How winner cells cause the demise of loser cells

Cell competition causes apoptosis of suboptimal cells: Their dregs are removed by hemocytes, thus preserving tissue homeostasis


  • Fidel-Nicolás Lolo,

    1. Molecular Oncology Programme, Spanish National Cancer Centre (CNIO), Madrid, Spain
    2. Vascular Biology and Inflammation Department, Spanish National Cardiovascular Centre (CNIC), Madrid, Spain
    Search for more papers by this author
  • Sergio Casas Tintó,

    1. Molecular Oncology Programme, Spanish National Cancer Centre (CNIO), Madrid, Spain
    2. Cajal Institute (CSIC), Madrid, Spain
    Search for more papers by this author
  • Eduardo Moreno

    Corresponding author
    1. Molecular Oncology Programme, Spanish National Cancer Centre (CNIO), Madrid, Spain
    2. Institute of Cell Biology, University of Bern, Bern, Switzerland
    • Molecular Oncology Programme, Spanish National Cancer Centre (CNIO), Madrid, Spain
    Search for more papers by this author


Recent results show that, during the process known as cell competition, winner cells identify and kill viable cells from a growing population without requiring engulfment. The engulfment machinery is mainly required in circulating macrophages (hemocytes) after the discrimination between winners and losers is completed and the losers have been killed and extruded from the epithelium. Those new results leave us with the question as to which molecules allow winner cells to recognize and impose cell death on the loser cells during cell competition.


PSR, phosphatidylserine receptor; SPARC, secreted protein acidic and rich in cysteine; Wasp, Wiskott–Aldrich syndrome protein.

Introduction: What is cell competition?

The competitive process among cells is the result of gene mutations that produce cell populations with different proliferation rates 1 (for an extensive review see 2). As a consequence of “fitness” comparison (Box 1

The “fitness” of cells

The idea of “fitness” is widely used in ecology and evolutionary biology to define which organisms survive at the expense of others during natural selection and evolution 20. But even in ecological terms fitness is, somehow, an abstract concept, because it is always revealed in retrospect: only after one individual dies is it assumed that it was less fit. Therefore, normally it is very difficult to know in advance which traits make a living organism more or less fit. Likewise, mutations that negatively affect the fitness of cells are also identified in retrospect. Most of those mutations have been recognized in an assay where viable, but mutant, cells are eliminated, but only when they grow surrounded by wild-type cells 1, 3, 4. It is important to note that these mutant cells would not die if they were surrounded by neighbors that were equally mutant 1, 3, 4. Therefore, it is implicit in the definition of “fitness” that the cells that survive are more fit than the cells that disappear and are replaced.

), cells that acquire a proliferative advantage, termed winners, will grow at the expense of less advantaged cells, termed losers. The latter will die by means of apoptosis and their corpses will be engulfed and eliminated. In contrast, the winners will continue growing and eventually spread along the tissue in such a way that the total number of cells and tissue morphology are preserved 3 (Fig. 1).

Figure 1.

The competitive process: Competition versus supercompetition. A mutational change turns a cell within a given tissue into a suboptimal one or into a supercompetitor. The first one will behave as a loser (when compared with the surrounding ones that behave as winners in this case) and will be eliminated by apoptosis. Neighboring cells proliferate to occupy the empty space left by the dead loser cell. The supercompetitor, however, will behave as a winner and will expand throughout the tissue killing the surrounding cells (that behave in this case as losers) by apoptosis. In both cases survivor cells proliferate to maintain morphology and number of cells of the tissue/organ.

Cell competition was first described in the wing imaginal disc of Drosophila melanogaster in 1975, when Gines Morata and Pedro Ripoll studied the phenotype of a group of mutations called Minute. They observed that Minute heterozygous cells, M/+, that appear to be viable in a homotypic environment, disappeared from the tissue when they were confronted with wild-type cells 4. From then on, more and more studies showed that cell competition was not only restricted to mutations in ribosomal proteins. In fact, there are several mutations that can affect the competitive behavior of cells 1, 3–6. For the sake of clarity, four gene categories can be proposed to encompass the different cell competition scenarios:

  • (1)Genes implicated in the regulation of cell metabolism, e.g. the aforementioned Minutes 4. Mutations within this group also include alterations of the insulin pathway, as in the case of chico, the Drosophila homolog of the insulin receptor substrate (IRS) in mammals 7, 8. Cells with defective signaling are eliminated by cell competition.
  • (2)Genes implicated in signaling pathways required for growth and survival. For example, the decapentaplegic/bone morphogenetic protein/tumor growth factor-β (Dpp/BMP/TGF-β) pathway, can induce cell competition 1. Similarly, cells with relative differences in Wingless (Wg) signaling lead to cell competition 9.
  • (3)Genes that encode apico-basal polarity proteins. This is the case for “disc large1” (dlg1), “lethal giant larvae” (lgl), and “scribbled” (scrib). The proteins encoded by these genes are localized to the basolateral region of epithelial cells in polarized tissues. Cells mutant for one of these genes behave as losers and are eliminated when surrounded by wild-type cells 10–13. Mutations that disrupt cell morphology, such as the case of RhoGEF2 or Rac1 overexpression, could also be included in this category as they have been shown to undergo cell death in mosaic tissues 14.
  • (4)Genes that control cell proliferation. A case in point is the well-known proto-oncogene dMyc (the Drosophila homolog of the mammalian Myc). Cells with higher levels of dMyc are able to outcompete surrounding wild-type cells 3, 5. Another example occurs with mutations of the Hippo–Salvador–Warts (HSW) pathway where cells mutant for one of the components of the pathway can induce apoptosis in the surrounding wild-type tissue 6, 15, 16. In contrast to the other cell competition scenarios, dMyc and HSW-induced competition are examples of the so-called supercompetition, where the mutant cells are the winners and the wild-type cells are the losers (Fig. 1). Other examples of this supercompetitor behavior occur with Axin mutants (that have deregulated Wg signaling) 9 and with the upregulation of Stat signaling 17. Interestingly, a role for the Stat signaling in winner cells has been recently proposed, in the sense that normal cells would require Stat signaling for the elimination of scrib mutant cells in mosaic tissues 18. It is worth noting that supercompetition possibly resembles what happens in pretumoral lesions, where tumor cells expand in a wild-type environment 3, 19.

In summary, cell competition relies on the ability of cells to compare their relative fitness status, so that suboptimal or damaged cells are effectively eliminated, whereas the best-qualified cells remain to populate the tissue.

The Flower wars

The Flower Wars (xochiyaoyotl) is the name given to the battles fought in Central America between the Aztecs and their neighbors, before the arrival of the Europeans 28, 29. This was a peculiar type of ancient war in which the Aztec warriors were trained to prefer capturing their enemies in battle rather than killing them. Therefore, losers were not killed immediately, but captured, then marked as “losers” with blue paint and eventually sacrificed (or saved) later during an independent ritual 21, 22, 28.

Results in Drosophila suggest that the gene flower works similarly during “cellular wars”. It is required for distinguishing loser cells from winner cells after competition for resources 21, 22. The different isoforms of Flower tag cells as Losers (expressing the Lose-A and/or Lose-B isoforms) or winners (cells expressing the ubi isoform) but the eventual death of the loser cells depends on the context and a cell–cell comparison on the relative levels of Lose and/or ubi expressed by neighboring cells 21 (Fig. 2).

Figure 2.

The Flower code. It is composed by three isoforms (Ubi, Lose-A, and Lose-B) that establish a code to identify winner and loser cells. Cells among neighbors with similar fitness proliferate and survive normally (A and B) however, a loser cell tagged with Lose-A or Lose-B among winner cells is eliminated from the tissue (C).

Consequently, the gene was called flower after the “Flower war” between the Aztecs and their neighbors 21. This name for the gene was also given, in the context of synaptic vesicles, as it reflects the fact that the mutant synapses have a form resembling a flower 30. Thus, both studies 21, 30 agreed on the name but for different reasons.

Recent discoveries

Cell competition: A step-wise process

New insights into cell competition have revealed that it proceeds through a series of consecutive events. It all starts by mutations that alter the cellular fitness, as the ones presented in the previous section. These differences in fitness are then translated by the Flower code into cell labels that mark cells as winners or losers 21 (Box 2). These labels are membrane tags that give information on the cellular “health status”; however, as yet the molecules that allow cells to recognize this code are still not known. The corollary in loser cells is the upregulation of the protein SPARC (secreted protein acidic and rich in cysteine), which transiently protects losers from being eliminated by apoptosis 22. Both Flower and SPARC have been recently related with cancer in the context of cell competition 23, 24, supporting the connection between the competitive process and tumor formation 19. Specifically, Flower-deficient mice have been shown to present a significant reduction of skin papilloma formation upon the tumorigenic agent 7,12-dimethylbenz[a]anthracene/12-O-tetradecanoylphorbol-13-acetate (DMBA/TPA) treatment, supporting the idea that the Flower code is evolutionary conserved and plays a role in tumor expansion 24 (for a recent review see 25). Complementarily, SPARC has been shown to be highly expressed in normal tissues at the boundaries of neoplastic lesions (as in the ductal adenocarcinoma and mucinous cystic neoplasia of the pancreas) and in tumor areas (such as in adenoma and adenocarcinoma of the colon and urothelial carcinoma) 23. These differences in expression are in agreement with the model of cell competition (Fig. 1).

In the end, loser cells undergo apoptosis and are eliminated from the tissue 1. This elimination is initially mediated by basal extrusion 3, a process by which apoptotic cells shorten their apico-basal axis, lose their contact with the surface of the epithelium and are finally extruded 26, 27. Cell competition, therefore, is a multistep process that begins with the establishment of the winner/loser status and ends with getting rid of the loser cells.

The winner cells

All the steps enumerated above occur in the loser cells, but what about the winner cells?

Previous studies showed that winner cells required engulfment genes such as draper, wasp, and psr to eliminate loser cells 31, hence evoking the concept that winner cells “eat” their way through mosaic compartments, consuming loser cells along the way 31. Similar observations were presented in another study where scrib- or dlg-deficient clones were eliminated by surrounding wild-type cells through the activation of ELMO/Mbc (engulfment and cell motility/myoblast city)-mediated engulfment 32.

However, we recently re-evaluated the role of the genes draper, wasp, and psr from the perspective of the behavior of winner cells during cell competition 33; these results suggest a different interpretation from the previous model of neighboring cells disposing of loser cells. First, and most importantly, we found that draper, wasp, and psr were not essential in the winner cells during cell competition, since analysis of the whole mutant background for these genes in different cell competition contexts revealed that loser cells still died by apoptosis. Additionally, we observed that winner clones did not show any significant change in their growth advantage, thus supporting the idea that cell competition is triggered independently of these genes. All in all, the evidence gathered strongly supports the concept that cell competition is a mechanism that identifies and kills viable cells from a growing population in the absence of psr, wasp, and draper. The implication is, therefore, that the relevant molecules that allow winner cells to recognize and eventually impose cell death on the loser cells during cell competition have yet to be identified.

Hemocytes: Scavengers of cell competition

What is the fate of loser cells after they have been killed by winner cells and extruded from the tissue?

Engulfment by epithelial cells has been reported 34, and it was also proposed to be performed exclusively by winner cells during cell competition 31. Although we cannot rule out the possibility that some apical apoptotic cells could, in fact, be engulfed by epithelial cells, we have put forward evidence that this “apical engulfment” would be a small proportion of the whole process. To be precise, we found that in the complete absence of engulfment genes, dead loser cells were mostly accumulated basally in the epithelium. On the other hand, specific downregulation of draper in epithelial cells during cell competition did not lead to a significant accumulation of basal apoptotic bodies, indicating that this gene does not have a specific role in this regard.

So, if winner cells are not engulfing loser cells, which cells assume the role of eliminating dead losers? The obvious cell types to be considered were the hemocytes (circulating macrophages), as they have been shown to seek out the tissue to remove apoptotic corpses and debris 35–39. Our data suggested that most of the clearance of basal apoptotic bodies is performed by these cells. Moreover, we showed that hemocytes are recruited during cell competition and draper functions in these cells to engulf the apoptotic bodies of loser cells – once they are extruded from the epithelium 33. The discrepancy with previous results might come from an over-interpretation, since the idea of apical engulfment was based on experiments in which a fully mutant background for engulfment genes was used. In this context other cells besides epithelial cells are also mutant and, thus, could well be implicated in this regard. We have shown that under these circumstances, specific restoration of draper function in hemocytes appeared to rescue the accumulation of apoptotic bodies observed. This again supports the key role of these cells in the clearance of cellular corpses during cell competition.

Examples of hemocytes associated with mutant cells have been shown elsewhere. For instance, Pastor-Pareja et al. 40 found that hemocytes associate with Rasv12/scrib tumors, and that they impair the growth of tumors from scrib mutant larvae. Cordero et al. 41 demonstrated that this effect is partly due to Eiger/tumor necrosis factor α secretion by hemocytes in the case of scrib- and lgl-deficient clones. Interestingly, they also discovered that Eiger had the opposite effect, i.e. promoting tumor growth, when scrib mutant cells also express oncogenic Ras 41.

We have shown that hemocytes also associate with loser cells during dmyc- and Minute-induced competition to play their function as cell wreckage cleaners. Our results could then broaden previous observations by considering hemocytes as important players not only for the killing of mutant cells but also for their removal. In the case of Minute-induced competition, on the other hand, Eiger seems to be dispensable since M/+ cells still die in an eiger mutant background 32. A possible explanation to cover this discrepancy is that hemocytes would take a different role in this context and rather than kill M/+ cells they would just engulf apoptotic bodies, according to our data. Therefore, hemocytes would interact differently with cells undergoing apoptosis, depending on the cell competition scenario.

The connection between hemocytes and cell competition could shed new light onto tumor progression as they both have potential implications in this field. Cell competition resembles different features of cancer development 42, especially where the progression of pretumoral lesions is concerned 19. Hemocytes, on the other hand, have been related to the so-called “extrinsic tumor suppression” 43, where they have been found to secrete Eiger to limit the survival within lgl and scrib mutant cells 41. Interestingly, macrophages (the human hemocytes) have also been associated with different aspects of tumor formation. Specifically, TAMs (tumor-associated macrophages) have been shown to have a role in carcinogenesis, invasion, and metastases of different tumors. There are two main subsets of TAMs: M2 macrophages, which favor tumor progression and angiogenesis 44, 45, and M1 macrophages, which play antitumoral roles by eliminating tumoural cells (for extensive reviews of macrophages and cancer see 46, 47). Hemocytes may possibly resemble M1-like macrophages as they appear to engulf “potentially dangerous cells”, thus further sustaining the connection between cell competition and tumorigenesis.

Our data deepen our understanding of the cell competition timeline, emphasizing the importance of hemocytes as key players in the elimination of dead losers once the cellular fight sets in.


The previous view of cell competition where loser cells were killed and engulfed by winner cells as they expand throughout the tissue has now been reinterpreted. A new picture has emerged where loser cells are killed by winners but their leftovers are engulfed by recruited hemocytes (Fig. 3). The engulfment machinery is therefore mostly required in circulating hemocytes after the discrimination between winners and losers is completed and the losers have been killed and extruded from the epithelium. These new results imply that, during cell competition, the molecules that allow winner cells to recognize the Flower code and impose cell death on the loser cells are yet to be discovered.

Figure 3.

Cell competition models. The upper part shows the previous model of cell competition in which winner cells kill loser cells using the engulfment machinery. The lower part shows the new scenario in which winner cells kill loser cells by still unknown mechanisms; the apoptotic corpses are then basally extruded from the epithelium and finally engulfed by hemocytes.


Work in our laboratory is funded by the European Research Council, Caja Madrid, Mutua Science Foundation, Swiss National Science Foundation, Josef Steiner Cancer Research Foundation and the autonomous community of Madrid. S.C.T. was recipient of a JdC fellowship from the Spanish MICINN.

The authors have declared no conflict of interest.