Rapid Crypt Cell Remodeling Regenerates the Intestinal Stem Cell Niche after Notch Inhibition

Highlights

•Dynamic intestinal crypt remodeling induced by short-term Notch inhibition

• •Intestinal stem cells are maintained but exhibit a transient loss of function
• •Paneth cells are rapidly lost by apoptotic cell death
• •A regenerative response replenishes Paneth cells to repair the crypt

Summary

Intestinal crypts have great capacity for repair and regeneration after intestinal stem cell (ISC) injury. Here, we define the cellular remodeling process resulting from ISC niche interruption by transient Notch pathway inhibition in adult mice. Although ISCs were retained, lineage tracing demonstrated a marked reduction in ISC function after Notch disruption. Surprisingly, Notch ligand-expressing Paneth cells were rapidly lost by apoptotic cell death. The ISC-Paneth cell changes were followed by a regenerative response, characterized by expansion of cells expressing Notch ligands Dll1 and Dll4, enhanced Notch signaling, and a proliferative surge. Lineage tracing and organoid studies showed that Dll1-expressing cells were activated to function as multipotential progenitors, generating both absorptive and secretory cells and replenishing the vacant Paneth cell pool. Our analysis uncovered a dynamic, multicellular remodeling response to acute Notch inhibition to repair the niche and restore homeostasis. Notably, this crypt regenerative response did not require ISC loss.

Graphical Abstract

Keywords

• cellular plasticity/ DLL1/ Paneth cell/ regeneration/ secretory progenitor cell
• DLL4
• CBC stem cell
• facultative stem cell

Introduction

In spite of our understanding of CBC niche regulation during homeostasis, little is known about niche responses after injury. Earlier studies of genetic Notch disruption or pharmacologic Notch inhibition via administration of a gamma-secretase inhibitor (GSI) such as dibenzazepine (DBZ), were limited by the reduced animal viability observed when inhibition extended over several days, which impeded analysis of the regeneration process. Indeed, the intestinal toxicity imparted by Notch inhibitors limits use in the clinic despite their great therapeutic potential for treating Notch-driven cancers (e.g., T cell acute lymphoblastic leukemia) and other diseases. Short-term Notch inhibition may be one approach to maintain ISCs and minimize toxicity in human patients, yet there is little known about ISC responses to short-term Notch interruption. Here, we introduce an intestinal crypt disruption model based on short-term niche factor inhibition. We probe the setting of pharmacologic inhibition to investigate the acute cellular response to Notch niche disruption. We demonstrate that short-term Notch disruption leads to transient ISC dysfunction and dynamic crypt cell remodeling. This process is highlighted by rapid Paneth cell loss, a novel contrast to previous findings established by studies using longer time points of Notch inhibition that demonstrated Paneth-like cell expansion. Furthermore, after short-term Notch disruption we observed an expansion of cells expressing Notch ligands and increased Notch signaling, with a regenerative response characterized by a proliferative surge. We show that Dll1+ cells are activated to function as multipotential progenitors to maintain the epithelium during ISC dysfunction, including replenishment of the vacant Paneth cell population. Our study sheds light into how niche pathways and crypt cell plasticity orchestrate intestinal repair.

Results

 Impaired CBC Function after Acute Notch Inhibition

We analyzed the immediate effect of Notch inhibition on ISC function and characterized the crypt regenerative response after short-term pathway disruption in adult mice treated with a single dose of DBZ. Intestinal tissue was isolated at various times post-DBZ, which has a plasma half-life of less than 12 h (

Milano et al., 2004), to characterize the dynamic crypt cell response (Figure 1A). We observed rapid loss of expression of the CBC marker (van der Flier et al., 2009) and Notch target gene (VanDussen et al., 2012) Olfm4 as early as 12 h post-DBZ, with expression returning at day 3 (Figures 1B, 1C, and S1A). In contrast, expression of the CBC Wnt target gene Lgr5 was not changed (Figures 1B and 1C), suggesting that the dynamic changes to Olfm4 reflected loss of CBC Notch signaling rather than stem cell depletion.

To assess the effect of acute Notch inhibition on CBC function, we measured lineage tracing using two different CBC-specific Cre driver strains (Olfm4-GFP-CreER^T2 and Lgr5-GFP-CreER^T2) crossed to the ROSA26-LSL-tdTomato (Tom) reporter. The Tom lineage mark was activated in CBCs by treatment with tamoxifen (TX), followed by DBZ or vehicle (Veh) treatment, with analysis 1 day later (Figure 1D). We observed significantly fewer lineage-traced cells in DBZ-treated mice compared with Veh-treated controls (Figure 1E). Quantification of the number of Tom-labeled cells per crypt showed that DBZ-treated Olfm4 and Lgr5 reporter mice had an approximately 2-fold reduction in lineage tracing, demonstrating impaired CBC function (Figure 1E). Interestingly, the Tom-labeled cells were clustered at the crypt base in a pattern distinct from the Veh-treated controls, suggesting crypt cell remodeling post-DBZ (Figure 1E).

 Rapid Paneth Cell Apoptosis after Acute Notch Inhibition

Histological analysis of the crypt post-DBZ showed dynamic cellular remodeling. Remarkably, granule-filled Paneth cells at the crypt base were lost within 12 h of DBZ administration, together with the appearance of delaminated cells (Figure 2A, arrowheads). To further examine this effect, we analyzed the expression of Paneth cell-specific markers by immunostaining (lysozyme) and qRT-PCR (cryptdins), showing that both were markedly reduced in DBZ-treated crypts as early as 12 h after administration (Figures 2B and 2C). To determine whether the loss of Paneth cell marker expression was due to cellular remodeling or cell loss, we used Defensin α4-Cre;Tom mice, which permanently label Paneth cells with a Tom lineage mark. We observed a marked loss of Tom-labeled cells 1 day post-DBZ in these mice, confirming that Notch inhibition led to rapid Paneth cell loss (Figure 2B, insets). Analysis of apoptosis by staining for cleaved caspase-3 showed a significant increase in apoptotic cells in the crypts, which peaked at 1 day post-DBZ (Figure 2D). Co-staining for the Paneth cell marker MMP7 and cleaved caspase-3 showed that the apoptotic cells were Paneth cells (Figure 2E).

 

Our findings suggest that Notch signaling is required for Paneth cell maintenance, which is unexpected as Paneth cells do not express Notch receptors or exhibit Notch signaling. To confirm that Paneth cell loss is a consequence of Notch inhibition and not due to the inhibition of another GSI target, we analyzed Defensin α4-Cre;Tom mice that were treated with a mixture of inhibitory antibodies to DLL1 and DLL4, the two Notch ligands required for crypt homeostasis (Pellegrinet et al., 2011). Similar to our findings after DBZ treatment, we observed reduced Paneth cell numbers after combined DLL1 and DLL4 Notch-signaling blockade, showing that Paneth cell loss is due to Notch inhibition (Figure S1). Paneth cell loss may explain reduced CBC function, shown by loss of Olfm4 and diminished lineage tracing post-DBZ (Figure 1), as Paneth cells are known to express several ISC niche factors (Sato et al., 2011). Notably, Paneth cells are returning by 3 days after acute Notch inhibition, with an apparent increase in numbers of granule-containing cells and Tom-marked cells by day 7, suggesting a rebound effect (Figures 2A and 2B). Interestingly, Paneth cells returning at day 3 corresponds to the timing of return of expression of Olfm4 (Figure 1B). The coincident reemergence of Olfm4 expression with Paneth cells was confirmed by co-imaging lysozyme and GFP in DBZ-treated Olfm4-GFP-CreER^T2 intestine, with GFP expression detected in stem cells adjacent to lysozyme-expressing Paneth cells (Figure 2F).

 Increased Notch Activity and Cell Proliferation during Crypt Regeneration

We assessed proliferation after DBZ treatment to determine if crypt remodeling included a regenerative response. Acute Notch inhibition resulted in a marked increase in proliferation at day 3, with a 1.6-fold increase in 5-ethynyl-2′-deoxyuridine (EdU+) cells (Figures 3A and 3B ). This proliferative surge was accompanied by dynamic changes in crypt cellularity. Decreased crypt cellularity was observed at 1 day post-DBZ (Figure 3C), a time point marked by loss of Paneth cells and diminished CBC stem cell function. However, at the time of the proliferative surge at day 3, crypts were expanded with increased cellularity compared with Veh-treated mice (Figures 3A and 3C).

 

Given the well-characterized role of Notch signaling to regulate intestinal proliferation (Fre et al., 2005, Pellegrinet et al., 2011, Riccio et al., 2008, Stanger et al., 2005, van Es et al., 2005, VanDussen et al., 2012), we investigated Notch activity to determine if the hyperproliferative response coincided with the return of Notch function. Immunostaining for the NOTCH1 intracellular signaling domain (NICD) showed loss of Notch activity 12 h post-DBZ (Figures 3D and 3E), which is consistent with the loss of expression of the Notch target gene Olfm4 at that time point (Figure 1B). NICD+ cell numbers increased over time, and by day 3 the number of NICD+ cells was markedly increased in comparison with Veh control, which coincided with the timing of increased proliferation and crypt cellularity. Interestingly, co-staining for NICD and EdU showed that proliferating cells at day 3 post-DBZ were commonly Notch-signaling cells, while most EdU+ cells in Veh control were NICD^–(Figure 3F). Furthermore, NICD+ cells were preferentially localized in the mid-crypt region when they are returning, in contrast to their normal localization at the crypt base. Occasional cells at the crypt base are NICD+ at day 3, in accordance with the timing of the return of Paneth cells and Olfm4 expression (Figure 3D, arrowheads).

 Rapid Expansion of Dll1- and Dll4-Expressing Cells during Crypt Regeneration

The rebounding Notch activity after acute Notch inhibition is associated with increased mRNA abundance of key Notch components, including ligands Dll1 and Dll4 (Pellegrinet et al., 2011), and receptors Notch1 and Notch2 (Carulli et al., 2015, Riccio et al., 2008) (Figure 4A). We made use of Dll1-mCherry and Dll4-mCherry transgenic reporter mice (Chakrabarti et al., 2018, Tikhonova et al., 2019) to follow the cellular pattern of Notch ligand-expressing cells during intestinal remodeling post-DBZ. Analysis of these reporter mice showed that Dll1- and Dll4-mCherry+ cells are normally present in both intestinal crypt and villus compartments in a scattered pattern consistent with secretory cell distribution (Figure 4B). We identified the cell types by co-imaging mCherry with markers of differentiated Paneth (lysozyme), goblet (mucin 2), and endocrine (chromogranin A) cells. This analysis showed that all three secretory cell types are marked with the mCherry reporter in both transgenic strains (Figure S2). This cellular expression pattern matches that previously characterized in Dll1-GFP-CreER^T2 mice, with GFP expression observed in differentiated secretory cell types as well as secretory progenitor cells (van Es et al., 2012). The cellular expression pattern was similar in DBZ-treated Dll1- and Dll4-mCherry mice, with the exception of the Paneth cell loss (Figure S2).

 

Numbers of Dll1- and Dll4-mCherry+ cells were markedly increased after Notch inhibition, accounting for increased mRNA abundance (Figure 4B). Notch ligand-expressing cell expansion corresponds to a surge in goblet cell numbers detected by PAS/Alcian blue staining (Figure S3). This finding is consistent with previous reports of secretory cell hyperplasia after several days of continuous Notch inhibition (

van Es et al., 2005

, 

VanDussen et al., 2012

). Dll1- and Dll4-mCherry+ cell expansion was profound, with expanded crypts at day 3 extensively composed of Notch ligand-expressing cells (Figure 4B). Quantification of Dll1-mCherry+ cells by flow cytometry showed that cell number increased 2.6-fold at 1 day post-DBZ (Figure 4C), with ligand-expressing cells primarily located at the crypt base (Figure 4B).

 Increased Progenitor Cell Function in Dll1-Expressing Cells after Acute Notch Inhibition

Interestingly, both Dll1-mCherry+ cells and Lgr5-expressing CBCs were localized to the crypt base 1 day post-DBZ (Figures 1 and 4), suggesting that CBCs might activate Dll1 expression in response to Notch inhibition. To examine this possibility, we treated Lgr5-GFP-CreER^T2;Dll1-mCherry mice with DBZ and co-imaged GFP and mCherry by histological and fluorescence-activated cell sorting (FACS) analysis. This work identified GFP+/mCherry+ co-expressing cells at the crypt base, suggesting that Lgr5 CBCs turn on Dll1 after Notch niche inhibition (Figures 5A and S4). Furthermore, we observed increased proliferation of Dll1+ cells 1 day post-DBZ, showing a 5-fold increase in EdU+/mCherry+ double-positive cells compared with Veh control (Figures 5B and 5C). We next tested whether Dll1-mCherry+ cells had increased stem/progenitor cell function by measuring organoid-forming potential of single Dll1-mCherry+ cells isolated by flow cytometry 1 day post-DBZ. This analysis showed a 2-fold increase in organoid-forming efficiency compared with Dll1-mCherry+ cells isolated from Veh-treated mice (Figure 5D). Together these findings suggest that Dll1+ cells exhibit enhanced progenitor cell status 1 day post-DBZ.

 Multiple Progenitor Cell Populations Generate Paneth Cells after Acute Notch Inhibition

We next studied whether FSCs are activated by Notch niche interruption to maintain the epithelium and replenish the lost Paneth cell population. In particular, we focused on Dll1-expressing secretory progenitors as this crypt cell population was previously demonstrated to regenerate CBCs after irradiation-induced crypt injury (

van Es et al., 2012

). We used a lineage-tracing approach in Dll1-GFP-CreER^T2;Tom mice (Figure 6A). Note that this analysis differs from our studies above in that Tom marks cells prior to DBZ treatment, thus allowing us to study pre-existing Dll1+ cells rather than the new cells that emerge after Notch inhibition. At 8 days after TX treatment, the only Tom+ cells in Veh controls were Paneth cells, due to their long half-life; all other Tom+ cells had turned over by this time (Figure 6B). In contrast, DBZ-treated crypts contained numerous Tom+ cells, including lineage stripes (Figure 6B, inset). Quantification showed that DBZ-treated mice commonly exhibited lineage stripes, while Veh-treated mice exhibited none (Figure 6C). We determined which cell types were generated by the activated Dll1+ cells by staining for markers of absorptive (enterocytes) and secretory (goblet, endocrine, Paneth) cells with Tom co-expression (Figure 6D). This analysis showed that DBZ-activated Dll1+ cells become multipotential progenitors, capable of generating all of the major epithelial cell types. Thus, they function to maintain the intestinal epithelium while CBCs are impaired.

Importantly, we found that the Tom+ cells generated from activated Dll1+ multipotential progenitors included Paneth cells (Figure 6D). Quantification of lysozyme/Tom double-positive cells showed increased numbers of lineage-labeled Paneth cells in DBZ-treated crypts compared with Veh control (Figure 6E). In light of the DBZ-induced Paneth cell loss (Figure 2), the observation of numerous Tom+ Paneth cells demonstrates that activated Dll1+ multipotential progenitors are replenishing the depleted Paneth cell pool.

The observation that Dll1+ cells contributed to differentiated cell lineages after acute Notch inhibition is consistent with FSC activity. However, lineage stripes from Dll1+ cells were short lived, with no lineage traces detected at 2 months post-DBZ (Figures 6C and 6F). Thus, these activated multipotential progenitors do not contribute to the CBC pool. This result is in agreement with our finding that CBCs are not lost after DBZ treatment (Figure 1). Further evidence that CBCs were maintained after DBZ was shown by the resumption of CBC lineage tracing in Lgr5-GFP-CreER^T2;Tom mice, with full crypt-villus Tom-labeling observed 7 days post-DBZ treatment (Figures 6G and 6H). Thus, CBC function was only transiently impaired after DBZ treatment. The data suggest that DBZ treatment activates Dll1+ cells to become multipotential progenitors to contribute to crypt regeneration after injury induced by Notch inhibition but not to displace resident CBCs.

Notably, we also observed Tom+ Paneth cells in Lgr5-GFP-CreER^T2;Tom mice (Figure 6H, inset, arrowhead). Together with the Dll1+ cell lineage-tracing data, this shows that multiple stem/progenitor cell populations can replenish the depleted Paneth cell pool. To test if other progenitor cell populations capable of FSC activity were activated after acute DBZ treatment, we carried out lineage-tracing studies in Hopx-CreER^T2;ROSA26-LSL-mTmG mice. In contrast to our result with Dll1+ cells, we did not observe enhanced lineage tracing from Hopx+ cells after DBZ treatment (Figure S5). This result suggests that the damage induced by Notch inhibition activates specific progenitor populations to contribute to the crypt repair process.

Discussion

Our findings differ in many respects from previous studies of Notch inhibition, which analyzed longer-lasting pharmacologic (Tian et al., 2015, van Es et al., 2005, VanDussen et al., 2012) or genetic (Carulli et al., 2015, Pellegrinet et al., 2011, Tsai et al., 2014) Notch depletion. In these earlier studies, long-lasting Notch inhibition led to CBC loss and a marked decrease in crypt proliferation, together with a surge of secretory cell types. Thus, our observation of Lgr5+ CBC stem cell retention and increased proliferation after short-term Notch disruption were unexpected. There is general agreement, however, that continuous Notch signaling is required to maintain CBC stem cells, with pathway interruption leading to rapid stem cell dysfunction and ultimately cell loss, depending on the duration of pathway inhibition.Our observation of rapid Paneth cell death was in stark contrast to the robust expansion of lysozyme-expressing cells seen in long-term Notch inhibition studies. However, these were not “true” Paneth cells as they exhibited altered secretory granule morphology and co-expressed goblet cell markers (VanDussen et al., 2012). Interestingly, Paneth cell loss was also observed in a distinct study of CBC niche disruption that blocked R-spondin signaling in CBCs using an adenoviral construct that targeted LGR5 (Yan et al., 2017). However, Paneth cell loss with this niche disruptor was considerably slower, involving a process of ballooning cellular degeneration and upward migration that took several days, instead of the rapid apoptotic cell death we observed with Notch niche inhibition. Together, our study and the Yan et al. study demonstrate that Paneth cells are very sensitive to CBC niche dysfunction, including both Notch and R-spondin pathway inhibition. The findings suggest that CBC niche disruption affects Paneth cell maintenance through an indirect mechanism as both studies targeted receptors specifically localized to CBCs.Paneth cells have been proposed to be Notch niche cells due to their close physical association with CBCs at the crypt base and their expression of Dll1 and Dll4, the key Notch ligands regulating CBC function (Pellegrinet et al., 2011, Sasaki et al., 2016, Sato et al., 2011). In addition, Paneth cells generate other niche factors, including Wnt ligands and growth factors, as well as metabolic products, suggesting a more expanded niche role for Paneth cells to support their stem cell neighbors (Rodriguez-Colman et al., 2017, Sato et al., 2011). However, niche function for Paneth cells has been controversial, with some studies showing that Paneth cells enhance stem cell function (Farin et al., 2012, Sato et al., 2011), while others demonstrate apparently normal CBC function after Paneth cell depletion (Durand et al., 2012, Garabedian et al., 1997, Kim et al., 2012, van Es et al., 2019). Our observation of reduced CBC lineage-tracing activity with Paneth cell loss, and a return to stem cell function with Paneth cell return, would support a functional niche role for Paneth cells. Notably, Notch signaling and Olfm4 expression at the crypt base did not recover until Paneth cells were restored at day 3, suggesting that Paneth cells are the key Notch niche cells presenting ligand to CBCs. Interestingly, we observed that Lgr5 CBCs activated Dll1expression upon Notch disruption, suggesting rapid stem cell remodeling. Whether this response is key to maintaining CBCs until Paneth cells return or is, perhaps, a marker of impending CBC differentiation will be important future questions to pursue.Crypt hyperproliferation and activation of multipotential progenitors are hallmarks of the intestinal regenerative response. The damage induced by Notch niche interruption is different from previously described crypt injury methods that induce CBC loss, including genetic, radiation, or chemotherapeutic drug treatment approaches. In our study, a robust regenerative response was observed although CBCs were retained. Our observations are in accordance with another study that observed modest increases in proliferation and crypt height after depletion of all secretory cells, including Paneth cells, from the adult mouse (Kim et al., 2012). Whether this response relates to Notch-signaling changes was not examined.We observed that acute Notch inhibition activates Dll1+ cells to function as multipotential progenitors to generate all types of differentiated epithelial cells while CBC function is diminished. The generation of enterocytes from Dll1+ cells is unusual as these progenitors are normally restricted to secretory cell types (van Es et al., 2012). However, the activated Dll1+ multipotential progenitors did not contribute to the CBC pool and thus did not function as FSCs. This contrasts with a previous study showing that Dll1+ secretory progenitors can regenerate CBCs lost after crypt damage caused by radiation (van Es et al., 2012). Thus, our study suggests that the two FSC functions of activation to maintain the epithelium and regenerate lost CBCs can be disconnected. Importantly, we observed that CBCs also contribute to restoration of the depleted Paneth cell pool. Thus, multiple progenitor cell populations can replenish these cells to repair the Notch niche. Our findings suggest that activation of epithelial cells to effect crypt repair is dependent on the specific cellular damage induced by the injury, underscoring the cellular plasticity and exquisite drive to regain homeostasis in the crypt.

To conclude, our studies highlight the exceptional cellular plasticity of the intestine and characterize the remodeling response to transient CBC niche inhibition. We propose a process whereby acute Notch inhibition stimulates a regenerative response stemming from rapid Paneth cell loss and impaired CBC activity, which is fueled in part by Dll1+ cell expansion and activation. Our study introduces acute Notch inhibition as a novel crypt disruption model which opens the door to studying mechanisms of CBC and Paneth cell responses during crypt repair.

Experimental Procedures

Author Contributions

A.J.C. and L.C.S. initially conceived the project, with subsequent conceptual contributions from N.B. and T.M.K. Methodology was developed by T.M.K., N.B., A.J.C., E.M.W., and L.C.S. The majority of the investigation and analysis was carried out by T.M.K. and N.B., with contributions from A.J.C., E.M.W., E.A.C., J.C.J., C.D.B., and P.J.D. Resources were provided by C.W.S., J.G., I.A., M.W.R., and M.G.M. N.B. and L.C.S. wrote the manuscript and all authors provided critical feedback. L.C.S. provided supervision and obtained funding.

Acknowledgments

We thank Y. Abushukur for technical help, E. Collin and L. Griffin for maintaining the mouse colonies, and E. Hibdon, K. McGowan, and N. Razumilava for critical comments on the manuscript. N.B. was supported by the Cellular and Molecular Biology Program , a Rackham research grant and the Benard L. Maas Fellowship. A.J.C. was supported by the Medical Scientist Training Program ( T32-GM07863 ), the Center for Organogenesis ( T32-HD007515 ), and a Ruth L Kirschstein NIH award ( F30-DK095517 ). The research was funded by NIH R01-DK118023 to L.C.S., and Core support from the University of Michigan Gastrointestinal Research Center NIH P30-DK34933 .