|Home | About | Journals | Submit | Contact Us | Français|
To function as an intact barrier, epithelia must maintain constant cell numbers despite high rates of turnover. If the rate of death exceeds proliferation, epithelial barrier function could become compromised; if it lags behind proliferation, cells could amass into tumors. Although the balance between cell death and division is critical for preventing pathology, most studies focus on each process in isolation. Loss of contact inhibition is a hallmark of cancer cells and cell contacts appear important for linking rates of cell division and death. However, epithelial cells continuously divide and die while maintaining contacts with each other, so other factors must control this balance. Recent studies find that cell crowding forces from cell proliferation can drive cells to die by extrusion from the epithelium. Factors that alter this response to cell crowding may lead to barrier function diseases or promote hyperplasia and cancer.
Epithelia are composed of tightly adherent cells that coat and protect our organs and body. Cells within epithelia have some of the highest rates of turnover in the body [1–3], where the number of dividing cells is tightly balanced by a similar number of dying cells. Since most cancers arise in cell populations with high turnover rates such as the blood and epithelia, it is likely that they arise from misregulation of cell number homeostasis. Because most solid tumors originate from epithelia, understanding what controls the link between cell division and cell death in epithelia is critical to our understanding of how tumors initiate. However, most studies have focused separately on either what controls cell death in response to apoptotic stimuli or what controls cell division in response to mitogens. Remarkably few studies investigate how these two processes are coordinated in vivo to maintain overall cell numbers.
The studies most relevant to cell number homeostasis started over sixty years ago with the discovery of contact inhibition. Contact inhibition actually refers to two separate fundamental findings: contact inhibition of growth and contact inhibition of locomotion. The former is based on the fact that cells dramatically reduce their rate of mitosis when they contact each other and establish a monolayer [4, 5]. Contact inhibition of locomotion instead refers to the fact that migrating cells will stop moving once they contact each other to form a monolayer . By contrast, cancer cells are not contact inhibited either in their growth or motility and will instead pile upon one another and continue dividing [7–9]. The absence of contact inhibition forms the basis for testing if cells are transformed by their ability to grow in soft agar. But what does contact inhibition mean in vivo for epithelial cells that are continuously migrating and dividing, despite the fact that they must maintain tight contacts with each other to preserve their function as a barrier? How, then, do epithelial cells adhered to one another in an epithelium maintain constant cell numbers?
Although establishment of cell contacts may not be sufficient to control cell numbers in epithelia, the extent of cell contacts with each other and their substratum may, instead, control whether cells divide or die. Contact inhibition of growth and migration in vivo may depend on cells reaching a threshold number of cell-cell or cell-matrix contacts, since both types of inhibition are dependent on cell density rather than the formation of cell-cell contacts . However, when cells reach still higher densities, crowding may lead to fewer engaged cell-cell and/or cell-matrix adhesions compared to neighboring cells, which can promote anoikis, or cell death by loss of adhesion-based cell survival signaling . In mathematical models, epithelial crowding can lead to increased mechanical tension on cells, which promotes cell loss [12, 13] to regulate epithelial tissue homeostasis . Thus, the mechanical strain and alterations of cell contacts in crowded epithelial regions may be at the heart of contact inhibition. Here, we discuss how cell contacts and density within epithelia may impact whether cells divide or die (Figure 1A).
Exactly how cell contacts affect the decision for a cell to proliferate is not entirely clear, however, hints come from discovery of the Hippo pathway. In Drosophila and mammals, mutations in the Hippo pathway lead to tissue overgrowth [15–22], suggesting that this pathway is critical for regulating proliferation and cell numbers. E-cadherin and alpha-catenin, essential proteins for cell-cell adhesion, control proliferation in response to changes in cell density by regulating the subcellular localization of critical Hippo downstream effectors, yes-associated protein (YAP) and TAZ [23, 24]. YAP is predominately cytoplasmic in confluent cells, but at lower cell densities accumulates in the nucleus where it acts as a transcriptional coactivator to promote proliferation [23–26]. Disrupting cell-cell contacts or cell-extracellular matrix (ECM) interactions is sufficient to shift YAP to the nucleus [23, 27]. Thus, the Hippo pathway is emerging as the long sought-after pathway for controlling cell proliferation in response to cells contacts, yet its regulation does not depend merely on cells making contacts with one another but upon the density of contacted cells. Mechanical forces generated from matrix stiffness can also activate YAP/TAZ independently of the Hippo pathway , suggesting that crowding forces independent of cell-cell contact could also signal density-dependent proliferation. We do not elaborate on contact-dependent regulation of cell proliferation here as a recent review covers this topic in depth , but instead consider how cell density affects cell death.
In addition to cell-cell contacts and density controlling whether a cell within an epithelium will proliferate, they also control whether it will die. Recent studies from a variety of epithelia, including developing Drosophila and zebrafish, human adult colon epithelia, and tissue culture epithelial monolayers, show that cells in crowded regions of epithelia extrude and later die [30, 31]. Cells routinely were found to proliferate at defined regions within the epithelium where cells were least crowded. Cells then migrate away from sites of proliferation within the monolayer to converge at regions of high density, where they get shoved out, or extruded. This conveyor belt model of cells within an epithelium is exemplified by intestinal epithelia (Figure 1B), but is also apparent in different epithelia in vivo, such as the edge of the fin epidermis in developing zebrafish . The surprising finding from these studies was that cells rarely died by apoptosis in situ. Instead, live epithelial cells in crowded regions extrude [30, 31] (Figure 1C) and later die by anoikis [11, 32, 33], or cell death due to detachment from the matrix. These studies support the idea that mechanical stresses and tension promote cell death by extrusion independent of the apoptotic pathway . Although it is not clear what marks a particular cell for extrusion, loss of a threshold number of cell contacts with other cells and the underlying matrix may be a factor.
Cell extrusion was originally identified as a process that maintains the barrier function of an epithelium when cells within that layer die . All epithelia observed have been found to extrude dying cells, ranging from C. elegans, Drosophila, mouse, chick, zebrafish, and human epithelia [30, 31, 34–37]. Induction of apoptosis in cell culture monolayers and zebrafish has been critical for elucidating the mechanisms that control apoptotic cell extrusion in vertebrate epithelia. From these studies, we know that inducing apoptosis in epithelia through either the intrinsic or extrinsic pathways  results in extrusion of cells (Figure 2A). Additionally, these experiments defined a conserved mechanism for cell extrusion in vertebrate epithelia. To extrude, the cell destined to die produces and secretes sphingosine-1 phosphate (S1P)  to signal its live neighboring cells to form a ring of actin and myosin IIA around the dying cell . Contraction of the multi-cellular actomyosin ring ejects the dying cell out of the tissue and closes any gaps that may have resulted from its exit. It is not clear if this mechanism and signaling pathway is conserved throughout all species since cells can extrude apically or basally, depending on the organism. In Drosophila, cells extrude predominantly basally, in a process typically referred to as cell delamination [31, 34, 37], whereas vertebrates extrude epithelial cells predominantly apically. Whether a cell is shed apically or basally could depend on different signaling pathway or molecular alterations in a conserved pathway [40, 41]. Mutant or transformed cells may also use extrusion to exit an epithelium. Expression of oncogenic K-Ras Src, or mutant DPP/BMP cells leads to preferential elimination of those cells when mixed with wild type neighboring cells in cell culture or the epidermis of developing Drosophila and zebrafish [42–46], although it is currently unclear if these delaminations use the same mechanism as apoptotic and homeostatic extrusions. Regardless of whether a common mechanism is responsible for extrusion in all organisms, extrusion appears to serve the same purpose throughout all species: it ejects cells while preventing any breaches to the barrier function, a function essential to all epithelia.
How does the apoptotic extrusion pathway relate to the live cell extrusion that occurs in response to cell crowding during normal homeostasis in the body? Live cell extrusion, like apoptotic extrusion, requires S1P signaling to activate ROCK-mediated actomyosin contraction . Yet, blocking apoptosis by Bcl-2 over-expression, which blocks apoptotic cell extrusion , does not inhibit live cell extrusion. Likewise, blocking cell death in developing Drosophila by overexpressing p35 does not affect extrusion [31, 42]. Instead, live cell extrusion from crowding strain during homeostasis requires stretch-activated signaling, presumably upstream of S1P signaling. Inhibiting stretch-activated signaling or knocking down Piezo1, a recently identified stretch-activated channel (SAC) that transmits calcium currents [47, 48], prevents live cell extrusion in vivo. Importantly, SAC inhibition also leads to accumulations of epithelial cell masses at sites where extrusions would have occurred, suggesting that extrusion is the chief mechanism for controlling epithelial cell death in vivo . Thus, while the S1P pathway controls both apoptotic and homeostatic cell extrusion, SACs control live cell extrusion during homeostasis (Figure 2B). This raises the question of how cell-crowding strain may activate SACs to induce extrusion of live cells.
How do epithelia sense crowding and activate stretch-activated signals in response to strain? Interestingly, extrusion in vivo always occurs in regions of the epithelium that are 1.8-fold more crowded than other areas . Importantly, this is a scaling effect, as the crowding response is relative to the size of other cells within the epithelium rather than an absolute cell size. This suggests that cells, like people, have a sense of personal space that is defined by the tissue, and when they become too crowded, some must leave for the group to return to comfortable densities. Interestingly, mathematical simulations have predicted a similar threshold in tension to promote “pressure-induced apoptosis” in response to cell growth  (Figure 3A). Pressures near extrusion-induced apoptosis have been measured by membrane recoil following laser ablation and suggest that the amount of local mechanical tensions influences the frequency of extrusions [31, 34, 37]. Crowding driven extrusion was also shown experimentally where cells are crowded in a stretching device used in reverse. Over a six-hour period following crowding, epithelia equilibrate to homeostatic cell densities by activating cellular extrusion . Increasing the amount of cellular crowding, both in mathematical models  and in experiments , increases the amount of live cell extrusion (Figure 3B&C). As the increase in cell density surpasses a fold change of 1.4–1.6, the number of live cells ejected from the tissue dramatically increases (Figure 3C). In vivo, extrusion may actually initiate at similar crowding densities, but the lag in extrusion activation may make it appear to occur at the 1.8-fold density measured. The in vitro crowding studies establish a critical crowding concentration where cells activate extrusion that may be used to predict situations where cells will be extruded.
How do cells sense crowding at a molecular level? One might expect cells experiencing compression in crowded zones of the epithelium to become thinner and taller. However, crowding instead causes the total cell volume to decrease . Because SACs likely play a role in translating the crowding strain force into activation of proteins controlling extrusion, changes in cell volume could ultimately affect the expression or activity of SACs [49, 50]. A key requirement for mechanosensitivity is that membrane stress reaches the channel so it can change shapes between open and closed states [51–54]. Therefore, cell density could directly impact SACs in the membrane. Either aqueous influx or resistance due to the state of the cytoskeleton could counteract increased pressure on a cell [55–57]. Interestingly, we have found an early role for potassium channels in promoting apoptotic cell extrusion, as addition of 4-aminopyridine, a potassium channel inhibitor, blocks cell death and extrusion following UV-C exposure . Although a role for potassium channels has not yet been established for crowding-activated live cell extrusion, these channels might also enable cells to condense and activate SACs. Further, both actin/myosin contraction and destabilized microtubules  have been noted early in the extruding cell, both of which could compact the cell and activate SACs. Likewise, rearrangements in the actin and microtubule cytoskeleton occur in both the extruding cell and its neighbors during cell delamination in the Drosophila amnioserosa . Alternatively, cell condensation could affect the number of cell-cell or cell-matrix contacts, which could target particular cells in crowded regions for extrusion. Although the Hippo pathway plays a role in controlling contact inhibition of proliferation, mediators of this pathway, YAP and TAZ, could also act to control extrusion in response to crowding, as cytoplasmic phosphorylated YAP increases when cells are plated in a small, restricted pattern of matrix . Recent mathematical simulations also show the activity of the Drosophila YAP homolog, yorkie, can be influenced by mechanical stresses during development . Further, activation of the S1P/S1P2/Rho pathway that controls extrusion also activates YAP . While YAP and TAZ regulate proliferation through transcriptional activation in the nucleus, during crowding, they could act in the cytoplasm to regulate extrusion.
As extrusion appears to be critical for regulating overall cell numbers during epithelial homeostasis, alterations of any step in pathway controlling cell extrusion could lead to epithelial pathologies that either disrupt barrier function or lead to hyperplasia and cancer (Figure 4). Here, we discuss how defects in the ability to sense and respond to cellular crowding may promote common epithelial diseases.
An inability to sense engaged cell contacts and crowding could also prevent extrusion of cells, leading to the accumulation of defective cells that then develop into a carcinoma (Figure 4A). In support of this, colon adenomas (polyps) are comprised of densely crowded epithelia that lack extruding cells . The idea that preventing the ability to sense crowding could lead to hyperplasia is supported by the fact that blocking the stretch-activated channel Piezo1 results in epithelial cell mass formation in zebrafish epidermis. Additionally, changes in the intrinsic density of cells or their surrounding matrix could also affect the ability of cells to sense crowding. Tumors are denser than the surrounding normal tissue [60, 61], a property that is evident in the ability to palpate tumors. Increased tumor density is due to a stiffer matrix underlying the tumor cells and denser cells that comprise the primary tumor, as measured by atomic force microscopy (AFM) of tumor tissue sections . Stiffer tumor tissue would likely be less responsive to crowding pressures that normal cells promote extrusion in healthy tissue, thereby preventing their death by extrusion. Alternatively, AFM measurements have also determined that metastatic tumor cells isolated from pleural effusions are softer than neighboring benign or primary tumor cells . A comparatively softer cell might have a higher probability of getting extruded from the tumor or epithelium from which it originates and allow it to invade and migrate through the body better. In support of this, increased physical pressure on tumors can promote the shedding of cells , and mechanical stimulation enhances cancer cell invasion [65, 66]. Moreover, tumor cells can be more contractile than their normal cells  and may be primed to extrude at a lower threshold than normal. Changes in density of tumor cells could even work together to promote invasion. Primary tumors could induce enough pressure to promote extrusion and metastasis of softer tumor cells.
A similar concept to how differences in cells could selectively kick out some cell types at the expense of others is exemplified genetically in the phenomenon of cell competition, seen in Drosophila and cell culture epithelia [68, 69]. In Drosophila, mosaic patches of cells with compromised growth  or altered polarity  will be eliminated by surrounding wild-type tissue through cell competition. Conversely, patches of mutant cells that have enhanced growth rates compared to surrounding wild type cells can act as super competitors to eliminate their neighbors [72, 73]. During cell competition, winning cells could promote extrusion of less fit cells by causing increased pressure through faster growth and proliferation. Mathematical models also support this idea, as patches of mutant clones with enhanced growth cause “pressure-induced apoptosis” on surrounding neighbors , a process that may likely be due to crowding-induced live cell extrusion. In addition to altering the ability of a cell to sense crowding, mutations in the extrusion pathway could alter the ability of a cell to extrude in response to increased pressures. For instance, several regulators of the extrusion pathway, such as sphingosine kinase-1, a precursor to S1P, S1P receptor 2, and RhoA and C, are misregulated in numerous tumors [74–79]. Future work will determine how alterations in the extrusion pathway could cause cells to accumulate or promote invasion.
Although extrusion of live cells normally promotes their death by anoikis, it is important to note that most aggressive, metastatic tumors upregulate survival signaling to override anoikis [33, 80, 81], a property which allows them to colonize elsewhere in the body. When cells can no longer die by anoikis, the direction a cell extrudes could have a dramatic impact on its later fate. Typically, cells extrude apically into the lumen [30, 35, 40, 41], which is a dead space, so even if cells continue to survive after extrusion, they would still be essentially eliminated by extrusion. In this way, extrusion could act to suppress tumor formation. In fact, cells expressing oncogenic K-Ras or Src, or mutant DPP/BMP preferentially get eliminated from epithelia in cell culture and from the epidermis of developing Drosophila and zebrafish [42–46]. On the other hand, if cancer cells with upregulated survival signaling extrude basally underneath the epithelium [40, 41], they could potentially invade the stroma and gain access to other sites in the body (Figure 4B). Interestingly, loss or mutation of the tumor suppressor adenomatous polyposis coli (APC) shifts cells to extrude predominantly basally , suggesting an additional way wild-type APC may act to suppress tumor formation. Therefore, aberrations in any step of sensing crowding, activating extrusion, or ensuring that extrusion occurs in the correct apical direction could contribute to the formation or progression of tumors.
Epithelia provide a barrier to the outside world, and disruption of this barrier results in a variety of diseases. While hyper-immune response is often thought to be the basis of diseases such as asthma, coeliac disease, and irritable bowel syndrome (IBS), an alternate hypothesis is that the primary cause of these diseases is poor epithelial barrier function. In support of this idea, many of these diseases are associated with mutations in cell-cell adhesion genes [82–86]. Epithelia in these situations are thought to have poor barrier function simply due to poor adherens and tight junctions. However, alterations in cell adhesion could also affect the ability of a cell to extrude effectively. Additionally, mutations in the extrusion pathway could disrupt the ability of an epithelium to extrude in response to pathogens or insults that trigger cell death. For instance, perturbation of Rho activity downstream of the apoptotic pathway can block cell extrusion but still allow death to occur [35, 39, 40], which could produce gaps in the epithelial layer (Figure 4C). Loss of epithelial barrier function in colitis or asthma would then lead to inflammation and further exacerbation of the disease.
Another possibility is that epithelial barrier lesions could arise from excessive extrusion. For example, bronchoconstriction of airway smooth muscle during asthma may lead to excessive over-crowding of the attached bronchial epithelia. Excess epithelial crowding could cause hyper-extrusion could lead to the epithelial denuding seen in asthma [87–90]. The resulting poor barrier could then lead to increased inflammation and infection commonly seen in asthmatics following an attack of bronchoconstriction [91–93]. Further, excess extrusion caused by pathogens such as Vibrio parahaemolyticus may contribute to the disintegration and inflammation seen in intestinal villi . Future studies will define what roles, if any, extrusion plays in the pathobiology of epithelial barrier diseases.
Despite its importance to human physiology and disease, how epithelial tissues maintain overall cell numbers has been a mystery. Previous studies implicated a role for cell contacts controlling proliferation of cells. Recent findings now add a role for epithelial cell densities in controlling cell division and death, suggesting that crowding forces and relative levels of cell contacts may be critical for regulating overall numbers. This emerging view of an old problem is also revealing new molecules that might translate changes in density, force, and contacts into whether a cell will live or die. These findings should, hopefully, also bring new models for how epithelial diseases initiate and new targets to treat or prevent these diseases.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.