SUMMARY
Cell extrusion is a process of cell elimination in which a cell is squeezed out from its tissue of origin. Extrusion occurs in organisms as diverse as sponges, nematodes, insects, fish and mammals. Defective extrusion is linked to many epithelial disorders, including cancer. Despite broad occurrence, cell-intrinsic triggers of extrusion conserved across phyla are generally unknown. We combined genome-wide genetic screens with live-imaging studies of C. elegans embryos and mammalian epithelial cultures and found that S-phase arrest induced extrusion in both. Cells extruded from C. elegans embryos exhibited S-phase arrest, and RNAi treatments that specifically prevent S-phase entry or arrest blocked cell extrusion. Pharmacological induction of S-phase arrest was sufficient to promote cell extrusion from a canine epithelial monolayer. Thus, we have discovered an evolutionarily conserved cell-cycle-dependent trigger of cell extrusion. We suggest that S-phase-arrest induced cell extrusion plays a key role in physiology and disease.
INTRODUCTION
During development and homeostasis, cells are eliminated by a variety of mechanisms. One such mechanism is cell extrusion, in which a cell is expelled from a layer of cells while the continuity of the layer is maintained. Cell extrusion has been observed in and studied using a wide variety of organisms, including the sponge H. caerulea, C. elegans, D. melanogaster, zebrafish and mammals, suggesting that cell extrusion is an evolutionarily conserved mechanism of eliminating unnecessary or harmful cells (De Goeij et al., 2009; reviewed by Gudipaty and Rosenblatt, 2017 and Ohsawa et al., 2018). Vertebrate epithelial tissues use cell extrusion as the primary mode of cell elimination (Gu and Rosenblatt, 2012; Günther and Seyfert, 2018). Cell extrusion plays a key role in epithelial defense mechanisms that remove oncogene-transformed cells from epithelial layers (reviewed by Kajita and Fujita, 2015). Excessive cell extrusion can produce epithelial layer breaches, like those observed in asthma and Crohn’s disease (Gudipaty and Rosenblatt, 2017). Decreased cell extrusion leads to the formation of epithelial cell masses and confers resistance to cell death (Eisenhoffer et al., 2012; Gu et al., 2015). Intestinal polyps, which can develop into colon cancers, lack clearly identifiable cell extrusions (Eisenhoffer et al., 2012), suggesting that extrusion might be important for the prevention of polyps and intestinal cancers. Disruption and subversion of the cell-extrusion process likely promotes tumor growth and metastasis in pancreatic, lung and colon cancer (Gu et al., 2015).
While several mechanisms of cell extrusion have been described for Drosophila and vertebrates, these mechanisms have focused on the cell-cell interactions and cellular contexts, such as crowding, topological defects, cell competition, etc., that induce cell extrusion (reviewed by Fadul and Rosenblatt, 2018, Ohsawa et al., 2018). A cell-intrinsic trigger that can induce extrusion in organisms of different phyla has not been identified.
C. elegans is an excellent organism for the study of evolutionarily conserved mechanisms of cell elimination. The discovery of a conserved set of genes regulating caspase-mediated apoptosis in C. elegans has been fundamental to the understanding of programmed cell death in metazoa (reviewed by Fuchs and Steller, 2011). Cell extrusion can eliminate cells fated for death in C. elegans (Denning et al., 2012). Embryos with mutations in the caspase-mediated apoptosis pathway, e.g. loss-of-function mutants of the caspase gene ced-3, eliminate by extrusion a subset of cells that are otherwise eliminated by caspase-mediated apoptosis and engulfment in wild-type embryos. Denning et al. (2012) determined that the PAR-4 – PIG-1 (mammalian homologs LKB1 – MELK) kinase cascade is required for cell extrusion by C. elegans ced-3(lf) embryos. However, LKB1 (mammalian homolog of PAR-4) was found to prevent extrusion from mouse embryos (Krawchuk et al., 2015), indicating that LKB1/PAR-4 is likely not a driver but a regulator of cell extrusion in nematodes and mammals. No “caspase-equivalent” cell-intrinsic driver of cell extrusion is known.
To seek a conserved cell-intrinsic driver of extrusion, we first comprehensively identified genes and pathways that control cell extrusion by C. elegans and then tested the corresponding pathways for a role in mammalian cell extrusion. Briefly, we performed a genome-wide RNAi screen for defective cell extrusion by C. elegans and used confocal microscopy to analyze the effect of RNAi against the identified genes on the cell extrusion process. From this analysis, we found that cell extrusion by C. elegans requires cell-autonomous cell cycle entry and subsequent S-phase arrest and that circumventing S-phase arrest blocks extrusion. We then tested and confirmed that pharmacological induction of S-phase arrest with hydroxyurea (HU) (Timson, 1975; Bianchi et al., 1983) promotes cell extrusion of mammalian epithelial cells. We conclude that S-phase arrest is a conserved cell-intrinsic trigger of cell extrusion in C. elegans and mammals.
RESULTS
Genome-wide RNAi screen identified cell-cycle genes as candidate regulators of cell extrusion
In wild-type C. elegans embryos, 131 cells are eliminated by caspase-mediated apoptosis and engulfment. By contrast, in C. elegans ced-3 caspase mutants, a few of the cells that would normally undergo programmed cell death instead are extruded from the developing embryo (Denning et al., 2012). Of the approximately six cells extruded from ced-3(lf) embryos, the cell ABplpappap is most frequently extruded (Denning et al., 2012 and unpublished). If extrusion fails to occur, ABplpappap (or descendants of ABplpappap) survive(s) and differentiate(s) into one (or two) supernumerary excretory cell(s), producing mutant animals with the two (or three)-excretory-cell (Tex) phenotype; by contrast both wild-type and ced-3(lf) animals have one excretory cell (Denning et al., 2012; Figure 1A). Mutations that reduce ABplpappap extrusion and produce the Tex phenotype also reduce the extrusion of other cells (Denning et al., 2012), making the Tex phenotype a convenient marker for defective cell extrusion.
We performed a genome-wide RNAi screen for the Tex phenotype in ced-3(lf) animals expressing the GFP excretory-cell reporter Ppgp-12::4xNLS::GFP (Figure 1B; Denning et al., 2012). We screened 11,511 RNAi clones (targeting about 55% of the ∼20,000 C. elegans genes by feeding (Rual et al., 2004)) and found 30 clones targeting 27 unique genes that consistently produced a Tex phenotype. Three RNAi clones identified genes previously reported to function in cell extrusion, grp-1, arf-1.2 and arf-3 (Denning et al., 2012), confirming that this RNAi screen could identify cell extrusion mutants. Unexpectedly, 10 of the RNAi clones targeted genes that control cell-cycle progression (Figure 1C), suggesting a possible role for the cell cycle in controlling cell extrusion.
We then tested a nearly complete set of C. elegans cell cycle genes for functional roles in cell extrusion (van den Heuvel, 2005) using an RNAi library of 61 publically available and, when necessary, newly generated clones with each clone targeting a unique gene (Kamath et al., 2003; Rual et al., 2004; Materials and Methods). We found that RNAi against four additional cell cycle genes produced a Tex phenotype (Figure 1C, Supplemental Table 1).
Most of the 14 cell-cycle genes we identified are well-characterized regulators of S-phase entry and progression. These genes include cdc-25.2, which encodes a homolog of the CDK-activating phosphatase CDC25 (Lee et al., 2016); cdk-1 and cdk-2, which encode homologs of CDK1 and CDK2, respectively (Boxem, 2006); cye-1 and cya-1, which encode homologs of S-phase cyclins E and A, respectively (Fay and Han, 2000; Kreutzer et al., 1995); and psf-1, psf-2 and psf-3, which encode homologs of pre-replicative and replicative complex components PSF1, PSF2 and PSF3, respectively (Ossareh-Nazari et al., 2016; Figure 1C). All of these cell cycle genes with S-phase function are required for C. elegans viability. However, it is unlikely that a general reduction in embryonic fitness causes the Tex phenotype, as RNAi against essential genes involved in other phases of the cell cycle (e.g., metaphase-to-anaphase transition genes mat-1, mat-2, etc. (reviewed by Yeong, 2004)) or those involved in transcription (e.g., cdk-7 and cdk-9 (Wallenfang and Seydoux, 2002; Bowman et al., 2013)) did not produce the Tex phenotype despite producing extensive lethality (Supplemental Table 1). Furthermore, as with RNAi targeting pig-1 or other genes generally required for cell extrusion (Denning et al., 2012), RNAi against 13 of the 14 identified cell cycle genes produced the Tex phenotype in ced-3(lf) mutants but not in wild-type animals (Supplemental Table 2). Altogether, we found that defects in S-phase entry and progression are specifically associated with a synthetic ced-3-dependent Tex phenotype, suggesting a role for these cell-cycle genes in promoting cell extrusion.
S-phase entry genes function to promote cell extrusion
Whereas defects in the extrusion of ABplpappap cause the Tex phenotype, it is possible that the supernumerary excretory cell(s) of some mutants could arise from other cell lineages. To directly determine whether genes functioning in S-phase entry are important for cell extrusion, we used time-lapse confocal microscopy to monitor the extrusion of ABplpappap from ced-3(lf) embryos deficient in cye-1 or cdk-2. To assess extrusion events, we imaged live embryos with ABplpappap in focus over a 50-minute period ending at the completion of ventral enclosure, when epidermal cells meet at the ventral midline following a dorsolateral migration; ventral enclosure coincides with the cell extrusion that occurs in ced-3(lf) embryos (Denning et al., 2012). For clarity, we refer below to embryos from ced-3(lf) parents treated with RNAi against a gene, say gene-x, as “gene-x(RNAi) embryos” and to embryos from ced-3(lf) parents treated with RNAi against the empty vector as “control embryos.”
In control embryos, cells neighboring ABplpappap on the ventral surface gradually disappeared from view until ABplpappap was left completely isolated (Figure 2A, Movie 1), indicating that ABplpappap had been extruded from the embryo. By comparison, in cye-1(RNAi) embryos (Figure 2B, Movie 2) or cdk-2(RNAi) embryos (Figure 2C, Movie 3), ABplpappap was surrounded by cells throughout the imaging period and hence remained within the embryo, failing to detach. Using dorso-ventral confocal sections from the live embryo imaging, we reconstructed sagittal views of the embryos during ventral enclosure. These views confirmed that ABplpappap was extruded ventrally from 10 of 11 control embryos (Figure 2D, Supplemental Figure 1), whereas ABplpappap failed to detach and was incorporated into the body of cye-1(RNAi) embryos (11 of 11 embryos; Figure 2E, Supplemental Figure 2) or cdk-2(RNAi) embryos (10 of 11 embryos; Figure 2F, Supplemental Figure 3). These findings demonstrate that the S-phase entry genes cye-1 and cdk-2 are required for ABplpappap extrusion in ced-3 embryos.
Entry into S phase is required for and precedes cell extrusion
Since genes that promote S-phase entry are required for ABplpappap extrusion, we tested if cells that undergo extrusion enter S phase. We used a previously characterized reporter transgene that expresses a truncated human DNA Helicase B (tDHB)-GFP fusion protein optimized for expression in C. elegans and that changes its intracellular location in response to CDK1 and CDK2 activity (van Rijnberk et al., 2017; Spencer et al., 2013). tDHB-GFP is enriched in the nuclei of quiescent or post-mitotic cells, whereas it exhibits an increasing cytoplasmic bias as cells progress from S-phase through mitosis (Figure 3A; Spencer et al., 2013). In control embryos, tDHB-GFP was mostly absent from the ABplpappap nucleus (10 of 10 embryos) both before ventral enclosure (Figure 3B) and as it was extruded (Figure 3E), indicating that ABplpappap entered S phase prior to its extrusion during the period of ventral enclosure. Cells extruded from other sites of the embryo also displayed low levels of nuclear tDHB-GFP (Figures 3I-L). By contrast, in cye-1(RNAi) embryos (Figures 3C, 3F) or cdk-2(RNAi) embryos (Figures 3D, 3G) the ABplpappap nucleus scored positive for the tDHB-GFP fusion protein during the period around ventral enclosure (10 of 10 embryos each for each RNAi treatment), suggesting that RNAi against these genes prevented both the entry of ABplpappap into S-phase and its extrusion. Quantification of the nuclear-to-cytoplasmic ratio of tDHB-GFP fluorescence intensity in ABplpappap in control, cye-1(RNAi) and cdk-2(RNAi) embryos at varying stages with respect to ventral enclosure confirmed these observations (Figure 3H). Some cells were still extruded in cye-1(RNAi) and cdk-2(RNAi) embryos, likely reflecting incomplete inhibition of gene function by RNAi. Consistently, such cells displayed low levels of nuclear tDHB-GFP, indicating that those cells entered the cell cycle (Supplemental Figure 4). Thus, cell-cycle entry appears to be a functionally critical step in the process of cell extrusion.
To define more precisely the cell-cycle phase that facilitates cell extrusion, we used a second reporter transgene (GFP::PCN-1), which expresses an N-terminal translational fusion of GFP to the C. elegans homolog of the DNA replication processivity factor PCNA (Brauchle et al., 2003). PCNA in mammalian cells and early C. elegans embryonic cells exhibits a punctate, sub-nuclear localization only during S-phase (Figure 4A; Brauchle et al., 2003; Zerjatke et al., 2017). The localization pattern of GFP::PCN-1 in cell cycles of cells close to ABplpappap on the embryonic ventral surface matched that described for early embryonic cells and contrasted with the continuous accumulation of GFP::PCN-1 observed during the C. elegans germline cell cycle (Supplemental Figure 5; Movie 4; Brauchle et al., 2003; Kocsisova et al., 2018). We found that in control embryos GFP::PCN-1 was localized in bright sub-nuclear foci in ABplpappap immediately prior to the initiation of ventral enclosure, indicating that this cell was in S phase (5 of 5 embryos) (Figure 4B). By contrast, ABplpappap showed a diffuse nuclear localization of GFP::PCN-1 in cye-1(RNAi) (Figure 4C, 4F) and cdk-2(RNAi) (Figure 4D, 4G) embryos, both before (5 of 5 embryos each) and after ventral enclosure (5 of 5 embryos each), indicating a failure to enter the cell cycle. We conclude that cells to be extruded must enter S phase for extrusion to occur and that blocking S-phase entry prevents cell extrusion.
Extruding cells exhibit an arrested S phase and experience replication stress
Given that cells required cell-cycle entry for extrusion, we examined the extent of cell-cycle progression in these cells as they were extruded. In control embryos, we found that GFP::PCN-1 was localized in bright sub-nuclear foci in ABplpappap both before (5 of 5 embryos) and after extrusion (5 of 5 embryos) (Figure 4B, 4E), indicating that ABplpappap entered but did not exit S phase. We observed no significant changes of GFP::PCN-1 localization in ABplpappap up to and after its extrusion over a period of 35 min (Figure 4H), indicating that it remained arrested in S phase. A second unidentified extruding cell showed a similarly unchanging GFP::PCN-1 localization pattern (Figure 4H). To determine if the S-phase arrest observed in ABplpappap and the other extruding cell is a general feature of cell extrusion, we examined other extruded cells in the embryo. Nearly all cells extruded by ced-3(lf) embryos displayed bright sub-nuclear foci of GFP::PCN-1, consistent with an arrested S phase in these cells (Figures 4H-J).
Arrest of DNA replication during S phase can occur as a result of replication stress, which can arise for many different reasons (reviewed by Zeman and Cimprich, 2014). Replication stress triggers the replication stress response, which stabilizes stalled replication forks, halts cell-cycle progression and prevents further firing of replication origins (reviewed by Zeman and Cimprich, 2014). As cells undergoing extrusion are arrested in S-phase, we asked if triggering the replication stress response was important for extrusion. Core components of the replication stress response pathway in C. elegans and other metazoans include ATR, Chk1, Rad17, Rad9, Rad1, Hus1, Replication Protein A, TopBP1, Timeless, Tipin and Claspin proteins (Stevens et al., 2016; Yazinski and Zou, 2016). RNAi against 7 of the 11 C. elegans genes encoding these core components of the replication-stress checkpoint (atl-1, chk-1, hpr-9, mus-101, tim-1, tipn-1 and clsp-1) produced a Tex phenotype in ced-3(lf) animals (Figure 1C, Figure 4L), indicating an involvement of replication stress response in cell extrusion. These findings suggest that the replication stresses underlying the S-phase arrest in extruding cells trigger the replication stress response, which promotes cell extrusion.
Bypassing S-phase arrest and completing the cell cycle prevents cell extrusion
After determining that S-phase arrest is a key feature of cell extrusion, we asked whether the previously identified C. elegans cell extrusion regulators pig-1 (homolog of the mammalian kinase gene MELK) and grp-1 (homolog of the mammalian ARF GEF gene CYTH3) (Denning et al., 2012) also play a role in producing this S-phase arrest. We monitored the fate of ABplpappap in pig-1(RNAi) embryos by time-lapse confocal microscopy. Strikingly, we found that instead of undergoing S-phase arrest, ABplpappap completed the cell cycle and divided before ventral enclosure in these embryos (Figure 5A; Movie 5). By examining virtual lateral sections, we found that instead of undergoing extrusion, as in the case of control embryos (Figure 5B), ABplpappap in pig-1(RNAi) embryos divided to generate daughters that were not extruded (5 of 6 embryos) (Figure 5C). The same fate of ABplpappap was observed in grp-1(RNAi) embryos (5 of 5 embryos) (Figure 5D). These findings indicate that in addition to the genes cye-1 and cdk-2, the genes pig-1 and grp-1 are required to produce the S-phase arrest that precedes cell extrusion.
Next, we investigated why ABplpappap completed the cell cycle in pig-1(RNAi) and grp-1(RNAi) embryos. In several cell lineages, such as the Q neuroblast cell lineage, the genes pig-1 and grp-1 are required for unequal cell divisions that generate apoptotic cells (Cordes et al., 2006; Teuliere et al., 2014). Consistent with their function in controlling unequal cell division, RNAi against each of the genes pig-1 and grp-1 perturbed the ratio of ABplpappap’s size to that of its sister and generated an abnormally large ABplpappap (Figures 5E,F; Supplemental Figures 6A,D,E). These findings indicated that unequal cell division plays an important role in producing the S-phase arrest that precedes cell extrusion. However, despite the requirement for cye-1 and cdk-2 to produce this S-phase arrest (Figures 3B, 4B), RNAi against cye-1 or cdk-2 did not affect unequal cell division. Neither the size of ABplpappap relative to that of its sister nor the absolute size of ABplpappap showed a difference among cye-1(RNAi), cdk-2(RNAi) and control embryos (Figures 5G,H; Supplemental Figure 6A,B,C). Together, our data indicate that genes required for cell extrusion function to produce the S-phase arrest preceding extrusion either by promoting S-phase entry (via cye-1 and cdk-2) or by preventing cell-cycle completion (via pig-1 and grp-1).
Another difference consistent with the distinct ways by which the cell-cycle genes and unequal-cell-division genes promote cell extrusion was observed in the Tex phenotype caused by RNAi against these genes. Some pig-1(RNAi) and grp-1(RNAi) animals with the Tex phenotype displayed two supernumerary excretory cells (Figure 5I), presumably because both daughters of ABplpappap adopted the excretory-cell in these animals. By contrast, cye-1(RNAi) and cdk-2(RNAi) animals with the Tex phenotype displayed only one supernumerary excretory cell (Figure 5I), presumably because ABplpappap did not enter and then complete the cell cycle but rather differentiated directly into an excretory cell.
In short, these findings indicate that the unifying feature of all genes required for cell extrusion is that they function to produce the S-phase arrest observed in cells to be extruded, supporting the conclusion that S-phase arrest is a key requirement of cell extrusion. Either preventing cell-cycle entry or bypassing the S-phase arrest to complete cell division prevented cell extrusion in developing C. elegans embryos.
S-phase arrest drives cell extrusion from mammalian epithelial cultures
Since S-phase arrest is the central feature of cell extrusion by C. elegans embryo, we asked if S-phase arrest also induces mammalian cell extrusion. We used a monolayer of Madin-Darby Canine Kidney (MDCK) cells as a model of mammalian epithelia and hydroxyurea (HU) as a chemical agent for inducing S-phase arrest. MDCK cells are a simple epithelial system for studies of mammalian cell extrusion in culture (reviewed by Ohsawa et al., 2018). HU causes S-phase arrest by inhibiting the enzyme ribonucleotide reductase and depleting deoxyribonucleotides during DNA replication, resulting in stalled DNA replication forks (Timson, 1975; Bianchi et al., 1983). We treated MDCK monolayers with either 2 mM HU or vehicle (negative control) for up to 24 h and obtained time-lapse micrographs of the monolayers from this period to assess cell extrusion. Strikingly, cells extruded apically from the MDCK monolayer into the culture medium were three to four times higher in number after HU treatment when compared to the number of cells extruded apically after vehicle treatment (Figure 6A, 6B; Movie 6,7). Next we used MDCK-Fucci cells to determine the cell cycle phase distribution of HU- and vehicle-treated extruded cells (Streichan et al., 2014). MDCK-Fucci cells produce a fluorescent signal that varies with the phase of the cell cycle (G0/G1-Red, S/G2/M-Green; Sakaue-Sawano et al., 2008). As expected, most of the HU-treated extruded cells displayed a green fluorescent signal (Figure 6C), indicating that they were in a cell cycle phase subsequent to the onset of DNA replication. We noted that stochastically extruded cells (from vehicle treatment) mostly exhibited red fluorescence indicative of the G0 or G1 phase (Figure 6D), consistent with the phase of cells that are naturally extruded from post-mitotic zones, such as the tips of intestinal villi (Carroll et al., 2018; Eisenhoffer et al., 2012).
Since HU is known to increase the rate of apoptosis (Timson, 1975), and agents that promote apoptosis increase the rate of cell extrusion (Andrade and Rosenblatt, 2011), we examined the role of apoptosis in HU-induced cell extrusion. Surprisingly, the fraction of extruded MDCK cells that were apoptotic was not increased by HU treatment (Figure 6D), indicating that HU-induced cell extrusion is not a consequence of an increase in the rate of apoptosis. In addition, we reseeded the extruded MDCK cells in fresh media to measure their viability following HU treatment. We found that the number of viable adherent cells at 2 hours post reseeding was proportional to the number of extrusions for both the HU- and vehicle-treated groups (Figure 6E). Additionally, the number of HU-treated cells doubled at 24 hours post reseeding when compared to 2 hours post reseeding (Figure 6E). Thus, cells extruded by HU treatment were not only viable but also able to resume and complete the cell cycle.
Taken together, the above findings indicate that S-phase arrest drives the extrusion of cells from mammalian epithelia and establish that cell extrusion caused by S-phase arrest is an evolutionarily conserved phenomenon.
DISCUSSION
Here we report that cell cycle S-phase arrest is a cell-intrinsic trigger of cell extrusion and can induce the extrusion of cells of organisms from two divergent branches of the phylogenetic tree - nematodes and mammals. Using RNAi screens and transgenic cell-cycle reporters, we determined in developing C. elegans embryos that perturbations that prevented S-phase arrest blocked cell extrusion. As summarized in Figure 7A, cells destined for extrusion from ced-3(lf) embryos are always the smaller daughters of unequal cell divisions. These smaller daughter cells enter S phase of the cell cycle and undergo S-phase arrest, likely because of a deficiency in the energetic and metabolic resources required for DNA synthesis (e.g., nucleotides, replication proteins, etc.). Thus, either bypassing S-phase arrest (by perturbing the process of unequal cell division) and hence completing cell division or preventing S-phase arrest (by blocking cell-cycle entry) prevents cell extrusion in C. elegans embryos.
To test the generality of our findings, we used mammalian epithelia treated with HU and showed that S-phase arrest is sufficient to promote cell extrusion. Thus, cell extrusion triggered by S-phase arrest is an evolutionarily conserved mechanism of cell elimination. These observations also demonstrate that cells can be extruded from mitotically active mammalian epithelial tissues. Previous studies of mammalian cell extrusion focused on extrusion from post-mitotic tissues (Rosenblatt et al., 2001; Eisenhoffer et al., 2012; Gudipaty et al., 2017; Saw et al., 2017; Kocgozlu et al., 2016) or oncogene-driven extrusion from growth-suppressing epithelial environments (Anton et al., 2018; Hogan et al., 2009; Kajita et al., 2010; Leung and Brugge, 2012; Slattum et al., 2014; Wu et al., 2014).
Our mechanistic model for the evolutionarily conserved process by which S-phase arrest promotes cell extrusion is presented in Figure 7B. In this model, a mitotically active cell destined for extrusion (i) enters the cell cycle, (ii) arrests in S phase, (iii) loses cell adhesion (see below), and (iv) is extruded as a result of reduced adhesion and forces generated by external morphological or physiological processes.
Why are S-phase arrested cells susceptible to extrusion?
We observed previously that cells extruded by C. elegans embryos do not express the classical E-cadherin HMR-1 and other cell-adhesion molecules (Denning et al., 2012). The absence of such adhesion molecules likely allows cells to be squeezed out of the embryo by morphological forces generated by migrating hypodermal cells and neighboring neuroblasts during ventral enclosure (Chisholm and Hardin, 2005; Wernike et al., 2016). Consistent with this view, the loss of E-cadherin-mediated adhesion caused by cleavage of the extracellular part of a cell’s E-cadherin molecules is sufficient to drive cell extrusion from an MDCK monolayer (Grieve and Rabouille, 2014). We speculate that a signaling pathway initiated by S-phase arrest downregulates the expression of cell adhesion molecules in cells destined for extrusion. Indeed, in HeLa cell cultures, cell adhesion increases during S phase via the activity of the cell-cycle regulator CDK1 and decreases later in the cell cycle upon inhibition of CDK1 by Wee1 (Jones et al., 2018). Interestingly, activation of the replication stress response, which is required for cell extrusion in C. elegans (Figure 1C, 4L), blocks CDK1 activity (Jin et al., 2003; Mailand et al., 2002; Xiao et al., 2003). Furthermore, HU-mediated S-phase arrest also inactivates CDK1 in MDCK cells (Anton et al., 2018). We therefore propose that a reduction in CDK1 activity following S-phase arrest decreases cell adhesion, thereby facilitating the extrusion of cells subjected to external forces from cellular neighbors.
Extrusion of S-phase arrested cells is likely tumor-suppressive
Replication forks under prolonged S-phase arrest can collapse and produce DNA damage, genomic rearrangements and ploidy defects, all of which are associated with oncogenesis (reviewed by Gaillard et al., 2015). The human genes that promote replication stress and S-phase arrest are frequently amplified, overexpressed or activated by mutations in tumors (Otto and Sicinski, 2017). Cells in such tumors experience persistent replication stress that can lead to S-phase arrest (Gaillard et al., 2015). Hence, tumor cells and cells with oncogenic potential might well be eliminated via cell extrusion, in which case the extrusion of cells arrested in S phase would be tumor-suppressive. We propose that cell extrusion driven by S-phase arrest is a checkpoint mechanism that functions to eliminate cells at all stages of the oncogenic transformation process, ranging from precancerous cells in S-phase arrest to tumor cells in a malignant tumor.
Subversion of cell extrusion driven by S-phase arrest might contribute to metastasis
Inactivation of either S1P2 or the tumor suppressor APC or expression of oncogenic K-Ras changes the direction of cell extrusion from apical, which favors cell elimination by extrusion into the lumen, to basal, which favors dissemination of extruded cells to surrounding tissue (Gu et al., 2015; Marshall et al., 2011; Slattum et al., 2014). Mutations in these genes are hallmarks of metastatic tumors. While cell extrusion caused by S-phase arrest likely can suppress tumor development, the same mechanism of cell extrusion might paradoxically promote cancer metastasis if the extrusion direction changes from apical to basal. We observed that cells subjected to HU-mediated extrusion failed to die (Figure 6D) and instead were capable of reentering the cell cycle and proliferating (Figure 6E). Thus, basal extrusion of cells arrested in S-phase might facilitate metastasis by disseminating live tumor cells arrested in S phase to other tissues and organs. We propose that mutations in genes, such as S1P2, APC and K-Ras, promote metastasis by facilitating the basal extrusion and spread of tumor cells arrested in S phase.
In summary, we have discovered a novel conserved mechanism that links a cell-cycle vulnerability to the process of cell extrusion. We suggest that cell extrusion mediated by S-phase arrest is a mechanism of cell elimination common to all metazoa. These findings have implications for the field of cancer biology, as cell extrusion caused by S-phase arrest likely regulates both the survival and spread of tumor cells.
AUTHOR CONTRIBUTIONS
H.R.H. supervised the project. V.K.D. and H.R.H. conceptualized the project. V.K.D. and H.R.H. designed the experiments that used C. elegans. V.K.D. and R.D. performed the experiments that used C. elegans. V.K.D., R.D. and D.P.D. generated reagents. C.P. and J.R. designed the experiments that used mammalian cells. C.P. performed the experiments that used mammalian cells. V.K.D., D.P.D. and H.R.H. wrote the original manuscript draft. All authors contributed to data analysis, interpretation, and reviewing and editing of the manuscript.
Declaration of Interests
The authors declare no competing interests.
Supplemental Movies
Movie 1 Control embryos extrude ABplpappap as it undergoes ventral enclosure
Time-lapse video of a ced-3(lf); control(RNAi) embryo undergoing ventral enclosure over a period of 50 minutes shows ABplpappap (circled at the beginning and end of video) was extruded from this embryo. All nuclei are labeled with GFP and membranes of egl-1–expressing cells are labeled with mCherry (magenta). Time-lapse images used for this video were obtained using confocal microcopy. Video playback is at 600x real speed. The embryo shown carried the transgenes stIs10026 and nIs632.
Movie 2 cye-1(RNAi) embryos do not extrude ABplpappap as it undergoes ventral enclosure
Time-lapse video of a ced-3(lf); cye-1(RNAi) embryo undergoing ventral enclosure over a period of 50 minutes shows ABplpappap (circled at the beginning and end of video) was not extruded from this embryo. All nuclei are labeled with GFP and membranes of egl-1–expressing cells are labeled with mCherry (magenta). Time-lapse images used for this video were obtained using confocal microcopy. Video playback is at 600x real speed. The embryo shown carried the transgenes stIs10026 and nIs632.
Movie 3 cdk-2(RNAi) embryos do not extrude ABplpappap as it undergoes ventral enclosure
Time-lapse video of a ced-3(lf); cdk-2 (RNAi) embryo undergoing ventral enclosure over a period of 50 minutes shows ABplpappap (circled at the beginning and end of video) was not extruded from this embryo. All nuclei are labeled with GFP and membranes of egl-1–expressing cells are labeled with mCherry (magenta). Time-lapse images used for this video were obtained using confocal microcopy. Video playback is at 600x real speed. The embryo shown carried the transgenes stIs10026 and nIs632.
Movie 4 GFP::PCN-1 shows continuous change in fluorescence intensity during embryonic cell cycles
Time-lapse video of a ced-3(lf) embryo expressing GFP::PCN-1 in all cells shows continuous change in GFP::PCN-1 fluorescence intensity as cells progress through the cell cycle, similar to that observed for early embryonic cell cycles (Brauchle et al., 2003). Membranes of egl-1–expressing cells are labeled with mCherry (magenta). Time-lapse images used for this video were obtained using confocal microcopy. Video playback is at 180x real speed. The embryo shown carried the transgenes isIs17 and nIs861.
Movie 5 ABplpappap divides before ventral enclosure and is not extruded in pig-1(RNAi) embryos
Time-lapse video of a ced-3(lf); pig-1 (RNAi) embryo undergoing ventral enclosure over a period of 57 minutes shows ABplpappap (circled at the beginning) divided to generate daughters (circled at the end of video) before ventral enclosure was complete in this embryo. All nuclei are labeled with GFP and membranes of egl-1–expressing cells are labeled with mCherry (magenta). Time-lapse images used for this video were obtained using confocal microcopy. Video playback is at 600x real speed. The embryo shown carried the transgenes stIs10026 and nIs861.
Movie 6 A few cells are extruded from a vehicle-treated MDCK monolayer
A time-lapse video of mammalian MDCK monolayer treated with vehicle control for 21.25 h shows that a few cells are extruded during this period. Extruded cells can be identified as bright, white, rounded spots rising from the epithelial plane. Video playback is at 7200x real speed. Scale bar, 100 µm.
Movie 7 A large number of cells are extruded from an HU-treated MDCK monolayer
A time-lapse video of mammalian MDCK monolayer exposed to HU for 21.25 h shows that many more cells are extruded during this period as a result of HU treatment. Extruded cells can be identified as bright, white, rounded spots rising from the epithelial plane. Video playback is at 7200x real speed. Scale bar, 100 µm.
Materials and Methods
Plasmids
L4054 was a gift from Andrew Fire (Addgene plasmid # 1632; http://n2t.net/addgene:1632; RRID:Addgene_1632). pDD111 - Pegl-1::mCherry::PH::unc-54 3’UTR was generated with the following steps: i) 6.8 Kb of the egl-1 promoter was amplified from genomic DNA with Phusion DNA polymerase using the primers DPD660 and DPD661; ii) the amplicon was digested with PstI and SacI (New England Biolabs) and ligated into pPD122.56, which encodes 4xNLS::GFP to generate Pegl-1::4xNLS::GFP::unc-54 3’UTR; iii) mCherry-PH (Pleckstrin Homology) sequence was amplified from pAA173 using DPD647 and DPD648 and digested with EcoRI and AgeI (New England Biolabs) and ligated into the pDD122.56 - Pegl-1::4xNLS::GFP::unc-54 3’UTR, which generated the plasmid pDD122.56 - Pegl-1::4xNLS::mCherry::PH::unc-54 3’UTR; iv) the 4xNLS sequence was removed with the primers DPD695 and DPD696 using QuikChange Site-Directed Mutagenesis (Agilent) to generate pDD111 - Pegl-1::mCherry::PH::unc-54 3’UTR.
RNAi clones were constructed for atl-1, mat-2 and lin-15B. Genomic regions of about 1 kb were amplified from wild-type genomic lysates using Q5 Hot Start high-fidelity polymerase (New England Biolabs) with the following primers: atl-1
RD105 TCGAATTCCTGCAGCTCCTCGAACCCATCATCCCT
RD106 TGACGCGTGGATCCCATGAAGCTGCGTGGTTGTTG
mat-2
RD103 TCGAATTCCTGCAGCCTGGAACTCATCCCATACGC
RD104 TGACGCGTGGATCCCCATTGGAACCTCCAGATGCT
lin-15B
RD101 TCGAATTCCTGCAGCGCTGACACAATTGCGAACAT
RD102 TGACGCGTGGATCCCCGTGTGCATAAAGACCAAGG
These inserts were cloned into the pL4440 vector linearized with XmaI (New England Biolabs) using the In-Fusion HD cloning kit (TaKaRa) according to manufacturers instructions. The cloned vector was then transformed into competent HT115 bacterial cells. Correct RNAi clones were identified by Sanger sequencing. Geneious 10.2.6 (Biomatters, Inc.) was used to guide all plasmid design and construction.
Contact for Reagent and Resource Sharing
Further information and resource sharing requests should be directed to and will be fulfilled by the lead contact, H. Robert Horvitz (horvitz{at}mit.edu).
Strains, transgenes and mutations
C. elegans hermaphrodite strains were maintained on Nematode Growth Medium (NGM) plates containing 3 g/L NaCl, 2.5 g/L peptone and 17 g/L agar supplemented with 1 mM CaCl2, 1 mM MgSO4, 1 mM KPO4 and 5 mg/L Cholesterol with E. coli OP50 as a source of food (Brenner, 1974). All strains were derived from Bristol N2 and are listed in Table S2. ced-3(lf) refers to the n3692 deletion allele of ced-3 (Denning et al., 2012). C. elegans strains carrying the transgenes nIs861 and isIs17 were maintained at 25°C. All other strains were maintained at 22°C. The transgenes and mutations used are listed below:
LGI: nIs433[Ppgp-12::4xNLS::GFP::unc-54 3’UTR; p76-16B(unc-76(+))]
LGII: heSi192[Peft-3::tDHB::eGFP::tbb-2 3’UTR + Cbr-unc119(+)]
LGIII: unc-119(ed3)
LGIV: ced-3(n3692, n717)
LGV: unc-76(e911), ltIs44[Ppie-1::mCherry::PH(PLC1delta1) + unc-119(+)]
Unknown linkage: stIs10026[Phis-72::HIS-72::GFP], isIs17[pGZ295(Ppie-1::GFP::pcn-1(W03D2.4)), pDP#MM051 (unc-119(+))], nIs861[pDD111(Pegl-1::mCherry::PH::unc-54 3’UTR)], nIs632[pDD111(Pegl-1::mCherry::PH::unc-54 3’UTR), pML902 (dlg-1::GFP),p76-16B(unc-76(+))]
Extrachromosomal array: nEx2188[pDD111(Pegl-1::mCherry::PH::unc-54 3’UTR), pML902 (dlg-1::GFP), unc-76(+)]
nIs632 and nIs861 express membrane-localized mCherry from the egl-1 promoter, which facilitated the identification of ABplpappap (an egl-1 expressing cell). nIs632 does not express dlg-1::GFP, presumably as a result of partial transgene silencing (Hsieh et al. 1999; Grishok et al. 2005; Fischer et al. 2013). stIs10026 (Boeck et al., 2011) ubiquitously expresses a GFP-tagged histone HIS-72 from its endogenous promoter, which produces fluorescence in the nuclei of all cells and facilitates in providing the context in which extrusion events are observed.
Germline transformation
Transgenic lines were generated using the standard germline transformation procedure (Mello et al., 1991). Extrachromosomal array transgene nEx2188 was generated by injecting pML902 at 3 ng/µL, pDD111 at 40 ng/µL, p76-16B (unc-76(+)) at 60 ng/ul and 1Kb Plus DNA ladder (Thermo Fischer Scientific) at 50 ng/µl into ced-3(n717) IV; unc-76(e911) V double mutant animals. nIs632 was generated by gamma-ray irradiation (4,800 rads) of nEx2188-carrying L4 animals and was identified by the 100% transmission of the transgene from transformed parent to progeny. nIs861 was a spontaneous integration in a germline cell of an animal injected with pDD111 at 10 ng/µL and 1 kb DNA ladder at 90 ng/µL, and was identified by the 100% transmission of the transgene from transformed parent to progeny.
RNAi treatments and genome-wide RNAi screen
Previously described feeding RNAi constructs and reagents were used to perform RNAi feeding experiments (Fraser et al., 2000; Rual et al., 2004). Briefly, HT115 Escherichia coli bacteria carrying RNAi clones in the pL4440 vector were grown for at least 12 h in Luria broth (LB) liquid media with 75 mg/L ampicillin at 37°C. These cultures were seeded onto 6 cm Petri plates with Nematode Growth Medium (NGM) containing 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG) (Amresco) and 75 mg/L ampicillin and incubated for 24 h at 22°C. For imaging experiments using confocal microscopy, 10 L4 animals were added to each RNAi plate and imaging of progeny embryos was performed on the next day as described in Microscopy below. For excretory cell counts, five L4 animals were added to each RNAi plate and L3-L4 progeny were scored for number of excretory cells, as described in Excretory cell count below. In case a bacterial clone targeting a certain gene was not available in previously constructed libraries (Kamath et al., 2003; Rual et al., 2004), we generated our own RNAi clone as described in Molecular biology above.
The ORFeome RNAi library was used to conduct a genome-wide RNAi screen (Rual et al., 2004). For each day of the RNAi screen, all bacterial colonies from two 96-well plates were cultured for at least 12 h at 37°C in LB with 75 mg/L ampicillin. These cultures were then pre-incubated with 1 mM IPTG (Amresco) for 1 h to maximize induction of dsRNA production. 24-well plates with each well containing 2 mL NGM medium with 1 mM IPTG (Amresco) and 75 mg/L ampicillin were prepared in advance and stored at 4°C until needed; they were brought to room temperature a few hours before seeding. Each bacterial colony culture was then seeded onto an individual well of a 24-well plate and incubated for 24 h at 20°C. Three L4 animals were picked into a 10 µl drop of M9 medium, which facilitated their transfer into a well using a pipette. The progeny of these 3 animals were screened 3 days later. Each set of RNAi clones screened also included a pig-1 RNAi positive control and an empty pL4440 vector negative control. The scorer was blinded to the identity of the RNAi clones. Excretory cell counts were performed as described in Excretory cell counts below. Sanger sequencing was used to confirm the identity of RNAi clones that reproducibly generated a Tex phenotype for more than 10% of the animals scored.
Microscopy
All RNAi screens scoring excretory cells were performed using a Nikon SMZ18 fluorescent dissecting microscope. DIC and epifluorescence images were obtained using a 63x objective lens (Zeiss) on an AxioImager Z2 (Zeiss) compound microscope and Zen Blue software (Zeiss).
For confocal microscopy, embryos staged at the 200-300-cell stage were picked and mounted onto a glass slide (Corning) with a freshly prepared 2% agarose pad. Embryos with ventral surfaces facing the objective were selected for imaging. Confocal images were obtained using a 63x objective lens (Zeiss) on a Zeiss LSM800 confocal microscope.
For observing extrusion (or absence of extrusion), we focused particularly on the cell ABplpappap, the identification of which is facilitated by its central position on the ventral surface (Sulston et al., 1983). The fluorescent transgene nIs861[Pegl-1::mCherry::PH] or nIs632[Pegl-1::mCherry::PH; dlg-1::GFP], which express the Pleckstrin homology domain of PLC-δ fused to mCherry from the promoter of egl-1, was used to label the membrane of the ABplpappap cell, an egl-1 expressing cell (Denning et al., 2012), to further facilitate cell identification. Another fluorescent transgene stIs10026[his-72::GFP], which expresses GFP-tagged HIS-72 histone protein, was used to label the nuclei of all cells to help define ABplpappap’s location within the embryo. Time-lapse confocal microscopy was used to monitor the location of ABplpappap in embryos, keeping the cell in view by refocusing on it every 30 sec. Confocal imaging during a period of about 50 min during which ventral enclosure (migration and meeting of hypodermal cells on the ventral surface of the embryo) occurs was sufficient to determine whether ABplpappap did or did not undergo extrusion.
For determining whether ABplpappap and other cells that are extruded entered the cell cycle, the transgene heSi192[Peft-3::tDHB::eGFP::tbb-2 3’UTR] was used to express a codon-optimized (for C. elegans) C-terminal fragment of Human DNA Helicase B, which translocates from the nucleus to the cytoplasm in response to the activity of the cell cycle CDKs 1 and 2 (van Rijnberk et al., 2017). nIs861 was used to label the membrane of ABplpappap with mCherry to facilitate cell identification.
For determining the cell cycle phase of ABplpappap and other extruded cells, isIs17[Ppie-1::GFP::PCN-1] was used to express GFP-tagged PCN-1 protein, which produces a phase-specific fluorescence intensity and localization pattern. nIs861 was used to label the membrane of ABplpappap with mCherry to facilitate cell identification.
Images were processed with ImageJ software (NIH), Photoshop CC 2019 (Adobe) and Illustrator CC 2019 (Adobe) software. The Time Stamper function in the Stowers ImageJ plugin was used to mark elapsed time on time-lapse videos.
Excretory cell counts
Excretory cell counts were performed using a dissecting microscope equipped with fluorescence at a total magnification of 270x. For the genome-wide RNAi screen, roughly 50 animals were examined in each well of a 24-well plate and any well with more than 5 animals with two excretory cells was marked for confirmatory testing. Excretory cell counts in confirmatory RNAi experiments, candidate RNAi experiments and experiments with genetic mutants were conducted using 6 cm Petri plates with appropriate media. Animals were first immobilized by keeping the Petri plates on ice for 30 min. At least 100 animals at the L3-L4 larval stage were scored for each genotype or RNAi experiment unless there was extensive lethality or a growth defect, in which case a lower number or earlier-stage animals, respectively, were scored. A cell was scored as an excretory cell if it was located in the anterior half of the animal and its nucleus had strong GFP expression.
tDHB-GFP fluorescence intensity quantification
The ABplpappap nuclear boundary, cell membrane boundary and the tDHB-GFP fluorescence signal were determined from DIC, mCherry and GFP channels, respectively, of confocal images of RNAi treated ced-3(lf) embryos expressing the transgenes heSi192 and nIs861. Mean tDHB-GFP fluorescence intensities inside the nuclear region, entire cell and background were quantified using Fiji software. Mean cytoplasmic tDHB-GFP fluorescence intensity was calculated by the following formula
Icytoplasm, Icell and Inucleus denote the mean tDHB fluorescence intensity in the cytoplasm, cell and nucleus, respectively. The ratio of nuclear-to-cytoplasmic tDHB fluorescence intensity in Figure 3H was adjusted for background fluorescence (measured from a random area outside the embryo boundaries), i.e., the background fluorescence intensity was subtracted from both nuclear and cytoplasmic fluorescence intensity values before calculating the ratios.
Calculation of cell size
Confocal micrographs were obtained for multiple focal planes starting at the ventral surface and ending at the dorsal surface of the embryo, with each plane separated by a distance of 0.37 µm. The greatest area occupied by a cell in any plane was designated the “maximum area” of a cell.
Cell culture
MDCK and MDCK-Fucci (Streichan et al., 2014) cells were cultured in DMEM supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin in a humidified incubator at 37°C with 5% CO2.
Chemicals
2 mM HU (Millipore Sigma, Cat#H8627) was prepared in culture medium prior to each experiment.
Mammalian cell imaging
These assays were performed using 6-well plastic plates. 20,000 MDCK cells were seeded in each well and grown to confluence for 72 h. The day of the experiment, cells were washed twice with PBS and treated with fresh medium or 2 mM HU in medium. After equilibration, plates were imaged at 15-min intervals for up to 24 h, using an Evos M7000 imaging system equipped with a humidified onstage incubator (37°C, 5% CO2). Several positions per well were imaged in the phase contrast and green and red fluorescence channels available in this system.
Mammalian cell extrusion quantification
In time-lapse phase contrast images, extruding cells are easily identifiable as bright, white, rounded spots emerging from the epithelial plane. We counted the number of cells with these features for each condition using the Cell Counter plugin of Fiji (Schindelin et al., 2012). Extrusions are reported as number of extruding cells/h for comparison between experiments of different duration.
Mammalian cell cycle phase determination
The Fucci system differentially labels the nuclei of cells in G1 (red) and S/G2/M (green) (Sakaue-Sawano et al., 2008). Images of MDCK-Fucci cells with HU or control treatment were obtained in the phase contrast, red and green fluorescence channels as per Mammalian cell imaging above. For each position, a multi-channel stack was built using Fiji (Schindelin et al., 2012). After identifying an extruded cell in the phase contrast channel, the cell cycle phase was determined using the fluorescence channels.
Mammalian re-seeding experiments
At the end of an imaging experiment, supernatants were collected and centrifuged (1200 rpm, 5 min, room temperature). Pellets were re-suspended in 50 µL of PBS, and 10 µL of the suspension was used for cell counting with Trypan blue in a Neubauer chamber, allowing us to simultaneously calculate the number of cells being re-seeded and the fraction of cells that was apoptotic. The remaining cells were seeded with 1 mL of fresh medium in a 24-well plate and grown in the cell culture incubator. Pictures were taken at 2 h and 24 h for cell counting.
Statistical analysis
For calculation of statistical significance for ratios, the ratios were first transformed to logarithm values. Ordinary one-way ANOVA was performed to determine statistical significance of the ratios with the assumption that logarithm of ratios produced a normal distribution of values. The maximum area of ABplpappap was also assumed to have normal distribution under different RNAi conditions and ordinary one-way ANOVA was used to determine statistical significance. Normal distributions with unequal variances were assumed for rates of extrusion under HU and vehicle treatments, and Welch’s two-tailed t-test was performed to determine statistical significance. No assumptions were made about the distributions of the rates of apoptosis under HU and vehicle treatments, and hence the Mann-Whitney test was used to determine statistical significance. No assumptions were made about the distribution of fraction of extruded cells in different phases of the cell cycle after HU and vehicle treatments, and the Kruskal-Wallis test was used to determine statistical significance. Normal distribution was assumed for numbers of cells reseeded in fresh media after pre-treatment in different conditions, and ordinary one-way ANOVA was used to determine statistical significance. All statistical analysis was performed using Prism 7 (GraphPad Software).
ACKNOWLEDGMENTS
We thank S. van den Heuvel for providing strains; the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440), for providing some strains; N. An and T. Ljungars for strain management; S. Luo, S.R. Sando, A. Doi, and A. Corrionero and other members of the Horvitz laboratory for helpful discussions, and D. Ghosh, C.L. Pender, J.N. Kong, M.G. Vander Heiden, P.W. Reddien, R.O. Hynes for suggestions regarding the manuscript. V.K.D. was a Howard Hughes Medical Institute International Student Research fellow. C.P.P. is the recipient of a Human Frontiers Science Program postdoctoral fellowship (LT000654/2019-L). J.R. and C.P.P. were funded by King’s College London startup funds. H.R.H. is the David H. Koch Professor of Biology at MIT and an Investigator at the Howard Hughes Medical Institute.