Reserve Stem Cells in Intestinal Homeostasis and Injury

Abstract

Renewal of the intestinal epithelium occurs approximately every week and requires a careful balance between cell proliferation and differentiation to maintain proper lineage ratios and support absorptive, secretory, and barrier functions. We review models used to study the mechanisms by which intestinal stem cells (ISCs) fuel the rapid turnover of the epithelium during homeostasis and might support epithelial regeneration after injury. In anatomically defined zones of the crypt stem cell niche, phenotypically distinct active and reserve ISC populations are believed to support homeostatic epithelial renewal and injury-induced regeneration, respectively. However, other cell types previously thought to be committed to differentiated states might also have ISC activity and participate in regeneration. Efforts are underway to reconcile the proposed relatively strict hierarchical relationships between reserve and active ISC pools and their differentiated progeny; findings from models provide evidence for phenotypic plasticity that is common among many if not all crypt-resident intestinal epithelial cells. We discuss the challenges to consensus on ISC nomenclature, technical considerations and limitations inherent to methodologies used to define reserve ISCs, and the need for standardized metrics to quantify and compare the relative contributions of different epithelial cell types to homeostatic turnover and post-injury regeneration. Increasing our understanding of the high-resolution genetic and epigenetic mechanisms that regulate reserve ISC function and cell plasticity will help refine these models and could affect approaches to promote tissue regeneration following intestinal injury.

Renewal of the intestinal epithelium occurs approximately weekly and requires a tightly controlled balance of proliferation and differentiation to maintain proper lineage ratios and support absorptive, secretory, and barrier functions. The intestinal epithelium has evolved in a luminal environment that exposes its cells to microbiota, naturally occurring toxins, and physical stresses that continually challenge epithelial integrity. In addition, man-made stressors such as contemporary chemotherapies and radiation treatments for cancer can have devastating consequences on epithelial integrity and renewal. In response to physiologic challenge or injury, intestinal stem cells (ISCs) repair damaged tissue and replace lost cells. ISC researchers investigate the specific intestinal cell types and mechanisms involved in the homeostatic renewal and injury-induced regeneration of the intestine.

Decades of research have generated competing models for how ISCs fuel homeostatic renewal and regeneration in the intestine. Based on 1 model, all ISCs are essentially similar in identity and provide a common functionally homogeneous pool of actively proliferating ISCs (aISCs), which continuously self-renew and give rise to all differentiated intestinal lineages. The rate of intestinal turnover is determined largely by parameters of ISC self renewal, cell differentiation, and death. In a separate model, aISCs mediate the normal homeostatic turnover of the intestinal epithelium, whereas a population of normally non- or slowly-proliferating reserve ISCs (rISCs) fulfill regenerative tasks specifically after tissue injury or extreme physiologic stress. In the most recent model, differentiated epithelial cell types have a large degree of phenotypic plasticity and can therefore re-acquire stem cell characteristics and function in response to injury. There is evidence to support of each of these models, indicating evolutionary selection for different pathways of regenerative responses in the intestine.

We review the evidence and concepts related to rISCs in intestinal homeostasis and injury and discuss the technical advances made and challenges to studying these. We introduce main concepts in ISC heterogeneity, regarding actively cycling vs quiescent ISCs, and explain the development of a hierarchical model, in which aISCs and rISCs are functionally separate ISC pools. We review rISC identity, function, and regulation because many intestinal cell types, including those previously thought to be committed to post-mitotic differentiated secretory cell (Paneth, enteroendocrine, tuft, and goblet) and absorptive enterocyte lineages, might have the capacity to participate in regeneration after tissue injury. We also review approaches for discriminating among cells that mediate the ISC regeneration response, and for identifying the signaling mechanisms that activate and direct regeneration. The complex nature of ISC dynamics is becoming clear—we discuss the difficulties in building consensus on ISC nomenclature, heterogeneity, and function. In this article, term reserve ISC (rISC) refers to cells that make no or a minimal contribution to homeostatic epithelial renewal, but mediate epithelial regeneration following damaging injury or stress, whereas active ISC (aISC) refers to cells that divide regularly during homeostatic epithelial renewal and are sensitive to and selectively compromised by damage or stress.

Identity, Function, and Regulation aISCs

Homeostatic renewal of the intestinal epithelial monolayer is mediated by a pool of ISCs located at the base of micro-anatomical units called crypts. These ISCs, originally classified as crypt base columnar cells (CBCs), are a highly proliferative population of cells intercalated between Paneth cells in the crypt base^1,2 (Figure 1). Barker et al found that CBCs express high levels of the leucine rich repeat containing G protein-coupled receptor 5 gene (LGR5). Genetic lineage tracing of Lgr5-expressing CBCs demonstrated the long-term self renewal and multipotency of these cells in mice³, thereby defining CBCs as a bona fide ISC pool. ISC function in single, cultured Lgr5-expressing cells was confirmed based on their ability to give rise to intestinal organoids, which are stem cell-containing epithelial units with the capacity for multi-lineage differentiation and long-term passage^4,5. Although these studies established the Lgr5-expressing CBCs as a predominant aISC population responsible for the homeostatic renewal of the entire intestinal epithelium, surprisingly, ablation of Lgr5-expressing cells, through diphtheria toxin-induced cell death, had little effect on homeostatic renewal of the epithelium⁶. These findings provided evidence for a pool of reserve ISCs that could contribute to epithelial renewal when the Lgr5-expressing ISC population was lost.

Figure 1. Role of crypt-localized cells in epithelial homeostasis and injury-induced regeneration.

A) Homeostatic renewal of the intestinal epithelium (comprising ISCs, transit-amplifying progenitors, absorptive enterocytes, Paneth, goblet, enteroendocrine, and tuft cells) is mediated by active ISCs (aISC) in the crypt base. A rare pool of ISCs located near the +4 position is maintained in a quiescent state (+4 ISC), making only minimal contributions to the homeostatic maintenance of aISC and renewal of differentiated lineages. B) During injury-induced regeneration, when aISCs are compromised or lost, rISCs become a primary cell source that mediates regeneration. Plastic non-ISC cell types can also contribute to regenerating epithelial tissue (blue arrows), presumably through the generation of an aISC intermediate. The +4 position is an approximation, referring to a region of the crypt from +4 to +7 nuclear positions, counted upward from the bottom-most nucleus in the crypt base.

ISCs fuel a complete turnover of the epithelium approximately every 4–7 days⁷. How do ISCs divide to achieve and maintain such high rates of turnover without becoming depleted? Studies are underway to determine the mechanisms by which self renewal and cell-lineage allocation are balanced under normal physiologic conditions. Lgr5-expressing ISCs can self renew by symmetric cell division to yield 2 equipotent stem cell progeny^7,8,9. This mode of self renewal generates monoclonal crypts, over time, according to neutral drift kinetics, indicating that symmetric divisions are prevalent in the Lgr5-expressing population^8,9,10. Asymmetric ISC divisions occur frequently in the Drosophila gut¹¹, but the extent to which asymmetric division is a primary mechanism of intestinal epithelial turnover in higher organisms is not well-established¹².

Lineage tracing of individual Lgr5-expressing ISCs have indicated that the decision to remain a stem cell or differentiate might be affected by competition for signals that are spatially patterned within the crypt niche-microenvironment^12,13. In this case, Lgr5-expressing ISCs positioned near the niche boundary (border cells) are more likely to be passively displaced out of the niche by newly born cells in the crypt base. These displaced ISCs ostensibly become transit-amplifying cells without a requirement for prior cell division. Conversely, ISCs at the bottom of the crypt receive signals that make them more likely to undergo self renewal via symmetric division¹³. It will be important to learn whether and how specific signals within the niche promote one process vs another—this information could help us build more accurate models for studying regulation of intestinal turnover and the pathogenic effects of defects in ISC regulation.

Although many signaling pathways are involved in regulating ISCs, we focus on Wnt signaling and how alterations in this pathway affect stem cell function and intestinal homeostasis. For more detailed review of other signaling pathways involved, see^{14,15,16,17,18,19,20}. The functions of aISCs are regulated by the canonical Wnt to beta-catenin signaling pathway. Although there has been much attention on Paneth cells as a predominant source of Wnt in the crypt, there are other cell sources for Wnt in the intestinal epithelium and parenchyma, indicating its importance in the ISC niche^{21,22,23,24,25,26}. Wnt ligands associate with frizzled class receptors and LDL receptor related protein 5 (LRP5) and LRP6 to induce axin-dependent stabilization and nuclear translocation of beta-catenin, which complexes with lymphoid enhancer binding factor (LEF/TCF) transcription factors to induce expression of genes regulated by Wnt signaling²⁷. Tissue-specific deletion of the transcription factor 7 like 2 gene (TCF7L2 or TCF4), which encodes a protein that activates Wnt-dependent gene expression, resulted in loss of aISCs and blocked intestinal crypt development in mice²⁸. Likewise, transgenic expression of dickkopf WNT signaling pathway inhibitor 1 (DKK1) resulted in progressive and pervasive loss of Lgr5-expressing ISCs and subsequent failure in homeostatic epithelial maintenance in mice^29,30,31. These loss of function studies established Wnt signaling as a cornerstone of ISC function.

Gain of function studies have provided more evidence for the importance of Wnt in ISC functions. R-spondins 1–4 (RSPO1–4) are secreted agonists of Wnt signaling that associate with and inhibit transmembrane E3 ligase receptors ZNRF3 and RNF43—these otherwise promote internalization and degradation of frizzled class receptors and LRP5 and LRP6^{32,33,34,35,36,37}. Transgenic overexpression of RSPO1 stimulates crypt cell proliferation and increased Lgr5-expressing cell numbers, eventually resulting in intestinal hypertrophy in mice^30,33. Conversely, combined disruption of the RSPO receptor genes (Lgr4 and Lgr5) is lethal; more crypts and ISCs are lost in double mutants (LGR4 and LGR5 double knockout) than in single mutants^35,38,39. Wnt signaling therefore balances aISC proliferation to promote normal homeostatic turnover of the intestinal epithelium, and is likely required to re-establish aISC function following cell loss due to disease, damage, or injury.

History on the conceptualization of rISCs

The concept that there is a pool of replicating cells that are distinct from aISCs/CBCs arose from experiments designed to locate and track dividing intestinal cell populations. In these, radiolabeled nucleotide analogues incorporated into DNA during S-phase are used to characterize differentiation, division, and migration kinetics of replicating cells⁴⁰. These analyses led to a model in which most replicating intestinal cells cycle every 24 hours⁴⁰, and cells migrating upward from the crypt originate from approximately 4 nucleus positions from the crypt base (the +4 position)¹²⁸. Some cells at or near +4, moreover, had a slowly cycling or quiescent phenotype, indicated by their ability to retain DNA label for extended time periods⁴¹. Together these early studies provided evidence for the existence of a long-lived, slowly cycling, label-retaining cell (LRC) population that resides at or near the +4 position.

Additional characterization of +4 LRCs suggested that at least some LRCs asymmetrically segregate their DNA strands during division—the old DNA strand is retained in the mother cell and new DNA strand moves to the daughter progeny, presumably to protect the DNA of the replicating mother stem cell from the accumulation of mutations acquired during S phase^42,43,44,45. In this immortal DNA strand model, +4 cells are a dormant and potentially damage-resistant cell population. There is little evidence, however, beyond the original studies to support asymmetric DNA segregation as a prevalent behavior of these cells. In fact, multi-isotope mass spectrometry evaluation of DNA label retention in ISCs indicated that ISCs dilute label in a manner consistent with random DNA strand segregation. The studies found no evidence for label retention in any non-Paneth crypt cells regardless of anatomic position⁴⁶. It wasn 2013⁶⁷, almost 5 decades after the +4 model was proposed, that there was functional evidence for ISC activity in +4 LRCs.

A genetic marker for +4 ISCs

In validating the +4 ISC model, much effort has been directed toward identifying +4 cell-specific biomarkers, to allow studies of cell-cycling, label-retention, and lineage-tracing behaviors. The first direct evidence for the existence of +4-resident stem cells was reported in 2008⁴⁷. BMI1 protooncogene, polycomb ring finger (BMI1) is a member of the polycomb repressor complex 1, which regulates chromatin silencing and has been implicated in regulating stem cell function in numerous systems, including the intestine^{47,48,49,50,51,52}. In tracing the Bmi1-IRES-CreER lineage at specific time points soon after tamoxifen administration, researchers found Bmi1-expressing cells predominantly at the +4 to +5 locations of many crypts in the proximal small intestine³⁰. Progeny of Bmi1-expressing cells traced for extended periods gave rise to differentiated lineages in the villus, demonstrating self renewal and multipotency of the Bmi1-expressing population. Notably, the kinetics of lineage tracing from Bmi1-expressing cells was significantly slower than Lgr5-expressing ISCs, suggesting that Bmi1 marks ISCs that proliferate and differentiate less frequently than the Lgr5-expressing ISC pool³. Direct comparison between Lgr5-marked and Bmi1-marked ISCs incorporating 5-ethynyl-2 -deoxyuridine (EdU) confirmed that Bmi1-expressing populations were essentially quiescent, and far less likely than Lgr5-expressing cells to enter S-phase or allocate differentiated cells under homeostatic conditions³⁰.

Single isolated Bmi1⁺ cells could generate organoids that were self renewing and possessed all post-mitotic cell lineages, further implicating ISC function in Bmi1-expressing cells³⁰. Bmi1-expressing ISCs, however, were relatively insensitive to canonical Wnt stimulation, compared with Lgr5-expressing ISCs, revealing an important phenotypic difference between proposed +4 ISC and aISC populations³⁰. Together, these studies provided evidence that ISCs with nonequivalent proliferation profiles coexisted in the intestinal crypt.

Conversion of ISCs between active and quiescent states

The initial report on BMI1 was followed by the discovery of additional biomarkers for slowly replicating stem cells located at or near +4. These markers include, but are not limited to, the HOP homeobox (HOPX), telomerase reverse transcriptase (TERT), leucine rich repeats and immunoglobulin like domains 1 (LRIG1), phosphatase and tensin homolog (PTEN), doublecortin like kinase 1 (DCLK1), protein phosphatase, and Mg2+/Mn2+ dependent 1D (PPMID or WIP1), as well as high-level expression of the transcription factors SRY-box 9 (SOX9) and mex-3 RNA binding family member A (MEX3A) (reviewed in^53,54). We focus on expression of Hopx to illustrate the concept that active and +4 quiescent ISCs can switch between phenotypes. Hopx is an atypical homeobox gene—its expression marks a population of +4 ISCs along essentially the entire intestine⁵⁵. Studies of Hopx lacZ knock-in mice revealed enrichment of Hopx-expressing cell numbers around +4 to +5, and an overlap in Hopx-reporter expression in LRCs, consistent with a quiescent +4 ISC phenotype. Notably, these studies used a combination of lineage tracing, gene-expression profiling, and laser capture microdissection to demonstrate that Hopx-expressing cells were capable of generating Lgr5-expressing CBCs, in addition to all the differentiated lineages of the intestine⁵⁵.

As with the Bmi1-expressing cells, however, a direct comparison of lineage-tracing kinetics revealed that the progeny of Hopx-expressing ISCs take longer to populate crypts and villi than Lgr5-expressing CBCs³. Isolated Hopx-expressing cells were also shown to be multipotent in culture. Importantly, actively dividing Lgr5-expressing cells could also give rise to new quiescent Hopx-expressing progeny in culture⁵⁵. These collective observations supported a model in which 2 populations of stem cells coexist in the crypt, each capable of converting between actively cycling and quiescent phenotypes.

Aside from their capacity for self-renewal and multipotency, ISCs have been extensively classified based on their proliferative behavior (actively cycling vs quiescent). The identification and use of biomarkers, lineage-tracing tools, and ex vivo ISC culture models have refined our understanding of ISC heterogeneity in terms of anatomical and numerical relationships, marker-identity, and stem cell function. In the latest model, quiescent +4 ISCs and aISC/CBCs can be thought of as 2 anatomically and functionally distinct ISC pools. How is homeostatic turnover of the intestinal epithelium, or its regeneration after damage, fueled by these respective populations? The ability of putative quiescent +4 ISC to convert to and from aISCs suggests that each population of ISCs can sense and adapt to fluctuations in physiologic demand for stem cell activity⁵⁶ (Figure 1B, Figure 2). Such fluctuations might occur during homeostatic conditions such as during periods of nutrient deprivation or during circadian cycles^63,130,131, or alternatively under conditions of extreme injury and stress. The concept that +4 ISCs are dedicated to the reserve function of replacing lost or compromised aISCs to support epithelial regeneration is of great interest.

Figure 2. Models for Conversion Between aISCs and rISCs.

A) Canonical Wnt signaling mediates self renewal of the aISC pool, which is the predominant source of all differentiated lineages. Conversion between aISCs and rISCs can occur, but is rarely observed under normal physiologic conditions. B) When aISCs are selectively lost after injury, rISCs become activated and convert to aISCs via unclear mechanisms. C) Epithelial repair is achieved as the aISC pool is replenished by rISCs, and regeneration of the full complement of epithelial cell types is completed. The rISC pool at this stage is presumably re-established, though this has not been formally tested. Note: transit-amplifying populations are not shown in this diagram.

Identity, Function, and Regulation of rISCs

The rISC phenotype is defined as a dedicated feature of cells that maintain stem cell capacity in a quiescent state under homeostatic conditions, and can re-acquire a proliferative aISC state when the resident aISC population is compromised. Although rISCs have been primarily associated with +4 ISCs, there is increasing evidence that many intestinal cell types, including progenitors of differentiated lineages and differentiated cell types themselves, possess rISC-like activity. It is therefore unclear whether rISC refers to a single and homogenous stem cell type. Does a larger collection of cell types exist, with varying degrees of ISC activity? If so, these cells might be used in regenerative therapeutic strategies for gastrointestinal disorders. A range of cell types have rISC-like activity, supporting a model in which crypt-based cells have phenotypic plasticity.

+4 ISCs

Experiments in which Lgr5-expressing aISCs (with a diphtheria toxin receptor gene knocked into the Lgr5 locus) could be selectively ablated by administration of diphtheria toxin provided evidence for the existence of a reserve ISC pool that does not express Lgr5⁶. These experiments showed that Lgr5-expressing ISCs, which existed before diphtheria toxin administration, were dispensable for maintenance of the intestinal epithelium, and that other epithelial cells could compensate for their loss—specifically Bmi1-expressing cells⁶. Further evidence for this concept came from lineage tracing of +4 quiescent ISCs after radiation-induced damage. In this situation, Lgr5-expressing ISCs were particularly susceptible to radiation exposure, and could be almost completely ablated by high doses of radiation^30,58,59. In contrast, Bmi1-expressing ISCs were not only radio-resistant, but lineage tracing output increased following exposure to radiation³⁰, indicating that the +4 quiescent ISCs were at least partly responsible for the regenerative response. In addition to cells that express Bmi1³⁰, cell populations that express Hopx^55,60,61, mouse Tert^62,63, and other genes^59,64,65,66 were found to have regenerative, ISC functions after epithelial damage, supporting a model in which quiescent +4 ISCs can survive and participate in regeneration after loss of aISCs. Notably, regeneration of Lgr5-expressing ISCs was essential for subsequent post-radiation epithelial repair¹³², consistent with a model in which rISCs repopulate Lgr5-expressing aISCs⁶. Altogether, these studies provide evidence for rare damage-resistant and quiescent +4 rISCs that can re-activate to replenish the aISC pool and facilitate regeneration.

Label-retaining secretory progenitors

In 2013, researchers confirmed that Lgr5-expressing populations could be subdivided into either actively cycling or quiescent label-retaining ISC fractions, based on an ability to retain an inducible H2B-YFP label⁶⁷. Lgr5-expressing H2B-YFP⁺ LRCs had patterns of gene expression consistent with secretory progenitors of the Paneth and enteroendocrine cell lineages, indicating that these cells are transient intermediates on the path to terminal differentiation. Single H2B-YFP LRCs, isolated by flow cytometry, could form organoids in culture and had stem cell-like functions in vitro. Using a split-Cre lineage tracing approach, these authors showed that LRCs were capable of extensive proliferation and multilineage differentiation after chemical injury in mice⁶⁷. The LRCs produced multi-lineage clones distributed along the crypt–villus axis, similar to conventional Lgr5-expressing aISCs. This study provided intriguing evidence that secretory precursor populations could be captured in an Lgr5-expressing and label-retaining state, and that these cells can recall a functional stem-like regenerative program in response to damage. It was not shown whether multiple subtypes of secretory progenitors were capable of this behavior, or whether the secretory progenitors investigated are a common pool for all secretory cell types.

A subsequent study compared transcription profiles and stem-cell functions between so-called short-term (10 days post label-induction) and long-term (up to 3 months) H2B-LRCs, with other ISC populations traced using Lgr5, Bmi1, and Hopx promoters to induce CreER expression. Short-term H2B-LRCs had a secretory progenitor-like gene signature and stem cell capacity⁶¹. These findings were consistent with those from a previous study showing that secretory precursors are plastic prior to terminal differentiation. Long-term LRCs were largely Paneth cells, which were not shown to have stem cell activity in culture or injury-induced regenerative capacity in this report⁶¹.

LRCs have been associated with quiescent ISCs, but tools were not always available to directly test their stem cell function. The studies we described were the first to use tracing techniques to determine whether LRCs had injury-induced stem cell potential. It is important to remember that a heterogeneous mixture of cell lineages is likely to be captured in the LRC assay, depending on the specific labeling parameters used as well as the replication, label-dilution, and cell-differentiation characteristics of various cell types under study. Studies are underway to identify genetic and epigenetic mechanisms that regulate the quiescent label-retaining rISC state—these studies could uncover the mechanisms of plasticity in LRCs.

Dll1+ secretory progenitors

The allocation of stem cell progeny toward the secretory lineages requires, at least in part, that Notch signaling activity is reduced or blocked^68,69,70,71. The delta-like canonical Notch ligand 1 gene (DLL1) encodes a transmembrane protein involved in Notch pathway activation and signaling. DLL1 is thought to be normally expressed by a subset of transit-amplifying secretory precursors that are the immediate progeny of aISCs^72,73,74. In studies of Dll1^GFP-ires-CreER knock-in mice and in situ hybridization experiments, Dll1^highCD24^mid populations were enriched for markers of progenitor cells and multiple secretory lineages⁷⁵. Lineage tracing showed that these cells generate only short-lived proliferative clones, which quickly differentiate into secretory cell types. These results support the concept that transit-amplifying secretory precursors are committed to differentiation, and are not fully multipotent for all intestinal lineages or normally capable of prolonged self-renewal.

Sorted Dll1^highCD24^mid cells, however, could form organoids that contain Lgr5-expressing ISCs following addition of exogenous WNT⁷⁵. Thus, ex vivo, Wnt-pathway stimulation induces stem cell activity in an otherwise committed Dll1^high population of secretory precursor cells. In mice, when lineage-tracing of Dll1^high cells was initiated (using a tamoxifen-inducible R26R^LacZ transgene) just prior to administration of sub-lethal doses of radiation, rare tracing events consistent with stem cell multipotency were observed, in addition to LacZ⁺ Lgr5-expressing cells. These results indicate that at least some secretory progenitors are plastic and can regain stem cell identity and function in response to damage. It was not determined whether Dll1^high cells generate slowly replicating or otherwise label-retaining cells, in addition to the Lgr5-expressing aISCs⁷⁵.

SOX9⁺ secretory progenitor cells

The Sry-box-containing proteins are a family of conserved transcription factors that function broadly during embryonic development and in adult stem cells. SOX9, a member of this family, has been shown via lineage tracing to mark adult and/or facultative stem cells in the intestine, colon, pancreas and liver^76,77,78. Cell populations that express high levels of a Sox9^EGFP reporter transgene (Sox9^EGFP-HI), contain putative label-retaining ISCs as well as secretory progenitors, differentiated enteroendocrine cells, and tuft cells^79,80. Single cell transcriptomic profiling of Sox9^EGFP-HI cells revealed that a small population of cells from crypt-enriched epithelial fractions have a secretory progenitor signature similar to those reported to participate in regeneration after injury⁶⁵. There is evidence that SOX9 might have direct roles in regulating ISC behaviors in response to stimuli such as nutrient-dependent signaling and damage^{63,81,82,83,84}.

Isolated Sox9^EGFP-HI cells rarely proliferate or form organoids in vitro, but have increased organoid formation after stimulation with insulin like growth factor (IGF) and radiation exposure⁸⁴. IGF also stimulates DNA synthesis in Sox9^EGFP-HI cells and facilitates intestinal regeneration in mice exposed to radiation. Therefore, it appears that Sox9^EGFP-HI cells can respond to stimuli by activating the cell cycle and participating in regenerative processes⁸⁴. No lineage tracing experiments have been performed to confirm this interpretation, due to the lack of a specific marker for SOX9^HI cells.

Perhaps more importantly, SOX9 protein is required for intestinal regeneration after radiation injury in mice⁶⁵. Because the SOX9-deficient epithelium is defective in generation of LRCs, SOX9 might preserve radio-resistance in ISCs, at least in part, by promoting a slower cycling state, and thereby increasing protection of the genome. It is also possible that specific functions for SOX9 during Paneth cell differentiation could be required to generate mature Paneth cells that are support regenerative processes in the intestine^85,86. Studies are needed to determine the roles for SOX9 in regulating the ISC cell cycle, in conferring ISC radio-resistance, and in regulating Paneth cell differentiation from ISCs, to define the mechanisms through which SOX9 regulates intestinal homeostasis and regeneration.

Paneth cells

Differentiated Paneth cells might also provide a source for new intestinal tissue after damage⁸⁷. Researchers used villin-rtTA/TRE-H2B-GFP mice, with doxycycline dosing and withdrawal strategies, to label long-lived LRCs in the crypt base. These LRCs, which persisted in the crypt for up to 100 days, were shown to be differentiated Paneth cells. Although these label-retaining Paneth cells expressed low levels of the stem cell markers Lgr5, Bmi1, Tert, and Lrig1, they did express high levels of the stem cell marker musashi RNA binding protein 1 (MSI1)^87,88. Label-retaining Paneth cells in mice could re-enter the cell cycle following radiation damage, and the numbers of H2B-GFP label-positive cells increased over 3–5 days. The intensity of GFP also decreased in these cells over time, presumably due to replication-associated dilution of H2B-GFP signal. These findings indicate active regeneration from the Paneth cell population. Eventually, the label-retaining Paneth cells lost Paneth cell markers and began to express Bmi1 during regeneration⁸⁷. Therefore, Paneth cells and/or their immediate precursors might reacquire stem cell function after radiation exposure to participate directly in epithelial regeneration. Further studies are needed.

Enteroendocrine cells

Post-mitotic enteroendocrine cells have also been reported to have injury-induced stem cell activity^84,89—specifically enteroendocrine subsets marked by neurogenin3-Cre⁹⁰ or Nkx2.2-Cre⁹¹. Using a Bmi1-GFP knock-in allele to allow Bmi1-expressing cell isolation without relying on Bmi1-CreER-mediated lineage tracing^30,47, researchers showed that single GFP⁺ intestinal cells could undergo clonogenic organoid formation indicative of stem cell function⁹³. GFP⁺ cells also had a gene expression signature associated with enteroendocrine cells, indicating that quiescent and regeneration-competent Bmi1-expressing subpopulations might be relatively differentiated enteroendocrine cells^92,93. Although the formal equivalence of the Bmi1-GFP⁺ knock-in and Bmi1-CreER⁺ alleles, in terms of cell specificity, has not been demonstrated, analyses of transcriptomes of Bmi1-CreER⁺ cells revealed expression of enteroendocrine cell-associated genes, although this was accompanied by Paneth cell gene expression patterns not usually associated with Bmi1-GFP⁺ cells. This might be due to imperfect overlap between true Bmi1⁺ ISCs, Bmi1-GFP⁺ cells, and the cells that become labeled by Bmi1-CreER⁺ after exposure to tamoxifen.

Bmi1-GFP⁺ cells also expressed the homeobox transcription factor prospero homeobox 1 (PROX1)⁹³, which is expressed by differentiated enteroendocrine cells⁹⁴. In Drosophila, the Prox1 ortholog, prospero, is required for generation of midgut intestinal enteroendocrine cells^95,96,97. In intestines of mice, Prox1-expressing cells exhibit clonogenic growth into organoids and radiation-inducible lineage tracing similar to Bmi1-GFP⁺ cells⁹³. Bulk and single-cell RNA sequencing revealed that Bmi1-GFP⁺ cells were predominantly lineage-committed Neurod1⁺Neurog3⁻ enteroendocrine cells, whereas Prox1-GFP⁺ cells comprise cells with a Bmi1-like mature enteroendocrine phenotype, but also contain a subset that expresses tuft-cell associated genes with low levels of enteroendocrine gene expression. Separately, it was shown that the differentiated chromatin and enhancer occupancy profile of Bmi1-GFP⁺ cells and of CD69⁺CD274⁺ goblet cells reverts to a distinct Lgr5-expressing ISC-like pattern upon radiation exposure⁹². The post-mitotic and differentiated state of at least some enteroendocrine cells is therefore not hard wired, and dynamic chromatin changes accompany the plastic conversion of these cells toward a more ISC-like condition during regeneration.

Absorptive Enterocytes

Damage to Lgr5-expressing ISCs induces crypt regeneration—not only from ISCs and secretory progenitor/precursor populations, as described above, but also from the progenitors of absorptive enterocytes⁹⁸. Alkaline phosphatase, intestinal (ALPI) has been widely used as a marker for enterocytes and their progenitors^98,99. Upon diphtheria toxin receptor-mediated ablation of Lgr5-expressing ISCs, Alpi-expressing cells de-differentiate into proliferative Lgr5-expressing ISC and Paneth-like cells in mice⁹⁸. During their conversion, Alpi-expressing cells upregulate stem cell-specific genes and downregulate enterocyte-specific genes. Because absorptive enterocytes and their progenitors constitute a majority of the intestinal epithelium, the concept that they can serve as a reservoir for new ISCs after damage is potentially of high significance.

Relevance of Reserve Stemness and Cell Plasticity to Regeneration

With a body of evidence showing that various intestinal cell types can participate in epithelial regeneration, questions are being pursued as to which cell-type or types represent the primary sources for generating new stem cells and differentiated tissue. On the one hand, it is possible that requisite populations of dedicated rISCs are maintained in the niche so that they can become activated in response to damage and fulfill regenerative tasks. On the other hand, it is also possible that regeneration could be accomplished by a concerted effort among many if not all epithelial cell types, ostensibly through the injury-induced activation of a plasticity program that is shared across multiple stem and differentiated cell types. Data indicate that elements of both processes occur to at least some extent (Table 1).

Table 1.

Contribution of regenerative populations to epithelial homeostasis and repair

Contribution to epithelial lineages
Biomarker	Model		Homeostasis	Regeneration	Injury Type	Citation
LGR5	Lgr5EGFP-IRES-CreER; Rosa26lsl.tdTom or Rosa26lsl.YFP	Tracing Period	7 Days	4–7 Days	12 Gy Whole body IR	30
		Trace frequency	~95% C/V Units	Eradicated
		Trace Type	Ribbons	-------------
BMI1	Bmi1CreER; Rosa26lsl.YFP	Tracing Period	7 Days	7 Days	12 Gy Whole body IR	30
		Trace frequency	~18% C/V Units	‘more extensive’
		Trace Type	Ribbons	Ribbons
BMI1	Bmi1CreER;Rosa26R	Tracing Period	6 Days	6 Days	Diptheria Toxin ablation of Lgr5* ISCs	6
		Trace frequency	~3% fully labeled crypts	~15-fold increase
		Trace Type	Labeled crypts	Labeled Crypts
HOPX	HopxCreER; Rosa26lsl.tdTom or Rosa26R	Tracing Period	2 Months	4 Days	12 Gy Whole body IR	55,60,61
		Trace frequency	Rare	Broad contribution
		Trace Type	Ribbons	Labeled cells in all C/V units
PROX1	Prox1CreER; Rosa26lsl.tdTom	Tracing Period	2–395 Days	7 Days	12 Gy Whole body IR	93
		Trace frequency	~5% C/V Units	“Similar to Bm/130”
		Trace Type	Ribbons by 395 days	Ribbons and large cell-clusters
ALPI	AlpiCreER;Rosa26R	Tracing Period	14 Days	14 Days	Diptheria Toxin ablation of Lgr5* ISCs	99
		Trace frequency	None	500–900 per mouse
		Trace Type	None in crypts	Ribbons
DLL1	Dll1GFP-IRES-CreER; Rosa26R	Tracing Period	28 Days	14 Days	6 Gy Whole body IR	75
		Trace frequency	Frequently Paneth Cells	~100 crypts per duodenum
		Trace Type	Paneth cells only	Ribbons
MEX3A	Mex3atdTOM-P2A-CreER; Rosa26mTmG	Tracing Period	7–14 Days	7–14 Days	5-fluorouracil	120
		Trace frequency	Not reported	Not reported
		Trace Type	Varies, eventual ribbons	More rapid clone and ribbon formation

Note: Tabulation of reported lineage tracing behaviors from intestinal epithelial cell populations proposed to contribute to homeostasis and/or regeneration after injury or cell ablation interventions. Values are approximations in some cases. C/V units refers to “crypt-villus” units, which are ribbon-like stripes of lineage labeled cells extending from individual crypt-bases upward toward the tips of villi.

Lineage-tracing of Bmi1-CreER^6,30 and Hopx-CreER^55,60,61 cells after injury with high-dose radiation (12 Gy) showed that regeneration from these purported rISC populations occurs frequently and broadly within the intestine. Moreover, knockout of MSI1 from Hopx-expressing cells perturbs intestinal regeneration after injury in mice⁸⁸. These findings show that Bmi1 and Hopx are markers of cells that mediate epithelial tissue formation during regeneration. In contrast, secretory progenitor cells marked by Dll1-CreER⁷⁵, Prox1-CreER⁹³, or H2B-label⁶⁷ have many fewer tracing events after injury. However, a caveat to comparison of these distinct putative rISCs is the heterogeneous nature of different ROSA reporter alleles used in the various studies (Table 1). Moreover, studies in which Lgr5-expressing cells were ablated and regenerative Alpi1-expressing enterocytes were traced indicated that Alpi1-expressing cells only rarely acquire functional stem cell characteristics after injury⁹⁸. A recent report suggested that a slowly cycling, multipotent subset of Lgr5-expressing ISCs, located nearer the crypt base (at and below +4) and expressing high levels of Mex3a, were resistant to radiation and the chemotherapeutic agent 5-fluorouracil—these cells converted to fast-cycling ISCs and participated in regeneration after chemotherapy-induced injury in mice (Ref¹¹⁹) Therefore, subsets of Lgr5-expressing cells (presumably a subpopulation of aISCs) can participate in epithelial regeneration following injury. Development of rigorous metrics for regeneration, used in combination with focused comparative analyses based on markers of intestinal cell sub-populations and different types or severity of damage, will provide more information on the frequency with which individual cell types acquire stem cell potential in response to injury.

Technical Approaches and Limitations

Until recently, the identification of aISCs and rISCs, has been based primarily on the presence or absence of single or small numbers of biomarkers (such as Lgr5, Bmi1, etc.). A more complex picture of intestinal epithelial cell heterogeneity, however, is now developing based on global transcriptomic analyses at single cell resolution. Cells expressing biomarkers once proposed to be restricted only to aISCs were found to also be expressed by quiescent and putative rISCs and differentiated cell populations (Figure 3). Biomarkers such as Bmi1 and Hopx, originally proposed to mark +4 quiescent ISCs, were found to be broadly expressed within the crypt. Bmi1 and Hopx mRNAs are expressed in most, if not all CBCs, as well as in transit-amplifying populations^{71,100,101,102,103,129}.

Figure 3. Expression patterns of biomarkers commonly associated with rISCs.

Originally reported biomarker specificity is indicated in dark blue. Marker expression patterns reported afterward are indicated in light blue. It is not clear whether and how gene expression patterns are altered, especially within short time periods, during injury (*).

Moreover, there are questions about the reproducibility of lineage-tracing initiation from the +4 position using Bmi1-CreER lines^46,71. Other proposed markers for ISC subpopulations have met similar criticisms, indicating the limitations of reliance on biomarkers and CreER-based lineage tracing to define specific cells or populations^104,105,106. For example, the mouse Tert gene was proposed, based on analyses of mTert-EGFP and mTert-CreER mouse lines, to mark a rare subpopulation of multipotent +4 ISCs distinct from Hopx-expressing cells (but with overlapping expression of Bmi1)^62,107. Tert mRNA, however, was found throughout the crypt—even at relatively high levels in Lgr5-expressing CBCs^100,108. This finding is perhaps not surprising, given that telomerase protects the genomes of replicating cells¹⁰⁹.

Other proposed +4 markers such as PTEN, DCLK1, and WIP1 were identified based on their expression by rare populations of +4-localized cells with label-retaining characteristics^110,111,112. Subsequent validation of these markers by antibody labeling or gene-expression analyses, however, indicated less specificity in their +4 location than previously reported. In the case of PTEN, antibody labeling studies found that PTEN marks, in addition to +4 localized LRCs, cells expressing enteroendocrine markers in the crypt base¹¹³. DCLK1 was originally found to be highly expressed in micro-dissected samples of fixed tissue sections, with a protein localization pattern consistent with rare +4 LRC, but was later found to mark rare populations of tuft cells^{53,114,115,116}. This result was confirmed by lineage tracing using a CreER knock-in line regulated by Dclk1 promoter sequences¹¹⁷. Although knock-in studies have identified cell populations that are induced in response to injury, these and other cases indicate a need for caution in relying on biomarkers to define subpopulations of regenerative cells within the intestine biomarkers alone should not be used to define subpopulations of cells within the intestine.

ISC researchers have used flow cytometry and fluorescence-activated cell sorting methods to study stem cell identity and function. Cell sorting-based enrichment of cell populations, using fluorescent reporter lines such as Lgr5^EGFP, Bmi1^EGFP, Hopx^EGFP, and Sox9^EGFP cells, is common practice. These techniques have been critical for validating cell identity, function, and heterogeneity^53,118,120. Several protocols use fluorescent reporter lines in conjunction with cell-surface antibody labeling to identify and isolate specific populations of cells. Although few if any current primary antibodies are specific for single subpopulations of ISCs, markers such as CD24¹²², CD44¹²³, GRP78¹²¹ and CD166¹²⁴ have been used to identify crypt-localized stem cell populations in mouse, and in some cases human, tissues^118,122,123.

Other, non-fluorescence-based sorting methods, such as side population (SP) sorting, have also been used to isolate and characterize various stem cell populations. Briefly, side populations are sorted based on the presence of efflux transporters on the plasma membrane of stem cells, to isolate stem cells from bone marrow and many solid tissues¹²⁴. SP sorting has been used to isolate crypt base ISCs^125,126, and more recently to segment populations of ISCs into cycling a non-cycling or slowly-cycling states¹²⁷. As the repertoire of genetic and antibody markers for specific subsets of stem and differentiated cell types grows, our understanding of the potential multitude of mechanisms underlying intestinal renewal and regeneration will improve.

Future Directions

The seeming diversity of cell populations in the intestine that have the demonstrated capacity to participate in regeneration has raised numerous questions regarding the dedicated functions of specific cell types in the contexts of homeostasis and injury. It is important to establish whether dedicated active and reserve ISC populations exist and function in homeostatic renewal and regeneration after injury¹³², or whether it is more accurate to consider reserve stemness more broadly—as cellular plasticity that is a shared among many cell types in the intestinal epithelium. Multiple mechanisms appear to have evolved to maintain and quickly repair the epithelial lining after damage. It is therefore not surprising that multiple cell types in the damaged epithelium contribute to repair. However, just because 1 cell type can participate in regeneration, it does not necessarily mean that this cell type is dominant in or required for reestablishment of the epithelial lining.

To the extent possible, it will be necessary to rigorously establish the relative and essential contributions of distinct cell populations to the regenerative response, and especially to determine whether similar or different molecular mechanisms are involved in each. Moreover, focused investigations into stem cell heterogeneity and the existence of rISCs in colonic crypts are needed to further studies of cancer stem cells and tumor initiation, progression, and response to treatment. It will also be important to identify the signals and niche cell types that regulate conversion between rISCs and aISCs—these might vary during homeostasis and repair. Such investigations will continue to refine prevailing models, challenge overly simplistic definitions of reserve and active stem cell states, and inform strategies for regenerative medicine.