Learn more: PMC Disclaimer | PMC Copyright Notice
Col6a1+/CD201+ mesenchymal cells regulate intestinal morphogenesis and homeostasis
1Abstract
Intestinal mesenchymal cells encompass multiple subsets, whose origins, functions, and pathophysiological importance are still not clear. Here, we used the Col6a1Cre mouse, which targets distinct fibroblast subsets and perivascular cells that can be further distinguished by the combination of the CD201, PDGFRα and αSMA markers. Developmental studies revealed that the Col6a1Cre mouse also targets mesenchymal aggregates that are crucial for intestinal morphogenesis and patterning, suggesting an ontogenic relationship between them and homeostatic PDGFRαhi telocytes. Cell depletion experiments in adulthood showed that Col6a1+/CD201+ mesenchymal cells regulate homeostatic enteroendocrine cell differentiation and epithelial proliferation. During acute colitis, they expressed an inflammatory and extracellular matrix remodelling gene signature, but they also retained their properties and topology. Notably, both in homeostasis and tissue regeneration, they were dispensable for normal organ architecture, while CD34+ mesenchymal cells expanded, localised at the top of the crypts, and showed increased expression of villous-associated morphogenetic factors, providing thus evidence for the plasticity potential of intestinal mesenchymal cells. Our results provide a comprehensive analysis of the identities, origin, and functional significance of distinct mesenchymal populations in the intestine.
Supplementary Information
The online version contains supplementary material available at 10.1007/s00018-021-04071-7.
Introduction
The mammalian intestine is characterized by a unique architecture, which ensures both efficient nutrient and water absorption and rapid self-renewal of the intestinal epithelium. Self-renewal is mediated by Lgr5+ multi-potent crypt-base stem cells (CBCs) that progressively give rise to transit amplifying (TA) progenitor cells and differentiated epithelial cell populations with specific absorptive or secretive functions [1, 2]. The tight regulation of this architectural organization is mediated by a gradient of factors produced both by epithelial and stromal cells. Among stromal cells, intestinal mesenchymal cells (IMCs) have emerged as an important cell type for the development and homeostasis of the intestine, by providing both structural support and regulatory elements [3]. Of particular interest is their contribution to the maintenance of the stem cell niche via the production of soluble mediators [4]. Notably, in the absence of epithelial Wnts, production of stromal Wnts is sufficient for the maintenance of epithelial proliferation, while depletion of Foxl1+ telocytes or Grem1+ trophocytes, which produce niche-supporting factors led to the disruption of the intestinal structure [5–7]. Additionally, BMP production by villous mesenchymal cells inhibits proliferation and favors epithelial cell differentiation, which ensures epithelial homeostasis and supports specialized epithelial functions [8, 9]. The production of such signals is believed to be induced and maintained via the reciprocal communication with epithelial cells. This has been convincingly shown during embryonic development, where PDGF and Hh proteins secreted from the endodermal epithelium act on the underlying mesenchyme to induce the formation of PDGFRα+ aggregates, which express BMPs and regulate villification of the intestine [10].
Beyond their homeostatic functions, intestinal fibroblasts contribute significantly to tissue damage and inflammation. During such conditions, resident fibroblastic cells are activated to produce pro-inflammatory cytokines and chemokines, angiogenic factors, as well as extracellular matrix (ECM) components and remodelling enzymes to facilitate acute inflammatory responses. Deregulation of these processes or chronic injury can lead to chronic inflammatory disorders and fibrosis [11]. Indeed, recent data point to an important role of the microenvironment in shaping cellular programs and driving epithelial regeneration, as fibroblast-specific deletion of the Wnt regulator porcupine (Porcn) or R-spondin 3 led to severely impaired intestinal regeneration [12, 13].
Until recently, mesenchymal cells in the intestine, although known to constitute a group of cell types, were frequently studied as one, mainly due to difficulties in marking, isolating and genetically targeting specific populations. However, recent single-cell transcriptomic analyses of the normal mouse and human intestine revealed the underappreciated extent of mesenchymal heterogeneity and identified several fibroblast subsets with distinct expression profiles and functions [4, 7, 14–16]. However, their origin and spatial organization, as well as the mechanisms through which distinct subsets coordinate signaling gradients along the crypt–villus axis in homeostasis and disease remain elusive. The use of specific markers and Cre-expressing mouse lines has begun to provide such information and is crucial for addressing these issues. Examples include CD34, Foxl1, PDGFRα and Gli1-positive IMCs, which act as critical regulators of the intestinal stem cell niche through their production of Wnts, R-spondins and Gremlin 1, although they are not strictly restricted to single cell types [6, 7, 12, 13, 17, 18]. Lgr5+ villous tip telocytes were also recently shown to regulate epithelial villus tip gene expression programs [19]. Single-cell analysis of the inflamed intestine has further highlighted the prominent pro-inflammatory activation of fibroblasts [14, 15, 20]. Notably, activated fibroblasts along with immune cells were associated with resistance to anti-TNF therapy, indicating their potential utility in patient diagnosis, stratification, and therapeutic decisions [15, 20].
We have previously shown that the Col6a1Cre transgenic mouse targets a fraction of mesenchymal cells in the intestine and that NFκΒ signaling in this subset uniquely protects against colitis and colitis-associated cancer [21]. In this study, transcriptomic, imaging, and functional analysis of the cells targeted by the Col6a1Cre mouse revealed preferential targeting of colonic PDGFRαhi fibroblasts and perivascular cells, which can be further described using specific markers. Using reporter mice, confocal imaging and cell depletion approaches, we have further identified their relationship with mesenchymal clusters during embryogenesis, as well as their functional significance in development and adulthood. Notably, both in normal conditions and during tissue regeneration, CD34+ IMCs proliferated, and adopted different topologies and gene expression profiles, following the depletion of colonic Col6a1-cre lineage mesenchymal cells, revealing the plasticity of IMCs towards organ homeostasis.
Results
The Col6α1Cre mouse targets predominantly CD34− mesenchymal cells in the mouse colon
We have shown in the past that the Col6a1Cre transgenic mouse targets a fraction of mesenchymal cells in the intestine [21]. To define its specificity for distinct mesenchymal subsets, we crossed Col6a1Cre mice with a TdTomato-to-GFP replacement (mTmG) reporter strain (Col6a1mTmG) and after exclusion of immune, endothelial, erythroid, and epithelial cells using the lineage negative (Lin−) markers CD45, CD31, Ter119 and CD326 (EpCAM), we isolated Col6a1-GFP+ and Col6a1-GFP− IMCs by FACS sorting (Fig. 1A). As previously reported, the Col6a1Cre mouse does not target Lin+ cells [21]. We then performed 3′ mRNA sequencing of the Lin−GFP+ IMCs (GC), Lin−Tomato+ IMCs (TC) and the initial unsorted cells (UC). Gene Ontology (GO) analysis of the upregulated genes in the GC versus the TC and/or UC cells revealed enrichment in biological processes related to epithelial cell proliferation and differentiation, as well as regulation of vasoconstriction and blood pressure (Fig. 1B). Enriched genes associated with epithelial differentiation included Bmps (Bmp3, Bmp7, Bmp2, Bmp5), Wnt5a, the Wnt inhibitor Wif1 and genes related to the differentiation of epithelial cells (e.g. Fgf9) [8]. Conversely, the TC population expressed genes associated with the maintenance of the stem cell niche, such as Grem1, Wnt2 and Nog [7, 17] (Fig. 1B). These results suggest that Col6a1-cre lineage cells have distinct homeostatic functions, potentially associated with different topologies along the colonic crypt length.
The Col6a1Cre mouse targets predominantly CD34− IMCs in the mouse colon. A FACS sorting strategy for the isolation of Col6a1-GFP+ (GC) and Col6a1-GFP− (TC) mesenchymal cells from the colon. Single cell preparations from the colon (UC) were stained for the Lin+ markers CD45, EpCAM, CD31 and Ter119 and Propidium Iodide (PI) for dead cell exclusion. 3 samples from 4–5 mice each were subsequently analyzed. B Heatmap of differentially expressed genes in GC vs TC and UC samples, corresponding to GO terms related to epithelial proliferation/differentiation and blood vessel regulation. Log2-transformed normalized read counts of genes are shown. Read counts are scaled per column, red denotes high expression and blue low expression values. C Representative FACS analysis of CD34 expression in Lin- cells in the colon of Col6a1mTmG mice (n = 4–5 mice). D Immuno-histochemical analysis for CD34 expression in the colon of Col6a1mTmG mice (n = 9–10 mice, Scale bar: 50 μm). E Total number and size of intestinal structures after 3 days of co-culture with sorted Col6a1-GFP+ and GFP− colonic IMCs, with and without R-Spondin 1, respectively. Data represent mean ± SEM from one of four experiments performed in quadruplicates. *p < 0.05, **p < 0.01, ***p < 0.001. F Representative bright-field images of intestinal organoids co-cultured with Col6a1-GFP+ and GFP− IMCs at day 3, in the absence of R-Spondin 1 (Scale bar: 100 μm). G Mean expression (z-score) of genes signatures extracted from the different intestinal mesenchymal clusters identified in Kinchen et al. [14] in Col6a1-GFP+ bulk RNA-seq samples. MF, myofibroblasts; SMC, smooth muscle cells. H Heatmap of the top 50 differentially expressed genes in the Str2 population (Kinchen et al. [14]) and their relative expression in the GC and TC samples
A similar cell sorting and sequencing approach were also performed for the small intestine, where two populations of GFP+ cells were identified, a GFPhi (GIH) and a GFPlo (GIL) population, each accounting for 23% and 29% of the Lin− population, respectively (Figure S1A). Analysis of the genes that were previously identified in the colon showed that GIH cells were also enriched in genes related to epithelial differentiation and blood vessel function, while GFPlo (GIL) cells were similar to the GFP− population (Figure S1B). The presence of GFPlo cells in the small intestine indicates a GFP+ cell population with potentially different cellular properties [22], and could represent a distinct cell subset, an intermediate state between GFP− and GFP+ cells or the result of non-specific recombination. These results indicate that the Col6a1Cre mouse targets a broader set of mesenchymal subsets in the small intestine in comparison to the colon.
Interestingly, one of the genes enriched in the GFP− samples in both the small intestine and colon was Cd34 (Fig. 1B). Stzepourginski et al. [17] described a broad population of PDPN+CD34+ cells, which are located at the bottom of the crypts, express Gremlin1, Wnt2b and R-spondin, and play a role in stem cell maintenance. FACS analysis and immunohistochemistry showed that indeed the majority of GFP+ cells in the colon (84%) and GFPhi in the small intestine (82%) were CD34-negative and further indicated a preferential subepithelial localization for GFP+ cells outside the bottom of the crypts (Figs. 1C, C,DD and S1C, D). Consistent with these data, co-culture of either GFP+ or GFP− colonic IMCs with intestinal organoids showed that GFP+ cells were less potent than GFP− cells in supporting the growth of epithelial organoids in the absence or presence of R-spondin 1 (Fig. 1E, E,F).F). Comparison of the gene signature of mesenchymal subsets from the recently published single-cell RNA sequencing of the mouse colon by Kinchen et al. [14] and the gene expression profile of GFP+ cells revealed an association with the Str2 cluster (Fig. 1G, G,H).H). Importantly, this cluster appears to be the highest conserved between mice and humans [14]. It should be noted that Col6a1 is expressed in most mesenchymal cells, both in the healthy colon and in DSS colitis, as seen in the analysis of the above single-cell transcriptomic data, and therefore it does not correspond to the expression pattern of the Col6a1Cre mouse (Fig. S2). These results show that the Col6a1Cre mouse targets mostly CD34− mesenchymal cells outside the bottom of the colonic crypt, rendering it appropriate for the functional characterization of these cells.
Colonic Col6a1-GFP+ cells are CD201-positive and include PDGFRαhi fibroblasts and perivascular cells
We next searched our sequencing data for markers that could be used to detect, isolate and study Col6a1-GFP+ cells. We found that GFP+ cells express higher levels of CD201 (Procr or EPCR), which was further verified by immunohistochemistry (Fig. 2A). FACS analysis indeed showed that 67% of the Lin−GFP+ population in the colon were CD201+ (Fig. 2B). It should be noted that CD201 is also expressed by endothelial cells, which are excluded as Lin− in our analysis. Since not all GFP+ cells express CD201, we also isolated Lin−GFP+CD201+ and Lin−GFP+CD201− cells from the colon by FACS sorting. qPCR analysis showed that GFP+CD201+, but not GFP+CD201− cells, expressed genes related both to the regulation of blood vessel function and epithelial cell differentiation, similar to GFP+ cells (Fig. 2C). These results suggest that the remaining GFP+CD201− cells either have another yet unknown role in the intestine or are the result of non-specific targeting by the Col6a1Cre mouse.
GFP+/CD201+ mesenchymal cells comprise distinct subsets in the mouse colon. A Immunohistochemistry for CD201 in the colon of Col6a1mTmG mice (n = 4 mice, Scale bar: 50 μm). The dotted line delimitates the epithelial surface towards the lumen. B Representative FACS analysis of CD201 expression in Lin− cells in the colon of Col6a1mTmG mice (n = 10 mice). C Gene expression analysis of selected genes in FACS-sorted GFP+CD201+, GFP+CD201− and GFP− colonic mesenchymal cells from Col6a1mTmG mice. Expression is measured in relation to the Hprt housekeeping gene (n = 3), *p < 0.05, **p < 0.01, ***p < 0.001. D) tSNE plots showing the expression of CD34, CD201, PDGFRα and αSMA in Lin− colonic mesenchymal cells using FACS analysis and E quantification of CD34+ and CD201+ cells in PDGFRα+ subsets (n = 8 mice). F Immunohistochemistry for αSMA and CD31 in the colon of Col6a1mTmG mice, showing their localization around blood vessels (white arrow) (Scale bar: 50 μm). The bottom of crypts is shown. G Immunohistochemistry for PDGFRα and CD34 in the colon of Col6a1mTmG mice. Different planes are shown. White arrows indicate PDGFRαhi (upper panel) and PDGFRαlo (lower panel) mesenchymal cells (Scale bar: 50 μm). The dotted line delimitates the epithelial surface towards the lumen. H Confocal imaging of GFP+ cells at the top of Col6a1mTmG colonic crypts (Scale bar: 10 μm). I Immuno-histochemical analysis of αSMAlo cells in the colon. White arrows indicate co-localization with GFP along the crypt’s length. (Scale bar = 10 μm) (n = 4 mice). J Quantification of Col6a1Cre-GFP+ cells in PDGFRα+ (top) and PDGFRα- (bottom) mesenchymal subsets using FACS analysis (n = 6 mice). K Immunohistochemistry for CD31 in the colon of Col6a1mTmG mice (n = 3 mice, Scale bar: 10 μm)
Combination of the CD201 and CD34 markers was able to distinguish between the two distinct Lin− mesenchymal subpopulations (Figs. 2D and S3A). Additional co-staining with PDGFRα and αSMA also divided PDGFRαhi, PDGFRαlo, and PDGFRα− into CD201+ and CD34+ cell subsets (Fig. 2D, D,E).E). PDGFRαhi cells, which represent 22% of Lin− mesenchymal cells, are predominantly CD201+αSMAlo/− (93%), while PDGFRαlo cells, which account for 68% of Lin-mesenchymal cells, include 81% CD34+ and 19% CD201+αSMAlo/− cells (Fig. 2D, D,E).E). PDGFRα− cells are 10% of the Lin− mesenchymal cells and include a CD201+αSMAhi subset (24%) (Fig. 2D, D,EE and Figure S3B-D). Notably, CD34+ cells were mainly PDGFRαlo/− and only a limited number expressed PDGFRαhi and αSMA, in line with recent reports [7, 17] (Fig. 2D, D,EE).
GFP+PDGFRα−CD201+αSMAhi cells were found around blood vessels and were most possibly pericytes (Figs. 2F and S3D), as also indicated by qPCR analysis (Figure S3E) and previous results [4, 23]. GFP+PDGFRαhiαSMAlo/− were long thin cells in a subepithelial location found both clustered at the top of the colonic crypts and individually along the crypt’s length, most possibly corresponding to telocytes or subepithelial myofibroblast (SEMFs) (F(Fiigs. 2G–I and S3C, S3F) [4, 7, 13, 19]. Indeed, GFP+ cells expressed markers related to telocytes, including Foxl1, Bmp7, Wif1 and Wnt5a (Figs. 1B, S3E and S4A). SmFISH experiments further verified the expression of Foxl1 by GFP+ cells mainly at the top of the colonic crypt (Figure S4B). Finally, GFP+PDGFRαlo-αSMAlo/− cells appeared as flat cells with extended processes that surround the colonic crypt (Fig. 2G). FACS-based quantification of Col6a1-GFP+ cells in these subsets revealed targeting of almost all the PDGFRαhi cells (93.2%), and one-third of the PDGFRαlo stroma (27.8%), including mainly PDGFRαlo CD201+ cells (77.8% versus 11.7% of PDGFRαlo CD34+ cells) (Fig. S3G). Absence of Grem1 expression (Fig. 1B) indicates that GFP+PDGFRαlo cells do not include trophocytes. The Col6a1Cre mouse also targeted 13.8% of PDGFRα− cells, comprising mainly CD201+αSMAhi pericytes (86%) (Fig. 2J). However, the apparent low abundancy of this population suggests that expression of genes related to blood vessel function could be also associated with the localization of Col6a1-GFP+ cells in close proximity to the subepithelial capillary network, indicating thus a potential dual function for colonic PDGFRαhi fibroblasts (Fig. 2K). These results reveal the complexity of mesenchymal cells surrounding the colonic crypt, provide a detailed characterization of their markers and localization and the efficiency of their targeting by the Col6a1Cre mouse.
Col6a1-GFP cells are found as mesenchymal aggregates during development and orchestrate intestinal morphogenesis
To examine the specificity of Col6a1Cre mice also during embryonic organogenesis, we analysed Col6a1mTmG mice at different developmental stages. We found that GFP+ cells were absent until E13.5 and started to appear at E14.5–E15.5 as aggregates beneath the epithelial layer (Fig. 3A). As villi became more elongated, GFP+ cells extended toward the bottom of the crypts (Fig. 3B). GFP+ aggregates expressed PDGFRα, as shown by both confocal microscopy and FACS analysis, in line with previous reports [24–26] (Fig. 3A, B, D). Notably, similar GFP+ PDGFRα+ aggregates were also detected in the developing colon (Fig. 3A, A,B).B). FACS analysis showed that the majority of GFP+ cells corresponded to 28% and 38.3% of PDGFRαhi cells at E16.5 and E18.5, respectively (Fig. 3D, D,E).E). Contrary to adulthood, the Col6α1Cre mouse targeted a limited number of other mesenchymal populations, including PDGFRαlo and αSMA+ cells (Fig. 3D, Ε). It should also be noted that CD34 was not expressed at this stage in agreement with previous reports [17]. CD201 was broadly expressed, although it did include PDGFRαhi cells also targeted by the Col6α1Cre mouse (Fig. 3F). These results show a potential ontogenic relationship between mesenchymal villous and colonic clusters and PDGFRαhi telocytes.
The Col6a1Cre mouse targets mesenchymal cell aggregates during development, which are necessary for intestinal morphogenesis. Confocal images showing GFP and PDGFRα expression in A the small intestine and colon at E15.5 (Scale bar = 50 μm) and B the small intestine and colon of Col6a1mTmG mice at the indicated developmental stages (Scale bar = 50 μm), (n = 3 mice per developmental stage). C Confocal images showing GFP and αSMA expression in the small intestine and colon of Col6a1mTmG mice at the indicated developmental stages (Scale bar = 50 μm), (n = 3 mice per developmental stage). D FACS analysis of Col6a1-GFP+ intestinal mesenchymal cells at E18.5. E FACS-based quantification of GFP+ cells in PDGFRαhi and PDGFRαlo cells in E16.5 and E18.5. F FACS analysis of GFP+ cells in E18.5, showing that they all are CD201+PDGFRαhi cells (n = 3–4 mice per developmental stage in all FACS analyses). G Schematic representation of DT administration. Pregnant females received two injections of diphtheria toxin (DT) (5 μg) at E14.5 and E15.5, which was followed by ex vivo culture of the intestine from E16.5 to E18.5. H Lightsheet imaging (maximum projection, Scale bar = 100 μm) and confocal images showing GFP, PDGFRα and αSMA expression (Scale bar = 50 μm) in the small intestine of Col6a1DTR and control mice. I) quantification of villi/nm in the presence of DT (n = 7). All (GFP+ and GFP−) and only GFP− villi are presented in DT treated mice. ***p < 0.001, **p < 0.01
To define the physiological importance of these cells during intestinal embryonic development, we employed the iDTR strain [27] in combination with the Col6a1mTmG mice (Col6a1DTR). Administration of diphtheria toxin (DT) in Col6a1DTR mice at Ε14.5 and E15.5 and subsequent ex vivo culture of the embryonic intestine for 48 h resulted in depletion of Col6a1-GFP+ cells and a significant reduction in the number of developing villi (F(Fiig. 3G–I, Fig. S5 and Supplementary Videos 1 and 2). Total and GFP− villi were separately quantified in the Col6a1DTR mice to take into account potential inefficient or patchy deletion, while all villi appear to have GFP+ clusters in the control mice (Fig. 3H, H,II and Supplementary Videos 1 and 2). Staining with PDGFRα and αSMA did not reveal major differences in their expression patterns after GFP+ cell depletion, in agreement with the level of Col6a1Cre targeting at this stage (Fig. 3C–E, H). These results suggest that Col6a1-GFP+ mesenchymal cells act as orchestrators of intestinal morphogenesis and patterning.
Colonic Col6a1-cre lineage IMCs regulate homeostatic epithelial proliferation and enteroendocrine cell differentiation
To define the homeostatic roles of colonic Col6a1-cre lineage IMCs, we then depleted Col6a1-GFP+ cells in 4–6-month-old Col6a1DTR mice. Due to increased lethality upon systemic DT injection, we performed local intra-rectal DT administration, as shown in Fig. 4A. Efficient, albeit not complete, depletion of GFP+ cells (75% reduction) in the last 3–4 cm of the colon was verified by confocal imaging and FACS analysis (Fig. 4B, B,C).C). Further analysis of the remaining GFP+ cells showed that although all subtypes were depleted, there was a preferential loss of PDGFRαhi cells and a proportional increase in PDGFRαlo cells (Fig. 4D). Accordingly, and consistent with our previous data (Fig. 2), PDGFRαhiCD201+ fibroblasts were reduced by 90%, while PDGFRαlo stromal cells were proportionally increased by 11%, and PDGFRα− cells remained largely unaltered (Fig. 4E, E,F).F). These results indicate that PDGFRαhi fibroblast is the mesenchymal subpopulation most efficiently depleted using our approach.
Col6a1-cre lineage cell depletion leads to deregulated epithelial cell differentiation and proliferation during homeostasis. A Schematic representation of DT administration in homeostasis. Col6a1DTR and control (Col6a1mTmG, iDTRf/f/) mice received 3 daily intra-rectal administrations of DT (20 ng/g body weight) and mice were sacrificed after 5 days. B Confocal images of GFP expression in Col6a1mTmG and Col6a1DTR mice (Scale bar = 50 μm). C FACS analysis and quantification of GFP+ cells in the colon of Col6a1mTmG and Col6a1DTR mice after DT administration (n = 9). D FACS analysis and quantification of PDGFRα+ subsets in GFP+ cells in Col6a1mTmG and Col6a1DTR mice (n = 3–5). E FACS analysis and quantification of CD201+PDGFRαhi cells in Col6a1mTmG and Col6a1DTR mice (n = 3–4 mice). F FACS analysis of PDGFRα+ subsets in Col6a1mTmG and Col6a1DTR mice (n = 4–10). G H&E staining of Col6a1DTR and control (Col6a1mTmG, iDTRf/f/) mice (Scale bar: 100 μm). H Expression analysis of the indicated genes in colon samples from Col6a1DTR and control (Col6a1mTmG, iDTRf/f/) mice. Expression is measured in relation to the B2m housekeeping gene (n = 9–14). I Immuno-histochemical-based quantification of differentiated epithelial cell types per crypt in Col6a1DTR and control (Col6a1mTmG, iDTRf/f/) mice (n = 5–13). J Expression analysis of the indicated genes in colon samples from Col6a1DTR and control (Col6a1mTmG, iDTRf/f/) mice. Expression is measured in relation to the B2m housekeeping gene (n = 5–15). K Representative BrdU staining and L quantification of the ratio of BrdU+ epithelial cells in the top/bottom of the colonic crypts of Col6a1DTR and control (Col6a1mTmG, iDTRf/f/) mice (Scale bar: 100 μm), (n = 6–7). *p < 0.05, **p < 0.01, ***p < 0.001
Histopathological examination of Col6a1DTR mice showed that the intestinal structure was normal following DT administration (Fig. 4G). Immunohistochemistry and/or qPCR analysis for the quantification of specific intestinal epithelial subpopulations showed a reduction in enteroendocrine cell differentiation upon GFP+ cell depletion, while stem cell maintenance, as well as the differentiation of Tuft and Goblet cells was not affected (Fig. 4H, H,II and Figure S6). Quantification of telocyte markers in these tissue samples showed a reduction in Bmp7 and Wnt5α expression levels, while the expression of most other Bmps and Foxl1 was not altered (Fig. 4J). Notably, Bmp7, which forms heterodimers with Bmp2 and Bmp4, was previously shown to be one of the factors that is exclusively expressed by PDGFRα+ telocytes at the crypt–villous boundary of the small intestine [7]. In addition, we also detected a defect in the distribution of BrdU+ proliferating epithelial cells along the crypt axis, characterized by an increase toward the top of the crypt, which could be associated with the role of BMPs in the regulation of stem cells [28] (Fig. 4K, K,L).L). These results show that colonic Col6a1-cre lineage IMCs have distinct pathophysiological roles in epithelial cell differentiation and proliferation, although they are largely dispensable for normal tissue architecture and function.
Loss of Col6a1-GFP+ IMCs is followed by CD34+ mesenchymal cell plasticity
The similar expression levels of most BMPs in the Col6a1DTR mice and its normal tissue architecture indicated that other mesenchymal cell populations could mediate some of the functions of Col6a1-cre lineage IMCs under these conditions. Indeed, we found that colonic crypt tops in the Col6a1DTR mice were populated by GFP−PDGFRα+CD34+ cells, which however remained PDGFRαlo (Fig. 5A, A,B).B). CD34+ cells in this area were also Ki67+, indicating that these cells could proliferate and occupy the space, where Col6a1-GFP+ fibroblasts were previously located (Figs. 5C, C,D,D, I and S7). Injection of BrdU after DT administration and FACS analysis verified that CD34+ cells proliferated following GFP+ cell depletion (Figs. 5E and S8). They also expressed αSMA, a marker commonly upregulated in response to fibroblast activation (Fig. 5F–H and S9). In addition, they showed increased expression of genes associated with telocyte functions, including Bmp2, Bmp3, Bmp7, Wnt5a and Foxl1, but not Bmp5 (F(Fiig. 5I). SmFISH experiments further showed that Foxl1 is expressed by subepithelial CD34+GFP− cells, thus supporting the acquisition of telocyte markers by CD34+ under these conditions (Figure S10). These results show that following Col6a1-cre lineage IMC depletion, CD34+ cells become activated, they proliferate and occupy the space at the top of the colonic crypts, partly compensating for the loss of PDGFRαhi fibroblasts, providing thus evidence for mesenchymal plasticity in the intestine.
Loss of Col6a1-cre lineage IMCs is followed by CD34+ mesenchymal cell plasticity. A Immuno-histochemical and B FACS analysis of CD34 and PDGFRα expression in the colon of Col6a1DTR and Col6a1mTmG mice (Scale bar = 50 μm) (n = 6 mice). A zoomed-in view of the crypt top is shown in the merged image. C Immuno-histochemical analysis and D quantification of Ki67+CD34+ mesenchymal cells in the colon of Col6a1DTR and Col6a1mTmG mice (Scale bar = 50 μm) (n = 5 mice). Both broad and zoomed-in views of the crypt top are shown. E FACS-based quantification of BrdU+CD34+ cells in the colon of Col6a1DTR and control mice following three consecutives injections with BrdU at days 3–5 of the protocol (n = 8–10 mice). F Immuno-histochemical analysis of αSMA expression in the colon of Col6a1DTR and control mice (the top of the crypt is shown) (n = 5). G FACS analysis and H quantification of αSMA+ cells in the CD34+ subset (n = 4 mice). I Gene expression analysis of isolated CD34+ cells from Col6a1DTR and control mice (n = 5–6). *p < 0.05, **p < 0.01, ***p < 0.001
Topological and functional plasticity of mesenchymal cells during intestinal regeneration
To further explore the functions of Col6a1-GFP+ cells in non-homeostatic conditions, we subjected Col6a1mTmG mice to the DSS model of acute colitis and isolated Lin-GFP+ (GDS) and Lin-Tomato+ (TDS) cells by FACS sorting, as previously described (Fig. 6A). Comparisons between GFP+ and GFP− IMCs and the unsorted population (UDS), as well as cells from untreated mice and subsequent Gene Ontology analysis revealed that both GDS and TDS cells were enriched for functions related to inflammatory/immune responses (Fig. 6B). GDS and TDS samples also showed enrichment in genes related to epithelial proliferation/differentiation and blood vessel function, similar to the homeostatic situation, suggesting that these cells, although activated, retain their homeostatic properties and marker expression (Figs. 6B and S11A). FACS analysis of mesenchymal populations described in Fig. 2 verified their presence also during acute colitis, although CD201+PDGFRα−αSMAhi were increased in accordance with fibroblast activation (Fig. S11B, C). Confocal imaging further revealed that in sites of ulceration, GFP+ cells were located in the upper part of the damaged area, indicating that they also retained their topology (Fig. 6C). This was further verified by comparisons with the recently published single cell RNA-seq data by Kinchen et al., which confirmed that GDS cells corresponded mostly to the Str2 population [14] (Fig. 6D).
Col6a1-cre lineage cells retain their properties and topology during colitis but are dispensable for tissue regeneration. A FACS sorting strategy for the isolation of Col6a1-GFP+ (GDS) and Col6a1-GFP− (TDS) mesenchymal cells from the colon at the end of the acute DSS protocol. 3 samples from 4–5 mice each were subsequently analyzed. B Heatmap of differentially expressed genes in GDS vs TDS and UDS, as well as the respective untreated samples, corresponding to GO terms related to epithelial proliferation/differentiation, blood vessel regulation and inflammatory response. Log2-transformed normalized read counts of genes are shown. Read counts are scaled per column, red denotes high expression and blue low expression values. C Confocal images of GFP expression in Col6a1mTmG mice at day 8 of the acute DSS protocol, Scale bar = 50 μm. D Mean expression (z-score) of genes signatures extracted from the different intestinal mesenchymal clusters identified in Kinchen et al. [14] during DSS colitis in Col6a1-GFP+ bulk RNA-seq samples. MF, myofibroblasts. E Schematic representation of DT administration during acute colitis and regeneration. Mice received 2.5% DSS for 5 days, followed by regular water for 16 days. 100 μl DT (20 ng/g body weight) was administered intrarectally at days 4, 5 and 6 of the regime. F Quantification of GFP+ cells in the colon of control and Col6a1DTR mice after DT administration (n = 5–7 mice), G H&E staining and H) histopathological score of Col6a1DTR and control (Col6a1mTmG, iDTRf/f/) mice at the end of the protocol (Scale bar = 100 μm) (n = 10–11 mice). Immuno-histochemical analysis of I CD34 and PDGFRα expression and J CD34 and αSMA expression in the colon of Col6a1DTR and control mice (Scale bar = 50 μm) (n = 4 mice)
Depletion of GFP+ cells during DSS administration and evaluation of tissue morphology 21 days after the initiation of the protocol showed similar histopathological score to untreated mice, indicating normal regeneration of the intestine (Fig. 6E–H). Similar to homeostasis, GFP−PDGFRα+CD34+ cells were localized at the crypt tops, indicating the plasticity of CD34+ cells and their ability to support the normal re-epithelization of the colon (F(Fiig. 6I). Notably, at this late time point, CD34+ cells were not αSMA+, indicating their potential reversible activation following Col6a1-cre lineage cell depletion (Figs. 6J and S12). Therefore, these results further support the plasticity of intestinal mesenchymal cells and suggest that reciprocal signals between the epithelium and the underlying mesenchyme are more or equally important to specialized cell types for intestinal regeneration.
Discussion
The importance of the mesenchymal stroma in the maintenance of intestinal structure and function, as well as in the response to injury is now well established [3, 4]. Opposing signaling gradients along the colonic crypt length and the crypt/villous axis in the small intestine, including Wnts, BMPs, and their inhibitors, are considered crucial for the maintenance of this morphology and the presence of specialized mesenchymal cell types could explain this phenomenon [8]. Indeed, several studies have described specific cell subsets that act as regulators of the stem cell niche [6, 7, 13, 15–18]. Accordingly, although less studied, telocytes and specific fibroblast subsets identified by sc-RNA sequencing analyses have been shown to express BMPs and regulators of epithelial differentiation [6, 7, 14, 19, 29].
In this study, we focused on characterizing the identities, markers, origins and functional significance of IMCs outside the colonic stem cell niche in intestinal development, homeostasis and tissue damage/regeneration using the Col6a1Cre transgenic mouse [30]. To this end, we have performed transcriptomic, FACS and imaging analysis of the cells targeted by the Col6a1Cre mouse strain. It should be noted that Cre expression and cell targeting in this strain do not reflect Col6a1 expression, which is much broader, possibly due to the random integration of the transgene. It can, therefore, only be used as a tool. To partly overcome this limitation, we also identified a novel extracellular marker, CD201, and in combination with PDGFRα and αSMA we were able to show that the Col6a1Cre mice target almost all CD201+PDGFRαhiαSMAlo/− fibroblasts, around one-third of the PDGFRαlo stroma, and specifically CD201+PDGFRαlo cells, as well as perivascular cells that are CD201+αSMAhi. Among them, PDGFRαhi fibroblasts are the only subpopulation targeted to its entirety, and it was shown to localize subepithelially along the colonic crypts, and concentrating at their tops, in a similar fashion to telocytes/SEMFs in the small intestine [7]. In addition, they were in close proximity to the subepithelial capillary network, indicating a potential dual role in both epithelial and endothelial cell function. Notably, a subset co-expressing telocyte and pericyte gene expression signatures has been previously described in both the stomach and the small intestine [16].
Little is known about the developmental origins of the different mesenchymal subsets, although recent single-cell analyses of the mouse and human developing intestine have begun to shed light into embryonic mesenchymal heterogeneity [31–33]. In relation to villification and crypt formation, single cell analysis of the mouse hindgut identified a distinct subset corresponding to crypt top fibroblast precursors with high Pdgfra and Bmp2/5 expression [33]. A similar PDGFRAhi/F3hi/DLL1hi subepithelial population was also found in the human developing intestine [31, 32]. Previous studies have shown that PDGFRαhi cells form clusters that emerge in the murine small intestine in waves after E13.5, are Hedgehog responsive and express BMPs to regulate villous morphogenesis [10, 24–26]. Interestingly, our results showed that Col6a1Cre mice target PDGFRahiCD201+ mesenchymal clusters also during development both in the small intestine and colon, further supporting an ontogenic relationship between them and homeostatic PDGFRαhi subepithelial fibroblasts. Mesenchymal–epithelial crosstalk plays a crucial role in the process of villification [3, 34]. Deletion of either PDGFα or PDGFRα in mice resulted in abnormal villi development due to reduced mesenchymal proliferation [26]. Accordingly, inhibition of Hedgehog (Hh) signaling led to lack of cluster formation and Hh was recently shown to directly regulate mesenchymal clustering through the intrinsic GLI2-mediated activation of planar cell polarity genes, which is necessary for epithelial remodeling in the villous area [24, 35]. FoxL1 deletion also led to a delay in villification [36]. Some similarities can also be found during chick intestinal development, where smooth muscle differentiation drives villi formation through forces that generate localized pockets of high Shh, crucial for the expression of mesenchymal cluster genes, such as PDGFRα and BMP4 [37, 38]. Nevertheless, the importance of distinct embryonic mesenchymal subsets for villification and intestinal morphogenesis has not been addressed yet. Our functional cell depletion analysis clearly shows the significance of mesenchymal clusters in intestinal villification and patterning. Additional analysis is, however, necessary to assess the effect of such cell depletion on epithelial–mesenchymal communication and epithelial cell functions, including proliferation and differentiation during intestinal organogenesis.
Interestingly, depletion of this population during homeostasis did not affect the architecture and morphology of the colon. As we show, this could be explained by the plasticity of CD34+ cells that are able to proliferate and occupy the area, where Col6a1Cre+/CD201+PDGFRαhi cells were previously found. Notably, CD34+ cells showed increased expression of Col6a1-cre lineage enriched genes, such as Bmp2, Bmp3, Bmp7, Wnt5a and Foxl1, but not Bmp5. This is most possibly related to the almost complete ablation of Col6a1Cre+/CD201+PDGFRαhi cells in comparison to other subsets [4]. Nevertheless, several aspects of intestinal homeostasis were compromised, including the total expression levels of Bmp7 and Wnt5a, the normal differentiation of enteroendocrine cells and the distribution of proliferating cells along the crypt axis. Both the function of enteroendocrine cells and the stemness of Lgr5+ cells have been previously shown to be modulated by BMP signaling [9, 28], including Bmp7, which forms heterodimers with Bmp2 and Bmp4 [39]. The exact CD34+ cell subset that displays such plasticity is yet not known; however, Gremlin-1+ cells have been previously shown to have stem cell potential in the intestine [40]. The phenotype of Col6a1Cre+/CD201+ IMC depletion contrasts with FoxL1+ telocyte deletion or their secretion of Wnt ligands. However, the Foxl1Cre mouse has been shown to also target pericryptal telocytes that express Gremlin-1 [6, 13]. Although a small population, Gremlin-1+ cells have displayed stem cell potential and are crucial for intestinal homeostasis and structure, as recently shown [40]. In addition, a recent publication showed that genetic inhibition of Wnt secretion by Ng2+ mesenchymal cells, which include telocytes and other mesenchymal cells, leads to mild stem cell phenotypes, whereas its inhibition more broadly throughout gut mesenchymal cells induces striking developmental phenotypes, implying mesenchymal redundancy in the stem cell niche [16]. Therefore, our results, delineate the functional importance of Col6a1-cre lineage IMCs at the top of the colonic crypts, illustrate how they can affect epithelial cell differentiation and proliferation and provide evidence for mesenchymal plasticity towards tissue homeostasis.
Similar to homeostasis, Col6a1-cre lineage cell depletion during acute DSS colitis did not affect inflammation and regeneration of the intestine, despite their inflammatory gene signature and specific topology at the top of the ulcerated tissue and thus their proximity to the regenerating epithelium. As we show, this is also associated with the plasticity of CD34+ mesenchymal cells, which can replace Col6a1-cre lineage IMCs during the re-epithelization of the intestine, although the precise identities of these cells remain unknown. These results further suggest that reciprocal communication of the mesenchyme with the regenerating epithelium is crucial to orchestrate tissue regeneration, although the specific molecular pathways are not yet clear.
In conclusion, we have described the properties and identities of Col6a1-cre lineage colonic IMCs, including PDGFRαhi fibroblasts and perivascular cells, defined their ontogenic relationship with mesenchymal clusters and identified their role as orchestrators of intestinal morphogenesis and regulators of epithelial homeostasis. We have further supported the concept of mesenchymal plasticity both during homeostasis and tissue repair. In the future, it would be interesting to characterize the identities and role of these cells also in other intestinal disorders, including cancer, and define the molecular mechanism driving mesenchymal plasticity in homeostasis and disease.
Materials and methods
Mice and study approval
Col6a1Cre mice were described before [30]. Rosa26mT/mG and Rosa26iDTR mice were purchased from the Jackson Laboratory [22, 27]. All mice were maintained under specific pathogen free conditions in the Animal House of the Biomedical Sciences Research Center “Alexander Fleming”. All studies were performed according to all current European and national legislation and approved by the Institutional Committee of Protocol Evaluation in conjunction with the Veterinary Service Management of the Hellenic Republic Prefecture of Attika under the permissions 5759/15, 8443/17, 993/18, 448195/19.
DSS colitis induction
DSS-induced colitis was performed as previously described [41]. Briefly, 6–10-month-old mice received 2.5% DSS in their drinking water, followed by 1–14 days of regular water. Colitis induction was monitored by measuring weight loss.
Diphtheria toxin experiments
Col6a1DTR, control iDTR and Col6a1Cre mice were subjected to intra-rectal administration of 100 μl diphtheria toxin (Sigma-Aldrich) dissolved in 0.9% sodium chloride at 20 ng/g body weight. Pregnant mice were injected i.p. for two days (E14.5 and E15.5) with 5 μg DT.
Embryo manipulation
Timed pregnancies were set up by checking vaginal plugs to obtain E13.5, E14.5, E16.5 and E18.5 embryos. Pregnant females were sacrificed by cervical dislocation on the specific post coitum day and embryos were dissected in ice-cold PBS. Embryos were fixed overnight in 4% PFA/PBS, immersed in serial solutions of 15% and 30% sucrose/PBS and embedded in OCT for cryo-section preparations.
Isolation and culture of IMCs
Isolation and culture of IMCs were performed as previously described [42]. Briefly, the colon or small intestine was removed and digested as described above. The cell pellet was re-suspended in culture medium, consisting of DMEM (Biochrom), 10% FBS (Biochrom), 100 U/mL penicillin/100 mg/mL streptomycin (Gibco), 2 mM l-Glutamine (Gibco), 1 μg/ml amphotericin B (Sigma) and 1% non-essential amino acids (Gibco) and plated in cell culture flasks. The medium was changed after 3–24 h and cells were used after 2–4 days.
Lightsheet microscopy
For Lightsheet microscopy, the intestine of the embryos was isolated and fixed in 4% PFA/PBS overnight. Tissue clearing was achieved using the Scale A2 clearing solution for 2 weeks [43]. The Lightsheet Z.1 from ZEISS, equipped with sample chamber and Clr Plan-Apochromat 20x/1.0, Corr nd = 1.38 lens was used for experiments with tissue cleared by Scale medium, which has a refractive index of n = 1.38. Quantification of villi was performed using the Imaris Software.
Immunohistochemistry
For confocal microscopy and immunohistochemistry, mice were perfused with 4% PFA prior to the resection of the colon or small intestine. The tissue was then incubated in 4% PFA/PBS overnight and either immersed in serial solutions of 15% and 30% sucrose/PBS and embedded in OCT (VWR Chemicals) for cryo-section preparations or in 2% agarose for sectioning with a vibratome (Leica). Sections were subsequently blocked using 1%BSA in TBS containing 0.05% Tween 20 (Sigma) (TTBS) and stained with antibodies listed in Table Table1.1. Staining for CD201 was performed in unfixed tissue, embedded in 4% agarose in CO2 independent medium (ThermoFisher) and sectioned with vibratome. Sections were blocked in 5% FBS in CO2 independent medium and stained with anti-CD201 antibody. 2 h fixation in 4% PFA/PBS followed prior to the addition of secondary antibody. Mounting medium containing DAPI (Sigma-Aldrich) was used to stain the nuclei. Images were acquired with a Leica TCS SP8X White Light Laser confocal system.
Table 1
Antibodies used in flow cytometry and immunohistochemistry
| Antibody against | Conjugate | Clone/Cat. Number | Company | Use |
|---|---|---|---|---|
| CD45 | APC/Cy7 | 30-F11 | BioLegend | FC |
| CD45 | A700 | 30-F11 | BioLegend | FC |
| CD326 (EpCAM) | APC-efluor780 | G8.8 | eBioscience | FC |
| CD326 (EpCAM) | Unconjugated | G8.8 | eBioscience | IHC |
| Ter119 | Biotin | TER-119 | eBioscience | FC |
| Ter119 | APC-efluor780 | TER-119 | eBioscience | FC |
| CD31 | Biotin | MEC13.3 | BD Pharmingen | FC |
| CD31 | APC/Fire 750 | 390 | BioLegend | FC |
| Podoplanin | PE/Cy7 | 8.1.1 | BioLegend | FC |
| Podoplanin | Unconjugated | 14-5381-81 | eBioscience | IHC |
| PDGFRα (CD140α) | Unconjugated | #AF1062 | R&D Systems | IHC |
| PDGFRα (CD140α) | APC | APA5 | eBioscience | FC |
| α-SMA | FITC | 1A4 | Sigma | FC, IHC |
| α-SMA | efluor660 | 1A4 | Invitrogen | FC, IHC |
| α-SMA | Cy3 | 1A4 | Sigma | IHC |
| CD201 | APC | eBio1560 | eBioscience | FC, IHC |
| CD201 | PE/Cy5 | eBio1560 | Invitrogen | FC, IHC |
| CD34 | Biotin | RAM34 | eBioscience | FC, IHC |
| Streptavidin | A750 | S21384 | Invitrogen | FC |
| Streptavidin | A647 | S32357 | Invitrogen | FC |
| Streptavidin | PE | 554061 | BD Pharmingen | FC |
| Streptavidin | Brilliant Violet 421 | 405225 | BioLegend | FC |
| Rabbit IgG | A647 | A21244 | Invitrogen | IHC |
| Rat IgG | A594 | A11007 | Invitrogen | IHC |
| Rat IgG | A647 | A21247 | Invitrogen | IHC |
| Goat IgG | A647 | A11056 | Invitrogen | IHC |
| Hamster IgG | A647 | A21451 | Invitrogen | IHC |
| Dclk1 | Unconjugated | Ab109029 | Abcam | IHC |
| ChgA | Unconjugated | Ab15160 | Abcam | IHC |
| Ki67 | Unconjugated | Ab15580 | Abcam | IHC |
| BrdU | APC | 552598 | BD Pharmingen | FC |
FC flow cytometry, IHC immunohistochemistry
For histopathology, colon tissues were fixed in 10% formalin and embedded in paraffin. FFPE sections were stained with hematoxylin (Sigma-Aldrich) and eosin (Sigma-Aldrich) and colitis score assessment was performed as previously described [44]. Stainings for epithelial cell differentiation markers were performed in FFPE section using the antibodies listed in Table Table1.1. Signal detection and development were performed using Vectastain ABC-HRP Kit and ImmPACT DAB kit (Vector laboratories). Quantification of proliferating cells was performed in mice that were injected i.p. with 100 mg/kg BrdU (Sigma-Aldrich) 2 h prior to sacrificing them, using the BrdU detection kit (BD), according to the manufacturer’s instructions. The number of BrdU+ cells was quantified in at least 30 intact, well-oriented crypts per mouse.
Single molecule mRNA fluorescent in situ hybridization (smFISH)
Vibratome sections from the colon (70 μm) were used for smFISH staining. A smFISH probe set for Foxl1 was designed by the Stellaris FISH Probe Designer software coupled with the Quasar 670 Dye (Biosearch Technologies). Hybridization was performed using Stellaris hybridization and wash buffers, according to the manufacturer’s instructions (Biosearch Technologies). Images were acquired with the Leica TCS SP8X White Light Laser confocal system.
FACS analysis and sorting
Intestinal tissue preparations were prepared as previously described [21]. Briefly, colon or small intestine was removed, flushed with HBSS (Gibco), containing antibiotic–antimycotic solution (Gibco) and cut into pieces. Intestinal pieces were incubated with HBSS, containing 5 mM EDTA, DTT and Penicillin/Streptomycin (Gibco) for 30 min, at 37 °C, to remove epithelial cells. After vigorous shaking, the remaining pieces were digested using 300 U/ml Collagenase XI (Sigma-Aldrich), 1 mg/ml Dispase II (Roche) and 100 U/ml Dnase I (Sigma-Aldrich) for 40–60 min at 37 °C. For embryos, the intestine of the embryos was isolated, cut into pieces and digested using 100 μg/ml Collagenase P (Roche), 800 μg/ml Dispase II (Roche), 200 μg/ml Dnase I (Sigma-Aldrich) for 20 min at 37 °C. The cell suspension was passed through a 70 μm strainer, centrifuged and re-suspended in FACS buffer (PBS with 2% FBS). For stainings, 1–2 million cells/100 μl were incubated with the antibodies shown in Table Table1.1. For intracellular stainings, cells were fixed and permeabilized using the Fixation and Permeabilization Buffer Set (eBioscience), according to manufacturer’s instructions. Flow cytometric analysis of bromodeoxyuridine (BrdU) incorporation into IMCs was performed using the APC BrdU Flow Kit (BD Pharmingen) according to manufacturer instructions. BrdU (0.1 mg/g) was injected via intraperitoneal route into the mice for three consecutive days (days 3–5 of the DT administration protocol) and the levels of cell-associated BrdU were measured three days later (day 8). Propidium Iodide (Sigma) or the Zombie-NIR Fixable Viability Kit (Biolegend) was used for live-dead cell discrimination. Samples were analyzed using the FACSCanto II flow cytometer (BD) or the FACSAria III cell sorter (BD) and the FACSDiva (BD) or FlowJo software (FlowJo, LLC).
Crypt isolation and co-culture with IMCs
Intestinal crypts were isolated as described previously [45]. Briefly, the small intestine was flashed with cold PBS (Gibco), opened longitudinally and villi were scraped off using a coverslip. Then, it was cut into 5 mm pieces and washed extensively until the supernatant was clear. Ice-cold crypt isolation buffer (2 mM EDTA in PBS) was added to the fragments and stirred for 1 h at 4 °C. Fragments were allowed to settle down, the supernatant was removed and ice-cold 2 mM EDTA/PBS was added followed by pipetting up and down. Released crypts were passed through a 70-μm-cell strainer and the procedure was repeated until most of crypts were released. Crypt fractions were centrifuged at 300g for 5 min and re-suspended with ice-cold basal culture medium (Advanced DMEM/F12 (Gibco) supplemented with 2 mM GlutaMax (Gibco), 10 mM HEPES (Gibco) and 100 U/mL penicillin/ 100 mg/mL streptomycin (Gibco). Crypts were centrifuged again at 200g for 5 min, re-suspended in warm basal culture medium and counted. Crypts were mixed with Col6a1-GFP+ and GFP− IMCs sorted by FACS at passage 1 and subsequently re-suspended in Matrigel (BD Biosciences) at 250 crypts/50.000 IMCs/30 μl in 48-well plates. After Matrigel polymerization, culture medium was added in the wells, consisting of DMEM/F12 medium (Gibco), Glutamax (Gibco), Penicillin/Streptomycin, N2 supplement (Life Technologies, 1 ×), B27 supplement (Life Technologies, 1 ×), and 1 mM N-acetylcysteine (Sigma-Aldrich), 50 ng/ml EGF (Life Technologies), 100 ng/ml Noggin (PeproTech) and Rspo1 (PeproTech) where indicated. Images were acquired with the Zeiss Axio Observer Z1 microscope. Organoid measurements were performed using the ImageJ/Fiji software.
RNA isolation and qRT-PCR
RNA was isolated using the RNeasy mini kit or the RNeasy micro kit (Qiagen), depending on the number of cells, according to the manufacturer’s instruction. 100 ng–1 μg of RNA was used to generate cDNA using the MMLV reverse transcriptase by Promega and oligo-dT primers (Promega), according to the manufacturer’s instructions. For qRT-PCR, the SYBR Green PCR Master Mix (Invitrogen) was used according to the manufacturer’s instructions. Forward and reverse primers were added at a concentration of 0.2 pmol/ml in a final volume of 20 μl and qRT-PCR was performed on a CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad). The primer list can be found in Table Table22.
Table 2
List of primers used for real-time PCR
| Gene | Sequence (5′–3′) | Product (bp) |
|---|---|---|
| Bmp2 | F: ACCCCCAGCAAGGACGTCGT R: TGGAAGCTGCGCACGGTGTT | 137 |
| Bmp3 | F: TTTGCTGATATCGGCTGGAG R: TGGTGGCGTGATTTGATG | 125 |
| Bmp4 | F: CCCGCAGAAGGGCCAAAC R: TAGCCGGGTGGGGCCACAAT | 138 |
| Bmp5 | F: ACCTCTTGCCAGCCTACATG R: TGCTGCTGTCACTGCTTCTC | 169 |
| Bmp7 | F: TCCAAGACGCCAAAGAACCA R: TGCAATGATCCAGTCCTGCC | 140 |
| Wnt5a | F: GGTGCCATGTCTTCCAAGTT R: TGAGAAAGTCCTGCCAGTTG | 176 |
| Ednrb | F: TTGCGAGAGGCCTGTTTAGG R: GAGACCAACTCGTGCGGATT | 136 |
| Cspg4 | F: GGATGCCTCCAGGTCAGACT R: CTCCGTCAACAGACAGCACA | 142 |
| Casq2 | F: GCCCAACGTCATCCCTAACA R: CCCATTCAAGTCGTCTTCCCAT | 133 |
| Kcnj8 | F: CAAACCCGAGTCTGAGGACG R: TTCCTTTCACCATAGCCCGC | 81 |
| Foxl1 | F: ATAAACCAGGCTCCCCTTTG R: AGCCAAAGTACGTGCCAAAC | 75 |
| ChgA | F: CCAAGGTGATGAAGTGCGTC R: GGTGTCGCAGGATAGAGAGGA | 129 |
| Lgr5 | F: CCTACTCGAAGACTTACCCAGT R: GCATTGGGGTGAATGATAGCA | 165 |
| Ascl2 | F: AAGCACACCTTGACTGGTACG R: AAGTGGACGTTTGCACCTTCA | 115 |
| Hprt | F: TGCCGAGGATTTGGAAAAAGTG R: CACAGAGGGCCACAATGTGATG | 116 |
| Cox6c | F: CTGAGCCAAGAAAGAAGGCG R: TGAACTTCTTTGGAGCGCAA | 149 |
| B2m | F: TTCTGGTGCTTGTCTCACTGA R: CAGTATGTTCGGCTTCCCATTC | 103 |
3′ RNAseq sequencing and analysis
The quantity and quality of RNA samples from sorted cells were analyzed using the bioanalyzer form Agilent in combination with the Agilent RNA 6000 Nano. RNA samples with RNA Integrity Number (RIN) > 7 were further used for library preparation using the 3′ mRNA-Seq Library Prep Kit Protocol for Ion Torrent (QuantSeq-LEXOGEN™) according to manufacturer’s instructions. The quantity and quality of libraries were assessed using the DNA High Sensitivity Kit in the bioanalyzer, according to the manufacturer’s instructions (Agilent). Libraries were subsequently pooled and templated using the Ion PI IC200 Chef Kit (ThermoFisher Scientific) on an Ion Proton Chef Instrument or Ion One Touch System. Sequencing was performed using the Ion PI™ Sequencing 200 V3 Kit and Ion Proton PI™ V2 chips (ThermoFisher Scientific) on an Ion Proton™ System, according to the manufacturer's instructions. The RNA-Seq FASTQ files were mapped using TopHat2 [46], with default settings and using additional transcript annotation data for the mm10 genome from Illumina iGenomes (https://support.illumina.com/sequencing/sequencing_software/igenome.html). According to the Ion Proton manufacturers recommendation, the reads which remained unmapped were submitted to a second round of mapping using Bowtie2 [47] against the mm10 genome with the very-sensitive switch turned on and merged with the initial mappings. Through metaseqr R package [48], Genomic Ranges and DESeq were employed to summarize bam files of the previous step to read counts table and to perform differential expression analysis (after removing genes that had zero counts over all the RNA-Seq samples).
Downstream bioinformatics analysis and visualization tasks were performed using InteractiveVenn for Venn diagrams (www.interactivenn.net) [49] and the Functional Annotation tool from DAVID for Gene Ontologies (david.ncifcrf.gov) [50, 51]. Volcano plots and heatmaps were generated in R using an in-house developed script utilizing the packages ggplot2, gplots and pheatmap (https://cran.r-project.org/web/packages/pheatmap/index.html) [52–54]. RNA-seq datasets have been deposited in NCBI’s Gene Expression Omnibus [55] and are accessible through the GEO Series accession number GSE117308.
Comparison with single cell datasets
The positive cluster marker genes of the healthy and DSS-treated mouse, from the public dataset GSE114374, were used as gene signatures for each cell type identified in the single cell analysis. For each gene of a signature, z-scores of log2 normalized expression values were calculated for the bulk RNA-seq samples GC(GC1, GC2, GC3), TC(TC1, TC2, TC3), UC(UC1, UC2, UC3) and GDS(GDS1, GDS2, GDS3), TDS(TDS1, TDS2, TDS3), UDS(UDS1, UDS2, UDS3). In the boxplots of Figs. 1G and and5D,5D, the mean z-scores of GC1, GC2, GC3 and GDS1, GDS2, GDS3 are displayed, respectively.
Statistical analysis
Data are presented as mean ± SD. Statistical significance was calculated by Student’s t test or one-way ANOVA for multiple comparisons. The D’Agostino Pearson’s test was used to test if the dataset followed a normal distribution. Welch’s correction was used for samples that showed unequal variance. P values ≤ 0.05 were considered significant. Data were analysed using the GraphPad Prism 8 software.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
We would like to thank Michalis Meletiou for mouse genotyping. We would also like to thank the Genomics Facility of BSRC “Alexander Fleming”, and specifically Vaggelis Harokopos for performing all RNA sequencing experiments and Martin Reczko and Panagiotis Moulos for initial bioinformatics analyses. We acknowledge support of this work by the InfrafrontierGR Infrastructure, co-funded by Greece and the European Union (European Regional Development Fund), under NSRF 2014-2020, MIS 5002135), which provided mouse hosting and phenotyping facilities, including, histopathology, flow cytometry, and advanced microscopy facilities. This work was supported by a grant from the Stavros Niarchos Foundation to the BSRC “Alexander Fleming” as part of the Foundation’s initiative to support the Greek research center ecosystem, a grant from the Hellenic Foundation for Research & Innovation (H.F.R.I.) to MES (Grant no. 1687), the FP7 Advanced ERC grant MCs-inTEST (Grant Agreement no 340217) to GK, a grant from the European Crohn’s and Colitis Organisation to VK and a grant from the Fondation Sante to VK.
Author contributions
M-TM, AH, GK and VK contributed to the study conception and design. Material preparation, data collection and analysis were performed by M-TM, AH, AP, MES, NC, DM, SG, and VK. Bioinformatics analysis was performed by CT and PC. VK supervised the study. The first draft of the manuscript was written by M-TM, AH and VK. All authors provided comments on the manuscript, read, and approved the final version.
Funding
We acknowledge support of this work by the InfrafrontierGR Infrastructure, co-funded by Greece and the European Union (European Regional Development Fund), under NSRF 2014-2020, MIS 5002135), which provided mouse hosting and phenotyping facilities, including, histopathology, flow cytometry, and advanced microscopy facilities. This work was supported by a grant from the Stavros Niarchos Foundation to the BSRC “Alexander Fleming” as part of the Foundation’s initiative to support the Greek research center ecosystem, a grant from the Hellenic Foundation for Research & Innovation (H.F.R.I.) to MES (Grant no 1687), the FP7 Advanced ERC grant MCs-inTEST (Grant Agreement no 340217) to GK, a grant from the European Crohn’s and Colitis Organisation to VK and a grant from the Fondation Sante to VK.
Data availability
The RNA-seq datasets generated and analysed during the current study are accessible through the GEO Series accession number GSE117308.
Declarations
The authors have no relevant financial or non-financial interests to disclose.
All studies were performed according to all current European and national legislation and approved by the Institutional Committee of Protocol Evaluation in conjunction with the Veterinary Service Management of the Hellenic Republic Prefecture of Attika under the permissions 5759/15, 8443/17, 993/18, 448195/19.
Not applicable.
Not applicable.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Maria-Theodora Melissari and Ana Henriques contributed equally to this work.