Villification of the intestinal epithelium is driven by Foxl1

Abstract The primitive gut tube of mammals initially forms as a simple cylinder consisting of the endoderm-derived, pseudostratified epithelium and the mesoderm-derived surrounding mesenchyme. During mid-gestation a dramatic transformation occurs in which the epithelium is both restructured into its final cuboidal form and simultaneously folded and refolded to create intestinal villi and intervillus regions, the incipient crypts. Here we show that the mesenchymal winged helix transcription factor Foxl1, itself induced by epithelial hedgehog signaling, controls villification by activating BMP and PDGFRa as well as planar cell polarity genes in epithelial-adjacent telocyte progenitors, both directly and in a feed- forward loop with Foxo3. In the absence of Foxl1-dependent mesenchymal signaling, villus formation is delayed, the separation of epithelial cells into mitotic intervillus and postmitotic villus cells impaired, and the differentiation of secretory progenitors blocked. Thus, Foxl1 orchestrates key events during the epithelial transition of the fetal mammalian gut.


Introduction
The surface area of the human small intestine measures about 30 m 2 , or the size of half a badminton court 1 .This large area is required to enable effective digestion and absorption of nutrients.The major factor responsible for increasing the size of the intestinal epithelium is villi cation, or the formation of intestinal villi that project into the gut lumen, which increase the surface area compared to a at epithelium by close to 100-fold 1 .Remarkably, while villi are a feature of the small intestine in mammals and birds, they arise by divergent morphogenetic mechanisms.Thus, villus formation in birds is dependent on the sequential formation of rst inner circular, then longitudinal, and nally muscularis mucosa muscles which coincide with the appearance of epithelial ridges, zigzags, and villi, respectively 2 .
In contrast, in mammals, villus formation can proceed in the absence of tensile forces generated by intestinal muscles; rather, it is dependent on PDGFRa/BMP positive mesenchymal cell clusters 3 , which themselves are induced by epithelial hedgehog signaling 4 .
At the onset of villi cation, also termed 'epithelial transition', the endoderm-derived epithelium forms a simple tube of pseudostrati ed cells with a layer thickness of ~ 50 µm.This pseudostrati ed epithelium, characterized by cell cycle-dependent interkinetic nuclear migration 5 , is then changed to a columnar epithelium coincident with villus formation.BMP signals from villus cluster mesenchymal cells restrict proliferation in the overlying epithelium so that cycling cells become restricted to intervillus regions, the precursors of the small intestinal proliferative lcrypts present in the adult.The critical role of BMP signaling in villus cluster formation was established by explant cultures of presumptive small intestine from 13.5 dpc (days post conception) embryos treated with localized sources of BMP or the pan-BMP inhibitor dorsomorphin 6 .
The induction of villus formation and the formation of villus clusters in the underlying mesenchyme are dependent on epithelial to mesenchymal hedgehog signaling.Thus, both Sonic Hedgehog (shh) and Indian Hedgehog (Ihh) are expressed in the epithelium, while the hedgehog receptors (Ptch1 and Ptch2) as well as the downstream transcription factor Gli1 are expressed in the mesenchyme 7,8 .Because the two epithelial hedgehog proteins are partially redundant, Madison and colleagues addressed the role of hedgehog signaling in the developing intestine using epithelial expression of the pan-hedgehog inhibitor Hhip 9 .Suppression of hedgehog signaling impaired villus formation, which was partially due to decreased BMP expression in the underlying mesenchyme.
Previously, the winged helix transcription factor Foxl1 (formerly termed Fhh6) was shown to be expressed in the rst one to two cell layers of the mesenchyme juxtaposed to the epithelium before villus formation by mRNA in situ hybridization 10 , which was recently con rmed by immuno uorescence staining 11 , making it an excellent candidate as regulator of key developmental processes.Mice null for Foxl1 exhibit delayed villus formation and a defect in the inhibition of epithelial proliferation in nascent villi, suggesting Foxl1 as an important transcription factor controlling mesenchymal to epithelial signaling 10 .A link to epithelial hedgehog signaling was subsequently established by the identi cation of functionally relevant binding sites for the hedgehog-dependent Gli transcription factors within an evolutionarily ultra-conserved enhancer at the Foxl1 locus 12 .Foxl1 expression was induced in fetal gut mesenchyme explants treated with Shh, while its mRNA levels were reduced in mice de cient for the hedgehog dependent transcription factors Gli2 and Gli3.Taken together, these ndings suggested that Foxl1 is a critical mediator of epithelial to mesenchymal cross-talk during villus formation.Here, we set out to determine the molecular targets and pathways controlled by Foxl1 during intestinal villi cation.

Results
Histological analysis of the small intestine shows that as villus formation is well on its way in wild type fetuses at 15.5 dpc, while this process has barely been initiated in Foxl1 de cient mice (Fig. 1A,B).Even two days later, at 17.5 dpc, villi cation is abnormal in mutant mice, with apparent bridging of nascent villi across the gut lumen, likely re ecting persistent epithelial ridges (Fig. 1C, D), a phenotype that persists at 18.5 dpc (Fig. 1E,F).In order to investigate this phenotype in three dimensions, we performed scanning electron microscopy.As shown in Fig. 1G-L, in control mice villi cation occurs through the formation of regularly spaced short invaginations into the gut lumen, which by 18.5 dpc have progressed to form elongated villi.This process is dramatically altered by loss of Foxl1, with absence of the regularly spaced nascent villi and presence instead of long epithelial ridges at 15.5 dpc (yellow arrow in Fig. 1H), which persist until late gestation (Fig. 1J, red arrow).The presence of epithelial ridges instead of villi is reminiscent of the phenotype seen in small intestinal explants treated with the pan-BMP inhibitor dorsomorphin 6 , a connection we explore further below.

Fetal gut telocytes progenitors exist in two subpopulations
The inclusion of tdTomato in the recently developed Foxl1 mutant allele employed here 13 enabled us to follow the fate and determine the molecular properties of Foxl1-positive cells embryos heterozygous and homozygous for this Foxl1 null allele.Of note, the Foxl1 phenotype is recessive, as Foxl1 heterozygous mice are indistinguishable from wild type controls 10,[14][15][16] .We performed scRNAseq on the proximal half of the small intestine of 15.5 dpc FoxL1 CreER-tdTom/+ fetuses (Fig. 2A,B).As shown in the UMAP analysis in Fig. 2C and D, Foxl1 + cells segregate into two closely related cell clusters, which we termed 'telocyte progenitors 1 and 2'.A list of marker genes for all cell populations shown in Fig. 2C is given in Supplementary Table 1.Telocyte progenitors 1 express high levels of Pdgfra and multiple Bmp mRNAs (Fig. 1E), and thus most likely correspond to 'villus cluster cells', i.e the PDGFRa positive cells important in villus formation originally identi ed by Karlsson and colleagues 3 .By exclusion, the telocyte progenitor 2 population likely represents Foxl1 + cells directly adjacent to the intervillus epithelium, i.e. the presumptive future crypt cells.Interestingly, we identi ed Glp2r, encoding the GLP2 receptor, as another gene predominantly expressed in telocyte progenitor 1 cells, and con rm the localization of the Glp2r mRNA to villus cluster cells by RNAscope analysis (Fig. 2F,G).A further telocyte progenitor 1 marker is the Zinc-nger transcription factor Spalt like transcription factor 1 (Sall1), which is part of the NuRD transcriptional repressor complex (Fig. 2H,I).Sall1/Foxl1 double positive cells can be seen to also mark a cluster of cells that appears to be in the process of initiating a new villus (Fig. 2I).Importantly, this single cell analysis clearly shows that Foxl1 + cells are distinct from myo broblast, interstitial cells of Cajal, and pericytes, identi ed by the expression of Acta2 med /Myh11 med /Des med /Tagln med , Etv1 + /Kit + /Acta low , and Cspg4 + /Pdgfrb + /Abcc9 + or Abcc9 + /Ndufa4l2 + , respectively (Fig. 2C).Major hallmarks of the telocyte progenitor 1 and 2 populations are summarized in the graphic shown in Fig. 2J.

Expression of PDGFRa in telocyte progenitors is dependent on Foxl1
Next, we analyzed the expression pro le of the fetal small intestine of Foxl1 null (FoxL1 CreER-tdTom/CreER- tdTom ) mice and compared them to those of heterozygous fetuses.Note that in both control and Foxl1 null fetuses, Foxl1-expressing cells are localized to the mesodermal cell layer that is directly juxtaposed to the developing epithelium, and loss of Foxl1 does not affect telocyte progenitor number (Fig. 3A).The UMAP plot shown in Fig. 3B clearly demonstrates that the telocyte progenitor 1 population is dramatically reduced in abundance in the Foxl1 null intestine.As mentioned above, among the markers of the telocyte progenitor 1 cells is PDGFRa.The scRNAseq data shown in Fig. 3C as well as the immuno uorescence staining presented in Fig. 3D demonstrate that PDGFRa expression is Foxl1 dependent.To determine if Pdgfra is a direct target of Foxl1, we performed Cut-and-Run assays on telocytes from 15.5 dpc fetal gut.However, we found no Foxl1 binding event within 10 kb of the Pdgfra promoter (data not shown).
It was previously reported that the winged helix transcription factor Foxo3 is a transcriptional activator of Pdgfra, which binds to an evolutionarily conserved cis-regulatory element in its proximal promoter 17 .
Therefore, we hypothesized that Foxl1 might indirectly control Pdgfra via activation of Foxo3.Using Cutand-Run of sorted fetal telocytes, we indeed found several Foxl1 binding sites in the Foxo3 promoter (Fig. 3E).In addition, our scRNAseq data show a reduction of Foxo3 transcript levels in the telocyte progenitor 1 population of Foxl1 null fetuses (Fig. 3F).These data suggest indirect regulation of the Pdgfra gene by Foxl1 through a transcription factor cascade via Foxo3.

Foxl1 controls multiple Bmp genes in telocyte progenitor cells
Next, we focused on BMP proteins, as BMP signaling from the mesoderm to the epithelium is critical for villus cluster formation 6 .Telocyte-produced BMPs enriched in the villus cluster cells signal to the overlaying endoderm to inhibit Wnt signaling and limit proliferation (Fig. 4A).High expression of several BMP genes was present in particular in villus custer telocyte-progenitor 1 cells (Fig. 4B).In case of BMP4, expression also extends to Foxl1 negative GPX3 + FLCs; however, its levels are clearly reduced speci cally in Foxl1-de cient telocytes (Fig. 4B).To determine if BMP signaling to the epithelium is impaired by this reduction in telocyte BMP expression, we performed immunostaining for phosphorylated SMAD1/5 (Fig. 4C).Nuclear pSMAD1/5 is clearly detectable in epithelial cells in the villus tip but not intervillus regions in control embryos.In contrast, epithelial cells in Foxl1 null mice are devoid of signal, con rming loss of active BMP signaling to the epithelium.
When analyzing the pSMAD1/5 staining, we also noticed signal in nuclei of mesenchymal cells within the invaginating villi, with most of them negative for the Foxl1-tdTomato signal.These ndings suggest unexpected bi-directional signaling of villus tip telocytes to both epithelium and neighboring mesenchymal cells.At present, the signi cance of this observation is unknown; however, the pSMAD1/5 signal in mesenchymal cells is also Foxl1-dependent (Fig. 4C).
Loss of mesenchymal BMP signals is expected to result in de-inhibition of Wnt signaling in the epithelium overlying villus cluster telocytes.Indeed, we found expression of the Wnt target gene Sox9 expanded from the developing crypts to nascent villi in the Foxl1 null fetal intestine (Fig. 4D).Likewise, epithelial proliferation was not con ned to the nascent crypts but extended to the villus epithelium in mutant mice (Fig. 4D).Thus, Foxl1 is a critical factor required for the demarcation of the postmitotic villus from the mitotic intervillus epithelium.The schema in Fig. 4E summarizes these ndings.

Foxl1 is required for mesenchymal expression of planar cell polarity genes
Patterning of the developing gut epithelium is clearly perturbed in the absence of Foxl1, and the hyperproliferation of the epithelium due to lack of BMP signaling documented above regionally leads to an apparent multilayered epithelium, in which mesenchyme-distal cells undergo apoptosis as indicated by cleaved caspase 3 staining (Fig. 5A).Recently, planar cell polarity genes were identi ed among the mesenchymal Gli targets in the fetal gut, and it was demonstrated further that the GLI2 target gene Fat4 is required for villus development during the epithelial transition 18 .The discovery that the PCP pathway acts within the mesenchymal compartment to structure stromal cells was surprising, as typically PCP pathway function has been reported within epithelial cell layers 19 .As documented above, before the epithelial transition, Foxl1 + cells form a uniform cell layer with a depth of only one to two cells surrounding the primitive gut tube, which is then patterned into the telocyte progenitor 1 (villus cluster) and telocyte progenitor 2 (crypt base) cells, possibly with the involvement of the PCP system.We found that several PCP genes (Fat4, Wnt5a, Vangl1 and 2) exhibit reduced expression in the absence of Foxl1 (Fig. 5B).Using our Cut and Run data, we found Foxl1 binding in the promoter of Fat4, suggesting a direct regulatory relationship (Fig. 5C).
Rao-Bhatia and colleagues had found villi cation defects in mice with mutations in the PCP gene Fat4, which were worsened by simultaneous heterozygous loss of Vangl2 18 .This defect was preceded by a reduction in the number of epithelial T-folds, characteristic invaginations of the epithelium that form the boundaries of developing villi and that can be visualized by staining with the apical membrane marker Ezrin.In order to evaluate if the reduced expression of PCP genes in Foxl1 null mice impacts epithelial remodeling, we stained small intestinal sections from fetuses at developmental stages spanning villi cation (13.5 to 15.5 dpc) for Ezrin to identify T-folds and PDGFRa to label villus cluster cells.As shown in Fig. 5D-F, the number of T-folds is clearly reduced in the Foxl1-de cient intestine, coinciding with the loss of PDGFRa expression in telocyte progenitors.Next, we employed staining for F-actin to assess the orientation of stromal cells in the developing intestine.As shown in Fig. 5G, while mesenchymal cells in the control fetal gut reorient their major axis to be parallel to the invaginating villi, this process fails to occur in Foxl1 null mice, supporting the notion of failed planar cell polarity in stromal cells.

Loss of Foxl1 impacts epithelial gene expression pro les
As shown above, Foxl1 de ciency impacts the patterning of the overlying epithelium, with many villus tip epithelial cells remaining in the cell cycle (Fig. 4D).We had also noted a shift in the UMAP pattern of epithelial cells between control and Foxl1 null cells in 15.5 dpc embryos (green box in Fig. 3B).To address this issue further, we reclustered the epithelial cells via UMAP.Figure 6A shows that fetal gut epithelial cells of control embryos partition into two major groups, which we identi ed as 'secretory progenitors' and 'undifferentiated epithelial cells' based on their expression pro le.The heatmap in Fig. 6B shows the 225 most differentially expressed genes between these two clusters (false discovery rate < 10%; absolute fold-change > 2), while Fig. 6C indicates selected markers genes for each cell type.
Fetal secretory progenitors are characterized by high levels of the mRNAs for transcription factors Klf4, Spdef, and Sox4, known to be critical for secretory cell differentiation [20][21][22] .Undifferentiated epithelial cells in contrast exhibit strong expression of Sox9 (which marks them as proliferating intervilllus cells as seen in Fig. 4D) as well as markers of the absorptive enterocyte lineage (Alpi, the gene for intestinal alkaline phosphatase, Fabp1, encoding fatty acid binding protein 1, Apoa4, encoding Apolipoprotein A4 which is important in intestinal cholesterol absorption, and Slc16a1, encoding the monocarboxylic acid transporter for lactate).
Next, we added epithelial cells from Foxl1 null embryos to the UMAP plot and found that they are largely con ned to the undifferentiated epithelial cell cluster (Fig. 6D).When we quanti ed the proportion of cells in each cluster, we found a striking loss of secretory progenitor cells in Foxl1 null embryos (Fig. 6E).Finally, we performed gene set enrichment analysis to search for pathways that are differentially regulated in the absence of Foxl1.As shown in Fig. 6F, the response to BMP signaling, negative regulation of epithelial proliferation, and establishment of planar cell polarity were all strongly enriched among the genes more highly expressed in the control gut epithelium, con rming that loss of telocyte Foxl1 has a major impact on the development of the fetal intestinal epithelium.Finally, we con rmed these ndings by immuno uorescent staining for markers of the secretory cell lineage.Staining for Agr2 (Anterior gradient protein 2 homolog), a protein disul de isomerase required for the formation of mixed disul des in intestinal mucins (Fig. 6G), and its substrate Muc2 (Mucin2), the major mucin of intestinal goblet cells (Fig. 6H), are both expressed in the embryonic day 15.5 control intestine in secretory cell progenitors, but completely absent from the Foxl1 null gut, con rming the ndings from our single cell RNAseq analysis.

Discussion
Villus formation is a fascinating and obviously essential process in vertebrate gut development and depends on reciprocal epithelial-to-mesenchymal cross talk.Epithelial hedgehog proteins are among the earliest signals emanating from the endoderm during organogenesis of the gut.Consequently, inhibition of hedgehog signaling with neutralizing antibodies or ablation of either Shh (sonic hedgehog) or Ihh (Indian hedgehog) causes impairment of gut development and villus formation 7,8 .Hedgehog proteins signal via their receptor Ptch1 (patched 1), expressed exclusively in the gut mesoderm, to stabilize the DNA-binding transcription factors Gli2 and Gli3.In 2006, computational analysis of evolutionarily conserved enhancers led to the identi cation of an ultra-conserved putative enhancer located between the neighboring Foxl1 and Foxf1 genes 23 .We identi ed seven Gli binding sites in this genomic region, some conserved from Fugu to human, and showed through in vitro and in vivo studies that both Foxl1 and Foxf1 are Gli target genes 12 .Here, we demonstrate that Foxl1-expressing telocyte progenitors are partitioned into two major subpopulations with distinct gene expression pro les.Telocyte progenitors 1 and 2 correspond to telocytes in the villus clusters and those adjacent to developing crypts, respectively.
Villi cation is strongly impaired in the absence of Foxl1, and abnormal proliferation persist in epithelial cells in the developing villi.We attribute this to the loss of BMP signaling as multiple Bmp genes exhibit reduced expression in absence of Foxl1.BMP signaling had been established by Walton and colleagues as key factor of villus formation 6 .Recently, planar cell polarity genes were identi ed among the mesenchymal Gli2 targets in the fetal gut, and it was demonstrated that the Gli2 target Fat4 is required for villus development during the epithelial transition 18 .The discovery that the PCP pathway acts within the mesenchymal compartment to structure stromal cells is novel and exciting, as typically PCP pathway function had been reported within epithelial cell layers 19 .The model of villi cation proposed by Rao-Bhatia and colleagues states that activation of Gli2 in telocyte progenitors opposite to the hedgehog-secreting epithelium is su cient to activate Fat4 and other planar cell polarity genes directly 18 .Our data indicate that PCP induction in the developing gut mesenchyme is more complex than proposed by Rao-Bhatia and colleagues and depends not only on GLI proteins but also on winged helix transcription factors active in telocyte progenitors.We document by scRNAseq analysis of the proximal small intestine of 15.5 dpc fetuses that expression of planar cell polarity genes including Fat4 is enriched in Foxl1 + telocyte progenitors and reduced dramatically in the absence of Foxl1 (Fig. 5).Therefore, we propose a feed-forward loop for the regulation of stromal planar cell polarity genes, in which Gli2 activates both Foxl1 and Fat4 (and related targets), while Foxl1 also activates planar cell polarity gene expression in epithelium-adjacent stromal cells in a coherent feedforward loop (Fig. 7).Loss of the telocyte transcription factor Foxl1 has a major secondary impact on the overlying epithelium, were reduced BMP signaling leads to loss of secretory progenitor differentiation and retention of proliferating epithelial cells overlying villus cluster cells.
In conclusion, we have shown that the winged helix transcription factor, expressed in endoderm-adjacent mesodermal telocytes, is critical for the epithelial transition, epithelial gene expression, and ordered villus formation via the regulation of BMP, PDGFRa, and PCP signaling molecules.

Mice
All animal procedures were approved by Institutional Animal Care and Use Committee of the O ce of Animal Welfare at the University of Pennsylvania and conducted under protocol 804436.The mice used in this study were housed in a Speci c Pathogen Free (SPF) facility at the University of Pennsylvania's animal center.They were individually housed in ventilated cages and provided with controlled temperature, humidity, a 12-hour light-dark cycle, a standard rodent chow diet, and constant access to water.C57BL/6 wild-type mice were obtained from Jackson Laboratory (Stock number: 000664).The Foxl1CreERT2-TdTomato gene replacement allele was described previously 13 .
For embryonic studies, embryonic day 0.5 (0.5 dpc) was de ned as noon on the day when the copulatory plug was observed.Since homozygotes of the Foxl1CreERT2-TdTomato gene replacement allele are viable and fertile in adulthood, the Foxl1 null embryos were generated by either crossing two heterozygotes or crossing one homozygote with a heterozygote.Single cells for the Cut & Run experiment were isolated from dissected embryonic small intestines at 15.5 dpc, which were obtained by crossing Foxl1CreERT2-tdTomato heterozygous mice with C57BL/6 wild-type mice.Embryonic data were collected from developmental stages before (13.5 dpc), during (14.5 dpc), and after (15.5-18.5 dpc) the onset of villi cation in the developing small intestines.Due to the indistinguishable sex of the embryos and the absence of reported sex differences in villus morphogenesis, mice of both sexes were included in all experimental procedures.Therefore, random assignment of mice from either sex was conducted within each experimental group.

Tissue isolation and sectioning
Mouse embryonic small intestinal samples from different stages were carefully dissected and harvested from the body.All samples used for histological sectioning and subsequent staining were taken from duodenal sections.For para n section, samples were xed overnight at 4° C temperature in 4% paraformaldehyde in PBS buffer (Invitrogen) and incubated in 70% ethanol solution after three times' wash with PBS buffer for para n section, and then were submitted to the Molecular Pathology and Imaging Core of the Center for Molecular Studies in Digestive and Liver Diseases (P30 DK050306) for further process, embed and sectioning for sectioning at tum thickness and then sectioned slices were kept at room temperature.For frozen section, samples were incubated overnight in 30% sucrose in PBS buffer after overnighted xation at 4°C in 4% paraformaldehyde in PBS buffer (Invitrogen) until the intestinal tissues completely sink to the tube bottom, and then were embed in OCT for quick-frozen and stored at -80°C.OCT-embedded samples were performed the cryo-sectioning though a using a cryostat (Cryostar NX50, ThermoFisher Scienti c) at 10 µm, dried for 10 minutes at room temperature before the further immuno uorescent staining or stored at -80°C.

Scanning Electron Microscopy
The scanning electron microscope (SEM) experiments were conducted at the microscopy core facility of the UPenn Department of Cell and Developmental Biology.After dissection, the presumptive duodenal sections of the 15.5 and 18.5 dpc fetuses were washed three times with 50 mM Na-cacodylate buffer and xed overnight using a solution consisting of 2.5% glutaraldehyde in 50 mM Na-cacodylate buffer at a pH of 7.3.Subsequently, samples were dehydrated using a graded series of ethanol, gradually reaching 100% ethanol over a span of 1.5 hours.After dehydration, samples were incubated for 20 minutes in a solution containing 50% HMDS (Sigma-Aldrich) in ethanol, followed by three changes of 100% HMDS.After air-drying overnight, samples were mounted on stubs and coated with a layer of gold palladium using the sputter coating technique.Finally, we observed and photographed the specimens utilizing a Quanta 250 FEG scanning electron microscope manufactured by FEI (Hillsboro, OR, USA) with a 10 kV accelerating voltage.

Histology and Immuno uorescence
Para n sections were depara nized and rehydrated using xylene and descending ethanol gradients.For H&E staining, tissues were stained with Harris' Hematoxylin and alcoholic Eosin Y.For IF staining, para n sections were subjected to antigen retrieval.For the frozen slices, OCT was directly removed in sterile water.Then the slices were performed the incubation with primary antibodies overnight at 4°C after serum blocking for 1 hour at room temperature and then with appropriate secondary antibodies at room temperature in the dark for 1 hour.
Single-cell capturing and cDNA library construction Timed embryos were obtained by crossing homozygous male mice with heterozygous female mice.The small intestine was obtained through dissection, and half of the portion close to the stomach was collected after folding it in half.Additionally, the tail of each embryo was collected for genomic DNA extraction using the KAPA Mouse Genotyping Kit HotStart (Kapa Biosystems, KK7352).Subsequently, genotyping was performed to distinguish between heterozygotes and homozygotes.Isolated intestinal cells were digested with collagenase type II and DNase I to prepare the single-cell suspension and then sorted in phosphate-buffered saline with 0.05% BSA to enrich the living cells through FACS sorting (MoFlo Astrios Sorter).Then the cells obtained were measured for cell concentration and viability with Trypan blue using a Countess II Automated Cell Counter from Life Technologies.Then the single-cell suspension was diluted to appropriate concentration and loaded on a 10x Genomics Chromium Single Cell Controller (Pleasanton, CA) with a target of about 5,000 cells per sample.Single-cell library preparation was completed using the 10x Genomics Chromium Single Cell 3' Library & Gel Bead Kit v2 strictly following manufacturer's protocol.The quality and quantity testing of obtained short cDNA fragment libraries using an Agilent 2100 Bioanalyzer and Invitrogen Qubit Fluorometer.Finally, the single-cell cDNA Libraries were sequenced on an Illumina Novaseq 6000 instrument.
Cut & Run and DNA product sequencing Timed embryos were obtained by crossing homozygous or heterozygous male mice with C57BL/6 female mice.The entire length of the small intestine was digested with collagenase type II and DNase I to reach the single-cell state.Subsequently, the cells were sorted in phosphate-buffered saline with 0.05% BSA to enrich the tdTomato-positive cells using FACS (MoFlo Astrios Sorter).Isolated tdTomato-positive telocytes from the embryonic small intestines at 15.5 dpc were subjected to CUT&RUN experiments.CUT&RUN experiments were performed using the CUT&RUN Assay Kit (EpiCypher, Catalog No. 14-1048) following the manufacturer's Instructions using 40,000 telocyte progenitors.Anti-Foxl1, H3K4me3 positive control and rabbit IgG negative control antibodies (13-0042k) were used in these experiments.Puri ed CUT&RUN DNA products were subjected to the CUT&RUN Library Prep Kit (EpiCypher, Catalog No. 14-1002) for library construction, and the libraries sequenced on an Illumina Hiseq X Ten instrument.(E) UMAP plot of mRNA expression of multiple genes with differential activity between the two telocyte progenitor populations.Note the high expression of multiple BMP mRNAs in telocyte progenitor 1 cells.
(F) UMAP plot of mRNA abundance for Glp2r, encoding the receptor for the intestinotrophic hormone GLP-2.Glp2r expression is highly enriched in telocyte progenitor 1 cells.
(H) UMAP plot of mRNA abundance for Sall1, encoding the Spalt Like Transcription Factor 1, a zinc nger transcriptional repressor.Sall1 expression is highly enriched in telocyte progenitor 1 cells.
(I) Immuno uorescence staining fetal mouse intestine from E15.5 fetuses with antibodies speci c to Sall1 (green) and Foxl1 (red) show localized expression in villus tip telocytes, as well as in a forming villus cluster (white arrow).DAPI (blue) was used to visualize nuclei.
(J) Model of relative positioning and prominent marker genes of telocyte progenitors 1 and 2 during intestinal villi cation.(D) Immuno uorescence labeling con rms dramatic reduction in PDGFRa expression (red) in the absence of Foxl1.The apical membrane of epithelial cells is stained for Ezrin (green).The persistence of telocyte progenitors in Foxl1 null mice is con rmed by staining for the tdTomato transgene (yellow) which replaces the Foxl1coding region in this Foxl1 null allele.
(E) Cut and Run analysis of fetal telocytes indicates that the Foxo3 promoter is in an active state, as indicated by the strong H3K4me3 signal (red trace).Foxl1 is bound to the proximal promoter of Foxo3.
(F) Foxo3 expression in fetal telocyte progenitors is Foxl1-dependent as shown by reduced expression levels in the Foxl1null intestine by scRNAseq analysis.Foxl1 is required for full activation of planar cell polarity genes in small intestinal telocyte progenitors.
(A) Small intestinal epithelial structure of the fetal small intestine in E15.5 fetuses shows cuboidal organization of a single-cell layer epithelium.In Foxl1null mice, epithelial organization is disrupted, and the epithelium is partially multilayered, with frequent apoptotic cells, as visualized by cleaved caspase 3 (F) Gene set enrichment analysis (GSEA) identi es critical pathways as Foxl1 dependent.The normalized enrichment scores (NES) were 2.18 for the 'response to BMP', 1.97 for 'negative regulation of epithelial proliferation', and 1.90 for 'establishment of planar polarity of embryonic epithelium.
(G) Staining of the fetal (E15.5)small intestine with an antibody speci c to Agr2 (Anterior gradient protein 2 homolog), a marker of the secretory lineage, was used to localize secretory progenitor cells.In control mice, these cells are distributed throughout the gut epithelium, while their number is dramatically reduced in the absence of Foxl1.
(H) Staining of the fetal (E15.5)small intestine with an antibody speci c to Muc2 (Mucin 2), a marker of the goblet cell lineage, was used to localize goblet cell progenitors.In control mice, these cells are distributed throughout the gut epithelium, while their number is dramatically reduced in the absence of Foxl1.

Figures Figure 1
Figures (A-F) Hematoxylin and eosin stained small intestine from control (A,C,E), or Foxl1 null (B,D,F) fetuses at the developmental stages indicated.The yellow arrow in B indicates lack of invagination, and the red arrow in D marks persistent epithelial ridges.(G-L) Scanning electron micrographs of small intestine from control (G,I,K), or Foxl1 null (H,J,L) fetuses at the developmental stages indicated.In all panels, larger magni cation images shown on the right correspond to the areas outlined on the left.Magni cation is indicated by scale bars.