Flexibility Sustains Epithelial Tissue Homeostasis

Abstract

Epithelia surround our bodies and line most of our organs. Intrinsic homeostatic mechanisms replenish and repair these tissues in the face of wear and tear, wounds, and even the presence of accumulating mutations. Recent advances in cell biology, genetics, and live-imaging techniques have revealed that epithelial homeostasis represents an intrinsically flexible process at the level of individual epithelial cells. This homeostatic flexibility has important implications for how we think about the more dramatic cell plasticity that is frequently thought to be associated with pathological settings. In this review, we will focus on key emerging mechanisms and processes of epithelial homeostasis and elaborate on the known molecular mechanisms of epithelial cell interactions to illuminate how epithelia are maintained throughout an organism’s lifetime.

Introduction

Epithelial cells constitute the protective layers that line our internal organs including the respiratory and digestive tract, reproductive and urinary systems, endocrine and exocrine glands, as well as the external skin epithelium. These epithelia perform a diverse array of functions including selective absorption of nutrients, secretion of hormones and enzymes, and formation of essential protective barriers; as a result, epithelial integrity and homeostasis are of central importance to survival. However, exactly how all the individual cells within an epithelial tissue behave to uphold its functions and maintain homeostasis throughout a lifetime—especially in the face of injury or mutations—is not yet clear. Improved knowledge of these fundamental principles would inform the etiology of many pathological states. Recent advances in cell biology, genetics, and live-imaging techniques have revealed that epithelial homeostasis represents an intrinsically flexible process at the level of individual epithelial cells. A better understanding of the principles and boundaries of this homeostatic flexibility is essential to our study of the plasticity mechanisms that emerge after wounding or during cancer. In this review, we will focus on recent work that highlights this inherent flexibility, which we define as a cell’s ability to perform diverse behaviors in response to the needs of the tissue, and show how it serves as a foundation of the body’s response to pathological insults.

Cellular and Molecular Mechanisms Sustaining Homeostatic Equilibrium

Healthy epithelia tightly balance the gain and losses of cells, maintaining homeostasis via a dynamic equilibrium. An inability to properly control cell numbers over time can have severe consequences, leading to compromised function in cases of excess cell loss and the potential formation of tumors in cases of excess cell gain [1,2]. Maintaining this balance is further complicated by the high turnover rates of many epithelial tissues, where cell loss through differentiation and/or death and cell gain via proliferation are a constant occurrence [3]. Here, we review recent insights into the cellular and molecular mechanisms that underlie this homeostatic balancing act.

Response to mechanical cues:

It has long been known that stretching cultured cells stimulates epithelial cell division and survival [4,5]. Later studies elucidated many of the mechanosensitive pathways behind this phenomenon, reporting that cell stretching activates the Hippo pathway transcription factors Yap and Taz, which in turn promote cell proliferation [6,7]. In parallel, application of mechanical strain can also drive β-catenin into the nucleus through an E-cadherin dependent mechanism [8]. Interestingly, nuclear-localized Yap and β-catenin act independently and affect distinct stages of the cell cycle, with Yap driving exit from GO and β-catenin inducing the G1 to S transition [8], indicating that mechanical changes can influence proliferation through multiple parallel inputs. More recently, Gudipaty et al. found that a similar in vitro stretching approach can also activate Piezo1 channels, leading to calcium-dependent activation of ERK1 and a rapid transition from G2 to M phase [9] (Figure 1). Reduction of Piezo1 levels in the larval zebrafish epidermis also leads to a decrease in mitotic cells, suggesting that this type of stretch response may also occur in vivo [9], potentially allowing cells to respond rapidly to decreased local density stemming from nearby cell death or overall tissue expansion. Interestingly, Piezo1 in the Drosophila midgut can also respond to mechanical cues by increasing cytosolic calcium, but in this case, the calcium influx can trigger two different outcomes: proliferation or differentiation towards the enteroendocrine lineage, each likely via a distinct molecular mechanism [10] (Figure 1).

Figure 1. Cellular neighborhoods impact epithelial fate decisions.

During normal epithelial turnover in Drosophila and mammalian intestinal epithelium, mechanical crowding from cell proliferation activates the stretch-responsive Piezo1 channel to trigger the extrusion of live cells, which later die by apoptosis. (A) New epithelial cells in the intestinal epithelium migrate and differentiate along the villus and in response to crowding stress, and cells extrude at the villus tip to maintain homeostatic cell numbers. (B) An increase in cellular crowding forces promotes basal extrusion in Drosophila intestinal epithelium. Mechanical forces from cell stretching can also activate stretch-activated Piezo1 channels and increase cytosolic calcium. A calcium influx can trigger two different outcomes: proliferation through calcium-dependent activation of ERK and differentiation towards the enteroendocrine lineage through calcium-regulation of Notch signaling. Additionally, healthy cells inhibit intestinal epithelial cell division through E-cadherin (E-cad), which prevents the secretion of mitogenic epidermal growth factors (EGFs). Individual apoptotic cells promote division by the loss of E-cad, which releases β-catenin and p120-catenin to induce rhomboid (rho). Induction of rho triggers the activation of the EGF receptor (EGFR).

At the opposite end of the spectrum, epithelia during development can also respond when local density becomes too high by eliminating cells from the crowded region. In several vertebrate epithelial monolayer tissues, this elimination takes the form of live cell extrusion that again depends on Piezo1, which is thought to respond to compression and lead to the non-autonomous accumulation of a basally localized actomyosin ring in neighbors of the cell that is to be extruded [11] (Figure 1). Cell extrusion also takes place in Drosophila epithelia [12], although in this context, crowding is thought to first cause apoptosis, followed by subsequent basal delamination of dying cells through a yet to be understood mechanism [13] (Figure 1). Recent work from Miroshnikova et al. serves as an example of how density sensing can be important not only for maintaining epithelial density but for building a properly stratified epithelial structure. In the developing mouse epidermis, proliferation within the basal progenitor layer leads to crowding, cell-shape distortion, and uneven stress distribution, driving the eventual differentiation of a subset of cells [14]. A subsequent decrease in cortical tension and an increase in cell adhesion within these cells lead to their delamination, contributing to the formation of the suprabasal layers that will comprise the mature adult epidermis (Figure 2).

Figure 2. Flexible homeostatic mechanisms maintain epithelial homeostasis across time and in face of mutational insults.

In many epithelial tissues, flexible homeostatic mechanisms are in place to maintain these tissues in a state of functional normalcy through the development or acquisition of mutational insults. In embryonic mammalian skin epidermis (left), a single basal layer of epithelial cells adheres to an underlying basement membrane. In this layer, crowding and proliferation of cells lead to apical extrusion from the basal layer. In adult mammalian skin epidermis (middle), a single basal layer of epithelial cells underlies suprabasal layers of differentiated cells. In this case, differentiation of a cell, which causes it to leave the basal layer, is then followed directly by the division of an adjacent neighboring cell. After mutation (right), flexible homeostatic mechanisms can lead mutant cells to be actively eliminated out of the tissue or even incorporated into the normal homeostatic program to return the tissue to overall normalcy.

Spatial orientation of divisions:

In both developing and adult homeostatic epithelia, cell divisions must not only properly balance cell loss, but they must also produce daughter cells that preserve or build the tissue architecture correctly. One common solution to this problem is to tightly control the orientation of dividing cells [15]. In interphase cells, the long axis determines the division orientation. But as cells enter mitosis, they lose the axis as they round up to ensure faithful chromosome segregation. Bosveld et al. found that tricellular junctions—places where three cells meet—serve as spatial landmarks during mitosis of Drosophila epithelial cells, encoding information about interphase cell shape and acting through Mud (vertebrate NuMA) and astral microtubules to properly orient the plane of division [16].

Flexibility of stem cell fates:

Many adult epithelia depend on the activity of resident stem cells to sustain a lifetime of rapid turnover. Traditionally, these stem cells have been thought to participate in stereotyped hierarchies where each stem cell has an equal ability to give rise to all terminally differentiated cell types, and where differentiation occurs in a unidirectional fashion [17]. However, insights from a number of epithelial systems have begun to emphasize that the capacity to self-renew or to differentiate towards a given lineage may in some cases vary more dynamically over the lifetime of a stem cell in response to the demands of the tissue [18,19].

An example of this flexibility can be seen in the skin epithelium of the adult mouse ear and paw. Until recently, the upstream factors driving stem cell self-renewal in this tissue had been difficult to address due to an inability to study stem cell fates in the context of neighboring cell behaviors. Using intravital imaging and in vivo tracking approaches, Mesa, Kawaguchi, Cockburn et al. were able to show that differentiation and exit of a single stem cell from the basal compartment allow neighbors to expand, leading to the division of an adjacent cell to locally balance cell numbers [20] (Figure 2). These data suggest that the demand for differentiated cells is a central driving force of homeostasis and indicate that stem cells remain flexible enough to differentiate when the demand arises or to divide instead when needed to replace any neighbors that have been lost. While the molecular mechanisms that support this flexibility in the mouse epidermis are unknown, a similar coupling behavior can be seen in the fly gut epithelium where apoptosis induces cell division of neighboring stem cells [21]. In this case, the authors beautifully demonstrate an adhesion junction-mediated mechanism in which the loss of E-cadherin in the dying cells initiates a rhomboid-mediated signaling cascade that activates EGFR and cell division in the nearby stem cells [21] (Figure 1).

The mouse hair follicle provides another example of previously unappreciated stem cell flexibility. A group of progenitor cells, arranged around a mesenchymal niche known as the dermal papilla, generate the multiple distinct cell types of a mature hair shaft. Prevailing models in the field proposed that each of these progenitors were unipotent, producing only a specific differentiated cell type [22,23]. However, recent work showed that individual progenitors are in fact multipotent, moving along the dermal papilla as the hair follicle grows and flexibly producing distinct differentiated cell types that correspond to their changing positions around the niche [24]. Interestingly, recent single-cell RNA sequencing results indicate that progenitors located at different positions around the dermal papilla have distinct molecular signatures [23], suggesting that the local microenvironment can prime cells to produce different follicle lineages as they transit. Together, these examples highlight the emerging idea that even during homeostasis, stem cells retain a level of plasticity that allows them to respond to the changing needs of the tissue.

Maintaining homeostasis: Containing Oncogenic Mutations in Normal Tissue

A growing body of evidence now indicates that cells carrying oncogenic mutations are frequently found within phenotypically normal human tissues [25–27]. Although we primarily discuss epithelia in this review, genetic mosaicism is observed in tissues from blood to brain [25,26,28–32]. These findings raise the question of how a tissue maintains homeostasis despite aberrant cellular behaviors caused by damaging mutations (Figure 2).

An important homeostatic principle is cell competition, a process that regulates organ and tissue size via the elimination of less fit cells. In 1975, Morata and Ripoli first described cell competition in Drosophila wing imaginal discs containing Minute mutant cells among wild-type cells [33]. Minute heterozygous flies develop slowly but normally, however, Minute heterozygous clones are eliminated from wild-type discs. Similarly, studies of the sole Drosophila homolog of Myc showed that Myc mutant cells are outcompeted by their wild-type neighbors [34], whereas cells expressing higher levels of Myc behave as super-competitors and eventually outcompete wild-type cells [35,36]. Since this initial discovery, cell competition has been observed in a myriad of epithelial and non-epithelial tissues in both Drosophila and mammals [37].

A similar phenomenon has also been observed through genetic testing of how normal tissue responds to the induction of oncogenic mutations. In the mouse skin, an activating mutation of β-catenin in hair follicle stem cells leads to the formation of benign tumors that contain both wild-type and mutant cells [38]. By longitudinally tracking these outgrowths in living mice over time, Brown and Pineda made the surprising observation that these tumors eventually regress and disappear from the tissue [39]. This regression depends on the presence of wild-type cells, which surround and somehow outcompete the mutants, leading to their differentiation into a hair shaft-like structure that is expelled from the skin [39]. Interestingly, when this β-catenin mutation is induced in the hair follicle progenitor compartment instead of the stem cells, the natural movement of progenitors around the dermal papilla allows for the relocation of the mutants to a location where Wnt signaling is high, and where they can seemingly differentiate normally, preserving follicle architecture [24]. The remarkable capacity of the skin to cope with aberrancies is not merely specific to mutant β-catenin, but also extends to tissue thickening and structural deformities caused by oncogenic Hras, which also resolve over time [39]. Whether correction, in this case, involves the elimination of Hras mutant cells from the tissue through a mechanism similar to what is seen for β-catenin is not yet clear. This outcome is also seen with a number of additional mutations. Recent work has shown that clones of p53 mutant cells in the skin initially expand, out-competing their wild-type neighbors and causing an epidermal thickening [40]. Eventually, however, this expansion slows, and thickness partially resolves, indicating that the p53 competitive advantage is somehow lost before these clones take over entirely. Similarly, through in vivo imaging approaches, Ying et al. demonstrate that the oncogenic activation of PIK3CA biases basal progenitor cell fate towards differentiation, effectively forcing the elimination of the mutant cells and suppressing epidermal growth and expansion [41].

Interestingly, all of the mutational situations described above might be expected to give cells a competitive advantage, leading to a complete take-over of the tissue from their wild-type neighbors [37]. However, in each case, the mutant clones are kept at bay, either by active elimination [39] or by remaining and integrating into the normal homeostatic program of the tissue [40]. While the mechanisms underlying tissue’s ability to eliminate or tolerate mutated cells remain to be uncovered, these examples clearly indicate wild-type cells have a remarkable capacity to keep the growth of tumors at bay.

Restoring Homeostasis: Wound Repair

Throughout an organism’s life, constant exposure to harmful stimuli necessitates effective mechanisms that respond to and repair injury. To that end, epithelial cells can leverage the flexibility of homeostatic mechanisms while adopting new repair-specific behaviors to enable efficient regeneration and restoration of a tissue upon damage. A key example of this can be seen in wounded mammalian skin, where basal epidermal cells acquire new behaviors such as migration but continue to perform behaviors also seen in homeostatic tissue such as proliferation and differentiation [42–44]. Recent work is beginning to shed light on how these behaviors are properly integrated to facilitate an effective wound repair response.

After wounding, the skin epithelium organizes into two concentric zones: a migratory front immediately surrounding the wound edge and, further out, a ring of rapidly proliferating cells [45–48]. Recent live imaging of the wound repair process has revealed that these two zones spatially overlap to some extent, with some cells performing both behaviors simultaneously [49]. Understanding how wound-induced proliferation and migration cues are received and executed in the same cells over time may illuminate aberrant behaviors in settings such as cancer metastasis as well as contexts that are associated with cancer emergence. In addition, molecular signatures for the migrating and proliferating zones were recently elucidated, which will provide insight into the regulation and function of these two behaviors [50]. Live imaging has also revealed that the proliferation zone serves not only as a source of new cells, as it does in homeostatic tissue, but it also functions to limit the area of unwounded epithelium utilized for re-epithelialization [49]. Thus, homeostatic mechanisms can be integrated with new wound-specific behaviors and re-purposed to close an epithelial wound and ultimately restore homeostasis.

In addition to migration, other behaviors specific to the wound-repair program are beginning to be identified. Zebrafish superficial epithelial cells subjected to a mild exfoliation injury can rapidly undergo a shape change that allows them to increase their surface area, presumably allowing them to better maintain the epithelial barrier [51]. However, as repair proceeds through other mechanisms, this shape change is reversed, demonstrating a transition between homeostatic and repair states at the level of individual cells. It will be interesting to determine whether this type of behavior, as well as the molecular mechanisms underlying it, is conserved in other wounded epithelia.

Work from multiple systems has shown that wound repair also involves highly plastic contributions from cell populations not normally involved in the homeostatic regulation of a given epithelium. In the wounded mouse skin epithelium, for instance, hair follicle stem cells that normally contribute only to the maintenance of that appendage, provide progeny that can migrate out of the follicle and assist with re-epithelialization [52,53]. Vice versa, loss by injury of the hair follicle stem cells is compensated by epithelial cells of the upper epithelial layers [54]. A related phenomenon can be seen in the mouse intestine, where injury or ablation of either actively cycling Lgr5+ stem cells at the crypt base or the more quiescent +4 stem cells, both of which make distinct lineage contributions during homeostatic conditions [55], can allow the other population to repopulate the lost compartment [52,56,57].

Surprisingly, however, wounding can also lead to the contribution of more differentiated cell types not known for their regenerative potential during homeostasis. This can be seen in the systems such as the lung, where ablated airway basal stem cells can be replaced by the dedifferentiation of committed luminal cells [58], or the intestine, where transit-amplifying or even more differentiated enterocyte-progenitor cells can similarly compensate for the loss of Lgr5+ stem cells [59,60]. It can also be seen in the wounded mouse skin, where differentiated cells have recently been shown to migrate after wounding, actively contributing to the re-epithelialization process [49]. Recent work has even suggested the presence of a small population of sebaceous gland cells that can de-differentiate and contribute to the interfollicular epidermis after wounding [61]. These contributions from differentiated cell types are not only unexpected, but they also suggest it may be fruitful to re-examine how we think about the behaviors and roles of differentiated cell types in homeostatic tissue as well.

For large and chronic wounds, which are often characterized by poor circulation, aberrant inflammation, and a failure of epithelial cells to regenerate the outermost shield [62], the healing process can be inefficient. Capitalizing on recent advances in cellular reprogramming [63,64], Kurita et al. used viral vectors to express key epithelial reprogramming factors in nonhealing wounds. This approach was sufficient to convert mesenchymal cells into epithelial cells, promoting surface re-epithelialization and healing and providing a potential new avenue for treatment [65]. Thus, in systems where the natural plasticity of the repair program cannot fully resolve an injury, more direct genetic interventions can help to generate the cell types needed for healing.

Conclusion

The most significant pressure for our tissues is to maintain function in the face of continuous turnover and insults, be it from injury or the progressive accumulation of mutations. Thus, from early on when tissues are built, flexible and tailored homeostatic mechanisms ensure tissue maintenance and function. Work spanning from invertebrate to vertebrate systems support a model whereby behavioral flexibility of individual cells underlies normal tissue homeostasis. This flexibility/plasticity can be exploited or extended during wound repair and become aberrant in pathological situations. We propose that behavioral flexibility does not necessarily represent a vulnerability but is an asset to maintain epithelial tissue and to continuously eliminate pathologies before they even emerge.