Hippo signaling controls cell cycle and restricts cell plasticity in planarians

The Hippo pathway plays a key role in regulating cell turnover in adult tissues, and abnormalities in this pathway are consistently associated with human cancers. Hippo was initially implicated in the control of cell proliferation and death, and its inhibition is linked to the expansion of stem cells and progenitors, leading to larger organ size and tumor formation. To understand the mechanism by which Hippo directs cell renewal and promotes stemness, we studied its function in planarians. These stem cell–based organisms are ideal models for the analysis of the complex cellular events underlying tissue renewal in the whole organism. hippo RNA interference (RNAi) in planarians decreased apoptotic cell death, induced cell cycle arrest, and could promote the dedifferentiation of postmitotic cells. hippo RNAi resulted in extensive undifferentiated areas and overgrowths, with no effect on body size or cell number. We propose an essential role for hippo in controlling cell cycle, restricting cell plasticity, and thereby preventing tumoral transformation.


Introduction
The same developmental processes that drive embryogenesis also regulate the constant cell renewal required throughout the natural life span of the organism. Successful cell renewal relies on multiple events, including proliferation and differentiation of progenitor cell populations and death of unnecessary cells. Failure to correctly coordinate these events can lead to diseases such as cancer. Of the multiple molecular mechanisms involved in the control of cellular renewal, the Hippo signaling pathway has emerged as a key hub. Although first identified as a key regulator of organ size through the control of cell death and proliferation [1][2][3][4], growing evidence suggests additional pivotal roles in coordinating stem-cell maintenance, cell differentiation, cell fate decisions, and cell survival [5][6][7][8][9][10].
At the core of the Hippo pathway is a kinase cascade that phosphorylates the nuclear effector Yorkie (Yki) (YAP/TAZ in vertebrates) and targets it for degradation. When the pathway is inactive, dephosphorylated Yki enters the nucleus to regulate gene expression [11,12]. In most organs and tissues, such as the liver, heart, and skin, loss of Hippo signaling, or elevated activity of Yki/YAP/TAZ, is associated with stem-cell expansion, inhibition of cell differentiation, the appearance of overgrowths, and tumorigenesis [2,[13][14][15]. In line with these observations, YAP/TAZ is hyperactivated in most human cancers [16]. Importantly, in regenerative contexts, YAP/TAZ promotes regeneration of the same organs in which it produces tumors under homeostatic conditions [17,18]. The Hippo signaling pathway therefore appears to exert a general function, promoting stemness or amplifying the population of progenitors, that is beneficial in regenerating tissues but harmful in homeostatic conditions [2,8,14].
Despite its crucial role in everyday tissue renewal and in the maintenance of healthy organisms, the mechanism by which Hippo signaling promotes stemness remains unclear. Studies performed in several tissue types have extensively demonstrated the positive effect of Hippo inhibition on cell proliferation and the consequent expansion of the resident population of stem cells [7,[19][20][21][22]. However, recent studies in the liver and intestine, as well as in embryonic stem (ES) and induced pluripotent stem (iPS) cell cultures, have shown that both Hippo down-regulation and YAP/TAZ nuclearization increase the plasticity of differentiated cells, allowing their dedifferentiation towards a stem-cell fate [9,[23][24][25]. The plasticity of cells within the hierarchical organization of a tissue has major implications for regenerative medicine and cancer [26][27][28].
To better understand the role of the Hippo pathway in driving adult cellular renewal and specifically in promoting cell stemness, we studied its function in planarians. Owing to the presence of a population of pluripotent adult stem cells (called cNeoblasts) [29,30], planarians have the ability to constantly grow and degrow depending on food availability and to regenerate any missing body part within a few days. Several lineage-restricted cycling cells (or lineagerestricted neoblasts) and their postmitotic descendants can be identified in planarians based on the expression of tissue-specific transcription factors [31,32]. In addition to the presence of a stem cell population, the continuous activation of signaling cues that coordinate cell death and cell renewal and direct precise cell fate decisions allow planarians to maintain proportioned and functional organs during growth/degrowth and regeneration. This continuous active regulation of the stem cell and postmitotic cell compartments makes the planarian an ideal in vivo model of the different events underlying homeostatic cell renewal and tissue regeneration. Furthermore, in contrast to most models of regeneration, planarian regeneration is fueled directly by the expansion of an abundant stem cell population, precluding the need for dedifferentiation [33].
Here, we investigate whether down-regulation of Hippo signaling exerts its stemness-promoting effect by increasing the proliferation of resident stem cells or by promoting cell dedifferentiation. We show that inhibition of Smed-hippo (referred to hereafter as hippo) in planarians reduces apoptotic activity and increases mitotic rates. However, this imbalance between cell death and mitotic activity does not lead to an increase in planarian body size or cell number, possibly because hippo RNAi does not increase the number of cycling cells but rather blocks mitotic exit and increases necrotic cell death. hippo (RNAi) planarians develop overgrowths and extensive regions composed of undifferentiated cells, accompanied by a general decrease in the number of differentiated cells throughout the body. A detailed study of the epidermal lineage reveals that hippo is required to determine the hierarchical transitions necessary for proper epidermal differentiation from epidermal-restricted stem cells to differentiated epidermal cells. Finally, our results indicate that hippo is required to maintain the differentiated state in planarian cells, because hippo inhibition could promote dedifferentiation of postmitotic cells. Overall, our results indicate that the overgrowths and undifferentiated regions observed after hippo inhibition in planarians are not caused by an imbalance between cell death and proliferation but by the inability of cells to reach and maintain the appropriate fate. We propose an essential role for hippo in restricting cell plasticity and hence in preventing tumoral transformation.

hippo controls the number of apoptotic and mitotic cells in planarians
To study the role of the Hippo pathway in planarians, we conducted a functional analysis of hippo, the core element of the pathway, in Schmidtea mediterranea [34]. In situ hybridization (ISH) and in silico searches in the single-cell database Planaria SCS [35] indicated that hippo is expressed in cells of all types (S1A and S1B Fig). To decipher the possible function of hippo during homeostatic cell renewal in planarians, we injected animals with hippo double-stranded RNA (dsRNA) for 3 weeks (see Material and methods and S1C Fig). This resulted in a significant decrease in hippo mRNA levels beginning during the first week of treatment (S1D Fig). The appearance of unpigmented regions, mainly around the body margin, was observed during the third week of hippo inhibition. Over time, these regions became larger or evolved into unpigmented overgrowths (Fig 1A and S1E Fig).
To determine whether the appearance of overgrowths was caused by an imbalance between cell death and cell proliferation, we performed TUNEL and caspase-3 assays and quantified mitotic activity by anti-phospho-Histone 3 immunostaining (H3P). After 2 weeks of inhibition, hippo (RNAi) animals exhibited a reduction in cell death compared to controls that became more evident after 3 weeks of inhibition (Fig 1B and S2A and S2B Fig). After 3 weeks of hippo RNAi, the number of apoptotic cells was reduced compared to controls, and mitotic activity was increased (Fig 1C and S2C Fig). Nonetheless, measurement of body area revealed no difference between hippo (RNAi) planarians and controls (Fig 1A and 1D). Quantification of the total number of cells using fluorescence-activated cell sorting (FACS) or a Neubauer chamber also revealed no differences in cell number between hippo (RNAi) and control planarians ( Fig 1E and S2D Fig). Our results thus indicate that Hippo promotes apoptotic cell death and controls mitotic activity and that its inhibition leads to the formation of unpigmented regions and overgrowths, without affecting animal size or cell number.

Hippo is essential for proper cell cycle progression
Studies using other animal models have reported that hippo dysregulation results in defects in cell cycle progression [36,37]. This effect could provide a plausible explanation for our observation that the decrease in cell death and increase in mitotic activity in hippo (RNAi) animals do not lead to changes in cell number. To test this hypothesis, we treated cells with 5-ethynyl-2 0 -deoxyuridine (EdU) and analyzed the proportion of cells in M and S phase 16 h later in hippo (RNAi) animals and corresponding controls (S3A Fig). We found comparable numbers of EdU+ cells in hippo (RNAi) animals and controls, indicating that similar numbers of cells enter the cell cycle in both conditions (Fig 2A and S3B Fig). Moreover, double staining with anti-EdU and anti-H3P revealed an increase in the number of EdU+/H3P+ double-positive cells in hippo (RNAi) animals versus controls (Fig 2B and S3B Fig). This suggests that in hippo (RNAi) animals, cells either transition faster from S to M phase and/or fail to complete M phase. The higher number of EdU-/H3P+ cells in hippo (RNAi) animals versus controls ( Fig  2C and S3B Fig) indicates that the former group harbors a greater number of cells that are in M phase but have not passed through S phase in the previous 16 h. Thus, hippo RNAi results in the trapping of cells in M phase but does not affect the number entering the cell cycle ( Fig 2D). This could explain why the increase in the number of H3P+ cells does not translate to an increase in cell number. However, we cannot rule out the possibility that hippo inhibition results in faster transit of cells from S to M phase. To further understand the role of hippo in the cell cycle, we examined the organization of the mitotic spindle and chromosomes in dissociated cells from hippo (RNAi) animals by double immunostaining with anti-α-tubulin and anti-H3P antibodies. The results revealed abnormal microtubule organization of prophase/prometaphase spindles and spindle poles in hippo (RNAi) cells (Fig 2E). Similarly, metaphase cells showed disorganized and less dense arrays of spindle microtubules, as well as chromosome alignment defects ( Fig 2E). Furthermore, in hippo (RNAi) cells, more than a half of all anaphases appeared in an early phase, while in controls, all were classified as late anaphases. Half of the early anaphases in hippo (RNAi) cells were characterized by a disorganized and less dense microtubule array (Fig 2E). This result supports an essential role of hippo in the normal cell cycle progression and is consistent with the increase in the number of cells in M phase, but not S phase, observed after hippo RNAi.
In agreement with the role of Hippo in cell cycle progression and specifically in the mitotic phase, we found that several genes related to cytokinesis and mitotic spindle organization were differentially expressed in the hippo (RNAi) RNA sequencing analysis (RNAseq) (S1 Table). Importantly, these included genes already known to be regulated by Mst1/2 or Lats1/2 and mainly involved in spindle orientation (afadin, Drosophila discs large, polo-like kinase 1) [38][39][40] (S1 Table).

Hippo is required for maintenance of the differentiated cell population
Detailed analysis of the overgrowths caused by hippo inhibition (Fig 3A) revealed that they are caused by the accumulation of cells in the subepidermal region or the mesenchyme (Fig 3B  and S4A Fig) and in some cases arise from the submuscular plexus region (S1-S3 Movies). In line with the unpigmented appearance (Fig 3A), which indicates that epidermal cells cannot produce pigment and thus are not terminally differentiated, we observed abnormal distribution of β-catenin-2, a component of adherens junctions, in the epidermal cells of the overgrowths [41] (Fig 3B, S1-S3 Movies). The defects in cell differentiation also affected the neural plexus, as evidenced by the absence of anti-synapsin staining in the overgrown areas ( Fig 3B). The accumulation of smedwi-1+/ SMEDWI-1+ cells in subepidermal overgrowths and among some cells of the corresponding epidermis ( Fig 3B and S4A Fig) indicates that the overgrowths are primarily composed of undifferentiated cells. The accumulation of mitotic cells in or around the overgrowths (Fig 3B and S1-S3 Movies) confirms loss of the differentiated state.
In addition to the specific overgrown regions, broader unpigmented areas were also observed in hippo (RNAi) animals ( Fig 3A). Analysis with specific markers revealed that differentiated structures such as the nervous system, the eyes, and the digestive system were not properly renewed or maintained in those unpigmented areas ( RNAseq further indicated that most of the markers associated with terminal differentiation of multiple cell types (opsin, pantothenate kinase 1, synapsin, tropomyosin) were down-regulated (Fig 3D), consistent with a general defect in the process of cell differentiation or maintenance of the differentiated state. Quantification of the number of dopaminergic and serotonergic neurons by ISH with the corresponding markers (tyrosine hydroxylase and tryptophan hydroxylase 1) further confirmed the general decrease in the number of differentiated neurons in hippo (RNAi) planarian heads ( Fig 3E).
In several animals, the disappearance of differentiated structures coincided with the presence of cell-depleted regions (Fig 3B and 3C). Therefore, using in vivo propidium iodide (PI) incorporation analysis, we investigated whether a general increase in cell death occurred in hippo (RNAi) animals. The results revealed an increase in PI+ cells in hippo (RNAi) animals (S5 Fig). Because we previously showed that caspase-dependent cell death is decreased in hippo (RNAi) animals, we can conclude that necrotic cell death is increased. This increase in necrosis could compensate for the decrease in apoptosis and contribute to the maintenance of cell number in hippo (RNAi) animals.
Taken together, these data indicate that Hippo plays an important role in the maintenance or renewal of differentiated tissues in planarians and that its inhibition leads to the appearance of extensive regions of undifferentiated cells that accumulate in overgrowths.

Hippo-Yorkie signaling controls cell differentiation during planarian regeneration
In regenerative contexts, Yki/YAP nuclearization induced after Hippo inhibition or direct Yki/YAP overactivation is associated with increased proliferation and improved regenerative capacity [17], [18]. Analysis of the effect of hippo inhibition following head and tail amputation (S6A Fig) revealed an increase in mitotic rate but the production of smaller blastemas (Fig 4A  and 4B), in which new tissues such as the brain and digestive system failed to properly differentiate (S6B and S6C Fig). Accordingly, numbers of differentiated neural cell types, such as The total number of tph+ and th+ cells in the head region was normalized with respect to the total area of each animal. Images correspond to confocal Z-projections; n ! 5. All experiments were performed in planarians subjected to 3 weeks of hippo RNAi. Error bars represent standard deviation. Data were analyzed by Student t test. Ã p < 0.05; ÃÃÃ p < 0.001. Data used in the generation of this figure can be found in S1 Data. Scale bars: 150 μm (A, B, and C 0 ); 1 mm (C and C@); and 200 μm (E). Bcat2, β-catenin-2; FISH, fluorescent in situ hybridization; gfp, green fluorescent protein; H3P, phospho-histone-H3-Ser10; pk1, pantothenate kinase 1; RNAi, RNA interference; th, tyrosine hydroxylase (dopaminergic neuron); tph, tryptophan hydroxylase (serotonergic neuron). dopaminergic neurons (th+) and eye cells (ovo+ cells in the eye) [42] were reduced after hippo inhibition ( Fig 4C and S6D Fig). Interestingly, despite the increase in mitotic rates and the defects in differentiation, overgrowths were never observed in regenerating hippo (RNAi) animals.
Although Yki/YAP is a highly evolutionarily conserved downstream element of the Hippo pathway [5], it has been proposed that Yki does not act downstream of Hippo in planarians [34]. To investigate the possible conservation of Hippo-Yki signaling in the control of planarian cell differentiation, we generated a specific anti-Yorkie antibody to determine the levels and pattern of Yki expression. Both western blot and immunohistochemical analysis revealed decreases in Yki levels in yki (RNAi) animals, as expected (S7 Fig). By contrast, in hippo (RNAi) planarians, Yki protein levels were specifically up-regulated in the nucleus (S7 Fig), suggesting conservation of a Hippo-Yki signal in planarians. We next investigated whether yki plays a role in cell differentiation in planarians. Inhibition of yki in regenerating animals led to an increase in the numbers of differentiated neurons and photoreceptors ( Fig 4C and S6D Fig), in direct opposition to the phenotype observed following hippo inhibition. In agreement with our results, a recent study [43] reported a general increase in the numbers of several types of hippo silencing promotes dedifferentiation differentiated cells in yki (RNAi) planarians. Taken together, these results indicate that a conserved Hippo-Yki signal regulates cell differentiation during planarian regeneration. In contrast to the effect observed in vertebrate systems [17,18], in planarians, hippo inhibition, and hence Yki nuclearization, blocks differentiation and decreases the regenerative response.

hippo defines the expression boundaries of epidermal markers
The epidermal lineage is the most abundant in planarians and its progression and determination is well understood [44]. Epidermal maturation requires temporally correlated transition states in planarians, in which stem cells (smedwi-1+) become postmitotic and start to sequentially express nb21/32, agat, and vimentin (vim) [44,45]. In parallel, epidermal precursors migrate from the inner parenchyma towards the epidermis [44]. Thus, proliferating cells are mainly found in the inner part of the animal and postmitotic epidermal cells are found at the periphery. Interestingly, in hippo (RNAi) planarians we detected a large number of mitotic cells in the periphery, where overgrowths and unpigmented regions were mainly found ( Fig  5A). Because this region should mainly contain postmitotic epidermal precursors [44], we reasoned that the process of differentiation and/or cell fate maintenance of the epidermal lineage may be impaired in hippo (RNAi) animals. In agreement with this hypothesis, we found that hippo was expressed in all epidermal lineage cells (S1A Fig and S8A Fig).
To gain further insight into cell fate progression in hippo (RNAi) animals, we analyzed the number and distribution of mitotic cells (H3P+), stem cells (smedwi-1+), and postmitotic epidermal cells (nb21/32+, agat+, or vim+). First, we quantified H3P+ cells in 4 different regions (inner to outermost) in transverse sections (Fig 5A). We observed a significant increase in the number of mitotic cells (H3P+) in the 2 outermost regions of hippo (RNAi) animals as compared with controls (S5A and S8B Figs). vim+ cells were located mainly in the epidermis (fully differentiated cells) of control animals but were predominantly detected in the mesenchyme of hippo (RNAi) animals ( Fig 5B and 5B 0 ). This indicates that hippo RNAi impairs the acquisition or maintenance of epithelial fate or the migration of epidermal cells. We detected higher numbers of vim+/ SMEDWI-1+ cells in the mesenchyme of hippo (RNAi) animals versus control animals, in which these cells were virtually absent from the mesenchyme (Fig 5B-5B@). This finding, together with the presence of ectopic mitotic cells, suggests impairment of the differentiation or fate maintenance, but not the migration, of epidermal cells following hippo inhibition.
Quantification of cells that were double positive for different epidermal markers revealed an increase in smedwi-1+/agat+, nb+/agat+, and agat+/vim+ cells in hippo (RNAi) animals versus controls (Fig 5C). This increase in bivalent cells in hippo (RNAi) animals is consistent with failure of the epidermal cells to progress appropriately through the hierarchical transitions that occur during epidermal lineage specification and suggests that they are unable to maintain a defined fate. Interestingly, double labeling of H3P+/agat+, H3P+/nb32+, or H3P+/smedwi-1 + revealed that mitotic cells were always smedwi-1+ but never agat+ nor nb32+ in either hippo (RNAi) animals or controls (S8C Fig). Although we cannot exclude the possibility that ectopic mitotic cells could arise from a different lineage or from aberrant migration of stem cells, our findings suggest that mitotic cells never express postmitotic markers.
Supporting the view that hippo inhibition interferes with specification of the epidermal lineage, analysis of the hippo (RNAi) RNA sequence revealed an increase in the expression of markers of z-neoblasts (zfp-1 and p53), which are precursors of the epidermal lineage [31] ( Fig  5D). This increase could be caused by an increase in the number of z-neoblasts. However, we found that while expression levels of smedwi-1, a general marker of stem cells, decreased in the transcriptome (Fig 5D), the number of cells expressing this marker was unchanged in hippo hippo silencing promotes dedifferentiation (RNAi) animals (S8D Fig). As such, the increased expression of z-neoblast markers could also be due to increased expression of those markers in z-neoblasts or to incorrect acquisition of their expression by other cell types.
Taken together, our analysis of the epidermal lineage reveals an increase in the population of cells co-expressing cell markers that define different epidermal cell lineages in hippo (RNAi) animals, as well as mislocalization of these markers ( Fig 5E). Thus, precursors and differentiated epidermal cells are both improperly defined and mislocalized. These results indicate that hippo is essential for the acquisition and/or maintenance of cell fate in the epidermal lineage.

The Hippo pathway maintains the differentiated state in planarians
The presence of large numbers of undifferentiated cells and overgrowths following hippo RNAi could be explained either by defective differentiation of stem cells towards a specific fate or by a dedifferentiation process in which postmitotic cells are unable to maintain the committed state and begin to express stem-cell markers. To investigate the possibility that postmitotic cells undergo dedifferentiation, we depleted the stem-cell population in planarians through Xray irradiation followed by the inhibition of hippo over 4 consecutive days ( Because DNA damage induced by high doses of irradiation could mask the possible effect of hippo inhibition on cell dedifferentiation, we next investigated the effects of histone 2b (h2b) inhibition as an alternative means of depleting the neoblast population [46], followed by hippo RNAi (Fig 6A). qPCR analysis using specific primers for h2b and hippo confirmed inhibition of both genes (S10A Fig). ISH for smedwi-1 revealed no differences between hippo (RNAi) animals and controls (the "smedwi-1 WT" phenotype in Fig 6A), whereas in h2b (RNAi) animals, smedwi-1 expression was completely absent ("no smedwi-1" phenotype in Fig 6A) (45%), or reduced to a few scattered cells ("disperse smedwi-1+ cells" phenotype in Fig 6A) (55%), which corresponded to stem cells that escaped the effects of h2b inhibition ( Fig 6A). Strikingly, several animals in the h2b/hippo (RNAi) group had clusters of smedwi-1+ cells in the marginal region of the body ("smedwi-1 clusters" phenotype in Fig 6A)   Immunostaining of h2b/hippo (RNAi) animals and corresponding controls with anti-H3P antibody, and corresponding quantification. The total number of H3P+ cells in the whole animal was quantified and normalized with respect to total body area. Error bars represent standard deviation (n ! 14). Data were analyzed using a Student t test ( ÃÃÃ p < 0.001). Scale bars: 1 mm (A). (C) EdU incorporation followed by FISH of smedwi-1 in h2b (RNAi) and h2b/hippo (RNAi) animals. The procedure used and the corresponding quantification of smedwi-1+/EdU-and smedwi-1+/Edu+ cells are shown. In total, 15 sections corresponding to n ! 8 animals were analyzed. Data were analyzed by chi-squared test, applying a Bonferroni correction ( Ã p < 0.01). Data used in the generation of this figure can be found in S1 Data. EdU, 5-ethynyl-2 0 -deoxyuridine; FISH, fluorescent in situ hybridization; gfp, green fluorescent protein; hpo, hippo; H3P, phospho-histone-H3-Ser10; ISH, in situ hybridization; RNAi, RNA interference.
(58.3%) (Fig 6A). Interestingly, this corresponded to the region in which ectopic mitosis and overgrowths were observed. Moreover, despite the presence of smedwi-1 clusters, the number of H3P+ cells in h2b/hippo (RNAi) animals was comparable to that of h2b (RNAi) animals and was much lower than that of controls (Fig 6B). In addition, EdU incorporation analysis revealed that the percentage of smedwi-1+ cells that entered the cell cycle in h2b/hippo (RNAi) animals was lower than that of h2b (RNAi) animals (Fig 6C and S10B Fig). This result supports the view that smedwi-1+ cells originate from distinct populations in h2b (RNAi) and h2b/hippo (RNAi) animals. Taken together, our data suggest that the clusters of smedwi-1+ cells found in h2b/hippo (RNAi) animals are not the result of the proliferation of remaining stem cells that escape h2b inhibition but are produced by dedifferentiation of postmitotic cells that regain the expression of stem-cell markers.
To better understand the nature of smedwi-1+ cells in h2b/hippo (RNAi) animals, we investigated whether the expression of different neoblast markers was also increased. The results revealed that expression of the stem-cell marker bruli [47] was also increased, although not to the same degree as smedwi-1 (S10C Fig). Because our previous results showed that hippo is required to properly maintain epidermal fate and that the expression of z-neoblast markers is increased after hippo RNAi, we analyzed the smedwi-1+ clusters observed in h2b/hippo (RNAi) animals for expression of the z-neoblast marker zfp1. ISH showed that h2b/hippo (RNAi) animals were mostly depleted of zfp1, as were h2b (RNAi) animals (S10D Fig). This result does not allow us to determine whether smedwi-1+ clusters arise from the dedifferentiation of epidermal cells. It is possible that smedwi-1+ clusters arise from a different lineage or, alternatively, that lineage restriction is not maintained during dedifferentiation of epidermal cells. Nonetheless, our results suggest that Hippo is essential to maintain the postmitotic state in planarians and that its absence may induce the dedifferentiation of postmitotic cells towards a stem-cell identity.

Discussion
In the present study, we show that hippo is required to regulate the apoptotic and mitotic response in starved degrowing planarians. hippo RNAi resulted in an increase in the number of mitotic cells and a decrease in the number of apoptotic cells. Importantly, and in contrast to reports in other organisms [2,48,49], the imbalance between apoptosis and mitosis observed following hippo inhibition did not lead to changes in planarian body size or cell number. This observation may be explained by the increase in the number of necrotic cells after hippo inhibition. Furthermore, the fact that the increase in the number of cells in M phase in hippo (RNAi) animals was not accompanied by an increase in the number of cells in S phase may also contribute to the maintenance of the cell number in hippo (RNAi) planarians. Our findings demonstrate that in planarians, Hippo is essential for proper cell cycle progression and, specifically, for successful completion of mitosis. The role of core Hippo signaling elements in critical mitotic events, namely in centrosomal duplication, chromosomal alignment, spindle formation, and completion of cytokinesis, has also been reported in Drosophila and vertebrate species [36,40,[50][51][52][53]. Planarian hippo (RNAi) cells showed improper assembly of the mitotic spindle. This effect, together with the number of genes directly involved in mitosis that are differentially regulated in the hippo (RNAi) transcriptome, suggests a direct role of Hippo in regulating their expression. For instance, centriolin, which is required to complete cytokinesis, was down-regulated in the hippo (RNAi) transcriptome. Corroborating the important role of Hippo in the cell cycle in planarians, in the transcriptome we detected several genes previously reported to mediate the requirement of Hippo for proper cell cycle progression (S1 Table). Despite defective mitotic exit, hippo (RNAi) animals showed no tissue regression, the phenotype normally associated with loss of the cycling cell population in planarians [29].

Indeed, the total population of stem cells or neoblasts (smedwi-1+ cells) was maintained in hippo (RNAi) animals. This suggests that only a fraction of the cells that enter the cell cycle become arrested in M phase after hippo inhibition. Supporting this view, around 30% of mitotic figures in hippo (RNAi) cells were defective.
There is a second possible explanation for the observation that the imbalance between apoptosis and mitosis in hippo (RNAi) animals does not lead to changes in total cell number: a fraction of the cells found in M phase could arise from the dedifferentiation of postmitotic cells. Supporting this hypothesis, in hippo (RNAi) planarians we detected ectopic mitotic cells in the periphery, a region in which postmitotic cells are normally located. These mitotic cells also exhibited aberrant co-expression of committed epidermal cell markers and therefore are unlikely to be the result of incorrect migration of stem cells, which are normally located in the innermost part of the organism. Moreover, the boundaries of expression of the epidermal markers corresponding to the entire lineage were not conserved, indicating that hippo inhibition impairs the maintenance of cell identity. The poorly committed cells found in hippo (RNAi) animals could arise during the process of differentiation or could be the product of induced cell dedifferentiation. Importantly, we found that hippo inhibition in planarians depleted of cycling cells (or neoblasts) may promote dedifferentiation of postmitotic cells. It should be stressed that although a process of cell dedifferentiation that allows re-entry in the cell cycle is common in regenerative models, it has never been described in animals with the level of plasticity found in planarians, which fulfill their regenerative and renewal requirements by activating proliferation of their abundant population of totipotent stem cells [33,54]. Thus, our data strongly indicate that hippo plays a crucial role in the maintenance of the differentiated state and that its inhibition promotes dedifferentiation.
Hippo inhibition in planarians promotes the formation of overgrowths, as described in all models in which it has been studied to date [1,48,55]. Although early studies of the Hippo pathway in model organisms attributed the appearance of tumors to an imbalance between cell death and proliferation [1,55], our results indicate that the inability of cells to maintain the differentiated state following hippo inhibition may underlie the appearance of undifferentiated regions and the formation of overgrowths, which also consisted of undifferentiated cells. In agreement with this hypothesis, the formation of overgrowths was preceded by the appearance of large areas in which different tissues, including those pertaining to the neural, digestive, and visual systems, showed a loss of differentiated cell types. Those results are in good agreement with recently published findings supporting an essential role of hippo in restricting cell plasticity in the liver and intestine [9,23], an effect that has been linked to its role in the recruitment of chromatin-remodeling complexes [14,56]. Thus, nuclear Yki/YAP, rather than pluripotency, could exert a critical role in conferring plasticity, which is a crucial property of tumor cells [57].
The formation of overgrowths could be due to the inhibition of apoptosis induced following hippo RNAi, in addition to the increase in cell plasticity (Fig 7). In wild-type animals, senescent cells and cells with DNA disarrangements are eliminated by apoptosis. However, these cells could not be eliminated in hippo (RNAi) animals and may contribute to the acquisition of additional genetic or epigenetic changes (Fig 7). Furthermore, the number of cells with DNA rearrangements and aneuploidies is probably increased in hippo (RNAi) planarians, because, according to our results, those animals display dysregulation of critical mitotic regulators and fail to complete cytokinesis, as previously demonstrated in LATS2 knockout mouse embryos [36]. Importantly, the control of apoptosis is intrinsically linked to genomic instability and malignant transformation [58]. It must be noted that the cell death that is inhibited after hippo inhibition in our experimental conditions is the one induced by starvation. However, it is likely that the damage-induced apoptosis also depends on hippo, because both starvation-and damage-induced apoptosis depend on such common mechanisms as JNK activation [59].
Taking into account the location of the overgrowths in the subepidermal region, which is abundant in epidermal precursors and is also the location of ectopic mitotic cells, it is tempting to speculate that the epidermal lineage is the origin of the cell accumulations in hippo (RNAi) animals. Epidermal cell plasticity promoted by hippo inhibition could account for the appearance of bivalent cells that are susceptible to tumoral transformation. Furthermore, although the total number of stem cells was not altered after hippo inhibition, we observed a marked increase in the expression of markers of z-neoblasts, the precursors of epidermal cells, underscoring the important role of the pathway in restricting the fate of this specific lineage. However, the fact that hippo inhibition in planarians depleted of cycling cells (or neoblasts) resulted in the formation of smedwi-1+ cell clusters that did not express the z-neoblast marker zfp1 does not support an epidermal origin. A deeper analysis of the molecular features of these smedwi-1+ cells will be required to determine whether they arise from a nonepidermal cell type or, alternatively, whether dedifferentiation keeps the cells in a more pluripotent state. Nonetheless, the fact that almost all markers of differentiation were down-regulated in the hippo (RNAi) transcriptome and the appearance of undifferentiated areas affecting neuronal, epithelial, and digestive cells in hippo (RNAi) animals indicate that hippo controls the fate of several cell types and that different cell lineages could contribute to the overgrowths observed. To understand whether hippo exerts a predominant role in specific tissues, it will be necessary to determine whether precursor cell markers are also aberrantly co-expressed following hippo inhibition in other cell types.
In addition to its role in maintaining cell fate, hippo may also be required for proper cell differentiation, as reported in other organisms [10,13,14]. In our study, the role of hippo in cell differentiation was more evident in the context of regeneration, because the decrease in the number of differentiated cells observed in newly regenerating blastemas, despite the increase in mitotic cells, could not directly result from dedifferentiation. Moreover, the observed defects in differentiation may be caused by the dedifferentiation of postmitotic cells in preexisting tissue after hippo RNAi, as previously described in intact animals, leading to misexpression by differentiated muscle cells of signaling molecules required for proper differentiation [60]. This possibility is consistent with the phenotype generated after yki inhibition in planarians, in which an increase in signaling molecules and differentiated cells is observed [43] ( Fig 4C and S6 Fig).
The results presented here support an essential role for hippo in the acquisition and maintenance of differentiated cell fates. Thus, as demonstrated in other models, Hippo favors stemness in planarians. Importantly, although planarians possess an abundant population of totipotent stem cells, hippo inhibition promotes stemness not by increasing stem cell renewal and expansion of the stem cell compartment but rather by promoting cell plasticity and thus dedifferentiation of postmitotic cells. This finding has profound implications in the fields of regenerative medicine and cancer therapy, because the acquisition of plasticity by lineagecommitted cells favors wound healing but also promotes tumorigenesis [28,61]. Accordingly, the cell plasticity induced by hippo inhibition is associated with beneficial effects in the context of heart and liver regeneration in vertebrates [17,18]. However, hippo inhibition in planarians, which have a much greater regenerative ability than vertebrates, impairs rather than promotes regeneration. One explanation for these contrasting effects is that in vertebrate regenerative systems, a process of cell dedifferentiation that allows re-entry in the cell cycle fuels the regenerative response [62,63], whereas in planarians, expansion of the stem cell population is the only source of new cells [33]. In homeostatic conditions, when basal cell renewal is required, hippo inhibition is also deleterious in planarians, because tumoral transformation is facilitated by the sustained increase in cell plasticity and the probable chromosomal instability induced by the cell cycle defects and the inhibition of apoptosis. The fact that we observed no overgrowths in regenerating regions further underscores the importance of the sustained effect of hippo inhibition in specific cell types to promote transformation.

Conclusion
The present findings demonstrate that in a stem cell-based system such as the planarian, the main role of the Hippo pathway is not to control the balance between proliferating and dying cells nor to regulate body size or the stem-cell population; in planarians, hippo is required to successfully complete the cell cycle and to promote apoptosis. Furthermore, hippo is necessary to acquire and maintain cell fate, thus restricting cell plasticity. Consequently, long-term hippo inhibition prevents tissue renewal and leads to the formation of overgrowths.
Although further evidence is required, our results suggest that in planarians, as described in other organisms, hippo may be involved in maintenance of the chromatin state and the genome stability of stem cells and postmitotic descendants. This function would explain the systematic alteration of hippo signaling elements in many types of cancer, as well as the proregenerative effect of the induction of Yki nuclearization. However, it remains unclear why hippo elements are mutated in most human tumors but never at their onset, whereas in model organisms, the appearance of overgrowths is a consistent feature of hippo inhibition. Finally, our results highlight the potentially hazardous effects of manipulating the hippo pathway for medical purposes in regenerative medicine if the stem cells and progenitors induced after hippo inhibition are in fact poorly committed cells.

Planarian culture
Asexual planarians from a clonal strain of S. mediterranea BCN-10 were maintained at 20˚C in planarian artificial medium (PAM) water, as previously described [64]. Animals were fed with veal liver and starved for at least 1 week before beginning experiments.

RNA interference analysis
dsRNA was synthetized by in vitro transcription (Roche) and microinjection performed, as previously described [65], following the standard protocol of 3 × 32 nl injection of dsRNA for 3 consecutive days. To achieve stronger hippo inhibition, we performed the same protocol over 3 consecutive weeks in starved planarians. yorkie (RNAi) animals were injected for only 1 week with injection of gfp during the 2 previous weeks. Regenerating animals were injected with hippo or yorkie dsRNA for 2 weeks prior to amputation of the head and tail. Animals were fixed at different times postamputation, depending on the experiment. Control animals were injected with dsRNA for green fluorescent protein (GFP), a gene not present in planarians.
In the h2b/hippo experiment, animals were injected with gfp or h2b on 3 consecutive days and, after 2 rest days, received another injection of gfp or h2b. Beginning the following day, animals were injected with gfp or hippo for 3 consecutive days. Animals were fixed 4 days after the last injection.

Quantitative real-time PCR
Total RNA was extracted from a pool of 5 planarians each for hippo RNAi and gfp RNAi. Quantitative real-time PCR was performed as previously described [46], and data were normalized based on the expression of EF2 or URA4 as an internal control. All experiments were performed using 3 biological replicates. The following sets of specific primers were used: hippo mRNA, 5 0 -TTTGGTCTTTGGGAATCAC-3 0 and 5 0 -TGGAGGAGGTTGAGAAGG-3 0 ; h2b mRNA, 5 0 -GAGAAAGTTGAACGGCCC-3 0 and 5 0 -AAGATAATACGTACTTCAACG ACG-3 0 .

Whole-mount ISH and immunohistochemistry
RNA probes were synthesized in vitro using Sp6 or T7 polymerase (Roche) and DIG-, FITC-, or DNP-modified (Perkin Elmer) nucleotides. RNA probes were purified and precipitated with ethanol and 7.5 M ammonium acetate. For ISH and fluorescent in situ hybridization (FISH), animals were fixed and processed as previously described [66,67]. After probe development, neoblasts were visualized with the rabbit anti-SMEDWI-1 antibody (1:1,000; kindly provided by Kerstin Bartscherer, Max Plank Institute for Molecular Biomedicine, Münster, Germany) [47]. Nuclei were stained with DAPI (1:5,000) and mounted with 70% glycerol in PBS.

Whole-mount TUNEL and caspase-3 activity assay
For TUNEL analysis, animals were fixed and treated as previously described [70], using the ApopTag Red In Situ Apoptosis Detection Kit (Merck-Millipore Ref.S7165). To avoid technical variance, at least 2 independent TUNEL experiments were performed. For the caspase-3 assay, total protein was extracted from a pool of 5 planarians and processed as previously described [71]. Enzyme activity was measured in a luminescence spectrophotometer (Perkin-Elmer LS-50) (1 excitation, 380 nm). A unit of caspase-3 activity was defined as the amount of active enzyme necessary to produce an increase of 1 arbitrary luminescence unit after a 2-h incubation. The results are presented as units of caspase-3 activity per μg of protein. All experiments were performed using 5 biological and 3 technical replicates for each condition. To avoid technical variance, at least 2 independent experiments were performed.

Flow cytometry
Dissociation of planarians, cell labeling, and isolation of cells by FACS were performed as described previously [72]. Absolute cell counts were performed by adding beads of a known concentration to the sample. Beads and cells were detected simultaneously, and absolute counts of the cells were calculated from bead numbers (absolute counts by indirect method). Flow-Check Fluorospheres Polystyrene Fluorescent Microspheres (Beckman Coulter Inc, Indianapolis, IN) were adjusted at 1 × 10 6 fluorospheres/mL. A 1:100 dilution of the bead solution was applied to each sample to obtain a final concentration of 1 × 10 4 beads/mL. Beads and cells were identified according to their distinct patterns of scatter and fluorescence.

Edu staining of paraffin sections
F-ara Edu (32 nl; Sigma) was injected into gfp and hippo (RNAi) animals at a concentration of 60 μg/mL (diluted in 10% DMSO/PAM water). After 16 h, animals were processed and stained with the EdU Click-555 kit (Baseclick Gmbh; BCK-Edu555), in accordance with the manufacturer's recommendations, following pretreatment with proteinase K (20 μg/mL) for 10 min at RT. Samples were immunostained with anti-H3P, stained with DAPI, and mounted with 70% glycerol in PBS. In the h2b/hippo (RNAi) experiment, animals previously injected with gfp/h2b and/or gfp/hippo were injected with EdU and after 16 h were processed for FISH using the smedwi-1 probe, followed by EdU staining.

In vivo PI incorporation analysis
Animals were injected with gfp or hippo over 3 consecutive weeks and then injected with 3 × 32 nl of a mixture of PI (1.66 × 10 −3 μg/μL; Sigma) and Hoechst 33342 (3.3 × 10 −6 μg/μL; Sigma) diluted in PAM water. Immediately, planarians were soaked in the same solution for 4 h. Single planarians were placed on a microscopy slide and enveloped in a drop of 3% low melting agarose. Confocal imaging was performed within minutes of preparation of the samples.

Generation of the anti-YORKIE polyclonal antibody
The complete coding sequence of yorkie cDNA was cloned into a p-GEMT Easy vector (Promega). Subcloning, protein expression, and antibody production were performed using Pro-teoGenix (ProteoGenix, France). Briefly, 200 μg of recombinant protein was used as an immunogen to produce polyclonal IgGs in 2 rabbits. The postimmunization serum was purified using protein A affinity purification, precipitated in sodium azide buffer, and stored at 4˚C.

Western blot assay
Protein extracts were obtained as previously described [73]. After incubation with rabbit anti-YORKIE (1:2,000) and mouse anti-α-TUBULIN (1:10,000; Sigma) antibodies, the signal was developed using Clarity Western ECL Substrate (Bio-Rad) and chemiluminescence was detected using a C-DiGit Chemiluminescent Western Blot Scanner (LI-COR). Quantifications were performed with Image Studio Lite and normalized to the anti-α-Tubulin signal.

Transcriptomic analysis
For transcriptomic analysis, total RNA was extracted from gfp (RNAi) and hippo (RNAi) animals after 1, 3, or 4 weeks of inhibition. Three replicates were generated per condition from a pool of 5 organisms each. RNA was extracted with Trizol (Invitrogen), following the manufacturer's instructions. RNA was quantified with a Nanodrop ND-1000 spectrophotometer (Thermo Scientific) and quality assessment performed by capillary electrophoresis in an Agilent 2100 Bioanalyzer (Agilent Technologies) prior to preparation of the library. cDNA libraries were prepared using the Illumina TruSeq Stranded mRNA Library Prep Kit and sequenced as paired-end reads in an Illlumina HiSeq 2000 sequencer. Quality assessment of the reads was performed using the FastQC suite. Transcript abundances were calculated with kallisto v0.43.0 [74] on the S. mediterranea dd_Smed_v6 transcriptome assembly [75]. Differential expression analysis was carried out using the sleuth [76] and DESeq2 [77] statistical packages. Raw sequencing data in FASTQ format as well as the transcript abundances have been deposited in the NCBI Gene Expression Omnibus (GEO) [78] and are accessible using GEO Series accession number GSE95130.

Imaging
FISH and immunostaining samples were imaged using a MZ16F stereomicroscope (Leica) equipped with a ProgRes C3 camera (Jenoptik) or an SP2 confocal laser-scanning microscope (Leica). Images were processed using Fiji and Photoshop CS5 (Adobe) software. Brightness/ contrast and color balance adjustments were always applied to the entire image. Quantifications were performed by hand using the "multi-point selection" tool of Fiji. Colocalization quantification was performed using the equivalent areas using the "ROI-manager" tool in Fiji. Nuclear area in Edu and IP experiments was measured using the "threshold" tool with the "moments" mask for all samples. Signal quantification of Yorkie antibody immunostaining was processed using Fiji software. Two planes were used to build the Z-projection. Nuclearstained (DAPI) images were transformed into a mask using the "threshold" tool with the "moments" mask. The mask was used to obtain the nuclear signal with the Image calculator process. The nuclear signal obtained from the resulting image was measured to obtain the raw integrated density (RID). The nuclear area was obtained from the mask. Next, the RID was normalized to the respective nuclear area. Results were averaged per group and significant differences determined by 2-tailed Student t test. Cartoon illustrating the RNAi procedure in regenerating conditions. Animals were starved for 1 week before the experiment and then injected on 3 consecutive days. The following week, animals were injected again on 3 consecutive days, cut the next day, and fixed at different time points. (B) Anti-synapsin immunostaining of the brain region of control and hippo (RNAi) animals. Nuclei are stained with DAPI. Images correspond to planarians after 12 dR. (C) In situ hybridization with pk1 (digestive system) in hippo (RNAi) and control animals. Images correspond to planarians after 10 dR. (D) Fluorescent in situ hybridization for ovo to label eyes in hippo (RNAi), yorkie (RNAi), and control animals. Images correspond to planarians after 5 dR. The corresponding quantification is shown. Data were analyzed by Student t test. Black arrows indicate smedwi-1+ cells/clusters. The corresponding quantification of smedwi-1 expression, distribution, and classification into 4 categories is shown (n ! 18). Data were analyzed by chi-squared test, applying a Bonferroni correction ( Ã p < 0.05). (D) Anti-H3P immunostaining in gfp and hippo (RNAi) animals irradiated with 17.5 Gy. Corresponding quantification is shown. H3P+ cells in the whole animal were quantified and normalized to the total area of each animal. Error bars represent standard deviation (n ! 13). Data were analyzed by Student t test. Data used in the generation of this figure can be found in S1 Data. Scale bars: 1 mm. dsRNA, double-stranded RNA; gfp, green fluorescent protein; H3P, phospho-histone-H3-Ser10; n.s., not significant; RNAi, RNA interference. (TIF) S10 Fig. Molecular analysis of h2b/hippo (RNAi) animals. (A) Relative expression of h2b and hippo, as determined by qRT-PCR, in planarians 4 days postinjection. Values represent the means of 3 biological replicates. Error bars represent standard deviation. Data were analyzed by Student t test. Ã p < 0.05; ÃÃÃ p < 0.001. (B) EdU labeling in transverse paraffin sections combined with immunostaining with anti-H3P antibody in h2b/gfp (RNAi) and h2b/hippo (RNAi) animals. Arrowhead indicates smedwi-1+ cells, some of which are EdU+. Fifteen sections corresponding to n ! 8 animals were analyzed. (C) In situ hybridization for bruli in h2b/hippo (RNAi) animals and corresponding controls. Black arrows indicate bruli+ cells. Quantification of bruli expression and classification into 2 categories (n ! 6). Data were analyzed by chisquared test, applying a Bonferroni correction; Ã p < 0.05. (D) In situ hybridization for zfp1 in the 4 RNAi conditions, corresponding quantification of zfp1 expression, and classification into 4 categories (n ! 10). Data were analyzed by chi-squared test, applying a Bonferroni correction. Data used in the generation of this figure can be found in S1 Data. Scale bars: 15 μm (B); 1 mm (C and D). EdU, 5-ethynyl-2 0 -deoxyuridine; h2b, histone 2b; H3P, phospho-histone-H3-Ser10; RNAi, RNA interference. (TIF) S1 Movie. Bcat2 staining of a planarian following hippo RNAi. Movie showing anti-Bcat2 immunostaining of a region corresponding to an overgrowth formed following hippo inhibition. Nuclei are labeled with DAPI. Anti-Bcat2 antibody nonspecifically labels adherens junctions in epithelial cells and muscle fibers. Accumulation of nuclei can be observed from the innermost section, corresponding to the area above the muscle plexus, to the outer sections, corresponding to the outermost part of the animal (i.e., the epidermis). The cell-cell junctions of epidermal cells appear disrupted in (RNAi) animals (yellow arrow). In the inner sections, accumulation of cells can be observed among the muscle tissue (white arrow). Bcat2, β-catenin-2; RNAi, RNA interference. (AVI) S2 Movie. Bcat2 and H3P staining of a hippo (RNAi) planarian. Movie showing anti-Bcat2 and anti-H3P immunostaining of a region corresponding to an overgrowth in a hippo (RNAi) animal. Both antibodies are shown in red. The H3P signal but not the Bcat2 signal colocalizes with nuclear DAPI staining. Anti-Bcat2 nonspecifically labels adherens junctions in epithelial cells and muscle fibers. Accumulation of H3P+ nuclei is observed in the marginal portion of the planarian. In the overgrown area, epidermal integrity is maintained despite the loss of Bcat2 signal (which labels adherens junctions) (yellow arrow). Note the appearance of accumulated cells in the muscle layer (white arrow). Bcat2, β-catenin-2; RNAi, RNA interference. (AVI) S3 Movie. Bcat2 and H3P staining in a control planarian. Movie showing anti-Bcat2 and H3P immunostaining in a control animal. Both antibodies are shown in red. The H3P signal but not the Bcat2 signal colocalizes with nuclear DAPI staining. Anti-Bcat2 nonspecifically labels adherens junctions in epithelial cells and muscle fibers. H3P+ nuclei are observed in the innermost sections. Staining of adherens junctions with Bcat2 antibody is evident in the epidermis. Bcat2, β-catenin-2. (AVI) S1 Table. Up-and down-regulation in the hippo (RNAi) RNAseq of genes involved in the cell cycle. Selected up-or down-regulated genes in the RNAseq of hippo (RNAi) animals involved in cell cycle. Each column contains the following information: PlanMine code (74), gene name, gene symbol, E-val (indicating homology with the predicted gene), fold change, adjusted p-value, gene function, and bibliography. Studies demonstrating the role of each gene in the cell cycle or its relation with Hippo pathway are cited. The p-value is adjusted to an FDR of 0.05. FDR, false discovery rate; RNAi, RNA interference; RNAseq, RNA sequencing analysis. (XLSX) S1 Data. All individual numerical values that underlie the summary data shown in the following figures: Fig 1B, 1C, 1D and 1E; Fig 2A, 2B and 2C; Fig 3E; Fig 4A and 4C; Fig 5A,  5B 0 , 5B@ and 5C; Fig 6A, 6B and 6C; S1D and S1E Fig; S2B and S2D Fig; S5 Fig; S6D Fig;  S7A and S7B Fig; S8D Fig; S9C and S9D Fig; and S10A, S10C and S10D Fig.  (XLSX)