Cell plasticity in regeneration in the stomach and beyond

Abstract

Recent studies using cell lineage-tracing techniques, organoids, and single-cell RNA sequencing (sc-RNA-Seq) analyses have revealed: 1) adult organs use cell plasticity programs to recruit progenitor cells to regenerate tissues after injury, and 2) plasticity is far more common than previously thought, even in homeostasis. Here, we focus on the complex interplay of normal stem cell differentiation and plasticity in homeostasis and after injury, using the gastric epithelium as a touchstone. We also examine common features of regenerative programs and discuss the evolutionarily conserved, stepwise process of paligenosis which reprograms mature cells into progenitors that can repair damaged tissue. Finally, we discuss how conserved plasticity programs may help us better understand pathological processes like metaplasia.

INTRODUCTION

Most adult organs lack dedicated, multipotent stem cells to replace lost cells after injury [1]. Some tissues such as the epithelial lining of the alimentary tract do have such stem cells to replace homeostatic cell loss, but certain injuries (e.g., chemotherapeutic agents and radiation) specifically kill those stem cells. In both tissue scenarios, regeneration of lost cells depends on cell plasticity: namely the repurposing of differentiated cells as regenerative progenitor cells. From a human health perspective, cell plasticity becomes important because chronic injury and inflammation can lead to a chronic pattern of aberrant differentiation where the tissue does not return to its homeostatic state. Such lesions, composed of normal (i.e., non-dysplastic and non-neoplastic) cells that are simply inappropriate for the tissue they are found in are called metaplasias. As metaplasia can increase risk for progression to cancer, understanding cell plasticity programs may help us better understand tumorigenesis [2–7].

Regenerative cell plasticity is universal across organs and among multicellular species; hence, it has been proposed to be executed by a conserved program of cellular-molecular mechanisms. One such program, paligenosis, depends on a dynamic, stepwise modulation of the cellular energy hub mTORC1 that ultimately reprograms differentiated cells into regenerative progenitors that express metaplastic, wound-healing, embryonic genes [8].

Even in normal homeostatic renewal of cells from constitutively active adult tissue stem cells, there may be far more cell plasticity than previously believed, with more committed progenitors capable of reversion to a multipotent state or prolonged existence as self-renewing, unipotent progenitors [9–11].

To illustrate both homeostatic and injury-responsive plasticity programs, we will focus on the gastric epithelium as a model. The stomach harbors both constitutively active progenitor populations and differentiated cell populations that can be recruited as stem cells (especially after injury and in metaplasia). Plasticity, cell identity, and stem cell behavior have been heavily investigated recently with new single-cell multi-omic technologies (chiefly single-cell RNA-Sequencing; sc-RNA-Seq) as well as various cell lineage tracing strategies. Here, we review related, recent findings on gastric epithelial stem cell differentiation and fate choice in development, homeostasis and during repair. We highlight examples that illustrate manifestations and aspects of cell plasticity.

The Stomach as a Model for Studying Constitutive and Recruited Stem Cells

The stomach has certain features that are useful for understanding homeostatic plasticity and differentiation as well as injury-induced plasticity (Fig. 1). Its epithelium is organized into repeating units characterized by clear zones of cell identity, with normal, constitutively active, multipotent stem cells well separated spatially from differentiated cells that can be recruited as stem cells after injury [12,13]. In the base of the glands, stem cell recruitment occurs via paligenosis, which can be induced in mice in a synchronous, orderly fashion [8,14]. The resulting metaplasia faithfully models the process that occurs in humans en route to gastric cancer. Such cancers (mostly adenocarcinomas) are among the leading causes of cancer death worldwide and occur via a relatively well documented stepwise progression starting with injury-induced plasticity and potentially becoming dysplastic and ultimately neoplastic [15–17].

Figure 1. The corpus and antrum gastric units comprise distinct cell lineages.

The antrum has three zones, including the pit zone containing foveolar cells (purple), the proliferative isthmus, and the base containing deep mucous cells (lime). Axin2+ (with some having LGR5 expression) in the isthmus mark the active stem cell compartment (light gray) [19,23,24,29]. The corpus gland includes a pit zone with mucus-secreting foveolar cells (purple), an isthmus with active stem cells (light gray) marked by Stathmin1 (STMN1), and IQ Motif Containing GTPase Activating Protein 3 (IQGAP3), a neck zone lined with mucous neck cells (green) and parietal cells (blue), and a base compartment with digestive enzyme-secreting chief cells (red). Rare endocrine cells (brown) are present in both glands with antral endocrine cells secreting factors, such as gastrin (GAST) in the antrum and histamine (generated via the HDC enzyme) in corpus [12,19–24,99].

In humans and mice, there are two glandular compartments of the stomach [18]. The proximal corpus primarily functions in early food digestion, containing epithelial cells that secrete acid or digestive enzymes into the lumen, while the glands of the antrum secrete primarily mucus into the lumen. Exocrine cell lineages (i.e., those secreting into the lumen) within the corpus glands include the mucous pit/foveolar cells, isthmal stem cells, acid-secreting parietal cells, mucous neck cells, and digestive enzyme-secreting chief (zymogenic) cells. Each corpus gastric unit has three distinct morphological zones progressing from gastric lumen to underlying muscularis: pit, neck, and base. In homeostasis, proliferative cells are largely confined to the narrowing (isthmus) between the pit and the neck. Cell lineage turnover times range from 3 days for pit cells, to nearly 2 months for parietal cells, and over 9 months for chief cells in wild type mice [14,19–22]. Antral glands harbor pit cells and gland base mucous cells; parietal and chief cells are rare in humans and absent in mice. The antrum also has an isthmus harboring actively proliferating cells at the intersection of pit and base (Fig. 1) [19,23,24]. Other cells such as hormone-secreting endocrine and inflammation-regulating tuft cells also inhabit both corpus and antrum.

In this short review, we will focus on plasticity of exocrine cells. Though plasticity is of course not exclusive to those cells, there are several advantages to using them to illustrate plasticity. They are usually long-lived, large cells with complex subcellular architecture (abundant rough endoplasmic reticulum and large secretory granules); thus, their reprogramming into much smaller, architecturally simpler progenitor cells is dramatic. They are also often found in discrete locations histologically so their behavior can be observed in situ.

Cell Fate Decisions and Identity Specification in Adult Gastric Epithelium

For context in understanding cell differentiation in the gastric epithelium with a focus on roles for cell plasticity, we begin by reviewing recent studies on key signaling pathways using organoids and single-cell multi-omic techniques. In both mouse and human systems, WNT, BMP, and Notch signaling have long been known to play roles in differentiation and maintenance of the gastric epithelium with new sc-RNA-Seq studies providing additional evidence and granularity for the importance of these pathways [25–28]. Sc-RNA-Seq of adult mouse antrum has shown with greater granularity that specific stromal cells at the base of antral glands increase WNT activity in basal epithelial cells via secretion of RSPO3; basal stromal cells also release BMP inhibitors like CHRDL1 and GREM1. Accordingly, gastric antral organoids treated with BMP4 displayed decreased proliferation and expression of stem cell markers. Furthermore, loss of BMP signaling specifically in Axin2+ cells led to hyperplasia [29]. This was consistent with earlier studies of deletion of the receptor for BMP2/4/7 (BMPR1a) in the mouse stomach, causing antral hyperplasia and tumors [30–32]. Notch signaling in the antrum promotes LGR5+ cell proliferation and represses differentiation, as demonstrated using organoids derived from both mouse and human antral tissue [33]. Specifically in the mouse antrum, basal cells expressing the Notch ligand Delta-like-1 (DLL1) can function as niche cells that promote proliferation in the adjacent LGR5+ basal cells [34].

In the stomach only a handful of the transcription factors that specify each cell lineage from its progenitors have been identified, most of which have been explored/validated in the mouse system. In the antrum, SPDEF is known to be essential for mucous cell maturation [35]. In the corpus, the transcription factor FOXQ1 promotes terminal differentiation of pit cells [36]. Transcription factors XBP1 and MIST1 are not completely necessary for chief cell differentiation but are necessary for maturation of their secretory system architecture [32,37]. HNF4α can act upstream of XBP1 to induce its expression in chief cells, but it also suppresses proliferation of progenitors [38]. The transcription factor Kruppel-like factor 4 (KLF4) likewise suppresses gastric stem cell proliferation, and its loss increases neck cell lineage while decreasing pit and parietal cell lineage within the gland [39,40].

It has been proposed that the large environmental and energetic variations present in the stomach result in more dynamic, reversible – i.e., plastic – cell fate decisions, which can respond rapidly to metabolic queues and not just traditional developmental signaling pathways and downstream transcription factors. For example, activation (e.g., by treating mice with metformin) of the cellular energy hub AMP Kinase (AMPK) increases parietal cell census via effects on parietal cell specification from progenitors as well as parietal cell maturation. AMPK appears to promote pit or parietal cell differentiation by activating KLF4 and further specifying parietal cells by inducing the mitochondria biogenesis master regulator PGC1α [41]. Metabolic promotion of parietal cell fate is corroborated by the loss of CD36, an AMPK target that mediates fatty acid uptake/scavenging, which decreases parietal cell regeneration after injury [42•,43].

Cell Plasticity and Gastric Epithelial Differentiation Following Injury

Most gastric adenocarcinomas arise in a setting of chronic inflammation (usually caused by infection with H. pylori) associated with loss of parietal cells from corpus gastric units. Along with parietal cell loss, the entire corpus gastric unit undergoes metaplastic change that has been termed “antralization” (Fig. 2) [7]. Most notably, chief cells undergo paligenosis to an embryonic-like progenitor state that is proliferative and expresses mucins (MUC6) and wound-healing proteins like Trefoil Factor 2 (TFF2, aka Spasmolytic Polypeptide), and Gastrokine 3 (GKN3) with diminished expression of chief cell markers such as the digestive enzyme pepsinogen C (PGC). These cells, often called Spasmolytic Polypeptide Expressing Metaplasia (SPEM) are similar to the normal deep antral (pyloric) mucous cells described above. Thus, the entire process with parietal cell loss and SPEM cells is known as pyloric or pseudopyloric metaplasia (Fig. 2) [1,7].

Figure 2. Mature cells across multiple organs dedifferentiate and various metaplastic populations arise following injury.

Gastric corpus glands lose parietal cells and exhibit increased proliferation in the isthmus and base following injury. Chief cells (red) downscale their mature secretory structures while SPEM cells (yellow) in the lower neck and base co-express GIF and TFF2 [8,53•,54••]. In ADM metaplastic ducts, mature acinar cells (red) express SOX4 and become proliferative. Tuft cells (tan) also arise from plastic acinar cells. In both stomach and pancreas, metaplastic cells (yellow) express SOX9 and CD44 [4,8,56••]. For sessile serrated lesions, a neoplastic MUC5AC+ population of serrated cells develop in upper region of the crypt (purple) and express embryonic gene MDK. Metaplastic cells (yellow) arise near the crypt base [89••]. In all three organs, mature cells (red) reprogram to a more fetal-like state (decreasing mature cell markers such as MIST1, CDX2, and ATOH), proliferating and expressing more mucins (expression of MUC6 and MUC5AC) as well as SPEM markers gastrokine3 (GKN3) and Aquaporin 5 (AQP5), whose function is still uncertain [8,53•,54••,56••,89••].

The cell plasticity of chief cells and their paligenosis into SPEM cells has been demonstrated primarily in mouse models using lineage tracing from multiple promoter-reporter combinations, some of which are expressed predominantly in chief cells such as MIST1, TROY (also expressed in parietal cells), LGR5, and GIF [44–48]. Plasticity has also been demonstrated using nucleotide labeling and pulse-chase tracing, and with promoters such as Runx1 eR1 and Iqgap3 that homeostatically express in isthmal, constitutively active progenitor cells and that are induced during paligenosis in chief cells [12,14,49]. More recently in both mouse and human tissue, Aquaporin5 (AQP5) which marks multipotent, self-renewing progenitor cells in the antrum is also shown to be activated in chief cells following gastric injury, further illustrating the plasticity of chief cells [50•,51].

Pyloric metaplasia, in particular SPEM, has been studied recently using sc-RNA-Seq and bulk transcriptomics in various mouse injury models [52••]. One emerging theme is that metaplastic cells are transcriptionally similar whether induced by chronic inflammation-inducing models that may more faithfully model the human situation (e.g., mouse-adapted pathogenic human H. pylori strains or transgenically-induced autoimmune gastritis) or in the acute, synchronous injury caused by High-Dose Tamoxifen [53•, 54••, 55•]. Sc-RNA-Seq has additionally clarified and defined subsets of metaplastic cells and new markers to help distinguish among them in both mouse and human metaplasia. The canonical immunohistological definition of a SPEM cell is co-expression of TFF2 (or labeling by the lectin GS-II) with co-expression of chief cell markers – at significantly decreased abundance than in mature chief cells – of GIF (applicable only to mice) or PGC (mice and humans). However, transcriptomics, in particular sc-RNA-Seq, have identified additional more SPEM-specific markers like AQP5, GKN3, and WFDC2 [50•,54••,56••,57•]. GKN3, for example, was originally identified as strongly expressed in the antrum with minimal expression in corpus mucous neck cells [58,59]. SPEM cells in sc-RNA-Seq express high levels of GKN3. Sc-RNA-Seq also reveals that mucous neck cells change during pyloric metaplasia, undergoing not just expansion (i.e., hyperplasia), but induced expression of GIF, as well as increased TFF2, MUC6 and, depending on the injury method, the induction of immunoregulatory genes as well [53•,54••]. Pseudotime analyses have shown a trajectory axis where mature chief cells reprogram into mucous neck cells and then become SPEM cells, again demonstrating the plasticity of mature chief cells and agreeing with labeled-nucleotide tracing studies of chief and stem cell behavior [60]. The molecular data thus support previous work in tissue using lineage tracing, and cytological criteria to demonstrate chief cell plasticity in response to gastric injury [1,61–63].

Paligenosis and the Overall Conserved Patterns of Plasticity in Injury Response

Paligenosis is the conserved program by which mature, fully differentiated cells access a more fetal or progenitor-like state to re-enter the cell cycle and repair damaged tissue (Fig. 3). The paligenotic program is characterized by three phases/stages: 1) an initial decrease in mTORC and induced autodegradation of mature cell machinery, 2) metaplastic expression of progenitor genes (Sox9 and CD44v), and 3) a reactivation of mTORC as cells re-enter the cell cycle [8,64]. The molecular network governing cell cycle re-entry involves suppression by p53 and requires reactivation of mTORC1 (Fig. 3). It is this phase that is most important for clearing cells with potentially tumorigenic mutations and chromosomal alterations [65,66]. Although metaplasia is normally a temporally limited means for regeneration of lost cells after injury, the Stage 2 to 3 sequence in paligenosis carries inherent risk of propagating clones of mutant cells. This suggests why metaplasia increases risk for cancer in comparison to normal, constitutively active stem cells, which are far more resistant to propagating mutations [18,67].

Figure 3. Paligenosis is a conserved program by which mature cells dedifferentiate, return to a more progenitor-like state, and re-enter the cell cycle.

Upon injury, mature secretory cells such as gastric chief cells or pancreatic acinar cells will shut down mTORC1 and upregulate autophagy in stage 1 to degrade secretory cell-specific machinery [8,64]. Differentiation factor MIST1, its target ELAPOR1, and miR-148a expression decrease in the initial stages as mature cells downscale even before the full autophagic response is induced [8,73,78•]. DDIT4 is necessary for the initial suppression of mTORC1 and for the induction of autophagy [64–66••]. Autophagy of mature vesicles and machinery during stage 1 is regulated by ATF3 which binds and activates lysosome trafficking protein RAB7 [74]. Concurrently, CD44v and as well as xCT expression increases to moderate ROS buildup [75, 78•]. During stage 2, transitioning cells exhibit metaplastic expression of various mucin, progenitor cell, and SPEM markers including TFF2, SOX9, GS-II, and GKN3 [8,53•,54••]. During the transition from stage 2 to 3, IFRD1 destabilizes p53 which relieves the suppression of mTORC [66••]. Once ADAR1 is upregulated to clear dsRNA, and mTORC is reactivated in stage 3, the cell can re-enter the cell cycle to proliferate and replace lost tissue [8,55•]. While the high dose tamoxifen (HDTAM) gastric injury and cerulein pancreatic injury have different timelines, this sequence of molecular changes that occurs within the tissue is conserved across both tissues. The timings for both injuries are shown below each stage.

The search is underway for genes that specifically induce and execute paligenosis in conjunction with the aforementioned network that monitors and eliminates cells with irreparable genomic damage. Such genes are expected to be largely devoted to paligenosis, and thus used in injury-induced proliferation but not in homeostasis nor in reversible stress responses – not even in the increased proliferation that normal, homeostatic progenitors can undergo in response to injury. Thus, p53 and mTORC1 clearly are not “paligenosis genes”, although they are important in numerous functions, including paligenosis.

More paligenosis-specific gene candidates have been emerging (Figure 3). For example, DNA-damage inducing transcript4 (DDIT4) (also known as REDD1) is conserved as a response injury gene across multiple species including axolotl, fruit flies, mice, and humans [68–70] and is largely dispensable for normal development or stem cell activity. It is induced in injury and regulates mTOR in various tissues including kidney cells, hepatocytes, dental pulp stem cells, and mesenchymal stem cells [8,71,72]. As shown in mice, DDIT4 is essential for the initial suppression of mTORC1 during Stage 1 of paligenosis. Cells with DNA damage are prevented from re-entering the cell cycle by suppression of mTORC1. However, since paligenotic cells lacking DDIT4 never undergo the initial mTORC1 suppression, they largely bypass the DNA damage checkpoint increasing risk for subsequent cancer. Thus, DDIT4 is required to prevent injured cells that have accumulated DNA damage and mutations from re-entering the cell cycle and proliferate in the third stage/phase [64–66].

The massive upregulation of lysosomes and autophagosomes that normally occurs in Stage 1 also depends in part on DDIT4 inducing suppression of mTORC1. Other key players early in paligenosis include activating transcription factor 3 (ATF3), which is upregulated early in (or potentially before) Stage 1 to suppress the pro-secretory differentiation factor MIST1 and induce the autophagosome and lysosome trafficking protein RAB7. Paligenotic cells lacking ATF3 die during the Stage 2 to 3 transition [73•]. Endosome/lysosome-associated apoptosis and autophagy regulator 1 (ELAPOR1), a MIST1 target that regulates secretory granule maturation, decreases accordingly with MIST1 indicating the loss of mature features in the dedifferentiating cell [74]. The autodegradative machinery in Stage 1 likely functions not only to clear mature-cell machinery but also to reduce reactive oxygen species (ROS) toxicity. ROS induce ATF3, and the cystine transporter xCT, which binds the cell surface protein, CD44v, alleviates ROS accumulation. xCT, like ATF3, is required for normal Stage 1 with xCT-deficient mice failing to induce autophagy and progress through paligenosis [75]. AQP5 also increases early, even before accumulation of lysosomes and autophagosomes [50•]. Finally, double-stranded RNA (dsRNA) accumulate in the early stages of paligenosis with clearance by Stage 3. The dsRNA clearance depends on the dsRNA responsive RNA editing protein ADAR1, as paligenotic cells lacking ADAR1 die in Stage 2 [55•].

MicroRNAs (miRNAs), which are involved in cell differentiation in various systems, may also play a role in de-differentiation of mature secretory cells to more progenitor state [76,77]. Like MIST1, mi-RNA148a is a chief-cell-enriched molecule that also decreases at the start of paligenosis. Suppression of miRNA-148a induced expression of SPEM marker, and xCT interactor, CD44v, suggesting that decreasing miRNA-148a is an important early event in paligenosis [78•]. One target of miRNA-148a in paligenosis is the DNA methylase DNMT1. DNMTs are known to reduce transcription of endogenous sequences that can generate cytoplasmic dsRNAs [79].

Another example of a paligenotic gene is Interferon-related developmental regulator 1 (IFRD1). IFRD1 functions at the end of Stage 2 to inhibit p53, allowing for the reactivation of mTORC, following prior suppression by DDIT4, so that cells may re-enter the cell cycle [66••]. Like DDIT4, IFRD1 is also conserved across species (from yeast to Drosophila to humans) and implicated in regeneration and/or paligenosis neurons and liver [66••,80,81].

Paligenosis As Part of a Broader Pattern of Tissue Plasticity and Metaplasia

The concept of a conserved, stepwise process of reprogramming mature cells into progenitors has been proposed only recently, but there is evidence that some of the key features described in paligenosis occur in many instances of plasticity. For example, during the forced reprogramming of mouse embryonic fibroblasts into induced pluripotent stem cells (iPSCs), early inhibition of mTOR (with either rapamycin or PP242) increases reprogramming efficiency [82], suggesting an initial low mTORC1/high autodegradative state is key as in paligenosis. Similar to the role of DDIT4 in paligenosis, Yamanaka factor SOX2 functions to suppress mTOR expression via the nucleosome remodeling and deacetylase (NuRD) complex, inducing autophagy and promoting reprogramming. Subsequently, mTOR levels are restored by the end of the iPSC reprogramming process [83].

Paligenosis itself may occur within more global patterns of plasticity in a tissue. Paligenosis-driven change from chief into SPEM cells occurs within a more global metaplastic program in the corpus gastric unit that includes loss of parietal cells and induced proliferation in the constitutive isthmal stem cells, and expansion of mucous neck cells. Eventually, pyloric metaplasia often progresses to include cells with intestinal differentiation (particularly goblet cells) in the process called intestinal metaplasia.

The pyloric metaplasia phenotype seems to be recapitulated in multiple organs following injury with even similar cohabitation of foveolar and metaplastic chief and neck cell lineages. In Barrett’s esophagus, the quintessential metaplastic glandular unit comprises SPEM-like cells at the base and mixed foveolar and goblet cells more superficially (pyloric with incomplete intestinal metaplasia) [84•-86]. In pancreas too, a pyloric-like metaplasia emerges after metaplastic injury. In mouse models, injection of the CCK analogue cerulein induces acinar-ductal metaplasia (ADM) in the exocrine pancreas, a lesion that increases risk for progression to pancreatic intraepithelial neoplasia (PanIN) and pancreatic ductal adenocarcinoma. And sc-RNA-Seq shows that the injured mouse acinar cells – which, like chief cells, secrete digestive enzymes – can undergo paligenosis to SPEM-like cells characterized by co-expression of MUC6, TFF2, and the normally exclusively gastric gene GIF. Cells expressing gastric foveolar/pit cell markers like MUC5AC also emerge during ADM (Fig. 2) In both mouse and human ADM samples, SPEM markers like CD44 and AQP5 (which is also normally exclusively expressed in stomach) are also expressed. [56••].

Recent sc-RNA-Seq analyses of human precancerous lesions known as sessile or serrated polyps have shown that similar metaplasia may also occur as a precancerous lesion in the intestines. These polyps tend to form in more proximal colon (right-sided), tend to be induced during chronic colitis, and are distinguished from the more common adenomatous polyps [87,88]. While the more common adenomatous polyps nearly invariably arise via aberrant WNT signaling and are composed of cells that are wholly intestinal in differentiation, serrated lesions are characterized by emergence of basal cells with SPEM-like differentiation and a superficial zone with admixed goblet and foveolar-like cells (Fig. 2). These lesions resemble pyloric metaplasia that has progressed to incorporate cells with intestinal differentiation, though presumably in this case the plasticity event is in the opposite direction with intestinal cells reprogramming towards a gastric phenotype. Gene expression analyses show specifically that serrated polyps exhibit decreased CDX2 (a critical intestinal epithelium-promoting transcription factor) and loss of intestinal secretory cell master transcription factor ATOH1. Gastric foveolar (MUC5AC) and SPEM (AQP5 and TFF2) markers are induced, and the tissue also appears characterized by reversion to a more fetal-like state with the induced expression of MDK, a gene only transiently expressed during colonic embryonic development [89••]. It is not clear if paligenosis plays a role in any of the plasticity events leading to serrated polyps.

Thus, injury can induce a pyloric metaplasia with SPEM-like cells in the distal esophagus, gastric corpus, pancreas, and colon. It is not clear why there is a reversion to a common antrum/pylorus-like phenotype during chronic inflammation and injury; however, it is possible that the antrum most resembles the simplest glandular units that line the alimentary tract during the embryonic stages, prior to rostral-caudal differentiation of each organ. Indeed, the antrum and proximal duodenum are the site of tremendous tissue fate specification embryonically, as gastric corpus, dorsal and ventral pancreatic buds, biliary tree and liver all emerge from this region [90]. Thus, in multiple adult organs, when injury requires regeneration, the epithelium may revert to an embryonic-like state where proliferation, cell plasticity and multipotent cell lineage differentiation choices are available to the regenerating tissue [91].

Conclusion

This review examines stem cell dynamics and differentiation, incorporating insights gleaned from the newest methodologies, including sc-RNA-Seq and new lineage-tracing mouse pedigrees, and emerging concepts of the universality of cell plasticity. We highlight glands in the stomach because they exhibit both constitutively proliferating stem cells and differentiated cells that can become progenitors via paligenosis. We conclude that plasticity may be important in stem cell behavior and differentiation in both the corpus and antrum of the stomach, even at homeostasis. It is particularly important after chronic inflammation of the corpus, when the entire gastric gland can undergo pyloric metaplasia. The chief cells can become SPEM cells via a highly conserved, stepwise process called paligenosis that involves massive rearrangement of cell architecture and switches in energy use. Paligenosis likely evolved to allow recruitment of stem cells for regeneration while decreasing tumorigenesis-related risks of reprogramming long-lived differentiated cells into progenitors during a time of tissue damage [1]. Surprisingly, the reprogramming capacity of the whole epithelium – not just of specific paligenosis-capable cells – in different organs might be similarly conserved, such that colon, esophagus, and pancreas can all undergo some version of pyloric metaplasia.

Though we have focused on plasticity in the gastric epithelium in this review, where different cell types are largely grouped into zones that allows investigators to follow cell behavior in the tissue, even by histochemical stains, there is abundant evidence for plasticity in other epithelia. For example, in small intestines, there are numerous genetic-based lineage tracing studies showing nearly all cell lineages can return to a stem cell state (reviewed recently in 92). Arguably, what has been more difficult in the intestine has been proving how specific cells actually undergo plasticity at the cellular and subcellular level. It is known that mTORC1 modulation is clearly critical for recruiting stem cells via plasticity in intestines [67,93,94]. It has also been shown that fasting (which would activate autophagy and lysosomes) before injury increases regeneration after stem cell damage by radiation or DNA damaging drugs [95–98] which would be consistent with autodegradation being a critical early player in a plasticity process to recruit replacement stem cells. However, to determine if any given more differentiated cell can undergo reversion to a stem cell state in intestines specifically via paligenosis (or some other stepwise plasticity program), we must be able to follow individual cells or groups of the same cell types at each stage in the process to show that autodegradation is induced, followed by new progenitor gene expression, followed by return to the cell cycle. The dynamic, intermixed cells in the intestinal crypts have hindered these sorts of cell biological studies, though there may be ways to get around these issues using prospective isolation via flow cytometry and/or by manipulating isolated crypts in organoids.

Our understanding of all the processes we address here has grown dramatically just in the last few years, as new perspectives of cell plasticity and new multi-omic, single-cell techniques have emerged and mutually reinforced each other. Single-cell multiplex analyses are revealing that cell identity may be a much more plastic, dynamic state than histology has suggested. Efforts to further define genetic regulation of conserved plasticity programs like paligenosis, to identify other plasticity programs that differ from paligenosis, and to determine which types of injuries and inflammatory signals dictate each type of plasticity response will be of great interest. Resulting insights could establish foundational concepts that will enable efforts to harness plasticity for regeneration while decreasing its inherent risk for tumorigenesis.