A stromal lineage maintains crypt structure and villus homeostasis in the intestinal stem cell niche

Background The nutrient-absorbing villi of small intestines are renewed and repaired by intestinal stem cells (ISCs), which reside in a well-organized crypt structure. Genetic studies have shown that Wnt molecules secreted by telocytes, Gli1+ stromal cells, and epithelial cells are required for ISC proliferation and villus homeostasis. Intestinal stromal cells are heterogeneous and single-cell profiling has divided them into telocytes/subepithelial myofibroblasts, myocytes, pericytes, trophocytes, and Pdgfralow stromal cells. Yet, the niche function of these stromal populations remains incompletely understood. Results We show here that a Twist2 stromal lineage, which constitutes the Pdgfralow stromal cell and trophocyte subpopulations, maintains the crypt structure to provide an inflammation-restricting niche for regenerating ISCs. Ablating Twist2 lineage cells or deletion of one Wntless allele in these cells disturbs the crypt structure and impairs villus homeostasis. Upon radiation, Wntless haplo-deficiency caused decreased production of anti-microbial peptides and increased inflammation, leading to defective ISC proliferation and crypt regeneration, which were partially rescued by eradication of commensal bacteria. In addition, we show that Wnts secreted by Acta2+ subpopulations also play a role in crypt regeneration but not homeostasis. Conclusions These findings suggest that ISCs may require different niches for villus homeostasis and regeneration and that the Twist2 lineage cells may help to maintain a microbe-restricted environment to allow ISC-mediated crypt regeneration. Supplementary Information The online version contains supplementary material available at 10.1186/s12915-023-01667-2.


Background
The small intestinal villi, tissue responsible for nutrient absorption, are renewed every 3-5 days by Lgr5 + intestinal stem cells (ISCs) located at the crypt base [1].ISCs and Paneth cells form the stem cell zone and shape the crypt architecture [2][3][4].Paneth cells are derived from ISCs and located at the base of crypts, where they secrete anti-microbial peptides (AMPs) including defensin/ cryptdin/CRS molecules to fend off the microbiota (at the order of 10 2 -10 7 cells/g in small intestines in human) [5,6].The villi are easily damaged by food-borne pathogens, carcinogenic agents, or radiation, which are repaired by ISCs or stem cells derived from committed or differentiated cells [7,8].Some injuries disrupt the physical barrier and result in microbial invasion and inflammation, which present an extra threat to ISCs [9].The ISC activities in homeostasis and regeneration are regulated by niche cells, e.g., Paneth cells and stromal cells, which secrete molecules such as R-Spondin, Wnts, and BMPs [10][11][12][13].Although the depletion of Paneth cells does not affect villus homeostasis [14,15], there is evidence that both epithelial cells and stromal cells cooperate to support ISCs [16][17][18][19].
Herein, we tested whether genetic markers for bone marrow mesenchymal stem/stromal cells (BM-MSCs), Prrx1, Twist2, Nestin, and Acta2, labeled intestinal stromal cells [37].We show that the Twist2 lineage cells represent the largest among the four lineages examined.Single-cell profiling suggests that Twsit2 is expressed in portions of the trophocyte and Pdgfra low stromal cell subpopulations, but not Acta2 + subpopulations.In contrast to the findings that deletion of Wls in Ng2-or Gli1-marked stromal cells do not affect the intestinal villus [16,25], ablation of even one Wls allele in Twist2 lineage cells, but not Prrx1, Nestin, or Acta2 lineage cells, disturbed the crypt structure and villus homeostasis.Moreover, upon IR-induced injury, Wls haplo-deficiency resulted in decreases in the number of Paneth cells and the production of antimicrobial proteins, accompanied by increased inflammation and impaired crypt regeneration.These defects were largely rescued by the eradication of commensal bacteria with antibiotics.In addition, Acta2-marked telocytes, myocytes, and pericytes, but not Prrx1 or Nestin lineage cells, play a niche role in crypt regeneration but not homeostasis.This study thus uncovered ISC niche components with previously unidentified functions.

Labeling of intestinal stromal subgroups by Twist2, Prrx1, Nestin and Acta2
To understand the roles of various intestinal stromal subpopulations in villus homeostasis and regeneration, we tested whether BM-MSC markers, Prrx1, Twist2, and Nestin marked intestinal stromal cells [38][39][40][41].We crossed ROSA-tdTomato reporter mice to the Cre mouse lines driven by the promoter of these genes to perform genetic tracing.We showed that Prrx1, Twist2, and Nestin marked 32.6%, 38.1%, and 11.8% of non-epithelial intestinal villus cells, respectively in adult mice, and some of these cells were located at the base of the crypts (Fig. 1a).We also included Acta2 lineage cells for comparison since the roles of myofibroblasts remain not well understood in crypt regeneration compared to homeostasis [42,43].We found that Acta2 marked 24.2% of non-epithelial intestinal cells in Acta2-Cre;Rosa-tdTomato mice (Fig. 1a).In these lineage tracing experiments, none of the EpCAM + epithelial cells were Tomato + (data not shown).These results indicate that Twist2 marks the largest portion of intestinal stromal/myofibroblast cells (38.1%) among the 4 marker genes.
Single-cell RNA-seq analyses suggest that intestinal stromal cells are heterogeneous and can be divided into telocytes/subepithelial myofibroblasts, myocytes, pericytes, trophocytes, and Pdgfra low stromal cells (reviewed in [34]), although they may have redundant functions in controlling ISC activities [24,25].We re-analyzed the scRNA-seq data of intestinal cells and found that intestinal stromal/myofibroblast cells indeed formed the 5 subpopulations (Fig. 1b and additional file: Fig. S1a) [25].Moreover, Twist2, Prrx1, Acta2, and Nestin were expressed only in stromal/myofibroblast cells but not in immune cells and these 4 marker genes showed distinct expression patterns (Fig. 1b and additional file: Fig. S1a).Notably, Twist2 was mainly expressed in portions of trophocytes (Pdgfra low CD81 + ) and Pdgfra low CD81 − stromal cells, but not much in telocytes/subepithelial myofibroblasts, myocytes, or pericytes (Fig. 1b).Prrx1 was expressed in a portion of myocytes, Acta2 was expressed in telocytes/subepithelial myofibroblasts, myocytes, and pericytes, and Nestin was expressed in portions of Acta2expressing subpopulations (Fig. 1b), consistent with the observation that Nestin marked fewer cells than Acta2 in tracing experiments.We also compared the expression of Gli1 and Twist2 cells in trophocytes (Pdgfra low CD81 + ) and Pdgfra low CD81 − stromal cells and found that their expression showed minimal overlapping (Additional file: Fig. S1b).We also compared the expression of surface markers on the intestinal Twist2 and Gli1 lineage cells and found that Twist2 lineage cells were positive for CD29 and CD106 but negative for CD44, CD73, CD105, and Sca-1, while Gli1 lineage cells in Gli1-CreERT;Rosa-tdTomato mice were positive for CD29, CD44, and CD73 but negative for CD105, CD106, and Sca1 (Additional file: Fig. S2).Overall, these data suggest that the Twist2 lineage cells are distinct from cells marked by Prrx1, Nestin, Acta2, or Gli1.Note that we used a Twist2-Cre line in this study which cumulatively marked the Twist2 lineage cells.However, the labeling of only 38.1% of intestinal stromal/myofibroblast cells and restricted expression pattern of Twist2 in trophocyte and Pdgfra low subpopulations suggest that the Twist2 lineage may not contribute to telocyte, pericyte, and myocyte subpopulations.Certainly, further verification will need CreERT lines for Twist2, Acta2, Prrx1, and Nestin.

Depletion of the Twist2 lineage disrupts crypt structure and villus homeostasis
To determine which stromal populations(s) play a role in villus homeostasis, we used an inducible diphtheria toxin receptor system (ROSA26-iDTR mouse) to transiently deplete each of these populations [44].Daily injection of diphtheria toxin (DT) for 4 days led to a great reduction in the numbers of BM-MSCs in Prrx1-Cre; iDTR mice, Nestin-Cre; iDTR and Twist2-Cre; iDTR mice (Additional file: Fig. S3a and b), and immunostaining showed that αSMA + cells were reduced in the intestines of Acta2-Cre; iDTR mice (Additional file: Fig. S3c).These mouse lines showed increased mortality compared to DT-treated control mice and survived for different periods of time (Additional file: Fig. S3d).Depletion of the Twist2 or Acta2 lineage for 4 days led to the shortening of small intestinal tracts while depletion of Prrx1 or Nestin lineage for 7 days did not produce obvious phenotypes (Fig. 2a and Additional file: Fig. S3e).Depletion of the Twist2 lineage caused red and swollen intestines, indicative of inflammation.
Depletion of the Twist2 lineage led to thinned villi (Fig. 2b).The morphology of the crypt was altered compared to DT-treated ROSA-iDTR mice or PBS-treated Twist2-Cre; iDTR mice (Fig. 2b and Additional file: Fig. S3f ).The crypts were thinner with the boundary between crypts and villus becoming indistinct, accompanied by decreases in Ki67-positive proliferating cells and the number of Paneth cells with some Paneth signals detected at the upper part of the crypts (Fig. 2b).Furthermore, depletion of Twist2 lineage cells led to a decrease in the number of Lgr5 + ISCs, with some of Lgr5 + ISCs being dislocated (Fig. 2b).Thus, depletion of Twist2 lineage cells altered the structure of the crypts.Depletion of Acta2 lineage cells, but not Prrx1 cells, also caused a decrease in villus size and cell proliferation (Additional file: Fig. S3g), although to a much lesser extent than depletion of Twist2 cells.
We then focused on the Twist2 lineage cells.The Twist2-Cre mouse line had the Cre cassette inserted into the coding sequence, which disrupted this gene.In this study, we used the heterozygous Twist2-Cre mice to perform genetic tracing or gene ablation.The heterozygous Twist2-Cre mice showed normal villus structure, normal cell proliferation, and differentiation of goblet and Paneth cells (Additional file: Fig. S4a), suggesting that Twist2-Cre mice behaved like wild-type mice and can be safely used to trace Twist2 lineage cells or to ablate Twist2 lineage cells.Single-cell profiling revealed that Twist2 is not expressed in CD45 + immune cells or CD31 + endothelial cells (Additional file: Fig. S1a).Further, our previous genetic tracing and bone marrow graft assays revealed that Twist2-Cre;Rosa-Tomato mice did not show labeling of immune cells under normal conditions or in DSSinduced colitis model mice [45].
We tested the potential involvement of Twist2 lineage cells in crypt regeneration after IR-induced damage.IR at 6.5 Gy induces modest crypt and villus injury in normal adult mice, which can be repaired quickly, as previously reported [46].We found that Twist2 lineage depletion mice could survive less than 4 days post IR and they showed more severe damage than control mice, manifested by decreases in the numbers of villi, crypts, proliferating cells, Paneth cells, and ISCs (Additional file: Fig. S4b).These results indicate that Twist2 lineage cells play a role in crypt regeneration as well.

Twist2 lineage-secreted Wnts are needed for crypt structure integrity and homeostasis
The niche cells provide signaling molecules including Wnts to support ISC proliferation [22,28].Following up on the lead obtained from our cell ablation experiments, we analyzed the expression of Wnt molecules in Twist2 lineage stromal cells and found that these cells expressed some Wnt molecules, especially Wnt 2B, 3A, 4, and 5A (Fig. 3a).scRNA-seq analysis also showed that Twist2 + subpopulations expressed high levels of Wnt2B, 4, and 5A, as well as Rspo1-3 (Additional file: Fig. S5).Both Wnt and R-Spondin (encoded by Rspo) molecules are critical niche signals for ISCs.However, the Wnt expression pattern did not entirely overlap with the intestinal mesenchymal mixtures [20], consistent with the notion that the Twist2 lineage does not represent all intestine stromal cells.
To determine the significance of Wnt molecules secreted by Twist2 lineage cells, we ablated Wls, which encodes a sorting receptor for Wnt secretion [47], by crossing Twist2-Cre mouse to Wls f/f mouse that also carries a Lgr5-GFP allele to label ISCs.Homozygous deletion of Wls in Twist2 lineage cells led to embryonic lethality at E18.5 whereas Wls heterozygous mice were viable.We found that deletion of even one Wls allele in Twist2 lineage led to a reduction of Wls expression and a decrease in β-Catenin, a downstream target stabilized by Wnt activation (Fig. 3b, c, and Additional file: Fig. S6).Note that Wls deficiency also led to a decrease of β-Catenin in the enterocytes of the villi (Fig. 3c), suggesting that Wnt molecules secreted by Twist2 lineage cells located at the upper part of the villi may also play a role.Phenotypically, the mutant mice showed significant decreases in the number and size of villi and the numbers of proliferating cells and goblet cells (Fig. 3d), and a portion of the crypts appeared thinner with the boundary between crypts and villi being obscure (Fig. 3d).Moreover, the number of Lgr5 + ISCs was reduced and some ISCs showed mislocalization (Fig. 3d).Thus, deletion of one Wls allele in Twist2 lineage disrupts villus homeostasis as well as the crypt structure.
Ablation of Wls in the Prrx1 lineage led to embryonic lethality while the heterozygous mice were viable.Prrx1-Cre; Wls +/f mice showed normal levels of β-Catenin in enterocytes and unaltered number and size of villi or crypts (Additional file: Fig. S7a and b).Acta2-Cre; Wls f/f mice, which were viable, showed no obvious defects in villus and crypt structure (Additional file: Fig. S7a  and b).One explanation is that the Prrx1 or Acta2 lineage stromal cells synthesize much less Wnt molecules than Twist2 lineage cells, which warrants further investigation.These results highlight the importance of Twist2 cells-secreted Wnts in villus homeostasis.

Wnts secreted by Twist2 lineage are required for villus and crypt development
Mouse intestinal villi first emerge at E14.5 [48].Twist2 lineage cells were detected in intestinal structures as early as E14.5 (Fig. 4a).An early study showed that Twist2-Cre; Wls f/f mice were lethal at E13.5 and the in vitro cultured small intestine showed a defect in epithelial cell proliferation [49].Our Twist2-Cre; Wls f/f mice died at E18.5, likely due to different genetic background.We analyzed the intestine of E16.5 embryos and found that deletion of two Wls alleles in Twist2 lineage cells led to decreases in Wls expression and β-Catenin (Fig. 4b and c), which were associated with smaller villi and decreases in proliferating cells and goblet cells (Fig. 4d).Moreover, the E16.5 Twist2-Cre; Wls +/f embryos also showed smaller villi and decreases in proliferating cells and goblet cells (Fig. 4bd).The intestinal crypts are formed around p12 [3,50].We found that P12 Twist2-Cre; Wls +/f mice exhibited decreases in Wls expression and β-Catenin (Fig. 4e and  f ), which were associated with decreases in the crypt size and the numbers of proliferating cells, goblet cells, and Paneth cells (Fig. 4g).Thus, Wnt molecules secreted by the Twist2 lineage play important roles in villus development and crypt formation.

Twist2 stromal cell-secreted Wnts are needed for crypt regeneration
Moreover, Twist2-Cre;Wls +/f mice showed defects in villus/crypt regeneration, manifested by decreases in the size of villi, the number of crypts, the number of proliferating cells, and the numbers of goblet cells and Paneth cells at day 4 or 7 following IR (Fig. 5a).The difference in the villi and crypts between the mutant and control mice at day 7 post IR was much greater than that without IR (Fig. 5a), suggesting that the repair process was impaired in Twist2-Cre;Wls +/f mice.
Prrx1-Cre;Wls +/f mice did not show obvious defects in crypt regeneration following IR (Additional file: Fig. S8).However, the Acta2-Cre;Wls f/f mice could survive only 4-5 days after IR and villus regeneration was compromised and the number and size of crypts and the numbers of proliferating cells and goblet cells were reduced (Fig. 5b).Thus, Wnt molecules secreted by Acta2-marked cells play critical roles in crypt regeneration but not in homeostasis.

Synthesis of AMPs and inflammatory cytokines during normal crypt regeneration
Paneth cells mainly exist in the small intestines, where they secrete anti-microbial peptides to fend off commensal bacteria.A recent study showed that Paneth cells may dedifferentiate to ISCs following IR [8].We found that after IR, apoptosis occurred to enterocytes and the number of Paneth cells went down but recovered later (Additional file: Fig. S9a).Interestingly, we observed a peak expression of AMPs including Crypt1, Crypt4, Plys, CRS1C, and CRS4C, but not mLys, Reg3g, TCF4, or MMP7 at day 2 post IR, which might be attributable to surviving Paneth or other cells (Additional file: Fig. S9b).We also found an increase in the levels of IL1β, IL6, IL17α, and IL22 at day 3 post IR, which went down at day 4, while the levels of TNFα and IFNγ were modestly increased, which persisted through the regeneration process (Additional file: Fig. S9c), suggestive of inflammation.AMPs secreted by Paneth cells may help restrict bacterial propagation and generate an environment suitable for crypt regeneration.

Decreased AMPs and increased inflammation in regenerating intestines of Twist2-Cre; Wls +/f mice
In accordance with a decrease in Paneth cells in the intestine of Twist2-Cre; Wls +/f mice following IR (Fig. 5a), immunostaining revealed that the signals for anti-microbial proteins Lysozyme and CRCS4C were reduced (Fig. 6a).Expression of Crypt1 was also reduced in Twist2-Cre; Wls +/f mice, under homeostatic conditions or at day 4 post IR (Fig. 6b).It has been established that loss or reduction of Paneth cells or AMPs production allows for commensal microbial infection, which led to aggravated intestinal inflammation [51].We indeed found that IR-induced injury led to an increase in the levels of inflammatory cytokines including IL6 and TNFα in the intestinal samples of Twist2-Cre; Wls +/f mice compared to control mice (Fig. 6c).Overall, these findings suggest that Wnts secreted by Twist2 stromal cells may help promote Paneth cell recovery and AMP secretion to contain microbes and inflammation.

Antibiotics partially rescue villus repair defects in Twist2-Cre; Wls +/f mice
Emerging evidence suggests that persistent inflammation impairs tissue stem cells including ISCs [52].We showed that LPS, a bacterial product, augmented cytokine synthesis and impeded crypt regeneration following IR (Additional file: Fig. S10a and b).We suspect that enhanced intestinal inflammation in Twist2-Cre; Wls +/f mice is likely caused by the invasion of commensal microbes due to decreases in the number of Paneth cells and the production of AMPs.To verify this, we pretreated these mice with a combination of antibiotics to eliminate commensal bacteria before IR.We found that the levels of inflammatory cytokines were suppressed in the intestines of antibiotics-treated Twist2-Cre; Wls +/f mice (Fig. 6c).Moreover, the crypt regeneration defects in Twist2-Cre; Wls +/f mice were partially rescued as well (Fig. 6d).These results collectively suggest that the Twist2 lineage-produced Wnt molecules help establish an inflammation-restricted environment to facilitate crypt regeneration.

Discussion
The current study shows that the small intestinal stromal/myofibroblast subpopulations can be genetically marked by Prrx1, Twist2, and Nestin in addition to Acta2.Single-cell profiling data suggest that these markers are expressed in different subpopulations of the small intestine.In particular, the Twist2 lineage cells form the Pdgfra low stromal cell and trophocyte subpopulations, while Acta2 lineage cells form the telocyte, myocyte, and pericyte subpopulations.While Nestin-expressing cells showed some overlapping with Acta2-expressing cells, Prrx1 is expressed in a portion of the myocytes.Moreover, Twist2-expressing cells and Gli1-expressing cells in the Pdgfra low stromal cells and trophocytes show little overlapping.While our findings support the clustering of intestinal stromal/myofibroblast cells [24,25,34], and that Twist2-marked cells are different from those marked by Prrx1, Acta2, or Nestin, the exact relation between these 4 subpopulations needs further investigation.Additionally, future studies are needed to reconcile the genetic tracing results and the scRNA-seq results.This may require more specific genetic markers and genetic tracing (at different times) in combination with singlecell profiling of the daughter cells.
We show that Twist2 stromal cells exhibit important niche activities in ISC-mediated villus homeostasis and regeneration, by secreting Wnt molecules.Moreover, deletion of one allele of Wls in Twist2 lineage cells generates more severe villus defects than depletion of Wls in Gli1 + cells [16].Killing Twist2 lineage cells also produces a more severe villlus phenotype than killing intestinal stromal cells marked by Gremlin1 [24], which was recently shown to be a marker of skeletal stem cells and trophocytes in the intestine [37].This can be explained that the Twist2 lineage cells comprise Pdgfra low stromal cells and trophocytes, both of which contribute to the regulation of ISCs by secretion of Wnt molecules.Comparison of Ng2-and Bapx1-marked intestinal stromal cells also supports the redundancy of stromal subpopulations in the regulation of ISC activity [25,35].Moreover, ablation of Wls in Twist2 lineage cells, but not the Prrx1, Acta2, or Nestin subpopulation, impairs villus homeostasis, highlighting the importance of the Twist2 lineage cells in this event.
Twist2 lineage cells appear to maintain the crypt morphology as well.We show that depletion of Twist2 lineage cells or deletion of one allele of Wls from Twist2 lineage cells alters crypt size and morphology including change in crypt length and the villus-crypt boundary and mislocalization of some ISCs.This is in contrast to telocyte cell ablation mice, telocyte-specific Porcn deletion mice, and mice with Wls deleted simultaneously in Gli1 stromal cells and Villin + enterocytes [16,28], which all caused shrinkage and collapse of the villi and crypts without distorting the crypt morphology.In addition, deletion of Wls or Porcn in Ng2 or Gli1 cells does not obviously affect the crypt morphology in small intestines [16,25].These findings suggest that Wnts secreted by Twist2 stromal cells localized in the upper part of the crypt may play a role in maintaining the crypt structure.
Findings of this study support the notion that crypt regeneration requires a more demanding niche.We show that Twist2 lineage cell-generated Wnts are required for proper formation and localization of Paneth cells, production of AMPs, and restriction of inflammation in injured intestines.Moreover, deletion of Wls in Acta2 lineage cells (telocytes, pericytes, and myocytes) impaired crypt regeneration but not villus homeostasis, this is inconsistent with the finding that deletion of Porcn in Foxl1 + telocytes causes collapse of the villi [28].One explanation is that Porcn and Wntless may have non-overlapping functions [31][32][33].Recent studies have shown that Ng2 + cells-secreted Wnt molecules play more important roles in crypt regeneration than in homeostasis [25] and that even macrophage-secreted Wnts are required for crypt regeneration [53].Overall, these findings suggest that under injury conditions, there might be a greater need for Wnt molecules than in homeostasis, which are supplied by multiple cell types and a defect in even one source may impair crypt regeneration [35].
During crypt regeneration, newly formed Paneth cells appear to shape the crypt morphogenesis and secrete anti-microbial peptides [54].Its importance is highlighted by its role in the pathogenesis of intestinal Crohn's disease.Notably, Paneth cell recovery and synthesis of anti-microbial peptides require Wnt signaling [3].Our findings provide evidence that Twist2 lineage stromal cells provide the Wnt molecules required for the activities of Paneth cells and imply that the interaction between Twist2 stromal cells and Paneth cells may generate a microbe-restricted and inflammation-restricted environment to support ISCs and crypt regeneration.Certainly, this warrants further investigation.

Conclusions
While Prrx1 and Nestin lineage BM-MSCs constitute niches for hematopoietic stem cells, we show that Twist2 lineage stromal cells act as a niche for ISCs during villus homeostatic and regeneration via secretion of Wnt molecules.We also show that Twist2 stromal cells help control the structure of the crypts and may provide a unique niche for ISCs during regeneration.

Mouse lines and maintenance
All animal work was in compliance with the recommendations in the National Research Council Guide for the Care and Use of Laboratory Animals, with the protocols approved.The ROSA-iDTR mouse was generated in Waisman's laboratory and Prrx1-Cre, Twist2-Cre, Nestin-Cre, Acta2-Cre, Gli1-Cre/ERT, Lgr5-GFP-Cre/ERT, and ROSA-tdTomato mouse lines were purchased from The Jackson Laboratory.
To deplete a given type of cells, 2-month-old iDTR mice expressing Cre were injected intraperitoneally with 100 ng of DT (Sigma) twice per day for several consecutive days.To irradiate the mice, lead plates were used to protect other parts of the mice, only the exposed belly received radiation with RAD SOURCE RS 2000 irradiator.Antibiotics treatment (ampicillin: 0.5 g/L, vancomycin 0.25 g/L, metronidazole 0.5 g/L, and neomycin: 0.25 g/L) of the mice was started two days before IR and sustained until the mice were sacrificed.LPS (2.5 mg/ kg) was given at the same time of IR by intraperitoneal injection.
To generate the villus/crypt regeneration model, adult male mice were irradiated at 6.5 Gy (only to the abdomen region) and the mice were euthanized after different periods of time.The wildtype mice survived well and the body weight was maintained, as previously reported [46].
For antibiotic treatment, the combined antibiotics (ampicillin: 0.5 g/L, vancomycin: 0.25 g/L, formazan nitazole 0.5 g/L, and neomycin: 0.25 g/L) were added to drinking water till the end of the experiment.

H/E staining and lineage tracing
After the mice were sacrificed, the full intestine was removed and the mesenteric fat was cleaned.The length of the intestine was measured.The upper part of the intestine (1 cm) was removed and fixed in 4% paraformaldehyde overnight.Samples were then dehydrated, embedded in paraffin, sectioned at 4 μm through a longitudinal crevice, and prepared for staining.For the cryostat section, intestine samples were embedded in OCT (Leica, 14,020,108,926) and frozen in liquid nitrogen.Sections were cut at 6 μm in thickness at − 20℃.To trace tomato-positive cells, cryostat sections were washed, counterstained with mixed DAPI & Mounting solution (1:1), and covered glasses for observation (Upright Microscope Nikon ECLIPSE 80i).H/E staining and ALP staining (Fast Blue RR Salt and Naphthol AS-MX phosphate, Sigma) were carried out following the standard protocols.

Immunohistochemical staining
The paraffin sections were dewaxed and rehydrated.The samples were treated with 3% H 2 O 2 for 20 min to quench the endogenous peroxidase activity, permeabilized with Triton X100 (0.1%) in PBS for 30 min.Antigen retrieval was performed with heated 1X sodium citrate.The samples were blocked, incubated with primary antibodies and secondary antibodies that were conjugated to horseradish peroxidase (HRP) as described for immunofluorescent staining.Afterwards, the samples were incubated with a DAB stain kit (Boster), washed, and incubated with the SABC solution at 37℃ for 1 h in darkness.The slides were then dehydrated through 2 changes of 95% alcohol and absolute alcohol, cleared in two changes of xylene, and mounted with resin for observation.Antibodies against β-Catenin (1:150; CST; 610153), Ki67 (1:200; Abcam; ab16667), lysozyme (1:200; Abcam; ab108508), and GFP (1:150; CST; 2956S) were used.

Paneth cell and goblet staining
The paraffin sections were deparaffinized with xylene and rehydrated through graded alcohol to deionized water.The samples were stained in Alcian Blue solution for 30 min, rinsed in running tap water for 5 min, placed in periodic acid solution (Periodic Acid-Schiff (PAS) Kit, Sigma) for 5 min, rinsed in several changes of deionized water, placed in Schiff Reagent for 15 min, and washed in lukewarm tap water for 10 min.The samples were then dehydrated through two changes of 95% alcohol and absolute alcohol, 2 min each, cleared in 2 changes of xylene, and mounted with resin for observation.

Intestinal stromal cell isolation
For intestinal stromal cell isolation, mouse intestines were cut into 0.5-1 cm pieces.Tissues were rinsed with cold PBS + P/S for 2 min, immersed into 10 ml 1 mM DTT solution (Sigma), and rocked at 200 rpm for 10 min.The tissues were rinsed again for 2 min, incubated in 10 ml 3 mM EDTA/1 mM HEPES Buffer with rocking at 200 rpm for 10 min at 37℃, transferred into new 6 cm plates, and incubated with digestion solution (100U/ ml collagenase VIII and 100U/ml DNase I) for 30 min.The samples were pipetted up and down to break up the tissue.The isolated cells were centrifuged with a low speed at 250 g for 5 min and single-cell suspension was obtained using a 70 μm strainer (BD Falcon).

FACS analysis and cell sorting
The isolated intestinal stromal cells were analyzed for MSC markers using flow cytometry using a mouse MSC marker antibody panel kit (R&D, Catalog Number SC018), following the protocol provided by the manufacturer.Analysis was done using a BD FACSCalibur.For cell sorting, the isolated cells were resuspended with Flow Cytometry Staining Buffer.Flow cytometry was performed using Beckman MoFlo ™ XDP high-speed cell sorter.

RNA isolation and quantitative PCR
Tissue samples or sorted cells were frozen in liquid nitrogen for storage and total RNA was isolated using TRIzol (Invitrogen).RNA was reverse transcribed to cDNA with PrimeScript ™ RT reagent Kit (Takara.Code No. RR037A).Quantitative PCR was carried out in triplicate on LightCycler ® 96 Real-Time PCR System (Roche) with a FastStart Universal SYBR Green Master (Roche), using the cDNA samples as templates.The primers for each gene of interest were shown in Additional file: Table S1.mRNA was quantified using the 2-ΔΔCt model, where ΔΔCt = ΔCtexperimental − ΔCtcontrol (ΔCt = Ctgene of intererst − Cthousekeeping gene).

Bone marrow colony forming unit assays
The bone marrow was flushed out with α-MEM culture medium from the femurs and tibiae.Red blood cells were cracked with red blood cell lysis buffer.Single-cell suspensions were prepared by filtration through a 40-μm strainer.1 × 10 6 cells bone marrow cells were plated in 3.5 cm dishes and allowed to grow in α-MEM medium containing 10% fetal bovine serum, 1% L-glutamine, and 1% penicillin-streptomycin (Gibco, Invitrogen Corporation, USA).After 3 days, the non-adherent cells were removed and the adherent cells were sub-cultured.Medium was replaced every 3 days.After 7 days, colonies were stained.

Single-cell RNA analysis
Raw reads were demultiplexed and mapped to the mouse reference genome with the Cell Ranger version 3.0.1 (10X Genomics) pipeline using the default parameters.The generated gene-cell expression matrix was used for subsequent analysis in R version 3.6.1 using Seurat version 3.1.5.

Quantitation and statistical analyses
To quantitate the cells in the villi, cells on ten views of three sections from at least 3 mice were counted.Data are given as mean and standard error of mean (SEM) of the results from > 3 samples in each experiment.Statistical analysis and plotting were processed with Graphpad Prism5.Differences between the two groups were measured by the Student's t-test.Two-way ANOVA was performed to compare two corresponding data points.P < 0.05 is defined as being significantly different.*P < 0.05, **P < 0.01, and ***P < 0.001.

ISCs
Intestinal stem cells AMPs Anti-microbial peptides scRNA-seq Single-cell RNA-sequencing Wls Wntless BM-MSCs Bone marrow mesenchymal stem/stromal cells • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year

•
At BMC, research is always in progress.

Learn more biomedcentral.com/submissions
Ready to submit your research Ready to submit your research ?Choose BMC and benefit from: ? Choose BMC and benefit from:

Fig. 1
Fig. 1 Lineage tracing and scRNA-seq analysis of Twist2 lineage cells in the intestinal villi.a Detection of stromal lineages labeled by Twist2, Prrx1, or Nestin, and Acta2 + myofibroblasts in the small intestinal villi of adult mice.Right panel: the percentage of Twist2, Prrx1, and Nestin lineage cells and Acta2 lineage myofibroblasts in non-epithelial (EpCAM + ) cells.Bar: 50 μm.The calculation was done by counting Tomato + cells on six views per mouse and divided by the number of non-epithelial DAPI + nuclei, N = 6.b Single-cell transcriptome analysis of small intestinal cells.A scRNA-seq dataset (PMID 32084389) was used to analyze the expression of various stromal/myoblast cell markers