1. Cell Biology
  2. Developmental Biology
Download icon

Radical and lunatic fringes modulate notch ligands to support mammalian intestinal homeostasis

  1. Preetish Kadur Lakshminarasimha Murthy
  2. Tara Srinivasan
  3. Matthew S Bochter
  4. Rui Xi
  5. Anastasia Kristine Varanko
  6. Kuei-Ling Tung
  7. Fatih Semerci
  8. Keli Xu
  9. Mirjana Maletic-Savatic
  10. Susan E Cole
  11. Xiling Shen  Is a corresponding author
  1. Duke University, United States
  2. Cornell University, United States
  3. Ohio State University, United States
  4. Baylor College of Medicine, United States
  5. University of Mississippi Medical Center, United States
Research Article
  • Cited 5
  • Views 1,163
  • Annotations
Cite this article as: eLife 2018;7:e35710 doi: 10.7554/eLife.35710

Abstract

Notch signalling maintains stem cell regeneration at the mouse intestinal crypt base and balances the absorptive and secretory lineages in the upper crypt and villus. Here we report the role of Fringe family of glycosyltransferases in modulating Notch activity in the two compartments. At the crypt base, RFNG is enriched in the Paneth cells and increases cell surface expression of DLL1 and DLL4. This promotes Notch activity in the neighbouring Lgr5+ stem cells assisting their self-renewal. Expressed by various secretory cells in the upper crypt and villus, LFNG promotes DLL surface expression and suppresses the secretory lineage . Hence, in the intestinal epithelium, Fringes are present in the ligand-presenting ‘sender’ secretory cells and promote Notch activity in the neighbouring ‘receiver’ cells. Fringes thereby provide for targeted modulation of Notch activity and thus the cell fate in the stem cell zone, or the upper crypt and villus.

https://doi.org/10.7554/eLife.35710.001

Introduction

Lgr5+ Crypt Base Columnar cells (CBCs) located at the bottom of the crypts constantly self-renew to maintain the small intestinal epithelium, which is one of the fastest regenerative tissues in the body (Barker et al., 2007; van der Flier et al., 2009). CBCs divide and move up the crypt into the progenitor or transit-amplifying zone where the cells rapidly proliferate and terminally differentiate into two major types: absorptive (enterocytes) and secretory (mainly Paneth and goblet cells). Enterocytes and goblet cells populate the villi while the Paneth cells move to the bottom of the crypt and provide a niche for the CBCs (van der Flier et al., 2009).

Notch signalling pathway primarily consists of Notch receptors (NOTCH1-4) and ligands (DLL1-4 and JAG1-2) (Bray, 2006). Upon activation of a Notch receptor by a ligand, it undergoes successive cleavages by ADAM/TACE and γ-secretase (Bray, 2006). The cleaved intracellular domain (NICD) translocates to the nucleus leading to the transcription of multiple genes such as Hes and Hey families (Kopan, 2002; Iso et al., 2003). The extracellular domain of the Notch receptor and ligands contain EGF-like repeats, some of which serve as substrates for O-fucosylation by POFUT1 (Rampal et al., 2007; Wang et al., 2001). The fucosylated product may be further modified within the Golgi network by Fringe proteins: Lunatic (LFNG), Manic (MFNG) and Radical Fringe (RFNG) (Moloney et al., 2000; Haines and Irvine, 2003). Fringe proteins are typically expressed in receptor-expressing ‘receiver’ cells (Haines and Irvine, 2003). Glycosylation of NOTCH1 by LFNG and MFNG increases its activation by DLL1 but decreases its activation by JAG1 (Haines and Irvine, 2003; Hicks et al., 2000; Panin et al., 1997). In contrast, glycosylation by RFNG increases the activation of NOTCH1 by both DLL1 and JAG1 (LeBon et al., 2014).

Notch pathway provides for spatial and context specific decision making in the intestinal epithelium. At the bottom of the crypt, Notch signalling is important for the maintenance of CBCs (Pellegrinet et al., 2011). In the upper crypt however, Notch activity, mainly through Hes1, is essential for the enterocyte differentiation (van der Flier et al., 2009; Fre et al., 2005). Inhibition of Notch signalling results in the loss of proliferative CBCs and progenitor cells and leads to their differentiation into goblet cells in the upper crypt and villus, indicating the importance of the pathway in maintaining the tissue (Pellegrinet et al., 2011; VanDussen et al., 2012; Riccio et al., 2008; Wu et al., 2010). Notch1, 2 and Dll1, 4 are known to be the necessary receptors and ligands in the intestine (Pellegrinet et al., 2011; Riccio et al., 2008; Schröder and Gossler, 2002). Although, the fringe proteins are known to be expressed in the intestine, their function has not been studied (Schröder and Gossler, 2002).

Here we show that Rfng and Lfng are expressed by the ligand-presenting secretory lineages, but at different locations. At the crypt base, Rfng expressed in Paneth cells modulates DLL1 and DLL4, which enhances Notch signalling and self-renewal of neighbouring CBCs. In the upper crypt and villus, Lfng is expressed by secretory cells including enteroendocrine, Tuft and goblet cells. LFNG promotes Notch signalling in the transit amplifying cells and impedes their differentiation into secretory cells. MFNG does not play any noticeable role in intestinal epithelial homeostasis.

Results

Rfng supports Lgr5+ stem cell self-renewal

Rfng transcripts have been detected in the crypt by in situ hybridisation (Schröder and Gossler, 2002). We analysed previously published microarray data on Lgr5+ CBCs and Paneth cells (Sato et al., 2011) and found that Rfng is significantly upregulated in Paneth cells (Figure 1—figure supplement 1A). We isolated CBCs and Paneth cells (CD24high/SSChigh) from Lgr5-GFP mice by FACS using an established protocol (Sato et al., 2011; Sato et al., 2009) and found that the Paneth cells are enriched for Rfng (Figure 1A). We validated that the isolated cells are indeed Paneth cells and CBCs by confirming their Lysozyme and GFP expression respectively (Figure 1—figure supplement 1B,C). We also confirmed that Rfng is enriched in the Paneth cells by RNA in situ hybridisation (ISH) (Figure 1B). We validated the specificity of ISH probes using Rfng null mouse intestinal sections (Figure 1—figure supplement 1D,E).

Figure 1 with 3 supplements see all
Rfng supports Lgr5+ stem cell self-renewal.

(A) RT-qPCR quantification of Rfng in Lgr5+ CBC and Paneth cells isolated from Lgr5-GFP mouse intestines. The experiment was performed in triplicate and presented as mean ± s.d. (standard deviation) (B) Representative image showing Rfng transcripts (red) and Lysozyme protein (green) expression at the bottom of the crypt of Lgr5-GFP mouse intestine. DAPI (Blue) labels the nuclei and scale bar represents 10 μm. Arrows point to CBCs. (C–D) Single Lgr5-GFP CBCs were transduced with either Sc. shRNA or Rfng shRNA. The experiment was performed in triplicate. (C) Colony forming efficiency measured after 7 days. Quantitative analysis calculated from 1000 cells/replicate presented as mean ± s.d. (D) Left: Representative flow cytometry plots indicating gated percentage of Lgr5+ (GFPhigh) and Paneth cells (CD24high/SSChigh). Right: Ratio of Lgr5-GFP+ CBCs/Paneth cells as determined by flow cytometry and presented as mean ± s.d. (E) Left: Representative plots indicating gated population of CBCs (CD166+CD24loCD44+GRP78-) from the intestine of Rfng+/+ and Rfng-/- mice. Percentage reflects fraction of total population. Right: Ratio of number of CBCs to Paneth cells of n = 3 mice and presented as mean ± s.d. (F) RT-qPCR quantification of Lgr5 in crypts extracted from Rfng+/+ and Rfng-/- mice. n = 3 mice. Data is presented as mean ± s.d. (*p<0.05; **p<0.01; ***p<0.001).

https://doi.org/10.7554/eLife.35710.002

We then established an in vitro knockdown (KD) model using organoid cultures of epithelial cells obtained from Lgr5-GFP mice. Single Lgr5-GFP+ CBCs were transduced with scrambled (Sc.) shRNA (control) or Rfng shRNA and propagated as organoids (Figure 1—figure supplement 1F). The colony formation efficiency of the Rfng KD CBCs was reduced compared to the control (Figure 1C). Flow cytometric analysis showed that the number of Lgr5+ CBCs decreased after Rfng loss, whereas the number of Paneth cells remained relatively unchanged (Figure 1D).

We confirmed the observation in vivo using previously published Rfng deficient (Rfng−/−) mice (Moran et al., 2009). Crypt cells were isolated from Rfng−/− and Rfng+/+ control mice and analysed by flow cytometry using a combination of surface markers to identify CBCs (CD24loCD44+CD166+GRP78-) (Wang et al., 2013). Analysis revealed that the Rfng−/− mice had almost a two-fold depletion in CBCs (Figure 1E). A reduction of CBC marker Lgr5 transcripts in the crypts harvested from Rfng−/− mouse intestines was observed by RT-qPCR measurement when compared to the control (Figure 1F). The number of Paneth and goblet cells remain largely unchanged and no other significant phenotype was detected in the epithelium (Figure 1—figure supplement 2A–F). Loss of Rfng in organoids seems to show a more significant phenotype in CBC reduction than its loss in vivo. This may be because CBCs in vivo also receive cues from the mesenchyme and not just the Paneth cells as in case of organoids.

To confirm that the loss of Rfng only in the Paneth cells can affect the CBCs, we performed the Organoid Reconstitution Assay (ORA) described previously (Rodríguez-Colman et al., 2017). FACS sorted Lgr5-GFP+ CBCs were incubated with Paneth cells from wild type or Rfng null mice for 10 min at room temperature and plated in Matrigel. We find that the colony formation ability of CBCs incubated with Paneth cells lacking Rfng was significantly lower than the control (Figure 1—figure supplement 3). It should be noted that not all CBCs associate with a Paneth cell during the incubation. Also, the Lgr5-GFP+ CBCs divide and give rise to Paneth cells with Rfng. Hence the result of this assay is not as significant as that shown in Figure 1C.

Rfng promotes Notch signalling in CBCs

We isolate, by FACS, the CBCs and Paneth cells from the Rfng KD and control seven days old organoids described earlier. Western blotting shows that Notch target genes Hes1 and Hey1 have reduced expression in the CBCs upon loss of Rfng. However, the levels of ligands DLL1, four and JAG1 on Paneth cells were not significantly altered, consistent with the post-translational modifying role of Fringe (Figure 2A). Rfng KD and control CBCs were then transfected with an RBPJκ-dsRed reporter (Hansson et al., 2006) as an indicator of Notch activity, cultured overnight and analysed by flow cytometry (Figure 2B). Rfng shRNA decreased the mean RBPJκ-dsRed fluorescent intensity, indicating reduced overall Notch signalling in CBCs.

Figure 2 with 1 supplement see all
Rfng promotes Notch signaling in Lgr5+ CBC.

(A) Western blot analysis of Notch signaling components in CBCs and Paneth cells FACS sorted from Rfng KD and control organoids. (B) Left: Representative plots for RBPJκ-dsRed and Lgr5-GFP expression indicating a gated double positive fraction for Rfng KD and control CBCs transfected with RBPJκ-dsRed reporter. Right: Mean fluorescence intensity (MFI) of RBPJκ-dsRed expression. The experiment was performed in triplicate and presented as mean ± s.d. (C) Ligand availability on Rfng KD and control Paneth cells. Representative traces (left) and MFI (right) showing ligand binding to NOTCH1 measured by flow cytometry. Unstained Paneth cells were used as a negative control. The experiment was performed in triplicate and presented as mean ± s.d. (D) Cell surface DLL1, DLL4, and JAG1 concentration on Rfng KD and control unpermeabilised Paneth cells. Left: Representative traces measured by flow cytometry. Right: MFI measurements. The experiment was performed in triplicate and presented as mean ± s.d. (E) Cell surface DLL1, DLL4, and JAG1 concentration on Rfng KD and control permeabilised Paneth cells. (**p<0.01).

https://doi.org/10.7554/eLife.35710.006

Fringes are known to modulate Notch signalling when present in receptor expressing cell (Haines and Irvine, 2003). But here we find Rfng in the ligand presenting cell promoting Notch activity in the neighbouring CBCs. We confirmed that the Paneth cells express the ligand Dll1 by RNA-ISH (Figure 2—figure supplement 1). To understand the mechanism behind this, we examined ligand availability and concentration on the cell surface according to a previously established method using flow cytometry (LeBon et al., 2014; Taylor et al., 2014; Yang et al., 2005). Seven days old Rfng KD and control organoids were dissociated and single unpermeabilised cells were labelled with CD24 to mark Paneth cells and NOTCH1-Fc to quantify ligand binding to NOTCH1 (Figure 2C). Mean fluorescent intensity (MFI) of NOTCH1 binding to Paneth cells with Rfng knockdown was reduced compared to the scrambled control. We further confirmed that the ligands available on the Paneth cell surface have reduced by using specific antibodies for DLL1, DLL4 and JAG1. Flow cytometry showed that DLL1 and DLL4 levels on the Paneth cell surface reduced after the loss of Rfng (Figure 2D), although the total expression level of DLL1 and DLL4 within Paneth cells is not changed by Rfng knockdown after the cells were permeabilised (Figure 2E). Western blotting also confirmed that the total DLL1 and DLL4 expression in Paneth cells does not change significantly after the loss of Rfng (Figure 2A). The outcomes of the ligand availability assays suggest that the DLLs available on the Paneth cell surface for NOTCH1 to bind to have been reduced after the loss of RFNG, which could decrease Notch activity in the adjacent CBCs.

Mfng plays an insignificant role

Mfng is expressed by scattered cells in the intestinal epithelium (Schröder and Gossler, 2002). To understand its potential function in maintaining the epithelium, we established an in vitro shRNA based Mfng knockdown model as before. Western blotting and RT-qPCR analysis validated Mfng knockdown (Figure 3A,B). Gene expression levels of Lgr5 and Notch components were comparable between Sc. Control and Mfng KD organoids (Figure 3B). Additionally, the colony forming efficiency (Figure 3C) and the expression pattern of Lgr5-GFP (CBC) and MUC2 (goblet cells) (Figure 3D) of Mfng shRNA-expressing CBCs were similar to the scrambled control. We quantified this observation using flow cytometry, which confirmed no significant change in the number of Lgr5-GFP+ CBCs and goblet cells (Figure 3E,F). Finally, the percentage of differentiated cells, identified by CK20 expression, was not significantly altered between Sc. control and Mfng KD organoids (Figure 3G).

Figure 3 with 2 supplements see all
Mfng plays an insignificant role.

Single Lgr5-GFP CBCs were transduced with either Sc. shRNA or Mfng shRNA. The experiment was performed in triplicate. (A) Western blot for Mfng expression. (B) RT-qPCR quantification of Mfng and Notch components in organoids. (C) Colony forming efficiency measured after 7 days. Quantitative analysis from 1000 cells/replicate. (D) Representative bright field and co-IF images indicating Lgr5-GFP (green) expression. MUC2 (red) marks Goblet cells. DAPI (blue) labels nuclei and scale bar represents 25 μm. (E) Representative flow cytometry plots indicating gated percentage of Lgr5+ CBCs (GFPhigh) and goblet cells (UEA-1+/CD24-). (F) Percentage of Lgr5+ CBCs and goblet cells as determined by flow cytometry and presented as mean ± s.d. (G) Left: Representative flow cytometry histograms indicating KRT20+ (CK20+) cells. Right: Percentage of KRT20+ cells and presented as mean ± s.d.

https://doi.org/10.7554/eLife.35710.008

We then analysed intestinal tissues from Mfng deficient (Mfng−/−) mice (Moran et al., 2009) (Figure 3—figure supplement 1A–C). IF microscopy showed similar MUC2 staining in intestinal sections of Mfng−/− and wild-type (Mfng+/+) mice (Figure 3—figure supplement 1D). Quantification in intestinal tissues based on IF expression indicated that the number of goblet cells was not significantly altered in Mfng+/+ mice compared to Mfng−/− mice. Finally, we examined the total number of CK20+ cells in intestines, which was similar in wild-type and Mfng null mice (Figure 3—figure supplement 2A).

We observed that goblet cells were slightly enriched in Mfng when compared to the CBCs (Figure 3—figure supplement 2B). We found no significant change in the cell surface expression of DLLs on goblet cells after the loss of Mfng (Figure 3—figure supplement 2C–F). Overall, these data suggest Mfng plays an insignificant role in intestinal tissues.

Lfng deletion leads to increased goblet cell differentiation

Lunatic Fringe is known to be expressed in the crypts and scattered cells in the villous epithelium (Schröder and Gossler, 2002). Immunofluorescence analysis of intestines from Lfng-GFP reporter mice confirmed that Lfng is expressed by a subset of cells in the upper crypt (transient-amplifying cell region) and villus (Figure 4A). We observe that Lfng-GFP+ cells are post-mitotic in the upper-crypt. (Figure 4A). Further analysis showed that these Lfng-GFP+ cells express ChgA, Dclk1 or Muc2 which are markers for enteroendocrine, Tuft or goblet cells respectively (Figure 4B–D). Secretory cells in the intestine, mainly enteroendocrine and goblet cells, are known to express Notch ligands, especially DLL1 (van Es et al., 2012). In the upper crypt, immunofluorescence analysis showed Notch1 activity in the cells adjacent to Lfng-GFP+ cells but not in themselves (Figure 4E). We performed RT-qPCR measurement of Lfng using goblet cells and CBCs isolated from Lgr5-GFP mice and confirmed that Lfng is in goblet cells and not CBCs (Figure 4—figure supplement 1A).

Figure 4 with 1 supplement see all
Lfng loss results in increased goblet cell differentiation in vitro.

(A–E) Representative IF images of the small intestine of Lfng-GFP reporter mice. (A) GFP (green) shows the Lfng expression and EdU (red) marks the proliferating cells. DAPI (blue) labels nuclei. Scale bar represents 50 μm. (B) GFP (green) shows the Lfng expression and DCAMKL1 (red) marks the Tuft cells. Scale bar represents 20 μm. (C) GFP (green) shows the Lfng expression and CHGA (red) marks the enteroendocrine cells. Scale bar represents 20 μm. (D) GFP (green) shows the Lfng expression and MUC2 (red) marks the goblet cells. Scale bar represents 20 μm. (E) GFP (red) shows the Lfng expression and NICD (green) identifies the cells with NOTCH1 activity. Scale bar represents 20 μm. (F) Representative bright field and co-IF images of Lfng KD and control organoids indicating Lgr5-GFP (green) expression. MUC2 (red) marks goblet cells. DAPI (blue) labels nuclei and scale bar represents 25 μm. (G) Representative plots indicating gated percentage of Lgr5+ (GFPhigh) and goblet cells (UEA-1+/CD24-) of Lfng KD and control organoids. (H) Ratio of goblet cells to Lgr5-GFP + CBCs as determined by flow cytometry. The experiment was performed in triplicate and presented mean ± s.d. (***p<0.001).

https://doi.org/10.7554/eLife.35710.011

We established an in vitro shRNA based Lfng knockdown model as before (Figure 4—figure supplement 1B). We observed only a slight decrease in colony forming efficiency of CBCs after Lfng knockdown and no significant change in the level of Notch activity in the CBCs (Figure 4—figure supplement 1C,D). However, we find that the number of goblet cells (MUC2+) increased after the loss of Lfng (Figure 4C). Quantification by flow cytometry showed that number of goblet cells (UEA-1+/CD24-) (Wong et al., 2012) was increased significantly in Lfng KD organoids (5.5% of the total population) when compared to the scrambled control (1.9%) (Figure 4D). Accordingly, the ratio of the number of goblet cells to Lgr5+ CBCs increased approximately three times in Lfng shRNA-expressing organoids (Figure 4E).

We confirmed these observations in vivo by examining the intestinal tissues from Lfng deficient (Lfng−/−) mice (Moran et al., 2009). We observed an increase in the number of goblet cells in the Lfng null mice as expected (Figure 5A and Figure 5—figure supplement 1A,B). Goblet cells were quantified in villus crypt units (VCU) of the small intestine (Ishikawa et al., 1997). Immunofluorescence analysis based on MUC2 expression in small intestinal tissues from Lfng−/− mice showed an increase in the number of goblet cells when compared to the control (Lfng+/+) mice (Figure 5B). Finally, using flow cytometry we quantified goblet cell numbers in Lfng−/− mice: 14.1% of small intestinal cells, which is significantly higher than the 7.9% goblet cells in the small intestine of wild-type litter-mate control mice (Figure 5C). We observed no change in the Paneth cell numbers after loss of Lfng (Figure 5—figure supplement 1C).

Figure 5 with 2 supplements see all
Lfng loss results in increased goblet cell differentiation in vivo.

(A) Representative H and E sections from the small intestine of Lfng+/+ and Lfng−/− mice. Scale bar represents 50 μm. (B) Left: Representative IF images of intestine of Lfng+/+ and Lfng−/− mice. MUC2 (red) marks goblet cells. DAPI (blue) labels nuclei. Right: Quantification of the number of goblet cells of n = 4 mice/condition and n = 500 VCU per mouse presented as mean ± s.d. (C) Left: Representative plots indicating gated percentage of goblet cells (UEA-1+/CD24-) from intestinal tissue derived from Lfng+/+ or Lfng−/− mice. Right: Percentage of goblet cells presented as mean ± s.d. The data represent n = 3 mice/condition. (D) Left: Representative plots indicating gated population of intestinal progenitors from the intestine of Lfng+/+ and Lfng−/− mice. Percentage reflects fraction of total population. Right: RT-qPCR measurements in progenitor cells from Lfng+/+ and Lfng−/− mice. The experiment was performed in triplicate presented as mean ± s.d. (**p<0.01).

https://doi.org/10.7554/eLife.35710.013

Lfng deletion reduces Notch signalling

Suppression of Notch signalling is known to increase the goblet cell numbers (van Es et al., 2005). We isolated and analysed intestinal progenitor cells (CD24loCD44+CD166+GRP78+) from Lfng+/+ and Lfng−/− mice using an established protocol (Wang et al., 2013) (Figure 5D). RT-qPCR measurements indicated decreased Hes1 and increased Atoh1 (transcriptional factor essential for generating secretory lineage (Shroyer et al., 2007)) expression in Lfng−/− progenitors compared to the control. We also confirmed reduced activated Notch1 (NICD) in the upper crypts of Lfng null mice intestines (Figure 5—figure supplement 2A,B). Therefore, Lfng silencing appears to lower Notch activity in the progenitors and promotes the secretory lineage leading to an increase in goblet cell numbers (Zheng et al., 2011; Kim and Shivdasani, 2011).

Secreted LFNG plays no apparent function

Previous reports have indicated that Lfng may be secreted into the extracellular space (Shifley and Cole, 2008Williams et al., 2016 ). We first examined the medium from intestinal organoid cultures derived from Lgr5-GFP mice using solid-phase ELISA. Secreted Lfng was detected at a concentration of approximately 315–325 ng/mL using two independent LFNG primary antibodies (Figure 6A and Figure 6—figure supplement 1A). The other Fringes, RFNG and MFNG, were not detected in the culture medium (Figure 6—figure supplement 1B,C). We tried to understand if secreted LFNG influences Notch signalling by affecting receptors in a non-cell autonomous manner. As before, single Lgr5-GFP + CBCs were transduced with Lfng shRNA and propagated as organoids followed by incubation with conditioned medium harvested from wild-type organoids that contained soluble form of secreted LFNG (sLFNG). After 24 hr, organoid cultures were analysed using flow cytometry, which showed that the percentage of goblet cells remained similar to the Lfng knockdown (shRNA) condition and significantly higher than scrambled shRNA-expressing organoids which express endogenous LFNG (Figure 6B). RT-qPCR revealed that the expression levels of Notch ligands DLL1 and DLL4 were similar between Lfng knockdown with and without soluble LFNG (Figure 6—figure supplement 1D).

Figure 6 with 1 supplement see all
Secreted LFNG plays no apparent function.

(A) ELISA of the secretion of LFNG in culture medium from Lgr5-GFP organoids. Culture medium (T = 0 days) was used as a control. The experiment was performed in triplicate and presented as mean ± s.d. (B) Left: Representative plots indicating gated percentage of goblet cells (UEA-1+/CD24-) for organoids under Sc. shRNA control, Lfng KD and Lfng KD treated with sLFNG conditions. Right: Percentage of goblet cells in each condition. The experiment was performed in triplicate and presented as mean ± s.d. (C) Left: Representative IF images of intestine of Lfng+/+ and LfngRLfng/+ mice. MUC2 (red) marks goblet cells. DAPI (blue) labels nuclei. Right: Quantification of the number of goblet cells of n = 4 mice/condition and n = 500 VCU/mouse. Data presented as mean ± s.d. (**p<0.01).

https://doi.org/10.7554/eLife.35710.016

We then examined intestinal tissues from mutant Lfng mice (LfngRLFNG/+ or RLfng) in which the N-terminal sequence of LFNG, which normally allows for protein processing and secretion, is replaced with the N-terminus of Radical fringe (a Golgi resident protein) (Williams et al., 2016) (Figure 6—figure supplement 1E). IF analysis based on MUC2 expression in small intestinal tissues from RLfng mice showed similar goblet cell numbers in villus crypt units compared to wild-type (Lfng+/+) mice (Figure 6C). Taken together, our in vitro and in vivo findings suggest that the effect of LFNG on goblet cell numbers and intestinal homeostasis is not dependent on its secreted form.

LFNG promotes DLL expression on the cell surface

In order to understand if LFNG, like RFNG, can affect the cell surface expression of DLL, we examined ligand availability and concentration on the cell surface. Seven days old Lfng KD and control organoids were dissociated and single unpermeabilised cells were labelled with CD24 and UEA-1 to mark goblet cells and NOTCH1-Fc to quantify ligand binding to NOTCH1 (Figure 7A). Mean fluorescent intensity of NOTCH1 binding to goblet cells with Lfng knockdown was reduced compared to the control, suggesting that the ligands available on the goblet cell surface for NOTCH1 to bind to were reduced after the loss of Lfng. Flow cytometry shows that DLL1 and DLL4 levels, detected using specific antibodies, on the goblet cell surface reduced after the loss of Lfng (Figure 7B), although the total expression of DLL1 and DLL4 by the goblet cells measured after the permeabilising the cells remained almost the same (Figure 7C). Western blotting also confirmed that the total DLL1 and DLL4 expression in goblet cells does not change significantly after the loss of Lfng (Figure 4—figure supplement 1D).

LFNG promotes cell surface expression of DLL.

(A) Ligand availability on Lfng KD and Sc. Control goblet cells. Representative traces (left) and MFI (right) showing ligand binding to NOTCH1 measured by flow cytometry. Unstained goblet cells were used as a negative control. The experiment was performed in triplicate and presented as mean ± s.d. (D) Cell surface DLL1 and DLL4 concentration on Lfng KD and Sc. Control unpermeabilised goblet cells. Left: Representative traces measured by flow cytometry. Right: MFI measurements. The experiment was performed in triplicate and presented as mean ± s.d. (E) Cell surface DLL1 and DLL4 concentration on Lfng KD and Sc. Control permeabilised goblet cells. (**p<0.01).

https://doi.org/10.7554/eLife.35710.018

Discussion

We report that Rfng is enriched in the Paneth cells and promotes cell surface expression of DLL1 and DLL4. This promotes Notch activity in the neighbouring Lgr5+ CBCs assisting their self-renewal. Mfng does not appear to contribute significantly in maintaining the epithelium. Lfng on the other hand is expressed by enteroendocrine, Tuft and goblet cells and suppresses the secretory lineage (Figure 8). Even though Fringe proteins do not appear to be essential, they provide another layer of spatial and lineage-specific modulation that might enhance robustness of intestinal homeostasis. This is consistent with the highly robust regulation of Notch activity in the intestinal epithelium as inhibition of Notch1 or Dll1 only causes minor defective phenotype, while inhibition of Notch2, Dll4, Jag1, Hes1, Hes3 or Hes5 causes no significant phenotype (Pellegrinet et al., 2011; Wu et al., 2010; Ueo et al., 2012).

Summary.

Rfng is enriched in the Paneth cells and promotes cell surface expression of DLL1 and DLL4. This promotes Notch activity in the neighbouring Lgr5+ CBCs assisting their self-renewal. Mfng does not appear to contribute significantly in maintaining the epithelium. Lfng on the other hand is expressed by enteroendocrine, Tuft, and goblet cells and suppresses the secretory lineage.

https://doi.org/10.7554/eLife.35710.019

We have observed that both RFNG and LFNG can increase the presence of DLL1 and DLL4 on the plasma membrane. This can potentially contribute to the increase in cis-inhibition of NOTCH1 by DLL1 in the presence of fringe (LeBon et al., 2014). Fringe modulation of ligands will be of significance in understanding Notch activity in cancer stem cell asymmetric division where LFNG, DLL1 and NOTCH1 are present in the same cell (Bu et al., 2013; Bu et al., 2016). However, the mechanism behind the increase in cell-surface expression of the ligands still needs to be understood. The glycosylation state of proteins has been known to affect their intracellular trafficking (Huet et al., 2003; Ohtsubo and Marth, 2006). It raises the possibility that fringe mediated glycosylation or the addition of Galactose and Sialic acid post fringe activity might affect the trafficking of DLLs to the cell surface. Lfng in the Dll1 expressing cell, in the presence of Dll3, is known to reduce Notch activity in the neighbouring cell (Okubo et al., 2012). This raises the possibility that ligands interact with each other in the presence of Lfng which might explain our observation. In vitro reductionist studies may need to be conducted in systems expressing single ligand and fringe to understand the mechanism in detail. Also, our experiments cannot completely rule out the possibility that low levels of Rfng expression in CBCs (in comparison to Paneth cells) can also contribute to the phenotype by directly modulating Notch receptors. We have observed some mesenchymal cells also express Rfng detectable by RNA ISH. We also observe that some of the mesenchymal cells also express Dll1 (Figure 2—figure supplement 1). Further studies are necessary to map the expression of all the Notch ligands in different mesenchymal cell types. This raises the possibility that the mesenchyme can also provide Notch ligands to the Lgr5 + CBCs in vivo. In case that is true, our proposed mechanism that Rfng promotes cell surface expression of Dll1 might be applicable to the mesenchymal cells too. Upon loss Rfng, reduced Dll1 expression on the cell surface of Paneth cells and mesenchymal cells would result in reduced Notch activity in the CBCs, as observed. However, the crypt cells are separated from the mesenchyme by the basement membrane. The efficacy of DLL mediated Notch signalling across the intestinal basement membrane needs to be explored.

Although we have observed that the Lfng expressing cells are found both in the upper crypt and in the villus, our data suggests that LFNG in NICD- post-mitotic secretory cells of the upper crypt promotes Notch activity in the neighbouring enterocyte progenitors. As Notch signalling is not active in the villus, the Lfng+ cells of the villi likely do not impact epithelial cell differentiation. It would be interesting to explore the reason secretory cells expressing Notch ligands and Lfng are found in the villi. The differences, other than functional consequence, between the Lfng+ cells of the upper crypt and villus needs to be explored.

Notch pathway is a potential therapeutic target, but blocking the pathway leads to serious GI related side effects (van Es et al., 2005). Targeting the Notch pathway through fringe appears to be a potentially viable strategy to exclusively modulate intestinal epithelial regeneration or its functions, absorption and mucus secretion, as Notch activity in the stem cells or progenitors can be specifically targeted by blocking Rfng or Lfng respectively.

Materials and methods

Mice

Lgr5-GFP (Jackson Lab #8875, RRID:IMSR_JAX:008875) strain has been described previously (Sato et al., 2009). Lfng null (Lfngtm1Rjo), Mfng null (Mfngtm1Seco, RRID:MGI:3849430) and Rfng null (Rfngtm1Tfv) mice were maintained as described here (Moran et al., 2009; Ryan et al., 2008). LfngRLFNG/+ mice were maintained as previously described (Williams et al., 2016). Littermates of Fringe mutants with wild-type gene expression were used as controls. Lfng-GFP (GENSAT # RRID:MMRRC_015881-UCD) were received as FVB/N - C57BL/6 hybrids and crossed to C57BL/6 mice for at least 10 generations (Gong et al., 2003; Semerci et al., 2017). All procedures were conducted under protocols approved by the appropriate Institutional Animal Care and Use Committees at Ohio State University (# 2012A00000036-R1), Duke University (# A286-15-11), Baylor College of Medicine (# AN-5004), Cornell University (# 2010–0100) or Research Animal Resource Center of Weill Cornell Medical College (# 2009–0029).

Organoid culture and flow cytometry

Request a detailed protocol

Organoids from Lgr5-GFP mouse intestines were cultured as previously described with minor modifications (Sato et al., 2011; Sato et al., 2009). Briefly, small intestines were harvested, washed with PBS and opened up longitudinally to expose luminal surface. A glass coverslip was then gently applied to scrape off villi and the tissue was cut into 2–3 mm fragments and incubated with 2 mM EDTA for 45 min on a rocking platform at 4°C. EDTA solution was then decanted and replaced with cold PBS. The tissues were vigorously agitated to release the crypts. Next, single cell dissociation was achieved by incubating purified crypts at 37°C with Trypsin-EDTA solution containing 0.8KU/ml DNase, 10 μM Y-27632 for 30 min. To isolate Lgr5-GFP+ cells, single cells were resuspended in cold PBS with 0.5% BSA and GFPhigh cells were sorted by FACS (Beckman Coulter/BD FACSAria).

Dissociated cells were also stained with anti-CD24 antibody and UEA-1. Paneth cells were sorted based on side scatter and CD24 expression (CD24high/SSChigh) and goblet cells were identified as UEA-1+/CD24- (Sato et al., 2011; Wong et al., 2012). Viable cells were gated based on 7-AAD or Sytox blue staining. Data analysis was performed using FlowJo software.

Single Lgr5-GFP+ CBCs were plated in Matrigel and cultured in medium containing: Advanced DMEM/F12 supplemented with Glutamax, 10 mM HEPES, N2, B27 without vitamin A, 1 μM N-acetylcysteine, 50 ng/mL EGF, 100 ng/mL Noggin, and 10% R-SPONDIN1 conditioned medium.

Lentiviral constructs containing Lfng shRNA (sc-39491-SH), Mfng shRNA (sc-39493-SH), Rfng shRNA (sc-39495-SH), or scrambled shRNA (sc-108060) were purchased from Santa Cruz Biotechnology. Lentiviral transduction of Lgr5-GFP CBCs were performed by ‘spinoculation’ method described previously (Koo et al., 2011). Transduced CBCs were cultured as organoids and analysed after 7 days. RBPJκ-dsRed reporter (Addgene #47683) was transfected into single Sc. shRNA-expressing or Rfng shRNA-expressing sorted Lgr5-GFP CBCs using Lipofectamine-2000 as described earlier (Schwank et al., 2013).

Organoid Reconstitution Assay was performed as described previously (Rodríguez-Colman et al., 2017). Briefly, FACS sorted Paneth cells and Lgr5-GFP+ CBCs were mixed, spun down and incubated at room temperature for 10 min. The pellet was then plated in Matrigel.

RT-qPCR and protein analysis

Request a detailed protocol

A Qiagen RNeasy kit was used to extract total RNA. RT-PCR primers from Genecopoeia were used for the following genes: β-Actin, Lgr5, Lfng, Mfng and Rfng. Taqman primers (ABI) were used for: Lgr5, Notch1, Hes1, Hes5, Dll1, Dll4, and Jag1. Gapdh or β-Actin was used as internal control. Protein isolation and western blotting were performed as previously described, using β-ACTIN for normalisation (Pan et al., 2008). ELISA kits for LNFG, RFNG, and MFNG were purchased from MyBioSource and assays were performed according to the manufacturer’s instructions similar to the following protocol. Solid-phase ELISA assays were independently conducted using LFNG, RFNG, and MFNG antibodies (referred to as antibody-2) purchased from Santa Cruz Biotechnology for verification of results obtained from the corresponding kits.

Ligand availability assay

Request a detailed protocol

Ligand availability assays were performed as previously described (LeBon et al., 2014). Briefly, blocking buffer (PBS, 2% FBS, 100 μg/mL CaCl2) and binding Buffer (PBS, 2% BSA, 100 μg/mL CaCl2) were prepared. Subsequently, cells were incubated in blocking buffer for 30 min at 37°C followed by incubation with 0.5 μg/mL NOTCH1-Fc (R and D #5267) diluted in binding buffer for 1 hr at 4°C. Cells were then washed three times in blocking buffer and incubated in secondary antibody diluted in binding buffer for 40 min at room temperature. Finally, cells were washed three times in blocking buffer and analysed by flow cytometry.

Immunofluorescence (IF) and immunohistochemistry (IHC)

Request a detailed protocol

Sections of paraffin embedded tissues were deparaffinised using Xylene and rehydrated. Antigen retrieval was performed using Proteinase K (Dako) or 10 mM Tris buffer at pH9. The sections were incubated in Protein Block (Dako) for 10 min at room temperature (RT). Primary antibodies diluted in Antibody Diluent (Dako) were added and incubated overnight at 4°C. Slides were then washed in PBS and incubated in secondary antibodies diluted in Antibody Diluent for 1 hr at RT and washed in PBS. The slides were then mounted using Vectashield mounting medium containing DAPI. Intestinal sections were stained with Haematoxylin and Eosin (H and E), Periodic Acid-Schiff (PAS), Alcian Blue (AB) or Nuclear Fast Red according to standard methods. Intestinal organoids embedded in Matrigel were fixed with 3% PFA for 15 min at room temperature and permeabilised using 0.2% Triton-X for IF according to the protocol described above. Antibodies used are listed in supplementary methods. Antibodies used are listed in Supplementary file 1.

Protocol was modified while staining for Notch1 intracellular domain (NICD). Antigen retrieval was performed using Trilogy (Cell Marque). Sections were then incubated in 3% hydrogen peroxide diluted in PBS for 10 min. Protein blocking, primary antibody and secondary antibody incubation were performed as described above. Signal was further amplified using TSA Plus Fluorescein kit (Perkin Elmer). To quantify, NICD+ nuclei and total number of nuclei (based on DAPI signal) were counted in each crypt (35 to 50 crypts from each section) to obtain the fraction of NICD+ nuclei.

0.5 mg EdU in 150 μl PBS (∼16.66 mg/kg) was injected intraperitoneally into Lfng-GFP mice two hours prior to euthanasia (Kabiri et al., 2014). Incorporated EdU was detected using Click-It EdU imaging kit (Thermo Fisher #C10640).

RNA in situ hybridisation (ISH)

Request a detailed protocol

RNA-ISH was performed using RNAscope 2.5HD duplex assay kit (ACDBio) (Wang et al., 2012) as per manufacturer’s instructions. Briefly, the assay was first validated by using a positive control probe for Polr2a and a negative control probe for a bacterial gene dapB. Probes, labelled with Alkaline Phosphatase, for Rfng or Dll1 were hybridised to the tissue sections. The signal was generated using Fast Red substrate. The slides were washed in water and then PBS and were stained for Lysozyme protein as described above in the Immunofluorescence section. Fast Red signal was detected by a fluorescence microscope as described in (Lauter et al., 2011).

Statistical analysis

Request a detailed protocol

The data is represented as mean ± S.E.M (standard error of mean) unless otherwise indicated. A Student t-test was applied to compare two groups using p<0.05 to establish statistical significance.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
    Notch: a membrane-bound transcription factor
    1. R Kopan
    (2002)
    Journal of Cell Science 115:1095–1097.
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
    Notch signaling in normal and disease States: possible therapies related to glycosylation
    1. R Rampal
    2. KB Luther
    3. RS Haltiwanger
    (2007)
    Current Molecular Medicine 7:427–445.
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50

Decision letter

  1. Fiona M Watt
    Reviewing Editor; King's College London, United Kingdom

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Spatially Specific Fringe Modulation of Notch Ligands Supports Intestinal Homeostasis" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Fiona Watt as the Senior and Reviewing Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Owen Sansom (Reviewer #2); Kevin Myant (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This manuscript describes interesting studies on the expression and possible function of Fringe proteins within the intestinal epithelium. The authors make the argument that several Fringe proteins are expressed highly in Paneth cells, where they increase the cell surface expression of DLL1 and DLL4, two Notch ligands. In this manner, they are proposed to regulate Notch signalling in Lgr5+ intestinal stem cells.

Essential revisions:

None of the reviewers was completely convinced that the authors have shown what they claim and further experiments and/or analysis are needed for the work to be convincing. In particular the authors need to more clearly demonstrate that Fringe proteins are expressed in Paneth cells and that their cell sorting technique is robustly sorting Paneth / Lgr5+ cells.

1) Expression of Fringe proteins in Paneth cells. Although this is a key issue for the study, the data is limited and needs to be strengthened. While Fringe proteins were shown by Schroder et al. to be expressed in the intestinal crypts, they did not appear to be particularly specific in their in situ studies for the location of Paneth cells. No additional localization by either ISH or IHC is shown. While the FACS studies "using an established protocol" suggest greater expression in Paneth cells compared to Lgr5+ cells, the methodology is not easy to decipher, and the relative specificity of the FACS studies is not convincing. The authors need to describe exactly which surface markers were used to achieve their claimed populations. Validation of the FACS studies is needed, together with immunolocalization studies.

2) Expression of DLL1 and DLL4 on Paneth cells. The authors suggest that the ligands DLL1 and DLL4 are well expressed on Paneth cells, based on Western blotting of "Paneth cells". These are presumably from cell sorting, although the figure legend describes the lysates as derived from "organoids". Stamataki D et al. (PLoS One 2011) used a DLL1-β-galactosidase transgene to mark DLL1+ cells and found it was expressed in goblet cells (96%) and enteroendocrine cells (66%) but not in Paneth cells. Van et al., 2012 used a different DLL1-GFP-IRES-CreERT2 knockin mouse; strong expression above the stem cell zone (e.g. position +5 – +10) and only weak expression in the Paneth cells was reported. Finally, Shimizu H et al. (PeerJ, 2014) used double immune staining to show that DLL1 was not expressed in Paneth cells. In contrast Shimizu et al. did confirm that DLL4 was expressed in Paneth cells. Given these published findings, the authors need substantial additional evidence (e.g. immunofluorescence, qPCR, in situ hybridisation) in order to claim DLL1 expression in Paneth cells.

3) The conclusion that Rfng expression on Paneth cells promotes Notch activity in neighbouring Lgr5+ CBCs is based on the finding that Rfng-/- intestines show reduced CBCs, although no other phenotype was detected, which is inconsistent with a reduced Notch phenotype. In addition, loss of Rfng led to a reduction in Lgr5+ CBCs in organoid culture. However, this conclusion that the changes are due to "Paneth cells" remains puzzling, given the reported absence of DLL1 in Paneth cells. In addition, several groups (Garabedian EM et al., J Biol Chem 1997; Kim TH et al., Proc Natl Acad Sci U S A 2012) have now shown that deletion in vivo of Paneth cells does not result in any significant in vivo phenotype. While Paneth cells likely express Dll4, depletion of Dll4 alone or administration of a Dll4-immunoneutralizing antibody has no effect on intestinal epithelial homeostasis (Ridgway et al., 2006). This would be consistent with the lack of a major role for Paneth cells in regulating the ISC compartment. To support the conclusion that Paneth cell numbers are reduced after Rfng deletion in vivo Figure 1, qPCR for LYZ needs to be done on material from Figure 1F. Also Lgr5 expression analysis would be better than OLFM4 and ASCL2 in terms of consistency. In all the authors' experiments CBCs are also depleted for RFNG (the organoids are derived from Rfng knockdown CBCs and the mice are whole body knockouts). Therefore, it is impossible to rule out a role for RFNG expressed in CBCs or other cell types. To address this the authors should directly mix Rfng knockdown Paneth cells with wildtype CBCs (and vice-versa) and determine whether colony forming capacity is impacted.

Other comments:

1) The authors state that "Notch signalling promotes self-renewal of CBCs" and reference the review by van der Flier and Clevers. In that review, inhibition of Notch leads to goblet differentiation, reduced proliferation, and fewer CBCs; however, this is often reversible. The authors should revise their statements accordingly.

2) The lysozyme expression appears very nuclear in Figure 1C. Conventional immunohistochemistry is required to assess the different cell populations clearly. The same nuclear localisation for the Lgr5-GFP is not in line with the previous published expression patterns and should be re-evaluated by the authors.

3) Given that goblet cell numbers vary along the length of the small intestine (SI) it is important the same parts of the small intestine are examined e.g. In some of the figures it appears that different parts of the SI have been analysed. The authors need to state which part has been analysed in their scoring and if not the same parts must be analysed.

4) To clearly show a reduction in Notch signalling activity after Lfng loss, the authors should examine the expression of Notch intra cellular domain (Nicd).

5) Given that Notch signalling is not active in the villus, the function of Lfng in goblet cells on Notch signalling within these cells is confusing. A more likely explanation is that Lfng positive progenitors contribute to the described effects on colony formation as well as differentiation. The authors should clarify whether this progenitor population is controlling the detected effects rather the expression in goblet cells. For example, by Lfng-GFP sorting and exclusion of Muc2 positive cells, or Lfng-GFP and CD44 double positive cell sorting, followed by functional assays.

[Editors’ note: this article was subsequently rejected after discussions between the reviewers, but the authors submitted a revised version at a later point, being the manuscript then accepted for publication.]

Thank you for submitting your work entitled "Spatially Specific Fringe Modulation of Notch Ligands Supports Mammalian Intestinal Homeostasis" for consideration by eLife. Your article was sent to the three peer reviewers of the previous version of the manuscript, and the evaluation has been overseen by a Senior/Reviewing Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Kevin Myant (Reviewer #3). Reviewer #1 did not agree to participate in the review process and so the decision regarding the manuscript was based on discussions between Reviewers #2 and #3 and the Senior Editor.

As you can from the verbatim comments, reviewer #3 was satisfied with your revisions. However, reviewer #2 felt that you had failed to address key points raised previously. In consultation, the Senior Editor agreed with reviewer #2 that your revisions were inadequate. Therefore we have agreed that the manuscript cannot be published in eLife.

Reviewer #2:

The authors have attempted to revise the manuscript in response to my comments. Unfortunately the response to my comments has not really allayed my problems with the manuscript.

The four points below have not, in my mind, been satisfactorily addressed and need to be.

Point 1: Expression of fringe.

The ISH data are not convincing for many reasons. First, RNAscope normally doesn't give dominantly signal in the nucleoli, based on the fact that mRNA is shuttled out of the nucleus after synthesis quickly. Additionally, why are Lysozyme positive cells in picture Figure 1B are not positive for Rfng? Quantification of Rnfg positive Paneth cells will be useful to understand the importance of Rnfg in Paneth cells. Second, cells outside the crypt are positive for Rnfg. How do the authors interpret these Rnfg-positive cells? And how do these cells contribute to the detected effects?

Point 2: Lysozyme expression.

The enlarged figure is not sufficient and doesn't show cytoplasmic lysozyme expression at all. This technical issue needs to be corrected and much more convincing data needs to be provided. Data shown in Figure 1B show expression patterns for Lysosyme as expected and indicates that correct staining is possible.

Point 4. To clearly show a reduction in Notch signalling activity after Lfng loss, the authors should examine the expression of Notch intra cellular domain (Nicd).

It appears that the intensity of the IF presented in the revised version of the manuscript of NICD is much stronger rather than a change in NICD positive cells. The authors need to quantify this data for both points to be able to show the effect of Lfng loss. Furthermore, double staining of Lfng-GFP and NICD will help to understand the interaction of Lfng positive and Notch positive cells.

Point 5. Given that Notch signalling is not active in the villus, the function of Lfng in goblet cells on Notch signalling within these cells is confusing.

Data which cell type is actually Lfng positive is not solid. Figure 4 doesn't show convincingly that Lfng is expressed in transit amplifying cells. This is of major importance since this would be the compartment that shows overlap with Notch activity and would explain the Lfng loss mediated effect on Notch signalling. Lfng positive Goblet cells in the villus can't explain this effect.

Reviewer #3:

My main concerns have been suitably addressed and I would now recommend publication of this study.

https://doi.org/10.7554/eLife.35710.029

Author response

Essential revisions:

None of the reviewers was completely convinced that the authors have shown what they claim and further experiments and/or analysis are needed for the work to be convincing. In particular the authors need to more clearly demonstrate that Fringe proteins are expressed in Paneth cells and that their cell sorting technique is robustly sorting Paneth / Lgr5+ cells.

1) Expression of Fringe proteins in Paneth cells. Although this is a key issue for the study, the data is limited and needs to be strengthened. While Fringe proteins were shown by Schroder et al. to be expressed in the intestinal crypts, they did not appear to be particularly specific in their in situ studies for the location of Paneth cells. No additional localization by either ISH or IHC is shown. While the FACS studies "using an established protocol" suggest greater expression in Paneth cells compared to Lgr5+ cells, the methodology is not easy to decipher, and the relative specificity of the FACS studies is not convincing. The authors need to describe exactly which surface markers were used to achieve their claimed populations. Validation of the FACS studies is needed, together with immunolocalization studies.

We agree with the reviewers on the need to strengthen the pivotal datum of the study. Firstly, we validated the FACS protocol used to obtain the Rfng expression pattern. We confirmed that the Paneth cells sorted-based on side scatter and CD24 expression are indeed positive for their marker Lysozyme (Figure 1—figure supplement 1A). Lgr5-GFP CBCs were sorted based on GFP signal (Figure 1—figure supplement 1B). We have updated the Materials and methods section to include the FACS protocol.

We have not been successful in immunostaining for RFNG using any of the commercially available antibodies. An antibody reported to be suitable for immunostaining (Ryan et al., 2008) showed non-specific staining on intestinal sections. We therefore performed in situ hybridisation (ISH) using RNAscope probes (ACDBio) to obtain Rfng expression pattern and the results of which show that Rfng transcripts are primarily localised to the Paneth cells (Figure 1B). Paneth cells were identified by Lysozyme expression. Z-stack images were taken to correctly identify the Paneth cells (Author response image 1). Also, our RNA-ISH assay using RNAscope produces similar images to those published previously (Lafkas D et al., Nature, 2015).

Author response image 1
3D projection (from z-stack spanning 9μm) was used to correctly identify the Paneth cells in Figure 1B.

Rfng transcripts (red) and Lysozyme protein (green) expression can be seen at the bottom of the crypts. Dapi (Blue) labels the nuclei.

https://doi.org/10.7554/eLife.35710.023

2) Expression of DLL1 and DLL4 on Paneth cells. The authors suggest that the ligands DLL1 and DLL4 are well expressed on Paneth cells, based on Western blotting of "Paneth cells". These are presumably from cell sorting, although the figure legend describes the lysates as derived from "organoids". Stamataki D et al. (PLoS One 2011) used a DLL1-β-galactosidase transgene to mark DLL1+ cells and found it was expressed in goblet cells (96%) and enteroendocrine cells (66%) but not in Paneth cells. Van et al., 2012 used a different DLL1-GFP-IRES-CreERT2 knockin mouse; strong expression above the stem cell zone (e.g. position +5 – +10) and only weak expression in the Paneth cells was reported. Finally, Shimizu H et al. (PeerJ, 2014) used double immune staining to show that DLL1 was not expressed in Paneth cells. In contrast Shimizu et al. did confirm that DLL4 was expressed in Paneth cells. Given these published findings, the authors need substantial additional evidence (e.g. immunofluorescence, qPCR, in situ hybridisation) in order to claim DLL1 expression in Paneth cells.

We agree that there is a lack of clarity in literature on the expression of Dll1 in Paneth cells. Studies based on Dll1 knock-in mice claim that Dll1 is expressed by Paneth cells. Samataki D et al. (PLoS One 2011, Figure 1D) show that a subset of Paneth cells express the β-galactosidase reporter driven by Dll1 promoter. Van et al., 2012 show that Paneth cells express DLL1 but the expression level is not as high as that in the secretory progenitors. Although Shimizu H et al. (PeerJ, 2014) claim that Paneth cells do not express DLL1 based on immunostaining experiments, upon closer observation of their data we can see in Figure 3A of their article that DLL1+ cell is found in the region below the progenitors suggesting that it might be a Paneth cell. Shorning BY et al. (PLoS One 2009) have also performed immunostaining for DLL1 albeit using a different antibody and shown that Paneth cells do express DLL1.

Pellegrinet et al., 2011 showed that Dll4 loss does not significantly affect the intestinal epithelium. However, they also showed that a combined loss of Dll1 and Dll4 leads to a near complete loss of stem cells. If Dll1 were not to be expressed at the bottom of the crypts as claimed by Shimizu H et al. (PeerJ, 2014), combined loss of Dll1 and Dll4 would not have such a severe phenotype.

We have been not been successful in immunostaining for DLL1 using commercially available antibodies. Hence, we performed ISH to obtain Dll1 expression pattern and find that Paneth cells express Dll1 (Figure 2—figure supplement 1A). We find that the expression level of Dll1 is very high in a few progenitors in coherence with the data in Van et al., 2012. We also observe that a subset of Paneth cells show higher expression of Dll1 transcripts compared to others which explains the data in Samataki D et al. (PLoS One 2011).

Together with RNA-ISH (Figure 2—figure supplement 1A), western blot (Figure 2A), flow cytometry (Figure 2D) and cited previous studies we can conclude that Paneth cells express Dll1 albeit at a lower level than the Dll1+ secretory progenitor cells.

We have modified the legend of Figure 2A to correctly indicate that the DLL1 and DLL4 expression data are based on Western blotting of Paneth cells FACS sorted from organoids. We thank the reviewers for pointing out the issue.

3) The conclusion that Rfng expression on Paneth cells promotes Notch activity in neighbouring Lgr5+ CBCs is based on the finding that Rfng-/- intestines show reduced CBCs, although no other phenotype was detected, which is inconsistent with a reduced Notch phenotype. In addition, loss of Rfng led to a reduction in Lgr5+ CBCs in organoid culture. However, this conclusion that the changes are due to "Paneth cells" remains puzzling, given the reported absence of DLL1 in Paneth cells. In addition, several groups (Garabedian EM et al., J Biol Chem 1997; Kim TH et al., Proc Natl Acad Sci U S A 2012) have now shown that deletion in vivo of Paneth cells does not result in any significant in vivo phenotype. While Paneth cells likely express Dll4, depletion of Dll4 alone or administration of a Dll4-immunoneutralizing antibody has no effect on intestinal epithelial homeostasis (Ridgway et al., 2006). This would be consistent with the lack of a major role for Paneth cells in regulating the ISC compartment. To support the conclusion that Paneth cell numbers are reduced after Rfng deletion in vivo Figure 1, qPCR for LYZ needs to be done on material from Figure 1F. Also Lgr5 expression analysis would be better than OLFM4 and ASCL2 in terms of consistency. In all the authors' experiments CBCs are also depleted for RFNG (the organoids are derived from Rfng knockdown CBCs and the mice are whole body knockouts). Therefore, it is impossible to rule out a role for RFNG expressed in CBCs or other cell types. To address this the authors should directly mix Rfng knockdown Paneth cells with wildtype CBCs (and vice-versa) and determine whether colony forming capacity is impacted.

Notch activity is responsible for the maintenance of Lgr5+ CBCs at the crypt bottom and differentiation of progenitors into enterocytes in the upper crypt (Noah TK et al., Annu Rev Physiol, 2013; Pellegrinet et al., 2011; Fre et al., 2005). Reduction of Notch signalling owing to the loss of ligands Dll1/4, receptor Notch1 or due to reduction of γ-Secretase activity (Noah TK et al., Annu Rev Physiol, 2013) affects both the upper and lower crypts. However, loss of Rfng reduces Notch signalling only in the lower crypts without affecting progenitor differentiation. This explains why we do not observe any goblet cell metaplasia typically associated with reduced Notch activity in the intestine.

We agree that the importance of Paneth cells in vivohas been contested. Lgr5+ CBCs are maintained mainly by Notch and Wnt signalling pathways. Both mesenchymal cells and Paneth cells are sources of Wnt ligands (Kabiri Z et al., Development, 2014; Sato T et al., Nature, 2011; Valenta T et al., Cell Rep, 2016). Paneth cells are known to express Notch ligands DLL1 and DLL4 to activate Notch receptors on the Lgr5+ CBCs (Sato and Clevers, Science, 2013; Van et al., 2012). Garbedian EM et al. (J Biol Chem, 1997) ablate mature Paneth cells using transgenic mice in which diphtheria toxin or SV40 T antigen is expressed as directed by Cryptidin-2 gene. They observe that the intestinal functions are not significantly affected. They also observe that the region previously occupied by Paneth cells is now occupied by what appear to be transit-amplifying cells. They have not investigated if Notch/Wnt activity has been affected in the Lgr5+ CBCs. They have not investigated if the expression of any stem cell markers have been affected (the study was published prior to the identification of Lgr5+ cells). It is possible that the “semi-differentiated” cells that replace the Paneth cells provide the necessary Notch ligands to support the neighbouring CBCs. Kim TH et al. (Proc Natl Acad Sci USA, 2012) and Durand A et al. (Proc Natl Acad Sci USA, 2012) ablate Paneth cells by using mice lacking Atoh1 (Math1) gene. It has been shown that the role of Notch activity in the intestinal crypts is to suppress Atoh1 expression (Kazanjian A et al., Gastroenterology, 2010). In the absence of Atoh1, suppression of Notch activity by using γ-Secretase inhibitors has no significant phenotype (Van et al., 2010; Kim TH, J Biol Chem, 2010; Kazanjian A et al., Gastroenterology, 2010). This explains why the loss of Paneth cells upon the loss of Atoh1 does not significantly affect the Lgr5+ CBCs. Sato T et al. (Nature, 2012) have shown that Paneth cell depletion leads to a significant reduction in stem cell numbers. Also, Yilmaz OH et al. (Nature, 2012) have shown that Paneth cells can boost Lgr5+ CBC function in response to calorie restriction.

However, we do not claim that the loss of Rfng leads to depletion of Paneth cells. We find that in the absence of Rfng, the availability of DLL1 and DLL4 on the surface of Paneth cells reduces which results in reduced Notch activity in the neighbouring CBCs leading to their reduced self-renewal. These results are consistent with the observations of Pellegrinet et al., 2011 where they find that a complete loss of both DLL1 and DLL4 led to a depletion of Lgr5+ cells without any significant reduction in Paneth cell numbers.

To provide the additional verification requested, we performed Organoid Reconstitution Assay (Rodriguez-Coleman, MJ et al., Nature, 2017) to show that loss of Rfng in Paneth cells affects the Lgr5+ CBCs. FACS sorted Lgr5-GFP cells were incubated with Paneth cells from wild type or Rfngnull mice for 10 minutes at room temperature and plated in Matrigel. We find that the colony formation ability of Lgr5+ CBCs incubated with Paneth cells lacking Rfng was significantly lower than the control (Figure 1—figure supplement 3A). It should be noted that not all CBCs associate with a Paneth cell during the incubation. Also the Lgr5+ stem cells divide and give rise to Paneth cells with Rfng. Hence the results of this assay are not as significant as that shown in Figure 1C.

As Rfng-/- mice do not have Lgr5-GFP reporter allele, we were unable to isolate CBCs from them. Hence we are unable to show if loss of Rfng only in CBCs affects their colony formation ability. We added this caveat in the Discussion section. As Rfng is expressed mainly in the Paneth cells we feel that we would not see a marked difference in colony formation ability in this case.

We have replaced Figure 1F (now Figure 1G) to show Lgr5 expression instead of Olfm4 and Ascl2, as suggested.

Other comments:

1) The authors state that "Notch signalling promotes self-renewal of CBCs" and reference the review by van der Flier and Clevers. In that review, inhibition of Notch leads to goblet differentiation, reduced proliferation, and fewer CBCs; however, this is often reversible. The authors should revise their statements accordingly.

We thank the reviewers for the comment. The statement has been revised to “Notch signalling is important for the maintenance of CBCs (Pellegrinet et al., 2011).”

2) The lysozyme expression appears very nuclear in Figure 1C. Conventional immunohistochemistry is required to assess the different cell populations clearly. The same nuclear localisation for the Lgr5-GFP is not in line with the previous published expression patterns and should be re-evaluated by the authors.

The staining appears nuclear in the figure shown owing to the smaller image size and the previous brightness and contrast settings. However, after we enlarged the image and increased the brightness and contrast, we can see that the staining is indeed cytoplasmic (Author response image 2). We have increased the size of the image in the main figure to avoid confusion. Also, our conclusions are mainly rely on the quantitative flow cytometric measurements.

Author response image 2
Lysozyme (red) and GFP (green) staining in Paneth cells and CBCs respectively are cytoplasmic as expected.

Figure 1D (previously Figure 1C) has been enlarged and brightness and contrast have been increased.

https://doi.org/10.7554/eLife.35710.024

3) Given that goblet cell numbers vary along the length of the small intestine (SI) it is important the same parts of the small intestine are examined e.g. In some of the figures it appears that different parts of the SI have been analysed. The authors need to state which part has been analysed in their scoring and if not the same parts must be analysed.

The intestine was divided into 9 equal parts (from the proximal to distal end: duodenum-1, 2 and 3, jejunum-1, 2 and 3 and ileum-1, 2 and 3) while harvesting. Corresponding regions from control and mutated mice were compared. Scoring was mainly performed on ileal sections. Regions next to Peyer’s patches were excluded from the analysis.

4) To clearly show a reduction in Notch signalling activity after Lfng loss, the authors should examine the expression of Notch intra cellular domain (Nicd).

We thank the reviewers for this suggestion. We have stained the intestinal sections from Lfng null and control mice for NICD and observed a reduction in its expression in the crypts of Lfng null mice (Figure 5—figure supplement 2A).

5) Given that Notch signalling is not active in the villus, the function of Lfng in goblet cells on Notch signalling within these cells is confusing. A more likely explanation is that Lfng positive progenitors contribute to the described effects on colony formation as well as differentiation. The authors should clarify whether this progenitor population is controlling the detected effects rather the expression in goblet cells. For example, by Lfng-GFP sorting and exclusion of Muc2 positive cells, or Lfng-GFP and CD44 double positive cell sorting, followed by functional assays.

Goblet cells are known to express Notch ligands DLL1 and DLL4, however their role has not yet been fully understood. We find that Lfng in goblet cells can increase the cell-surface expression of these ligands. However, we are unable to ascertain the functional importance of Lfng in goblet cells.

Loss of Lfng leads to reduced Notch activity in the progenitors (Figure 5D) and this leads to an increase in the number of goblet cells (Pellegrinet et al., 2011), which is consistent with the reviewers’ explanation.

[Editors’ note: the author responses to the re-review follow. As indicated above, the manuscript was subsequently accepted for publication.]

[…] As you can from the verbatim comments, reviewer #3 was satisfied with your revisions. However, reviewer #2 felt that you had failed to address key points raised previously. In consultation, the Senior Editor agreed with reviewer #2 that your revisions were inadequate. Therefore we have agreed that the manuscript cannot be published in eLife.

Reviewer #2:

The authors have attempted to revise the manuscript in response to my comments. Unfortunately, the response to my comments has not really allayed my problems with the manuscript.

The four points below have not, in my mind, been satisfactorily addressed and need to be.

Point 1: Expression of fringe.

The ISH data are not convincing for many reasons. First, RNAscope normally doesn't give dominantly signal in the nucleoli, based on the fact that mRNA is shuttled out of the nucleus after synthesis quickly.

We thank the reviewer for this observation. We agree that the result is puzzling as mRNA is shuttled out of the nucleus soon after synthesis. We consulted many experts to understand the contrariety. According to them, they have seen both (dominant signal from nucleus vs. cytoplasm) in many cases depending on the RNA species. The bright signals in the nuclei are likely to be pre-mRNA molecules or active transcription sites. Multiple nascent mRNA molecules are often found in a focus at transcription sites resulting in a bright spot (Raj A et al., Plos Biol, 2006). Mature mRNA molecules are often diffused in the cytoplasm and if the copy number of mRNA molecules is not very high it is possible that the only signals bright enough for detection would be from the transcription sites. We can see in (Figure 1, 2; Kwon S et al., Sci Rep, 2017) that the nuclear dots are quite brighter than those in the cytoplasm. We can see similar images in (Figure 1B, Bentovim L et al., Development, 2017; Figure S7, Levesque and Raj, Nat Meth, 2013). Accumulation of nascent mRNA in the nucleus, as the gene is transcribed, also results in a prominent nuclear signal (Lodish H et al., Mol. Cell Biol., 2000; Shermoen et al., Cell, 1991; Kwon S et al., Sci Rep, 2017). Stapel LC et al. (Development, 2016) also has shown that transcripts of certain genes are detected more often in the nucleus than in the cytoplasm.

That said, we have used RNAscope chromogenic assay for our studies. The chromogenic Fast Red signal obtained is also known to be fluorescent, and we acquired the data using a confocal fluorescent microscope. It should be noted that chromogenic RNAscope assay has more signal amplification steps when compared to the fluorescent assay. As a result, we may expect to see difference in the punctate signal in both cases. RNAscope chromogenic assays do often show a dominant signal in the nucleus. We have included a few snapshots of published images for the reviewers’ consideration to show that mRNA ISH using RNAscope can result in a dominant nuclear signal (please refer to Christensen AB et al, Mediators of Inflammation, 2015, Figure 4; Dominguez-Brauer C et al, Cell Stem Cell, 2016, Fig. S2D; Ziskin JL et al, Gut, 2012, Fig. 2F; Cheung EC et al, Genes and Dev., 2016, Suppl Fig. 2A; and Cammareri P et al, Cell Death Differ., 2017, Fig. 3A).

We counterstained the Rfng-ISH slides using haematoxylin and obtained a brightfield image (Author response image 3). As the tissue section is thick, was pre-treated with protease and haematoxylin was significantly diluted to prevent masking the ISH signal, stain appears weak. Nucleoli stain strongly with haematoxylin and we can see that ISH signal does not colocalise with bright blue spots. Although it is not possible to distinguish between heterochromatin and nucleolus in the image, it appears that ISH signal does not colocalise with either of them.

Author response image 3
Representative image of RNAscope ISH for Rfng counterstained using haematoxylin on Lgr5-GFP mouse intestine.

Arrows indicate regions with strong haematoxylin staining. Note that they do not overlap with the ISH signal. Scale bar represents 10μm.

https://doi.org/10.7554/eLife.35710.025

As the above explanations only suggest but not completely confirm the legitimacy of the assay, we confirmed the specificity of the Rfng ISH probes by performing the assay on Rfng null intestines. We see no specific signal in absence of Rfng mRNA (Figure 1—figure supplement 1D, E). We performed in situ HCR for Rfng and observed nuclear dots. It should be noted that the assay is not as sensitive as the RNAscope chromogenic assay (Author response image 4).

We felt that an independent validation would boost the confidence in this observation. We analysed published microarray data (Sato T et al., Nature, 2011) on Lgr5+ CBCs and Paneth cells and found that Rfng is enriched in Paneth cells (Figure 1—figure supplement 1A).

Author response image 4
Representative image of in situ HCR for Rfng on Lgr5-GFP mouse intestines shows signal primarily in the nuclei.

Arrows point to nuclei with HCR signal. Dotted line separates the epithelium from the mesenchyme. Scale bar represents 10μm. Assay was performed as per manufacturer’s recommendation (Choi, Beck et al., 2014, Shah, Lubeck et al., 2016). V3.0 Probes for Rfng were purchased from Molecular Instruments. Tissues were pretreated using RNAscope reagents. Probes were hybridised to the section overnight at 400C, washed using 5x SSCT buffer (5x SSC with 0.1% Tween-20) and signal was amplified overnight at 400C. Slides were washed in SSCT, mounted and imaged using a confocal microscope. Spectral imaging and linear unmixing using Zen (Zeiss) software was performed to detect Rfng signal.

https://doi.org/10.7554/eLife.35710.026

Additionally, why are Lysozyme positive cells in picture Figure 1B are not positive for Rfng? Quantification of Rnfg positive Paneth cells will be useful to understand the importance of Rnfg in Paneth cells.

Lysozyme positive cells in Figure 1B are positive for Rfng. As not all cells are centred on the same plane, the signal in any 2D image would vary from cell to cell. As it can been seen from the 3D rendering of the image for the reviewers’ consideration cited above (please refer to Christensen AB et al, Mediators of Inflammation, 2015, Figure 4; Dominguez-Brauer C et al, Cell Stem Cell, 2016, Fig. S2D; Ziskin JL et al, Gut, 2012, Fig. 2F; Cheung EC et al, Genes and Dev., 2016, Suppl Fig. 2A; and Cammareri P et al, Cell Death Differ., 2017, Fig. 3A), multiple punctate signals are present in different planes in different cells. And this would help us conclude that Lysozyme-expressing Paneth cells also express Rfng.

Second, cells outside the crypt are positive for Rnfg. How do the authors interpret these Rnfg-positive cells? And how do these cells contribute to the detected effects?

We thank the reviewer for this very interesting observation. We agree that some mesenchymal cells also express Rfng detectable by RNAscope assay. As Notch is a juxtacrine signalling pathway, we had confined the analysis to Paneth cells. Schröder et al., 2002 have shown, by RNA ISH, that the mesenchymal cells also express Rfng and this agrees with our images. We see signal in the mesenchymal cells by in situHCR (Author response image 4). We also observe that the mesenchymal cells also express Dll1 (Figure 2—figure supplement 1). Further studies are needed to map the expression of all the Notch ligands in different mesenchymal cell types. This raises the possibility that the mesenchyme can also provide Notch ligands to the Lgr5+ CBCs in vivo. In case that is true, our proposed mechanism that RFNG promotes cell surface expression of DLL1 might be applicable to the mesenchymal cells too. Upon loss of Rfng, reduced DLL1 expression on the cell surface of Paneth cells and mesenchymal cells would result in reduced Notch activity in the CBCs, as observed. However, the epithelium is separated from the mesenchyme by a basement membrane. The efficacy of DLL mediated Notch signalling across the intestinal basement membrane would be an interesting scientific question by itself which we hope to address in our upcoming studies. We have included this limitation in the Discussion section to help readers interpret the data appropriately.

Point 2: Lysozyme expression.

The enlarged figure is not sufficient and doesn't show cytoplasmic lysozyme expression at all. This technical issue needs to be corrected and much more convincing data needs to be provided. Data shown in Figure 1B show expression patterns for Lysosyme as expected and indicates that correct staining is possible.

We agree with the reviewer that lysozyme staining on organoids does not look as good as that on FFPE samples shown in Figure 1B. Whole organoids in Matrigel were fixed, stained and imaged in Figure 1D. The thickness of the tissue and the need to capture cells centred on different planes resulted in this appearance of Lysozyme signal. Similar staining has been reported previously (Basak et al., Cell Stem Cell, 2017). To avoid any confusion to the readers we have removed the image from the manuscript. Our conclusions are based on quantitative flow cytometry data in Figure 1E and F and hence would not be affected by removal of this image.

Point 4. To clearly show a reduction in Notch signalling activity after Lfng loss, the authors should examine the expression of Notch intra cellular domain (Nicd).

It appears that the intensity of the IF presented in the revised version of the manuscript of NICD is much stronger rather than a change in NICD positive cells. The authors need to quantify this data for both points to be able to show the effect of Lfng loss. Furthermore, double staining of Lfng-GFP and NICD will help to understand the interaction of Lfng positive and Notch positive cells.

We thank the reviewers for suggesting these experiments. We quantified the fraction of nuclei positive for NICD in the crypts of Lfng+/+ and Lfng-/- mouse intestines and found that Lfng loss indeed reduces the number of NICD+ nuclei (Figure 5—figure supplement 2A, B). We performed double staining of LfngGFP and NICD and found that Lfng-GFP+ cells do not express NICD but are adjacent to NICD+ cells (Figure 4E).

Point 5. Given that Notch signalling is not active in the villus, the function of Lfng in goblet cells on Notch signalling within these cells is confusing.

Goblet cells have been shown previously (Shmizu et al., PeerJ, 2014) to express Notch ligands Dll1 and Dll4, even though Notch signalling is not known to be active in the villus. We find that Lfng is also expressed by the Dll1+ Dll4+ goblet cells. We have found that LFNG, like RFNG, can promote DLL1 and DLL4 to their cell surface.

However, we agree with the reviewer that Lfng in goblet cells of the villi likely do not have a functional consequence in terms of Notch signalling or cell lineage determination. We further explored Lfng expression in the progenitor compartment, which is discussed in the following point.

Data which cell type is actually Lfng positive is not solid. Figure 4 doesn't show convincingly that Lfng is expressed in transit amplifying cells. This is of major importance since this would be the compartment that shows overlap with Notch activity and would explain the Lfng loss mediated effect on Notch signalling. Lfng positive Goblet cells in the villus can't explain this effect.

We thank the reviewer for encouraging us to probe further into the cells expressing Lfng in the crypts. We have since realised the importance of this experiment. We surprisingly found that the Lfng-GFP+ cells are not proliferating suggesting that they might be terminally differentiated (Figure 4A). As they are negative for activated Notch1 staining (Figure 4E), we hypothesised that they might be secretory cells. We find that they are positive for ChgA or Dclk1, which are markers for enteroendocrine or Tuft cells respectively (Figure 4B-D). Tuft, enteroendocrine cells and often goblet cells also are known to be present in the upper crypt (Gerbe et al., J Cell Biol, 2011; Barron et al., CMGH, 2017; VanDussen et al., 2012; Tian et al., Cell Rep, 2015). Enteroendocrine and goblet cells are known to express Notch ligand Dll1 (Shimizu et al., PeerJ, 2014, Van et al., 2012). These data suggest that LFNG in NICD- post-mitotic secretory cells of the upper crypt promotes Notch activity in the neighbouring enterocyte progenitors.

https://doi.org/10.7554/eLife.35710.030

Article and author information

Author details

  1. Preetish Kadur Lakshminarasimha Murthy

    1. Center for Genomics and Computational Biology, Department of Biomedical Engineering, Duke University, Durham, United States
    2. Sibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Writing—original draft, Project administration, Writing—review and editing
    Contributed equally with
    Tara Srinivasan
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9762-8376
  2. Tara Srinivasan

    Mienig School of Biomedical Engineering, Cornell University, Ithaca, United States
    Contribution
    Data curation, Formal analysis, Investigation, Visualization, Writing—original draft, Writing—review and editing
    Contributed equally with
    Preetish Kadur Lakshminarasimha Murthy
    Competing interests
    No competing interests declared
  3. Matthew S Bochter

    Department of Molecular Genetics, Ohio State University, Columbus, United States
    Contribution
    Resources, Maintained the mouse colonies and, harvested and processed intestinal samples
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9607-3770
  4. Rui Xi

    Center for Genomics and Computational Biology, Department of Biomedical Engineering, Duke University, Durham, United States
    Contribution
    Formal analysis, Investigation, Visualization
    Competing interests
    No competing interests declared
  5. Anastasia Kristine Varanko

    Mienig School of Biomedical Engineering, Cornell University, Ithaca, United States
    Contribution
    Investigation, Performed experiments to generate supporting data not shown in the manuscript
    Competing interests
    No competing interests declared
  6. Kuei-Ling Tung

    1. Center for Genomics and Computational Biology, Department of Biomedical Engineering, Duke University, Durham, United States
    2. Department of Biological and Environmental Engineering, Cornell University, Ithaca, United States
    Contribution
    Investigation, Performed experiments to generate supporting data not shown in the manuscript
    Competing interests
    No competing interests declared
  7. Fatih Semerci

    Department of Pediatrics, Baylor College of Medicine, Houston, United States
    Contribution
    Resources, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-0512-1827
  8. Keli Xu

    Department of Neurobiology and Anatomical Sciences, University of Mississippi Medical Center, Jackson, United States
    Contribution
    Resources, Writing—review and editing
    Competing interests
    No competing interests declared
  9. Mirjana Maletic-Savatic

    Department of Pediatrics, Baylor College of Medicine, Houston, United States
    Contribution
    Resources, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6548-4662
  10. Susan E Cole

    Department of Molecular Genetics, Ohio State University, Columbus, United States
    Contribution
    Conceptualization, Resources, Writing—review and editing
    Competing interests
    No competing interests declared
  11. Xiling Shen

    1. Center for Genomics and Computational Biology, Department of Biomedical Engineering, Duke University, Durham, United States
    2. Mienig School of Biomedical Engineering, Cornell University, Ithaca, United States
    3. School of Electrical and Computer Engineering, Cornell University, Ithaca, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Writing—review and editing
    For correspondence
    xiling.shen@duke.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4978-3531

Funding

National Institutes of Health (R35GM122465, R01GM114254)

  • Xiling Shen

Defense Advanced Research Projects Agency (HR0011-16-C-0138)

  • Xiling Shen

National Science Foundation (NSF 1350659 career award)

  • Xiling Shen

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We are grateful to Dr Brigid Hogan of Duke University, Dr Linda Samuelson of University of Michigan and Dr Robert Haltiwanger of University of Georgia for their valuable comments. We thank Dr Steven Lipkin of Weill Cornell Medical College for providing R-SPONDIN1 conditioned medium. We thank Dr Kameswaran Surendran of Sanford Research for sharing the protocol for NICD staining.

Ethics

Animal experimentation: All procedures were conducted under protocols approved by the appropriate Institutional Animal Care and Use Committees at Ohio State University (# 2012A00000036-R1), Duke University (# A286-15-11), Baylor College of Medicine (# AN-5004), Cornell University (# 2010-0100) or Research Animal Resource Center of Weill Cornell Medical College (# 2009-0029).

Reviewing Editor

  1. Fiona M Watt, King's College London, United Kingdom

Publication history

  1. Received: February 7, 2018
  2. Accepted: March 6, 2018
  3. Version of Record published: April 9, 2018 (version 1)

Copyright

© 2018, Kadur Lakshminarasimha Murthy et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,163
    Page views
  • 160
    Downloads
  • 5
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Cell Biology
    2. Physics of Living Systems
    Kaushikaram Subramanian et al.
    Research Article
    1. Cell Biology
    2. Developmental Biology
    Jean-André Lapart et al.
    Research Article