis of all peak sequences in the CUT&Tag data was conducted to reveal significant enrichment of Foxa3 and other Foxa family transcription factors (Fig. 6f). Concurrently, both samples demonstrated binding peaks of Foxa3 protein within the Foxa3 promoter region (Fig. 6e). Transcription Factor Binding Sites (TFBS) prediction analysis of the 3-kb upstream region of the Foxa3 gene further confirmed the presence of the Foxa3 binding motif “GTAAACA” (Fig. 6g). These findings not only validate the efficacy of the CUT&Tag targeting experiment, but also indicate that Foxa3 transcription and regulation plays a crucial role in the differentiation and development of small intestinal stem cells. Subsequently, a comparison was made between the peaks from CUT&Tag data and the chromatin-open peaks from scATAC-seq data, in order to ascertain the consistency between them. The results demonstrated that, at the gene level, all genes exhibiting Foxa3 protein-binding peaks within their coding sequences and regulatory regions also displayed chromatin-open peaks in their regulatory regions within scATAC-seq data (Supplementary Fig. 16a). Peak read density (reflecting peak signal intensity) at identical genomic positions across both datasets showed positive correlation (R = 0.51, p-value < 2.2 × 10⁻¹⁶) (Supplementary Fig. 16b).

In accordance with the preceding analysis, CUT&Tag peak signals were screened within the 3-kb upstream region of the TSS (candidate promoters) for goblet cell marker genes to identify specific Foxa3 binding sites. The results indicated the presence of Foxa3 protein binding peaks in the upstream region of the TSS of the Tff3 gene (Supplementary Fig. 17). This finding is consistent with the conclusions of previous studies, which have identified Foxa3 as a pivotal regulatory transcription factor in the differentiation of human intestinal goblet cells²⁹.

We further elucidated the potential epigenetic interactions between Foxa3 and PPARs by screening CUT&Tag data within the 3 kb region upstream of the transcription start sites (TSS) of PPAR genes. The results obtained revealed peaked signals in the upstream regions of multiple PPAR pathway genes (including Ppara, Ppard, Hmgcs1, and Fabp5), indicating Foxa3 protein binding at candidate promoter regions (Fig. 6h). The upstream regions of these genes, measuring 3 kb, were extracted for the purpose of predicting TFBS. The results obtained from the JASPAR database confirmed the presence of the Foxa3-specific binding motif “GTAAACA” within the 3 kb upstream regions of these genes (Fig. 6i, Supplementary Fig. 18). Motif enrichment analysis of the 3 kb upstream regions of all PPAR signaling pathway genes similarly revealed significant enrichment of Foxa family transcription factors, including Foxa3 (Supplementary Fig. 19).

The aforementioned results confirm the regulatory relationship between Foxa3 and the PPAR family genes at both the transcriptomic and epigenomic levels. It has been demonstrated by previous studies that Ppard promotes Paneth cell differentiation^50,51. In line with this, miR-802 knockout mice exhibited elevated numbers of Paneth cells and upregulation of Ppard, a known positive regulator of Paneth cell differentiation and maturation⁵². Collectively, these findings indicate a close association between PPARs and Paneth cell differentiation. Consequently, it can be concluded that Foxa3 predominantly promotes Paneth cell differentiation in the small intestine by regulating the expression of core genes in the PPAR signaling pathway, particularly members of the PPAR gene family (Ppard).

Discussion

Crypt-based Lgr5⁺ ISCs represent a crucial source of maintenance of small intestinal epithelial homeostasis and renewal of epithelial tissue following injury in vivo. It appears that these cell populations, which demonstrate differentiation and regeneration capabilities, are diverse. This raises several questions regarding the specific contributions of distinct cell types to the maintenance of homeostasis and the repair of injured tissue. The current published researches generally favor the use of two pools of intestinal crypt-based ISCs during homeostasis in vivo: actively proliferating ISCs (aISCs) and reserve ISCs (rISCs). aISCs are identified by Lgr5 expression and exhibit relative homogeneity, contributing to the daily turnover of epithelial cells. In contrast, rISCs engage in regenerative processes following tissue injury or extreme physiological stress^10,53.

Until recently, the identification of aISCs and rISCs was largely based on the presence or absence of a single or a small number of biomarkers (e.g., Lgr5, Bmi1). This approach of defining specific cell populations on the basis of biomarkers and lineage tracing has certain inherent limitations. For example, Lgr5, which was previously believed to be expressed exclusively in aISCs, has also been observed in +4 rISCs and TA progenitors^14,16. As a result, a more sophisticated map of the heterogeneity of small intestinal stem cells is needed. It is also important to clarify the contribution of different stem cell subpopulations to small intestinal differentiation and regeneration, as well as the specific molecular mechanisms involved. To this end, we analyzed the transcriptional profiles and chromatin accessibility of small intestinal crypt cells expressing Lgr5 using single-nucleus multi-omics sequencing.

Our single-nucleus multi-omics approach offers a more reliable definition of cell types than either data type alone, and enables the identification of the processes involved in the early fate determination of small intestinal stem cells. The present study maps small intestinal Lgr5⁺ stem cells at the single-cell level, identifying six distinct functional subpopulations and their characteristic genes. The developmental trajectory extrapolation demonstrated that ISCs can either differentiate initially into TA cells and subsequently into absorptive precursor cells or directly into secretory lineage progenitors, Lgr5⁺ LRCs. This outcome aligns with the findings of Basak, O. et al., who reported that TA cells undergo two to five divisions before maturing into intestinal epithelial cells, whereas secretory cells can mature directly from stem cells¹³. Our study validates and refines this process at the single-cell level, thereby providing a more accurate understanding of the underlying mechanisms.

The process of early fate determination of small intestinal stem cells remains unclear, as the available single-cell data on intestinal epithelial cells does not allow for a sufficient number of Lgr5⁺ ISCs to be identified. The present study employs single-cell multi-omics analysis of enriched Lgr5⁺ ISCs to define ISC subtypes with heterogeneous differentiation potential and progenitor cells with unidirectional differentiation potential (AP and Lgr5⁺ LRC), inferring differentiation trajectories among various ISC subtypes. Notably, Lgr5 expression levels do not show a strict correlation with developmental stemness among these ISC subtypes. This is consistent with previous studies that ISC subtypes with Lgr5-low expression have the relatively highest level of cellular potency^30,32.

A further analysis of key regulatory factors influencing the bidirectional differentiation of small intestinal stem cells was also conducted. The results demonstrate that Hnf4a regulation is a driving force behind the differentiation process in the absorptive lineage cells, which is in accordance with the findings of existing studies³⁷. Notably, this study has identified a critical role for Foxa3 in the differentiation process of secretory lineage progenitor cells. The overall expression levels of Foxa3 and its regulatory genes, as well as their chromatin opening, are elevated during ISC differentiation to secretory lineage progenitors. In vitro organoid experiments also revealed that the knockdown of Foxa3 disrupted the differentiation of Paneth cells. While previous studies have demonstrated the relevant role of the Foxa family in secretory cell differentiation in different tissues of different species^42,45,49, our study identifies the developmental significance of Foxa3 in mammalian small intestinal tissues.

Furthermore, we investigated the regulatory mechanism of Foxa3 in the initial stages of Paneth cell differentiation. Knockdown of Foxa3 downregulates the expression of genes associated with the PPAR signaling pathway, particularly the core transcription factors PPARs. The upstream regions of the PPARs gene TSS (candidate promoters) contain Foxa3-specific transcription factor binding sites and exhibit binding peaks for the Foxa3 protein. Existing studies have established that the PPAR gene family plays critical roles in the differentiation and physiological functions of Paneth cells in the small intestine. Specifically, Ppard has been shown to promote Paneth cell differentiation by regulating the Hedgehog signaling pathway⁵⁰. In aged Paneth cells, reduced Ppara activity leads to increased Notum expression, thereby impairing the regenerative capacity of senescent intestinal epithelial cells⁵⁴. Mice deficient in Pparg exhibit decreased expression of antimicrobial peptide-encoding genes and increased bacterial colonization between villi⁵⁵. However, the precise mechanisms by which PPARs and their associated regulatory factors form complex regulatory networks to govern Paneth cell differentiation remain unclear. The present study, focusing on the level of transcription factors, has identified Foxa3 as a potential upstream regulator of PPARs in Paneth cell differentiation, advancing our molecular understanding of this process.

Concurrently, the present study further confirms the role of Foxa3 in goblet cell differentiation, which is consistent with previous findings²⁹. Although both goblet cells and Paneth cells are classified as members of the small intestinal secretory lineage, notable differences exist between them in terms of physiological localization, specific functions, and differentiation regulatory mechanisms. Existing studies have established that these two cell types are regulated by shared transcription factors, which in turn activate distinct downstream transcription factors to modulate their respective differentiation and functions. For instance, ATOH1 is a pivotal initiator that specifies all secretory cell lineages in the small intestine: it drives goblet cell differentiation and function by activating SPDEF, while regulating Paneth cell differentiation through the activation of SOX9¹⁹. However, the specific molecular regulatory mechanisms governing the differentiation of small intestinal stem cells into these two secretory cell types remain poorly understood. Our study has identified Foxa3 as a transcription factor that simultaneously regulates both goblet and Paneth cell differentiation. Moreover, we found that Foxa3 modulates Paneth cell differentiation by regulating the expression of PPARs. The existing body of research suggests that PPARs appear to be dispensable for goblet cell differentiation and functional execution. Consequently, further elucidation is required to determine the precise role of Foxa3 in goblet cell differentiation. In summary, our integrated analysis identifies Foxa3 as a transcription factor enriched in secretory progenitor cells, capable of modulating marker gene expression during Paneth cell and goblet cell differentiation.

Over the past three decades, the incidence of IBD has been increasing worldwide, including in China⁵⁶. Despite extensive research, the cause of the disease remains unclear, and no effective therapeutic solution has yet been found for the eradication of IBD in the clinic⁵⁷. It has been demonstrated that the pathogenesis of IBD is accompanied by aberrant differentiation of intestinal secretory lineage cells. The inhibition of Atoh1 expression results in the impaired differentiation of goblet cells, which in turn leads to the loss of the colonic mucus layer and the subsequent development of ulcerative colitis (UC)⁵⁸. Similarly, the promotion of intestinal epithelial cell differentiation into Paneth cells has been demonstrated to reduce the severity of Crohn’s disease (CD)⁵⁹. The discovery of the role of Foxa3 in the differentiation of intestinal secretory cells may provide new insights into the clinical treatment of IBD.

In conclusion, the utilization of single-cell transcriptomics and single-cell chromatin accessibility co-assays of Lgr5-enriched small intestinal stem cells facilitated an investigation of the diverse populations of Lgr5⁺ ISCs and their distinctive attributes. The stem cell atlas allowed us to identify Foxa3 as a pivotal regulator in the early Paneth cell fate decision of small intestinal stem cells. Furthermore, the CUT&Tag technology revealed that Foxa3 promotes Paneth cell differentiation by regulating the expression of PPARs.

Methods

Mouse

Lgr5-EGFP-IRES-CreERT2 (C57BL/6) mice were purchased from the Jackson Laboratory (Strain no. 008875). Lgr5⁺ small intestinal cells were isolated from the 12-week-old male mice of this strain. Animal experiments were performed following relevant guidelines and approved by the Animal Care and Use Committee of Fudan University (Ethics approval number: IDM2024018). All mice were housed in the SPF (Specific-Pathogen-Free) animal facility with standard 12 h light/dark cycles and standard temperature (22–24 °C). Mice were provided with ad libitum access to standard laboratory food and water. We have complied with all relevant ethical regulations for animal use.

Isolation of Lgr5⁺ small intestinal cells

To isolate Lgr5⁺ small intestinal cells from mice, the following steps were performed: First, intestines from Lgr5-GFP mice were excised, dissected, and cleaned with cold PBS, then longitudinally cut, washed twice with cold PBS, and the villi were carefully scraped off using an operating scalpel. The remaining tissue was cut into 5 mm pieces and incubated in PBS containing 10 mM EDTA on ice for 30 min. After removing EDTA, the tissue pieces were vigorously suspended with cold PBS using a 10-ml pipette, and the crypt-enriched supernatant was passed through a 70 μm cell strainer followed by centrifugation at 600 rpm for 3 min to obtain pure crypts. Small intestinal crypts were digested using TrypLE (GIBCO, A1217701), and Lgr5⁺ cells were directly isolated and sorted by flow cytometry for downstream analysis.

Cell dissociation and library preparation

Single Nucleus Multi-Omics sequencing (RNA + ATAC sequencing) was conducted on Lgr5⁺ small intestinal cells derived from an Lgr5-EGFP-IRES-CreERT2 mouse⁴. Briefly, small intestinal crypts were isolated and digested, and GFP+ cells were sorted in PBS buffer via BD Fortessa flow cytometer. The isolated nuclei were processed immediately following the Chromium Next GEM Single Cell Multiome ATAC + Gene Expression protocol. A final 4700 nuclei/µL was used for the Chromium Next GEM Single Cell Multiome ATAC + Gene Expression Reagent Bundle (10x Genomics), following the CG000338 user guide for transposition, GEM generation, ATAC, and Gene Expression library construction. A total of 17,390 nuclei were loaded on Chromium. Libraries were quantified using the KAPA Library Quantification Kit and sequenced using the Illumina NovaSeq 6000 platform.

Processing and filtering of multi-omics sequencing data

Multi-omics sequenced files were processed for demultiplexing and analyzed using Cell Ranger Software Suite (10x Genomics Cell Ranger ARC v2.0.2). Genes were mapped and referenced using refdata-cellranger-arc-mm10-2020-A-2.0.0 as a reference to map reads on the mouse genome (GRCm38/mm10)⁶⁰. R4.2.2 was used to perform analyses. Datasets were analyzed using Seurat 4.3.0⁶¹ and Signac 1.9.0⁶². The Cell Ranger outputs were imported into Seurat by the ‘Read10X_h5’ function. Low-quality cells including doublets, dead cells and poor sequencing depth cells were removed by filtering out cells with low RNA counts (nCount_RNA < 2000) and low ATAC counts (nCount_ATAC < 1000); high RNA counts (>50,000) and high ATAC counts (>100,000); High mitochondrial ratio (>20); nucleosome signal > 2 and transcription start site (TSS) enrichment < 2. For the gene expression data, ribosomal genes were removed, and cell cycle genes were scored using ‘CellCycleScoring’. For ATAC data, peaks were called using MACS2⁶³, and the genomic positions were mapped and annotated with reference mouse genome EnsDb.Mmusculus.v79 (10.18129/B9.bioc.EnsDb.Mmusculus.v79) and mm10^60,64.

Weighted nearest neighbor (WNN) and unsupervised clustering

We next perform pre-processing and dimensional reduction on both assays independently, using standard approaches for RNA and ATAC-seq data. Gene expression data were processed with SCT transformations, and regressions of the difference between sequencing depth, mitochondrial proportion, and G2M and S phase scores were performed to remove these confounders. The principal components (PCs) were estimated by ‘RunPCA’. ATAC data were normalized and scaled using ‘RunTF-IDF’, and ‘FindTopFeatures’ was used to identify highly variable features. Lsi dimensions were obtained by linear dimensionality reduction using the ‘RunSVD’ function.

‘FindMultimodalNeighbors’ was used to generate a joint neighbor graph using the first 30 ‘pca’ based on gene expression and the second to 30th ‘lsi’ reductions based on ATAC.UMAP dimensionality reduction was then performed by the ‘RunUMAP’ to place cells based on the joint neighbor graph and visualize the distribution of cells. ‘FindClusters’ with the Leiden algorithm was used to identify clusters.

Definition of intestinal stem cell subpopulations

Cell types in the integrated datasets were identified by performing differential expression analysis. ‘Find All Markers’ using the ‘wilcox’ test method was used to determine highly expressed features (min.pct = 0.7, logfc.threshold = 0.3). Heatmap of the top 10 genes in each cluster generated by ‘DoHeatmap’. Differentiation potential of different single cell subpopulations assessed by CytoTRACE2⁶⁵ 1.1.0 package. Genome enrichment analyses were performed using the ‘erichGO’ in the clusterProfiler⁶⁶ package using the “org.Mm.eg.db” database from the Gene Ontology. The AUCell package was also used to score individual cells for gene sets of known function to complement the determination of each subgroup's function.

Trajectory analysis of scRNA-Seq data

Trajectory analysis was performed using Monocle3⁶⁷ 1.3.1 package. After integrating the UMAP dimensions and clustering information for each cell in the seurat object into the celldataset object, developmental trajectory inference was performed using the ‘learn_graph’ functions. Pseudotime was computed and determined using ‘order_cells’ by selecting the initial node pseudo-timepoint. A dynamic analysis of gene expression was conducted by performing differential expression feature analyses along the pseudotime trajectory. The ‘graph_test’ function was employed with the ‘neighbor_graph’ parameter set to ‘principal_graph’. Genes with a Moran’s index greater than 0.1, a log2-fold change >0.3, and an adjusted p-value of 0 (corrected by Benjamini–Hochberg method) were selected as highly differentially expressed genes for constructing gene expression modules. Then, the ‘aggregate_gene_expression’ function was used to cluster these genes into small modules based on their expression patterns via community analysis using the Louvain algorithm. These small modules were then merged through an appropriate number of iterations to form gene modules that were specifically expressed by each cell subpopulation during development.

Gene regulatory networks and hub TFs

Subsequently, we selected 7026 highly differentially expressed genes along pseudotime (q_value < 1e−3) for the analysis of regulatory elements of small intestinal differentiation. pySCENIC⁶⁸ package (Python Single-Cell rEgulatory Network Inference and Clustering) was used to construct a regulatory network connecting differentiation-related targets with transcription factors. Firstly, the GRNboost2 algorithm was used to infer the gene–gene co-expression relationships between transcription factors (TF) and their potential targets (pyscenic grn). Secondly, the regulon prediction step was performed (pyscenic ctx). Each regulon contained one TF, and its target genes were enriched for the motifs of the TF. Lastly, the single-cell transcriptomic data were analyzed using the AUCell algorithm to assess the activity scores of each regulon.

The calcRSS function was employed to calculate the specificity index for each regulon in a given cell type, using the RAS matrix from the pySCENIC analysis. This RSS value was then converted to a Z-score, enabling the comparison of the specificity of the same regulon across cell types. Regulons with a Z-score >0.8 were selected for visualization and subsequent analysis as cluster-specific transcription factors (TFs).

Downstream analysis of scATAC-Seq data

The downstream analysis was done in R 4.2.2 applying Seurat (v4.3.0), Signac (v1.9.0), CHIPseeker⁶⁹ (v.1.34.1), and chromVAR⁷⁰ (v.1.8). The R package ChIPseeker was used to annotate the positions of the mass-filtered ATAC peaks. Peaks were associated to different positions relative to the gene using the annotation reference files “TxDb.Mmusculus.UCSC.mm10.knownGene” and “org.Mm.eg.db”. The distance upstream and downstream of the TSS was defined as 3 kb, and the region downstream of the gene was defined as 50 kb. 46,863 peaks were obtained by removing the peaks located in the Exon, Intron, and Distal Intergenic regions that are less associated with gene regulation.TF activities on the position-filtered peaks were calculated using the Signac implementation of chromVAR using the RunChromVAR function, taking as a tested motifs dataset from HOCOMOCO v11 mouse TF binding models database (529 TFs). Motif enrichment information was incorporated as assay data in the integrated Seurat object files. The ‘wilcoxauc.Seurat’ function in the presto package was used to determine upregulated motif accessibility. Screening of motifs for each cluster-specific motif for subsequent joint analysis (|logFC| > 0.25, q value < 0.05).

Pearson correlation analysis

The lineage-specific TFs obtained from SCENIC and the TFs corresponding to the motifs open to lineage differences at the level of ATAC data were taken to intersect in order to obtain the lineage-specific TFs supported by the multi-omics data. Initially, Pearson correlation analyses were conducted on the AUC score and the motif scores of these TFs in order to verify the consistency of the transcriptome data and the ATAC data at the single-cell level. Subsequently, a Pearson correlation analysis was conducted to ascertain the consistency of the TFs in relation to transcription and chromatin opening at the stem cell subpopulation level, following the calculation of the mean AUC score and motif score for each subpopulation. TFs with correlation coefficients R > 0 and p-values < 0.05 were identified as key TFs regulating lineage differentiation. Heatmap showing the overall expression of regulators and the degree of motif opening of key TFs in each subpopulation, drawn with the ‘pheatmap’ package.

For each TF, its gene expression and motif scoring in each subpopulation were plotted separately with ‘VlnPlot’, and the significance of differences between the three clusters of Lgr5-low SC, AP, and Lgr5⁺ LRC was analyzed. The ‘computeMatrix’ function of deeptools⁷¹ (v3.5.1) was employed to calculate the binding scores for each TF and its regulatory gene region. The reference-point mode was selected for the calculation. The score profiles of this set of genomic regions in each stem cell subpopulation and different functional lineage cells were plotted using the ‘plotProfile’ function.

Small intestinal organoid culture

The small intestine was removed from adult male C57BL6/J mice, and the intestinal crypts were dissociated using 5 mM EDTA. Count the number of crypts followed by re-suspension with Matrigel at a concentration of 20 crypts/μl. Plate the crypts with Matrigel into a 24-well plate and put it in the incubator at 37 °C for 20 min. Then, the small intestinal organoid standard medium (advanced DMEM/F12 + N2 + B27 + N-Acetylcysteine + GlutaMax + P/S + EGF + Noggin + R-Spondin1) was added.

Passage of the small intestinal organoid

Passage can be performed when a few cells of small intestinal organoids die. The organoids with the medium are collected into a centrifuge tube, centrifuged, and washed with PBS 3 times. Use a pipette to dissociate organoids into fragments. Then the fragments were suspended with Matrigel and plated in the 24-well plate, passaging at a density of 1:2–1:3.

Construction of Foxa3 shRNA lentivirus plasmid

The steps of lentivirus plasmid construction were performed according to AddGene protocol (https://www.addgene.org/protocols/plko/). In brief, short hairpin RNA (shRNA) sequences targeting Foxa3 (sense: 5’-TCCTTCGTCCACACCTTATTT-3’; antisense: 5’-AAATAAGGTGTGGACGAAGGA-3’) were synthesized (Sangon Biotech). pLKO.1-TRC cloning vector was digested by restriction endonuclease AgeI and EcoRI. The shRNA oligos were annealed and ligated into pLKO.1 linearized plasmid by T4 Ligase. The recombinant plasmid was transformed into the competent cell and extracted after being identified by gel electrophoresis and gene sequencing. The recombinant plasmid was then transfected into HEK293T cells with packaging plasmid psPAX2 and envelope plasmid pMD2.G. Supernatant was collected 24 and 48 h later and added to the virus concentration solution to harvest lentivirus particles.

Virus infection of the small intestinal organoid

The protocol for virus infection of small intestinal organoids is outlined as follows: In brief, the organoids were dissociated into single cells using 0.25% Trypsin. Transfer the cells to a 48-well plate with an appropriate density and add high-titer lentivirus. Intestinal crypt organoids were infected with lentivirus expressing non-target shRNA (SHC002) as the control group. Put the 48-well plate containing cells in a pre-warmed centrifuge at 32 °C and rotate at 600×g for 1 h. Then, put the organoid-virus mixture in a culture incubator and incubate for another 1 h at 37 °C to allow transduction. Collect the mixture, and remove the supernatant after centrifugation. Re-suspend cells with Matrigel and plate in 24-well plates. After incubating at 37 °C for 20 min, add small intestinal organoid standard medium supplemented with 10 mM nicotinamide, 10 μM CHIR99021, and 10 μM Y27632⁷².

RNA extraction and RT-PCR analysis

To harvest organoid RNA, the organoids with the medium were collected into a centrifuge tube, centrifuged, and washed with PBS, then removed the supernatant. Use Buffer RLT Plus containing 1% β-mercaptoethanol to re-suspend precipitate, lyse cells, and release RNA. The total RNA of the organoid was extracted according to the procedure of RNeasy Plus Kits for RNA Isolation (QIAGEN, 74034). Reverse transcription was conducted using the GoScript™ Reverse Transcription kit (Promega, A5001). RT-PCR was performed, and relative mRNA levels were normalized to the Gapdh mRNA levels. The RT-PCR primer sequences employed in this study are detailed in the Supplementary Table 1.

Immunofluorescent staining and antibodies

Small intestinal organoids were fixed with 4% PFA and then embedded with 3% agarose. Agarose, including organoids, was dehydrated with gradient alcohol and embedded with paraffin. Paraffin sections of organoids can be prepared in 5μm. Paraffin sections were rehydrated with gradient alcohol, followed by antigen retrieval. Then sections were blocked in 5% BSA in PBS for 1 h at room temperature, and then incubated with primary antibodies Keratin 20 (D9Z1Z) XP® Rabbit mAb (1:200, Cell Signaling Technology, 13063S) or MMP-7 (D4H5) XP® Rabbit mAb (1:200, Signaling Technology, 3801S) at 4 °C overnight, followed by washes. Secondary antibodies and DAPI were incubated for 1 h at room temperature. After washing, sections were mounted with Fluoromount-GTM (Southern Biotech, 0100-01).

Bulk RNA-seq and analysis

Total RNA was isolated from small intestinal organoid cells using the RNeasy Plus Mini Kit (QIAGEN, 74134). Freshly extracted RNA was further purified to enrich mRNA via VAHTS mRNA Capture Beads (Vazyme, N401). Subsequently, the enriched mRNA was subjected to library construction with the Equalbit RNA HS Assay Kit (Vazyme, NR605) following the manufacturer’s standard protocols. The constructed RNA-seq libraries were sequenced on the Illumina NovaSeq 6000 platform. Each sample was accompanied by two independent biological replicates, and each replicate library generated no <40 million reads. The RNA-seq data were mapped to the mm10 genome⁶⁰ by TopHat⁷³ v2.0.8 with no more than 2 mismatches, and only uniquely mapped reads were used to estimate the expression values at the gene level in terms of RPKM⁷⁴. Statistical significance testing for differentially expressed genes was performed using the edgeR⁷⁵ 4.0.16 package in R. Differentially expressed genes were identified using the threshold of |log2FC| > 0.5 and an adjusted p-value (corrected by Benjamini–Hochberg method) < 0.05. KEGG pathway enrichment analysis was performed on significantly differentially expressed genes, with entries selected as significantly enriched if they had Benjamini–Hochberg (BH) adjusted p-values < 0.05, and the results were then visualized using ggplot2.

CUT&Tag-seq and analysis

CUT&Tag libraries were constructed using the CUT&Tag Assay Kit (ABclonal, RK20265) in accordance with the manufacturer’s instructions. Briefly, cells derived from small intestinal organoids were harvested, washed with cell wash buffer, and subsequently incubated with activated Concanavalin A-coated magnetic beads. The Foxa3 primary antibody (Santa Cruz, sc-74424) was then added, and the mixture was incubated overnight at 4 °C. After washing, the cells were fragmented with pAG-Tn5 transposase at 37 °C for 1 h. Following a 30-min incubation with proteinase K at 55 °C, genomic DNA was extracted and subjected to library preparation via PCR amplification. The prepared libraries were subjected to high-throughput sequencing on the Illumina NovaSeq 6000 platform using 150 bp paired-end reads. Each library generated at least 40 million reads to ensure sufficient depth for downstream analysis⁷⁶.

The sequencing data were mapped to the mm10 genome using Bowtie2^60,77. Peak calling was performed using MACS2 with parameters set to a p-value of 1e−5, a g value of 2.00e−9, and a genome index based on mm10^60,63. The analysis incorporated background correction and identified peaks with an FDR q-value threshold of 0.05 to control for false positives⁶³. The resulting peaks were analyzed for motif enrichment using HOMER with parameters set to a size of 200 and a genome index based on mm10^60,78.

Statistics and reproducibility

The sample size was not predetermined using statistical methods but was instead established based on previously published studies in this field. All quantitative analyses and experiments yielding representative images were performed with a minimum of three biological replicates (except for single-cell multi-omics sequencing, one biological replicate; bulk RNA–seq and CUT&Tag experiments, two biological replicates). Statistical analyses were performed using R (v 4.2.2). The Kruskal-Wallis test was employed to assess variations in stemness scores for Cytotrace2 across distinct cell types. Differentially expressed genes (DEGs) across conditions for all cell types were identified using the FindMarkers function in Seurat, employing a built-in Wilcoxon test with Benjamini–Hochberg correction of p-values. Wilcoxon rank-sum tests were used to analyze differences in the expression of specific transcription factors and their motif scores across the six ISC subtypes. The disparities in quantitative PCR analysis of indicated genes and quantification of IF staining between the experimental and control groups were appraised using Student’s t-test. The number of biological replicates is specified in the sample sizes (n) indicated in the figure legends.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Xinran Deng contributed to methodology, investigation, formal analysis, and writing; Shenfei Sun contributed to methodology, investigation, validation, and writing; Chenqi Lu contributed to conceptualization, methodology, and writing; Hanxiao Du, Xiaoyu You, Yumou Gong, Runrong Li and Xingyu Yu contributed to formal analysis; fujing huang contributed validation and writing; Kai Gong and Jingxin Guo contributed to validation; Bing Zhao contributed to conceptualization; Xinhua Lin contributed to conceptualization, and funding acquisition; Ning Jiang contributed to conceptualization, methodology, writing, and funding acquisition.

Peer review

Peer review information

Communications Biology thanks Natalia Soshnikova, Kylie James and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Christina Karlsson Rosenthal and David Favero. A peer review file is available.

Data availability

All sequencing data have been uploaded to the China National Center for Bioinformation (CNCB) Genome Sequence Archive (GSA) database under the accession number: PRJCA030631. Previously published snRNA-seq datasets originating from Hickey et al. (10.1038/s41586-023-05915-x) and reanalyzed in this study are accessible via 10.5061/dryad.8pk0p2ns8. The numerical source data underlying all main Figures in the manuscript can be found in the Supplementary Data 5–9. All other data supporting the findings of this study are available from the corresponding author upon reasonable request.

Code availability

All analyses in this paper were performed using R (v 4.2.2) and Python (v 3.7.16). The standard procedures provided by these software packages were followed in conducting the analyses. The scripts for data analysis and visualization are available on GitHub (https://github.com/Xinran-Deng/Single-Nucleus-Multi-Omics-Analysis-of-Mouse-Small-Intestinal-Lgr5-Cells).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-09736-2.