Cross-Journal Focus Stem Cells Cross-Journal Focus Stem Cells

EMBO Molecular Medicine

cross-journal focus Stem Cells cross-journal focus Stem Cells

EDITORS Thomas Schwarz-Romond Senior Editor [email protected] | T +49 6221 8891 407

Thomas studied Wnt-signal transduction in the lab of Walter Birchmeier at the Max-Delbrueck Center in Berlin and received his PhD from the FU-Berlin in 2003. He continued to investigate this exciting topic as a postdoctoral fellow with Mariann Bienz at the MRC-LMB in Cambridge and joined The EMBO Journal in 2006.

Barbara Pauly Senior Editor [email protected] | T +49 6221 8891 109

Barbara joined EMBO Reports in September 2008. She completed her PhD at the University of Munich, focusing on signal transduction in the fresh water polyp Hydra. She worked at the University of California at Berkeley as a post-doctoral researcher, studying the role of the actin cytoskeleton in endocytosis in mammalian cells.

Céline Carret Editor [email protected] | T +49 6221 8891 310

Céline Carret completed her PhD at the University of Montpellier, France, characterising host immunodominant antigens to fight babesiosis, a parasitic disease caused by a unicellular EMBO Apicomplexan parasite closely related to the malaria agent Plasmodium. She further developed Molecular her post-doctoral career on malaria working at the Wellcome Trust Sanger Institute in Cambridge, Medicine UK and Instituto de Medicina Molecular in Lisbon, Portugal. Céline joined EMBO Molecular Medicine as a Scientific Editor in March 2011.

Maria Polychronidou Editor [email protected] | T +49 6221 8891 410

Maria received her PhD from the University of Heidelberg, where she studied the role of nuclear membrane proteins in development and aging. During her post-doctoral work, she focused on the analysis of tissue-specific regulatory functions of Hox transcription factors using a combination of computational and genome-wide methods.

emboj.embopress.org | embor.embopress.org | embomolmed.embopress.org | msb.embopress.org full articles

The EMBO Journal Neutral competition of stem cells is skewed by proliferative changes downstream of Hh and Hpo. Amoyel M, Simons BD, Bach EA. DOI: 10.15252/embj.201387500 | Published 04.08.2014

EMBO Reports Review The Hippo signaling pathway in stem cell biology and cancer. Mo JS, Park HW, Guan KL. DOI: 10.15252/embr.201438638 | Published 13.05.2014 Article Heterotrimeric G proteins control stem cell proliferation through CLAVATA signaling in Arabidopsis. Ishida T, Tabata R, Yamada M, Aida M, Mitsumasu K, Fujiwara M, Yamaguchi K, Shigenobu S, Higuchi M, Tsuji H, Shimamoto K, Hasebe M, Fukuda H, Sawa S. DOI: 10.15252/embr.201438660 | Published 26.09.2014

EMBO Molecular Medicine Targeted gene therapy and cell reprogramming in Fanconi anemia. Rio P, Baños R, Lombardo A, Quintana-Bustamante O, Alvarez L, Garate Z, Genovese P, Almarza E, Valeri A, Díez B, Navarro S, Torres Y, Trujillo JP, Murillas R, Segovia JC, Samper E, Surralles J, Gregory PD, Holmes MC, Naldini L, Bueren JA. DOI: 10.15252/emmm.201303374 | Published 23.05.2014

Molecular Systems Biology Intercellular network structure and regulatory motifs in the human hematopoietic system. Qiao W, Wang W, Laurenti E, Turinsky AL, Wodak SJ, Bader GD, Dick JE, Zandstra PW. DOI: 10.15252/msb.20145141 | Published 15.07.2014

For further reading please see inside back cover

Article

Neutral competition of stem cells is skewed by proliferative changes downstream of Hh and Hpo

Marc Amoyel1,*, Benjamin D Simons2,3,4 & Erika A Bach1,5,**

Abstract resulting in chance dominance of a clone at the niche. Neutral competition has been established for both vertebrates and inverte- Neutral competition, an emerging feature of stem cell homeosta- brates and in several different tissues (Clayton et al, 2007; Klein sis, posits that individual stem cells can be lost and replaced by et al, 2010; Lopez-Garcia et al, 2010; Snippert et al, 2010; Doupe their neighbors stochastically, resulting in chance dominance of a et al, 2012; de Navascues et al, 2012). However, the fact that loss clone at the niche. A single stem cell with an oncogenic mutation and gain of stem cells occurs opens the possibility of a transformed could bias this process and clonally spread the mutation through- stem cell exploiting this process in its favor and achieving clonal out the stem cell pool. The Drosophila testis provides an ideal dominance. Such behavior theoretically could underlie the observa- system for testing this model. The niche supports two stem cell tion of tumor-initiating cells in certain types of cancer (Reya et al, populations that compete for niche occupancy. Here, we show that 2001) and has recently been reported for mouse intestinal stem cells cyst stem cells (CySCs) conform to the paradigm of neutral compe- (Vermeulen et al, 2013; Snippert et al, 2014). tition and that clonal deregulation of either the Hedgehog (Hh) or The Drosophila testis provides an ideal system for analyzing single Hippo (Hpo) pathway allows a single CySC to colonize the niche. stem cell behavior. The niche (called the hub) supports two stem cell We find that the driving force behind such behavior is accelerated populations, germ line stem cells (GSCs) and somatic cyst stem cells proliferation. Our results demonstrate that a single stem cell (CySCs) (Fig 1A and de Cuevas & Matunis, 2011; Hardy et al, 1979). colonizes its niche through oncogenic mutation by co-opting an GSCs give rise to sperm, while CySCs produce somatic cyst cells, underlying homeostatic process. which ensheath developing germ cells and are required for germ cell differentiation. Each testis niche harbors approximately 9–14 GSCs, Keywords competition; Hedgehog; Hippo; stem cell; testis which divide with oriented mitosis perpendicular to the niche, such Subject Categories Cell Cycle; Development & Differentiation; Stem Cells that one offspring, likely to remain in contact with the niche, self- DOI 10.15252/embj.201387500 | Received 25 November 2013 | Revised 2 July renews while the other, physically displaced from niche signals, 2014 | Accepted 7 July 2014 | Published online 4 August 2014 begins differentiation (Yamashita et al, 2003; Sheng & Matunis, The EMBO Journal (2014) 33: 2295–2313 2011). Serially reconstructed electron micrographs of wild-type testes revealed ~13 somatic cells, presumed to be the CySCs, in contact with See also: ER Morrissey & L Vermeulen (October 2014) the hub in young adults (Hardy et al, 1979). Most current studies rely on immunofluorescence of nuclear factors in presumptive CySCs and their daughters. The best molecular marker of CySCs is Zfh1, which Introduction labels the nucleus of ~44 cells in wild-type testes (Leatherman & Dinardo, 2008; Inaba et al, 2011; Amoyel et al, 2013). This value The ability of a stem cell to continually generate offspring for tissue substantially overestimates the true number of CySCs and includes maintenance depends on its ability to remain and renew at the post-mitotic daughter cells that no longer contact the niche. Finally, niche. A critical consideration is whether stem cells are eternal and there is no evidence for oriented division among CySCs (Cheng et al, always divide invariantly or whether they function as members of 2011), raising the possibility that this population may be subject to an equipotent population, within which a single stem cell could be different regulation than GSCs. Stem cell loss and replacement has lost and replaced stochastically by a neighbor. Recent work has been observed in Drosophila gonads, in both somatic and germ revealed that the latter, termed neutral competition, is an emerging lineages, but its significance remains under debate (Margolis & feature of stem cell homeostasis. This model states that individual Spradling, 1995; Xie & Spradling, 1998, 2000; Zhang & Kalderon, stem cells can be stochastically lost and replaced by their neighbors, 2001; Wallenfang et al, 2006; Nystul & Spradling, 2007). It remains to

1 Department of Biochemistry and Molecular Pharmacology, New York University School of Medicine, New York, NY, USA 2 Cavendish Laboratory, Department of Physics, University of Cambridge, Cambridge, UK 3 Wellcome Trust-CRUK Gurdon Institute, University of Cambridge, Cambridge, UK 4 Wellcome Trust-Medical Research Council Cambridge Stem Cell Institute, University of Cambridge, Cambridge, UK 5 The Helen L. and Martin S. Kimmel Center for Stem Cell Biology, New York University School of Medicine, New York, NY, USA *Corresponding author. Tel: +1 212 263 7787; Fax: +1 212 263 8166; E-mail: [email protected] **Corresponding author. Tel: +1 212 263 5963; Fax: +1 212 263 8166; E-mail: [email protected] [The copyright line of this article was changed on 24 October 2014 after original online publication.]

ª 2014 The Authors. Published under the terms of the CC BY 4.0 license The EMBO Journal Vol 33 | No 20 | 2014 2295 The EMBO Journal Gain of Hh or Yki cause niche competition Marc Amoyel et al Marc Amoyel et al Gain of Hh or Yki cause niche competition The EMBO Journal

A Figure 1. Characterizing the CySC pool. A Left: Schematic of the apical tip of the Drosophila testis. GSCs (red) and CySCs (dark blue) contact the hub (purple). Differentiating progeny move away from the ◂ hub to form germ cysts (red), which are ensheathed by two cyst cells (light blue). Center: Boxed enlargement showing that CySCs form a ring around the hub and contact the hub in between the GSCs. The CySC nucleus (dark blue) resides just ‘behind’ the row of GSCs. A marked CySC (green) will undergo division with possible outcomes depicted at right. Right: In asymmetric renewal (top), the two daughters of the clone give rise to one CySC and one differentiating cyst cell, which ensheaths a gonialblast along with an unmarked cyst cell (light blue). In duplication (middle), both marked daughters remain at the niche as CySCs, displacing an unmarked CySC (blue) in the process. This displaced unmarked cell differentiates into an ensheathing cyst cell. In differentiation (bottom), both daughters of the marked CySC differentiate into cyst cells, resulting in no marked CySCs at the hub. B A control testis labeled with Stat92E (green, single channel B’), Ptc (red, single channel B”), and Zfh1 (blue, single channel B’”) showing that while some Zfh1- positive cells co-labeled for Ptc and Stat92E (red arrowhead), others were only positive for one factor (yellow arrowhead) or for neither (arrow). C CySC MARCM clones labeled with membrane-targeted CD8-GFP (C’) showing identifiable single cells, some of which contacted the hub (DE-cadherin, blue) with membrane extensions (arrow in C’). D–F Clonal analysis, GFP (single channels D’–F’) indicates the clone, Vasa (red) labels germ cells and Zfh1 (blue) CySCs and early cyst cells; the hub is indicated by a dotted line. GFP-labeled control clones were generated by the MARCM technique and analyzed at 2 (D) and 14 dpci (E, F). Although clones were small at 2 dpci (D), they varied markedly by 14 dpci (E, F). G Variation of average size of control clones as a function of time. The data points (boxes) show the mean fraction of labeled CySCs in persisting clones. The black line shows a fit of the neutral drift model to the data using an induction frequency of CySCs at a ratio of one in 10 (i.e., 10%). The dashed orange line represents the predicted clonal evolution if only a single CySC clone was induced with a time-shift of 3 days with the same set of parameters. One may note that the clone sizes observed from multiple independent induction events and from a single induction event converge rapidly. For details of the neutral drift model and the notation, see Supplementary Materials and Methods. n = 83, 74, 73, 81 for 2, 7, 14, 28 dpci, respectively. Error bars denote SEM. H Comparison of observed (boxes) and predicted (line) frequency of clusters of somatic cell clones. Each cluster is presumed to represent an independent labeling event. The line was generated by a least-squares fit and suggests a labeling efficiency of 11%(q = 0.11). Error bars denote SEM. I Distribution of persisting clone sizes in wild-type testes. The boxes show experimental data, and lines show the predictions of the model. n = 83, 74, 73, 81 for 2, 7, 14, 28 dpci, respectively. Error bars denote SEM. BB′ B′′ B′′′

be resolved whether loss of stem cells reflects their loss of fitness or loss of ptc causes constitutive activation of the pathway. We found represents a normal homeostatic process of neutral competition. that ptc mutant CySCs outcompeted both wild-type CySCs and GSCs The molecular signals governing self-renewal at the testis niche for niche access. We determined that this phenotype was due to have been well characterized (de Cuevas & Matunis, 2011). GSCs biased competition, skewing normal behavioral dynamics in favor of are maintained by Bone Morphogenetic Protein (BMP) signals origi- the mutant cell. We showed that adhesion and JAK/STAT signaling nating from both the hub and CySCs (Shivdasani & Ingham, 2003; could not cause stem cells to acquire colonizing capabilities. Rather, Kawase et al, 2004; Leatherman & Dinardo, 2010). CySCs require at we showed that simply accelerating proliferation was sufficient to least two signaling inputs, the Janus Kinase/Signal Transducer and cause a single CySC and its descendants to outcompete wild-type CC′ DD′ Activator of Transcription (JAK/STAT) and Hedgehog (Hh) path- CySCs and GSCs. Furthermore, we established a critical role for the ways, in order to self-renew (Kiger et al, 2001; Leatherman & conserved growth regulatory Hippo pathway in regulating competi- Dinardo, 2008; Michel et al, 2012; Amoyel et al, 2013). Ligands for tion and self-renewal in CySCs independently of Hh signaling. Thus, both pathways, Unpaired (Upd) and Hh, respectively, are produced we demonstrate that proliferation is the key driver of somatic stem by the hub cells (Forbes et al, 1996; Kiger et al, 2001; Tulina & cell behavior and provide a model for how oncogenic mutations can Matunis, 2001; Dinardo et al, 2011). Two known targets are spread throughout a stem cell pool by exploiting a fundamental expressed in CySCs in response to JAK/STAT pathway activation, homeostatic process of stochastic stem cell replacement. Zfh1 and Chinmo. Overexpression in CySCs of the JAK Hopscotch (Hop) or of either pathway target results in autonomous hyper- E E′ FF′ proliferation of CySCs and non-autonomous hyper-proliferation of Results GSCs, due to BMP production by the CySCs (Leatherman & Dinardo, 2008, 2010; Wang et al, 2008; Flaherty et al, 2010). Conversely, Hh Characterizing the CySC pool activation only regulates the self-renewal and numbers of the CySCs, without affecting the GSC niche (Amoyel et al, 2013). We first attempted to use molecular markers to sub-divide the Although both stem cells co-exist at the same niche, and somatic population near the niche. We reasoned that only a subset although the CySCs are a necessary component of the niche for GSCs, of the ~44 Zfh1-positive cells could constitute the true stem cell pool. these two populations compete for access to the niche, as revealed We therefore examined whether markers of self-renewal pathways by analysis of the mutant phenotype of the JAK/STAT negative feed- in CySCs—Ptc for Hh and Stat92E for JAK/STAT—were co-expressed. GIH back regulator Socs36E (Issigonis et al, 2009; Singh et al, 2010). This We only found expression of these markers in Zfh1-positive somatic reduction of GSCs in Socs36E mutants was attributed to increased cells located one cell diameter from the hub. Within this group, only JAK/STAT signaling in Socs36E mutant CySCs, leading to upregula- a subset co-expressed Ptc and Stat92E (Fig 1B–B’’’, red arrowhead), tion of integrin-based adhesion and enabling the mutant cells to while others expressed only one or neither (Fig 1B–B’’’, yellow displace wild-type GSCs and CySCs from the niche. arrowhead and arrow, respectively). This analysis suggests that Here, we characterize CySC behavior by clonal analysis. We using the best available molecular markers may not be the most found that the behavior of CySCs was consistent with them being robust method to identify CySCs. Since membrane contact with the lost and replaced stochastically, as predicted by the neutral competi- niche appears to be the defining feature of stemness in the Drosophila tion model. For this study, we made clones homozygous mutant for testis (Hardy et al, 1979; de Cuevas & Matunis, 2011), we estimated patched (ptc), which encodes the Hh receptor (Chen & Struhl, 1996); the actual number of CySCs by generating single-cell control

2296 The EMBO Journal Vol 33 | No 20 | 2014 ª 2014 The Authors ª 2014 The Authors The EMBO Journal Vol 33 | No 20 | 2014 2297 The EMBO Journal Gain of Hh or Yki cause niche competition Marc Amoyel et al Marc Amoyel et al Gain of Hh or Yki cause niche competition The EMBO Journal

MARCM clones (Lee & Luo, 1999), expressing a membrane-targeted the total number of CySCs (~16–21 obtained by this method versus frequency of unlabeled testes at 2 dpci, we estimated a labeling effi- of GFP-expressing cells. By comparing the predicted frequency of GFP (Fig 1C). We used this clonal method because labeling all ~13 obtained above). Therefore, to circumvent this uncertainty, we ciency of around 10% for each of the 13 CySCs following the heat clusters with direct measurements at 2 dpci, a least-squares fit somatic cell membranes did not allow us to determine whether an monitored both the total number of GFP-labeled and unlabeled cells shock. At this level of induction, we therefore expect that testes will suggested a labeling efficiency of q = 11% (Fig 1H), consistent individual cell contacts the hub or not. Only 30.5% of Zfh1-positive considered to be contacting the hub and used these values to deter- experience multiple induction events, leading to isolated “clusters” with the observed frequency of unlabeled testes. With the labeling clones (29/95 single cell clones) had membrane extensions contact- mine the fraction of labeled CySCs as a percentage in each testis. ing the hub (Fig 1C’, arrow). Extrapolating this proportion to an At 2 days post-clone induction (dpci), we found few GFP-labeled average of 43 7 Zfh1-positive cells per testis that we counted in CySCs, consistent with a low clone induction rate (Fig 1D and H, AA A A Æ ′ ′′ ′′′ these samples (n = 59), we estimated 13 CySCs per testis, consistent Supplementary Materials and Methods, see below). To characterize with the 12.6 value that has been previously reported (Hardy et al, CySC dynamics, we separated testes according to whether they main- 1979). In the genotype we examined, there were 13.2 GSCs (n = 34). tained at least one GFP-expressing cell in contact with the hub In the Drosophila testis, stem cells are actively dividing, and within (termed ‘persisting’) and those in which all GFP-expressing cells had the somatic lineage, only CySCs divide (Hardy et al, 1979; Inaba detached from the hub (termed ‘differentiating’). We observed empir- et al, 2011). As further confirmation of the number of CySCs, we ically that the mean fraction of labeled CySCs in persisting clones examined markers of cycling cells, PCNA-GFP to mark cells in S-phase increased steadily as a function of time (Fig 1G), while the number of (Thacker et al, 2003) and Cyclin B (CycB) for G2/M. We found 11.2 labeled CySCs in individual clones varied considerably between somatic cells one cell diameter away from the niche undergoing samples at the same time point, as exemplified by the 14 dpci replication that were positive for PCNA (Supplementary Fig S1A–A”’, samples shown in Fig 1E and F. The increased number of labeled arrow). In the same testes, 9.2 out of 12.2 total GSCs on average CySCs in persisting clones is inconsistent with the model of invariant B B′ B′′ B′′′ expressed PCNA-GFP, suggesting a 1.3:1 ratio of CySCs to GSCs and asymmetric stem cell division as in this scenario this parameter by extrapolation a total of ~15 CySCs. Similarly, in an unrelated should not change over time. However, the change observed is genetic background that contained on average 7.9 GSCs, we consistent with CySCs undergoing loss and replacement (Fig 1G). observed 5.6 GSCs and 5.6 CySCs-expressing CycB (Supplementary We next subjected these data to a quantitative analysis, using a Fig S1B–B”’, arrows). Taken together, these data suggest that GSCs parallel approach to that developed to study stem cell dynamics in the and CySCs exist in a ratio close to 1:1. murine intestinal crypt (Lopez-Garcia et al, 2010). The assumptions Two different models have been proposed to explain stem cell contained in the model are the following: (1) CySCs form a single behavior in actively cycling homeostatic tissues; in the first, stem equipotent population in which any cell has an equal chance of being cells are invariant and divide asymmetrically to self-renew and are lost and replaced; (2) in line with the geometry of the testis, the only rarely lost in cases of damage or loss of fitness. In the other, ensemble of CySCs can be approximated as a one-dimensional ‘neck- asymmetry is achieved only at the level of the stem cell population. lace’ of cells around the niche; (3) as CySCs proliferate, some lose In the latter case, stem cell populations are dynamic, and their contact with the niche and differentiate, and this is perfectly compen- CD clonal make-up changes according to stochastic variations such that sated by the duplication of a neighboring CySC to maintain a constant some clones are lost entirely while others expand to occupy empty total number of CySCs. By contrast, an asymmetrical CySC division stem cell berths, through a process termed ‘neutral competition’ leaves the number of labeled CySCs unchanged (Fig 1A). As a simpli- (Simons & Clevers, 2011). fication, we do not take into account GSCs, which are simply regarded We tested these models by generating control FRT42D MARCM as a separate lineage with their own fate behavior. In this modeling clones that mis-expressed only membrane CD8-GFP and scored the scheme, clonal dynamics of the CySC compartment is dependent number of labeled somatic cells contacting the hub. While the upon only two parameters—the CySC loss/replacement rate, k, and membrane labeling of clones allows for direct identification of the total number of CySCs contacting the hub, N (see Supplementary CySCs, this methodology has two drawbacks. First, CySCs outside Materials and Methods). Since CySC divisions that lead to asymmetric the clones (which are unmarked) have to be scored more subjec- fate outcome do not change the number of marked CySCs in a clone, tively by their position relative to the hub. Second, once many cells the clonal fate data are insensitive to the CySC division rate. around the niche are labeled, it becomes difficult to distinguish the To implement the modeling scheme, it is important to define membranes of individual cells, resulting in a slight overestimation of the labeling efficiency of the FRT42D MARCM system. From the EF

Figure 2. Neutral drift dynamics are skewed by ptc mutant clones. A, B Clonal analysis, GFP (single channels A’,B’) indicates the clone, Vasa (red, single channels A”,B”) labels germ cells and Zfh1 (blue, single channels A’”,B’”) CySCs and early cyst cells; the hub is indicated by a dotted line. GFP-labeled ptc mutant clones were generated by the MARCM technique and analyzed at 2 (A) and 14 dpci ▸ (B). Arrow (B–B’”) shows displacement of GSCs by ptc mutant CySCs. C Variation of average size of ptc mutant clones as a function of time. The data points (boxes) show the mean fraction of labeled CySCs in persisting clones. The black line shows a fit of the neutral drift model, modified to have a bias in favor of the labeled cell, to the data using an induction frequency of 10%. The dashed orange line represents the predicted clonal evolution if only a single CySC clone were induced with a time-shift of 3 days with the same set of parameters. One may note that the clone sizes observed from multiple independent induction events and from a single induction event converge rapidly. For details of the biased drift model and the notation, see Supplementary Materials and Methods. n = 63, 81, 79, 66 for 2, 7, 14, 28 dpci, respectively. Error bars denote SEM. D Distribution of clone sizes of persistent ptc mutant clones. The boxes show experimental data, and lines show the predictions of the model. Error bars denote SEM. E Number of unlabeled CySCs at 14 dpci in testes containing either control (green) or ptc mutant (red) clones. Lines show mean and standard deviation. Asterisks denote statistically significant difference from control. n = 73 and 79 for control and ptc mutant, respectively. F Number of GSCs at 14 dpci when control or ptc CySC clones were present. ptc mutant CySCs displaced wild-type GSCs, leading to a significant decrease in the number of GSCs. Asterisks denote statistically significant difference from control. n = 48 and 49 for control and ptc mutant, respectively. Error bars denote SEM.

2298 The EMBO Journal Vol 33 | No 20 | 2014 ª 2014 The Authors ª 2014 The Authors The EMBO Journal Vol 33 | No 20 | 2014 2299 The EMBO Journal Gain of Hh or Yki cause niche competition Marc Amoyel et al Marc Amoyel et al Gain of Hh or Yki cause niche competition The EMBO Journal

efficiency and CySC number defined, only the CySC loss/replace- the number of testes harboring labeled clones decreased over time ment rate, k, remained to be determined. (Supplementary Fig S2G and H). Quantitative analysis of nuclear To fix the loss/replacement rate, we compared the predictions of GFP MARCM control clones (Supplementary Fig S2E–J) revealed the model with measurements of the average fraction of labeled that they obeyed similar neutral drift dynamics to the membrane CySCs in persisting clones using an induction frequency of 10%. CD8-GFP MARCM control clones (Fig 1D–I). Importantly, we were Adjusting the loss/replacement rate, we found that the mean frac- able to infer a CySC loss/replacement rate of around once per day, tion of labeled CySCs could be well reproduced by a loss/replace- which is comparable to the loss/replacement rate of 0.84 per day ment rate of k = 0.84 0.05 per day (Fig 1G, compare lines and inferred from the earlier modeling scheme (see above). Using two Æ boxes). However, alongside the mean fraction, the model also different labeling methods, generating clones on two chromosome DD′ EE′ predicts the variation in the size distribution of individual clones as arms, scoring CySCs by two independent methods, we reach a simi- a function of time post-induction. Taking the inferred loss/replace- lar conclusion: CySCs are lost and replaced stochastically and obey ment rate and induction frequency, we found that the model neutral drift dynamics. provides an excellent prediction of the measured cumulative clone size distribution (Fig 1I, compare lines to boxes), defined as the ptc mutant CySCs skew neutral drift dynamics and outcompete fraction of testes that have a fraction of labeled CySCs larger than wild-type CySCs the given value. (We note that N and k exist in a fixed ratio N2/k (see Supplementary Materials and Methods), meaning that the fit to The dynamics of neutral stem cell competition have been reported a model with a different N would generate quite reasonable agree- in mammalian and Drosophila stem cells (Wallenfang et al, 2006; ment with the data. However, for a larger N, we would require a Clayton et al, 2007; Klein et al, 2010; Lopez-Garcia et al, 2010; proportionately larger k, potentially in excess of the CySC division Snippert et al, 2010; de Navascues et al, 2012), but mutations that F F′ GG′ rate.) Taken together, these empirical and modeling data strongly co-opt the homeostatic mechanisms underlying this process for the suggest that the ~13 CySCs in a wild-type testis produce equivalent benefit of the mutant cell have only recently been described offspring which have stochastic fates. (Vermeulen et al, 2013; Snippert et al, 2014). We and others previ- Given the potential uncertainty of the membrane labeling method ously showed that Hh signal reception is required for the mainte- as a means to identify CySCs, we chose to challenge our findings by nance of CySC fate. CySCs that are unable to transduce the Hh signal following a second unbiased (albeit less direct) approach to measur- are lost from the niche and differentiate (Michel et al, 2012; Amoyel ing CySC number and scoring clones. We generated control MARCM et al, 2013). Here, we studied the effect of clonal gain of Hh signaling clones on two separate chromosome arms (FRT40A and FRT42D) that by making clones homozygous mutant for patched (ptc). Cells lack- mis-expressed only nuclear GFP and scored the size of CySC clones ing ptc function can no longer inhibit Smoothened activity and expe- relative to the total Zfh1-positive population and the clone recovery rience sustained ligand-independent Hh signal transduction (Ingham 42D I rate (percentage of testes with any marked clone positive for Zfh1) et al, 1991; Chen & Struhl, 1996). We examined FRT ptc mutant H at various times after clone induction. We assumed that each CySC CD8-GFP MARCM clones as compared to the appropriate control, contributes equally to the total Zfh1-positive pool and counted the that is, FRT42D CD8-GFP MARCM control clones. Similar to control, number of Zfh1-positive cells (N) that were labeled with the clone we found few GFP-labeled ptc mutant CySCs at 2 dpci (Fig 2A). In marker and expressed this as a fraction N/43 where 43 was the contrast to control clones, ptc mutant clones contained more labeled average number of Zfh1-positive cells found per testis (see above). CySCs on average by 14 dpci and were often seen to take over the At 2 dpci, we found few GFP-labeled Zfh1-positive cells, consistent entire somatic lineage, presumably by causing the displacement of with a low clone induction rate (q = 0.18 for FRT40A and q = 0.3 for wild-type CySCs (Fig 2B, compare boxes in Fig 2C to those in FRT42D (Supplementary Fig S2A and B and Supplementary Materials Fig 1G). We counted unlabeled CySCs in control and ptc samples and Methods)). At 14 dpci, individual clones varied considerably and found that there were significantly fewer when ptc mutant clones between samples at the same time point (Supplementary Fig S2C were present (Fig 2E, P < 0.004). These results indicated that ptc and D). Using this method, we obtained similar results to those mutant CySCs expanded at the expense of their wild-type neighbors. observed for membrane labeling: the mean fraction of labeled CySCs However, like control clones, the frequency of persistent ptc mutant increased as a function of time (Supplementary Fig S2E and F) and CySC clones decreased over time (Fig 8E, red line). These results

2300 The EMBO Journal Vol 33 | No 20 | 2014 ª 2014 The Authors ª 2014 The Authors The EMBO Journal Vol 33 | No 20 | 2014 2301 The EMBO Journal Gain of Hh or Yki cause niche competition Marc Amoyel et al Marc Amoyel et al Gain of Hh or Yki cause niche competition The EMBO Journal

Figure 3. Clonal overactivation of the Hh pathway, but not JAK/STAT, causes niche competition phenotypes. AA′ A′′ A′′′ Α No increase in Stat92E staining (red, single channel in A”) was seen in a ptc mutant CySC (green, arrow, single channel in A’) compared to neighboring wild-type CySCs (arrowhead) at 2 dpci. Somatic cells were labeled with Tj (blue, single channel in A’”). See also Supplementary Fig S5 for Stat92E staining in clones at 7 and ▸ 14 dpci. B–G MARCM clones at 14 dpci, with single channels showing the clone marker GFP in (B’–F’), Vasa in red and Zfh1 (B, C) or Tj (D–F) in blue. Control clones (B, C) showed variation in the number of cells labeled. Overexpression of CiAct (D) or RNAi against ptc (E) recapitulated the ptc mutant phenotype (compare with Fig 2B), but overexpression of Hop did not (F). Hop overexpression activated JAK/STAT signaling (G), as seen by stabilization of Stat92E protein (blue, single channel in G’) in the clones (green, arrows). H Number of GSCs at 14 dpci when CySC clones of the indicated genotype were induced. Hop overexpression did not affect GSC number, while CiAct and ptc RNAi- expressing clones caused loss of GSCs, similar to ptc mutant clones (see Fig 2F). Asterisks denote statistically significant change from control. n = 15, 24, 27, 8 for control, UAS-Hop, UAS-CiAct, UAS-ptc RNAi, respectively. Error bars denote SEM. Ι Number of GSCs at 14 dpci when control or ptc mutant CySCs were present, showing an enhancement of GSC loss when ptc mutant clones were induced in a background lacking one copy of the Stat92E gene. Asterisks denote statistically significant change from the ptc mutant clones alone. n = 48 (control), 36 (control; Stat92E/+), 49 (ptc), 20 (ptc; Stat92E/+). Error bars denote SEM. BB′ CC′ efficiency and CySC number defined, only the CySC loss/replace- the number of testes harboring labeled clones decreased over time ment rate, k, remained to be determined. (Supplementary Fig S2G and H). Quantitative analysis of nuclear To fix the loss/replacement rate, we compared the predictions of GFP MARCM control clones (Supplementary Fig S2E–J) revealed the model with measurements of the average fraction of labeled that they obeyed similar neutral drift dynamics to the membrane CySCs in persisting clones using an induction frequency of 10%. CD8-GFP MARCM control clones (Fig 1D–I). Importantly, we were Adjusting the loss/replacement rate, we found that the mean frac- able to infer a CySC loss/replacement rate of around once per day, tion of labeled CySCs could be well reproduced by a loss/replace- which is comparable to the loss/replacement rate of 0.84 per day ment rate of k = 0.84 0.05 per day (Fig 1G, compare lines and inferred from the earlier modeling scheme (see above). Using two Æ boxes). However, alongside the mean fraction, the model also different labeling methods, generating clones on two chromosome DD′ EE′ predicts the variation in the size distribution of individual clones as arms, scoring CySCs by two independent methods, we reach a simi- a function of time post-induction. Taking the inferred loss/replace- lar conclusion: CySCs are lost and replaced stochastically and obey ment rate and induction frequency, we found that the model neutral drift dynamics. provides an excellent prediction of the measured cumulative clone size distribution (Fig 1I, compare lines to boxes), defined as the ptc mutant CySCs skew neutral drift dynamics and outcompete fraction of testes that have a fraction of labeled CySCs larger than wild-type CySCs the given value. (We note that N and k exist in a fixed ratio N2/k (see Supplementary Materials and Methods), meaning that the fit to The dynamics of neutral stem cell competition have been reported a model with a different N would generate quite reasonable agree- in mammalian and Drosophila stem cells (Wallenfang et al, 2006; ment with the data. However, for a larger N, we would require a Clayton et al, 2007; Klein et al, 2010; Lopez-Garcia et al, 2010; proportionately larger k, potentially in excess of the CySC division Snippert et al, 2010; de Navascues et al, 2012), but mutations that F F′ GG′ rate.) Taken together, these empirical and modeling data strongly co-opt the homeostatic mechanisms underlying this process for the suggest that the ~13 CySCs in a wild-type testis produce equivalent benefit of the mutant cell have only recently been described offspring which have stochastic fates. (Vermeulen et al, 2013; Snippert et al, 2014). We and others previ- Given the potential uncertainty of the membrane labeling method ously showed that Hh signal reception is required for the mainte- as a means to identify CySCs, we chose to challenge our findings by nance of CySC fate. CySCs that are unable to transduce the Hh signal following a second unbiased (albeit less direct) approach to measur- are lost from the niche and differentiate (Michel et al, 2012; Amoyel ing CySC number and scoring clones. We generated control MARCM et al, 2013). Here, we studied the effect of clonal gain of Hh signaling clones on two separate chromosome arms (FRT40A and FRT42D) that by making clones homozygous mutant for patched (ptc). Cells lack- mis-expressed only nuclear GFP and scored the size of CySC clones ing ptc function can no longer inhibit Smoothened activity and expe- relative to the total Zfh1-positive population and the clone recovery rience sustained ligand-independent Hh signal transduction (Ingham 42D I rate (percentage of testes with any marked clone positive for Zfh1) et al, 1991; Chen & Struhl, 1996). We examined FRT ptc mutant H at various times after clone induction. We assumed that each CySC CD8-GFP MARCM clones as compared to the appropriate control, contributes equally to the total Zfh1-positive pool and counted the that is, FRT42D CD8-GFP MARCM control clones. Similar to control, number of Zfh1-positive cells (N) that were labeled with the clone we found few GFP-labeled ptc mutant CySCs at 2 dpci (Fig 2A). In marker and expressed this as a fraction N/43 where 43 was the contrast to control clones, ptc mutant clones contained more labeled average number of Zfh1-positive cells found per testis (see above). CySCs on average by 14 dpci and were often seen to take over the At 2 dpci, we found few GFP-labeled Zfh1-positive cells, consistent entire somatic lineage, presumably by causing the displacement of with a low clone induction rate (q = 0.18 for FRT40A and q = 0.3 for wild-type CySCs (Fig 2B, compare boxes in Fig 2C to those in FRT42D (Supplementary Fig S2A and B and Supplementary Materials Fig 1G). We counted unlabeled CySCs in control and ptc samples and Methods)). At 14 dpci, individual clones varied considerably and found that there were significantly fewer when ptc mutant clones between samples at the same time point (Supplementary Fig S2C were present (Fig 2E, P < 0.004). These results indicated that ptc and D). Using this method, we obtained similar results to those mutant CySCs expanded at the expense of their wild-type neighbors. observed for membrane labeling: the mean fraction of labeled CySCs However, like control clones, the frequency of persistent ptc mutant increased as a function of time (Supplementary Fig S2E and F) and CySC clones decreased over time (Fig 8E, red line). These results

show that ptc clones could differentiate, indicating that they are not and H), similar to the results obtained for the membrane labeling the niche. We previously found no epistatic relationship between constitutively active form of the Hh signal transducer Cubitus inter- locked in a perpetual state of stem cell self-renewal, which is consis- experiment. Finally, we carried out modeling using the same bias as Hh and STAT signaling in the testis (Amoyel et al, 2013). Consistent ruptus (CiAct) (Price & Kalderon, 1999) or of an RNAi hairpin against tent with a prior report (Michel et al, 2012). Indeed, we observed before (Supplementary Fig S4C–E, lines) and found that the compu- with this, we found that Stat92E levels were unchanged in ptc ptc recapitulated the ptc mutant phenotype (Fig 3D and E) and differentiating ptc mutant cyst cells ensheathing spermatogonial cysts tational output for number and distribution of CySCs as well as mutant CySCs (Fig 3A–A’’’, arrow, Supplementary Fig S5). We used caused a statistically significant reduction of GSCs (Fig 3H, similar to control clones (Supplementary Fig S3A and B). This repre- clone recovery rate was well matched to the experimental data for the MARCM technique to assess whether the clonal overexpression P < 0.0001 for both CiAct and ptc RNAi), thus validating our tech- sents a different situation from a previously observed instance of FRT42D ptc mutant nuclear GFP MARCM clones. Taken together, the of various factors could induce niche competition. As expected, nique. Surprisingly, clonal hyper-activation of Stat92E by mis- stem cell competition, where stem cells in the ovary that cannot data indicate that the behavior of ptc mutant clones is consistent control clones that only overexpressed GFP had variable clone sizes expression of Hop did not cause CySC clones to compete with either differentiate eventually replaced their wild-type neighbors (Jin et al, with a biasing of competition between stem cells. (Fig 3B and C). By contrast, clonal overexpression of either a wild-type CySCs or with GSCs (Fig 3F and H), despite clearly 2008). Thus, these data suggest that ptc mutant CySCs have a competitive advantage over wild-type CySCs in effecting stem cell ptc mutant CySCs outcompete wild-type GSCs replacement. To assess whether the dynamics of ptc mutant clones represent a As a readout for the competitive activity of ptc mutant CySCs, we AA′ BB′ biasing of the neutral competition process toward persistence, we also quantified the number of GSCs (defined as Vasa-positive cells sought for the simplest revision of the neutral drift model which in contact with the niche) in testes with control or ptc mutant could accommodate the observed behavior. In particular, we CySCs at 14 dpci. Indeed, we found that GSC number was signifi- assumed that, following the loss of a CySC (control or ptc mutant) cantly reduced (P < 0.0001) non-autonomously when ptc mutant through commitment to differentiation, a neighboring ptc mutant CySC clones were present (Fig 2F, red bar). At the same time, colo- CySC will have a higher chance of replacing it through symmetric nizing CySCs contacted the hub in place of the outcompeted GSCs cell division than a wild-type neighboring CySC. We also assumed (Fig 2B, arrow, Supplementary Fig S4B), similar to the phenotype that the competitive advantage of the ptc mutant CySC is sustained described for Socs36E (Issigonis et al, 2009). GSC loss was only since the loss rate of CySCs is not differentially affected by ptc muta- observed once the majority of CySCs were replaced by ptc mutant tion. Once again, using the frequency of unlabeled testes at 2 dpci, CySCs (Supplementary Fig S4F), suggesting a sequential outcompe- we inferred a CySC labeling efficiency of around 10%, similar to the tition of first wild-type CySCs and then wild-type GSCs by ptc CC′ DD′ control. Then, taking the loss/replacement rate of wild-type CySCs mutant CySCs. The fact that ptc mutant CySCs had normal levels of to be unperturbed from its control value, by adjusting the bias of ptc factors that mediate GSC extended niche function, that is, Stat92E mutant CySCs away from loss and toward replacement (by around and Zfh1 (Leatherman & Dinardo, 2010) (Fig 3A, Supplementary 35%), we found a good agreement between the model dynamics Fig S5, arrows, for Stat92E; Fig 2B, Supplementary Fig S4B for and the experimental data (Fig 2C, compare lines and boxes, and Zfh1), strongly suggests that GSC loss is not due to lack of appro- Supplementary Materials and Methods). Significantly, taking these priate support from ptc mutant CySCs. Thus, gain of Hh signaling model parameters, comparison of the cumulative clone size distribu- results in niche colonization by the mutant cell, as a consequence tion revealed an excellent agreement of the model prediction with of the displacement of resident wild-type CySCs and GSCs at the the data over the range of time points (Fig 2D, compare lines and niche. boxes). Once again, we repeated this experiment using the alternative JAK/STAT signaling, adhesion and cell competition factors are not EF G labeling and scoring method in which CySC number is estimated by causal to niche competition the labeled fraction of Zfh1-expressing cells. We analyzed FRT42D ptc mutant nuclear GFP MARCM clones as compared to FRT42D Several possibilities could explain niche colonization by ptc mutant nuclear GFP MARCM control clones (compare Supplementary Fig CySCs. We ruled out the trivial explanation that the niche size was S4A and B to Supplementary Fig S2B and D). We found that the altered in testes with ptc mutant clones (Supplementary Table S1). number of these ptc mutant CySCs increased faster than control We next tested whether an increase in integrin-based adhesion CySCs and that they were lost less frequently (compare boxes in downstream of Stat92E, as proposed for Socs36E mutants (Issigonis Supplementary Fig S4C and D to those in Supplementary Fig S2F et al, 2009), caused ptc mutant CySCs to anchor more securely to

Figure 4. Increased adhesion is not causal to niche competition. HI A, B ptc mutant clones did not upregulate adhesion molecules. No change in bPS-integrin (A, red, single channel in A’) or in DE-cadherin expression (B, blue, single channel in B’) was seen at the hub in testes with ptc mutant clones (green). Vasa labels germ cells in red (B), Tj labels somatic cells in blue (A). The hub is indicated with a dotted line. ▸ C, D GFP-positive MARCM clones (green, single channels in C’,D’) overexpressing bPS-integrin (C) or DE-cadherin (D) did not outcompete neighboring wild-type CySCs or GSCs. Vasa labels germ cells in red, Tj labels somatic cells in blue. The hub is indicated with a dotted line. E, F Control (E) and rhea mutant (F) MARCM clones showing marked CySCs which contacted the hub at 7 dpci (arrows). Vasa labels germ cells in red, Tj labels somatic cells in blue. The hub is indicated with a dotted line. G CySC clone recovery rates at 2 (blue bars) and 7 (red bars) dpci for control (left) and rhea mutant (right). The presence of rhea mutant clones at the niche at the 7-day time point indicates that rhea was not required in CySCs for self-renewal. n = 38 and 24 for control at 2 and 7 dpci, respectively, and n = 9 and 49 for rhea1 at 2 and 7 dpci, respectively. H Number of GSCs present when CySC clones of the indicated genotype were generated at 14 dpci. Overexpression of bPS-integrin, TalinH or DE-cadherin did not affect GSC numbers. n = 15, 19, 25, 17 for control, UAS-bPS-integrin, UAS-TalinH or UAS-DE-cadherin, respectively. Error bars denote SEM. Ι Number of GSCs when ptc mutant CySC clones were present along with a single mutant copy of the indicated genes at 14 dpci. Reduction of a-cat had no effect on the ptc phenotype, while one rhea allele partly suppressed GSC loss. n = 48 (control); 21 (control; rhea1/+); 49 (ptc); 26 (ptc; a-cat/+); 19 (ptc; rhea6-66/+); 35 (ptc; rhea1/+). Error bars denote SEM.

2302 The EMBO Journal Vol 33 | No 20 | 2014 ª 2014 The Authors ª 2014 The Authors The EMBO Journal Vol 33 | No 20 | 2014 2303 The EMBO Journal Gain of Hh or Yki cause niche competition Marc Amoyel et al Marc Amoyel et al Gain of Hh or Yki cause niche competition The EMBO Journal

Figure 5. ptc mutant CySCs proliferate faster than controls. AA′ A′′ A′′′ A, B There was an increase in the S-phase index in CySCs mutant for ptc. Quantification of S-phase in control (A) or ptc mutant (B) clones. Clones expressing GFP (green, single channel A’,B’) were labeled with Tj (red, single channel A”,B”) and EdU (blue, single channel A”’,B’”). Triply labeled cells (yellow arrowheads) were counted as ▸ a ratio of total cells double positive for GFP and Tj, with quantification shown in (E). C stg-GFP (green, single channel C’) was upregulated in ptc mutant CySCs (yellow arrowheads). Zfh1 (red, single channel C”) labels CySCs, and their offspring and clones are identified by loss of the bgal marker (blue, single channel C’”). D PCNA-GFP (green, single channel D’) was upregulated in ptc mutant clones. Clones are labeled by loss of bgal (blue, single channel D”’). Zfh1 (red, single channel D”) marks CySCs and their offspring. Arrow shows control CySC, and arrowhead shows a ptc mutant CySC. E S-phase index. See legend of (A) above. Asterisks denote statistically significant change from control. Error bars denote SEM. F Quantification of PCNA-GFP fluorescence intensity in control or ptc mutant CySCs. n = 11 for both genotypes. An asterisk denotes statistically significant change from control. Error bars denote SEM.

BB′ B′′ B′′′ elevated levels of Stat92E in the clone (Fig 3G, arrows), a well- predominate in a M/+ heterozygous background, contained a established readout of Stat92E activity (Chen et al, 2002). We normal complement of GSCs (Supplementary Fig S6B). Finally, we reasoned that if the ptc mutant phenotype was due to elevated found no evidence of cell death in testes with ptc mutant clones Stat92E levels in CySCs, we could suppress ptc-dependent GSC loss (Supplementary Fig S6C and D), and removing a copy of the pro- by removing a copy of Stat92E. The number of GSCs in Stat92E/+ apoptotic gene hid (which suppresses dMyc-dependent cell competi- heterozygotes was indistinguishable from wild type (Fig 3I, dark tion) did not suppress ptc-dependent competition (Supplementary green bar). However, GSC loss was actually enhanced when ptc Fig S6E, red bar). mutant clones were induced in a Stat92E/+ heterozygous background (Fig 3I, red bar, P < 0.008), presumably due to the role of ptc mutant CySCs proliferate faster than controls Stat92E in GSC-hub adhesion (Leatherman & Dinardo, 2010). We also examined whether the ptc mutant phenotype could be Having ruled out increased JAK/STAT signaling or adhesion as CC′ C′′ C′′′ ascribed to changes in cell-matrix (integrin) or cell–cell (cadherin) causal factors in niche competition, we reasoned that proliferation adhesion. We did not detect changes in bPS-integrin in ptc clones might be a driving force of clone dominance within the stem cell (Fig 4A, arrow), in contrast to the observations reported for Socs36E pool. We therefore tested whether ptc mutant CySCs had an altered mutants (Issigonis et al, 2009). Furthermore, clonal overexpression rate of proliferation relative to control clones and, in doing so, might of bPS-integrin, or a dominant-active form of Talin (TalinH), which outcompete wild-type CySCs in the race to replace neighbors. Label- strengthens integrin adhesion (Tanentzapf & Brown, 2006), neither ing with 5-ethynyl-20-deoxyuridine (EdU) revealed that ptc mutant recapitulated the ptc phenotype nor induced competition with CySCs CySCs had a higher S-phase index than control clones (Fig 5A, B and GSCs (Fig 4C and H). Importantly, we found that rhea, which and E, P < 0.0001). The E2f-responsive reporter PCNA-GFP is a encodes the Drosophila Talin, was dispensable for CySC self- marker of S-phase and was normally expressed at higher levels in renewal (Fig 4E–G). DE-cadherin levels did not change in ptc GSCs than in CySCs (Supplementary Fig S1A’). However, in ptc mutant clones (Fig 4B, arrow). Moreover, clonal mis-expression of mutant CySCs, PCNA-GFP was upregulated to the level observed in D D′ D′′ D′′′ DE-cadherin also did not cause niche colonization (Fig 4D and H). GSCs (Fig 5D, arrowhead, quantified in Fig 5F). ptc mutant CySC Furthermore, competition caused by ptc mutant clones was not clones also had increased M phase (Michel et al, 2012), suggesting altered by reducing the genetic dose of a-Catenin, which connects that ptc clones accelerate proliferation as opposed to shortening only DE-cadherin to the cytoskeleton (Sarpal et al, 2012) (Fig 4I). one phase of the cell cycle. We examined a protein-trap reporter Although one mutant allele of rhea partially suppressed the ptc for string (stg, the cdc25 homolog in Drosophila; Edgar & phenotype (Fig 4I, P < 0.65 for rhea1 and P < 0.051 for rhea6-66), O’Farrell, 1989). Stg-GFP was upregulated in ptc mutant cells this is likely to be an indirect effect of loosening the tethering of the (Fig 5C, arrowheads). hub to the muscle sheath and allowing more stem cells to surround the hub (Tanentzapf et al, 2007). Consistent with this, there were Increased proliferation downstream of ptc is necessary and more GSCs in testes from rhea/+ heterozygous animals (Fig 4I). sufficient for colonizing behavior These data strongly suggest that increased adhesion does not skew EF neutral drift dynamics in CySCs. We next addressed if the competitive behavior of ptc mutant CySCs An alternative explanation for the ptc phenotype is that ptc depended on their ability to increase their proliferation rate. To mutant CySCs induce death in neighboring wild-type cells, akin to accomplish this, we removed one copy of stg and counted the classical cell competition in which more robust cells kill and take number of labeled CySCs and of GSCs at the niche. In a stg/+ hetero- the place of weaker cells (Amoyel & Bach, 2014). A key process in zygous background, the number of ptc mutant CySCs was signifi- cell competition is ribosomal function, which in turn is dependent cantly reduced (Fig 6A, P < 0.034), suggesting that ptc mutant on optimal levels of the cellular growth regulator dMyc and of ribo- CySCs have a reduced competitive advantage when stg is limiting. somal subunits, encoded by Minute genes (M) (Morata & Ripoll, In addition, in a stg/+ background, the outcompetition of GSCs by 1975; de la Cova et al, 2004; Moreno & Basler, 2004). Clonal overex- ptc mutant CySCs was significantly suppressed (Fig 6B, red bar, pression of dMyc, which causes cell competition in imaginal discs P < 0.008). We note that the number of GSCs was not changed (de la Cova et al, 2004), did not cause niche colonization or loss of in stg/+ heterozygotes when control clones were present (Fig 6B, GSCs (Supplementary Fig S6A and B). Similarly, testes from a M/+ dark green bar). These data indicate that increased proliferation animal harboring wild-type clones (labeled M+), which normally downstream of Ptc is necessary for niche competition in the

2304 The EMBO Journal Vol 33 | No 20 | 2014 ª 2014 The Authors ª 2014 The Authors The EMBO Journal Vol 33 | No 20 | 2014 2305 The EMBO Journal Gain of Hh or Yki cause niche competition Marc Amoyel et al Marc Amoyel et al Gain of Hh or Yki cause niche competition The EMBO Journal

A B Figure 6. Increased proliferation downstream of ptc is necessary and sufficient for colonizing behavior. A, B Loss of one copy of stg suppressed the ptc mutant phenotype at 14 dpci. Graph showing number of labeled CySCs in the indicated genotypes (A). Lines in (A) show mean ◂ and standard deviation. n = 36 (control), 59 (ptc), 31 (ptc; stg/+). Graph showing the number of GSCs when CySC clones were present in the indicated genotype (B). Asterisks denote statistically significant change from the ptc mutant clones alone (A, B). n = 48 (control), 10 (control; stg/+), 49 (ptc), 17 (ptc; stg/+). Error bars in (B) denote SEM. C Graph showing the number of GSCs when CySC clones were present in the indicated genotypes at 14 dpci. An asterisk indicates statistically significant suppression of the ptc mutant phenotype. n = 48 (control), 25 (control; Akt/+), 21 (control; InR/+), 30 (control; S6k/+), 19 (control; cdk2/+), 21 (control, E2f/+), 49 (ptc), 21 (ptc; Akt/+), 30 (ptc; InR/+), 35 (ptc; S6k/+), 28 (ptc; cdk2/+), 30 (ptc, E2f/+). Error bars denote SEM. D Graph showing the total number of labeled cells within control clones (blue line) or clones overexpressing UAS-CycE + UAS-Stg (red line) at 48, 72, and 96 h pci. n = 21, 21, and 26 for control at 48, 72, and 96 h pci, respectively. n = 6, 12, and 15 for UAS-CycE+UAS-Stg at 48, 72 and 96 h pci, respectively. E Clonal overexpression of CycE and Stg caused CySC overproliferation and GSC loss. Clones are labeled by GFP expression (green, single channel E’), Vasa (red, single channel E”) marks germ cells, and Tj (blue, single channel E’“) marks the somatic lineage. The hub is indicated by a dotted line. F Quantification of GSC loss in the presence of CycE+Stg overexpressing CySC clones. Asterisks denote statistically significant difference from control. n = 15 and 22 for control and UAS-CycE, UAS-Stg, respectively. Error bars denote SEM.

C D were found with cdk2, which encodes a cyclin-dependent kinase the activity of the transcriptional co-activator Yorkie (Yki), the (Lehner & O’Farrell, 1990) (Fig 6C, P < 0.04, and P < 0.02, respec- Drosophila homolog of Yes-associated protein (YAP), which is onco- tively). We note that GSC number was not changed in E2f/+, cdk2/+ genic in flies and mice (Dong et al, 2007). We noted that this path- or in any of the heterozygous backgrounds tested below (Fig 6C). In way was active in the soma, as seen by expression of the pathway Drosophila, cellular growth and proliferation are genetically separable target expanded (ex)-lacZ (Hamaratoglu et al, 2006) (Fig 7A). Next, (Neufeld et al, 1998), so we also tested whether increased cellular we generated hpo mutant clones and measured the number of growth was required for CySC colonization. Removal of one copy of mutant CySCs at several time points. We note that FRT42D CD8-GFP the gene encoding the Drosophila Insulin receptor, InR, or genes hpo mutant MARCM clones were induced at rates comparable to encoding its effectors Akt1 and S6k, did not suppress niche coloniza- FRT42D CD8-GFP control MARCM clones (compare Fig 7B to tion by ptc mutant clones (Fig 6C) (Chen et al, 1996; Montagne et al, Fig 1D). Strikingly, hpo mutant clones displayed overproliferation 1999; Verdu et al, 1999). In fact, clonal mis-expression of Drosophila and colonized the niche at the expense of wild-type stem cells Phosphoinositide 3-kinase (PI3K) Dp110, or clonal loss of PI3K path- (Fig 7C and F). Importantly, hpo mutant cyst cells differentiated way inhibitors Tsc1 or Pten, was incompatible with CySC fate normally and were readily observed ensheathing spermatogonial (Supplementary Fig S7 and Supplementary Table S2) (Leevers et al, cysts (Supplementary Fig S3C). E E′ E′′ E′′′ 1996; Goberdhan et al, 1999; Potter et al, 2001; Tapon et al, 2001). We applied the same quantitative analysis to hpo mutant clones Consistently, we also recovered fewer dMyc-expressing CySC clones as described above for ptc mutant clones. Noting that the labeling at 14 dpci compared to control (Supplementary Table S2). Thus, cell efficiency of the CySC was comparable to that of the control and ptc cycle progression is essential for CySCs to gain an advantage over mutant (at around 10%), we used the same strategy to analyze the their neighbors at the niche, while excessive activation of cellular clonal fate data. In doing so, we found that the behavior of hpo growth pathways like PI3K and dMyc is detrimental to CySC mutant clones was consistent with a bias in neutral drift in favor of function. the mutant cell (Fig 7D and E, compare lines and boxes), quantita- We next tested whether increasing proliferation was sufficient to tively similar to the trend we found for ptc mutant clones (compare cause niche competition by expressing the G1/S-phase promoting Fig 7D and E with Fig 2C and D). Indeed, within error bars, we factor CyclinE (CycE) and G2/M-phase promoting factor Stg together could discern no distinction between the bias for ptc and hpo in a clonal fashion. In imaginal discs, clonal overexpression of these mutants. Furthermore, hpo mutant CySCs displaced GSCs from the factors together led to marked acceleration of the cell cycle and niche (Fig 7F, P < 0.0001 hpo versus control), similar to the compet- F increased cell number (Neufeld et al, 1998). We found that CycE+Stg itive behavior of ptc mutant clones (Fig 7F, P < 0.0001 ptc versus overexpressing clones with at least one labeled CySC grew at a faster control). Like ptc-dependent niche colonization, the loss of GSCs rate than control clones, indicating that CycE+Stg overexpression caused by hpo mutant CySCs could be suppressed by removing one also led to cell cycle acceleration in the testis (Fig 6D). Strikingly, copy of stg (Fig 7F, dark red bar, P < 0.0001 hpo versus hpo; stg/+). CycE+Stg overexpressing clones outcompeted both wild-type CySCs We next tested the role of the Hpo pathway effector Yki in niche and GSCs at the niche (Fig 6E and F, P < 0.004), in a manner remi- competition. Clonal mis-expression of an activated form of Yki niscent of ptc mutant clones. Combined, these data indicate that (YkiAct (Oh & Irvine, 2008)) resulted in CySC clones that out- proliferation downstream of ptc is necessary and sufficient to induce competed wild-type CySCs and GSCs at the niche (Fig 7G and H, competition at the niche. Thus, altering the rate of cell division skews P < 0.0014). Consistent with an essential role of yki in CySCs, we the stochastic process of stem cell loss and replacement at the niche found that yki was required autonomously for self-renewal in CySCs in favor of the faster proliferating CySCs, and disrupting the normal but not in GSCs (Fig 8A, E and F), the latter consistent with a prior homeostatic balance between GSCs and CySCs, in favor of the latter. report (Sun et al, 2008). Finally, we addressed whether Hh and Hpo, two proliferative pathways in CySCs, were epistatic. To test The Hippo pathway regulates proliferation, self-renewal, and this, we generated clones that were mutant for both ptc and yki, Drosophila testis and that CySC–CySC and CySC–GSC competitive pathways are normally active in CySCs, we examined other cellular niche competition independently of Hh with the expectation that loss of yki would suppress the competi- interactions are related, making GSC number a good readout for growth and proliferation factors for their ability to rescue the ptc tiveness in ptc mutant CySCs. Indeed, CySCs lacking ptc and yki CySC competitiveness. mutant phenotype when reduced. Removing one copy of the gene As a proof of concept for the central role of proliferation in niche did not overproliferate and colonize the niche (Fig 8B, C and E, To corroborate the hypothesis that increased proliferation is E2f, which encodes an S-phase regulator (Duronio et al, 1995), competition, we examined a universal regulator of proliferation, the compare red to purple line), indicating that Hpo is epistatic to necessary for niche competition by CySCs and to determine which partially suppressed the loss of GSCs; similar genetic interactions Hippo (Hpo) pathway (Pan, 2010) using clonal assays. Hpo restrains Hh signaling in the testis. However, we observed no change in

2306 The EMBO Journal Vol 33 | No 20 | 2014 ª 2014 The Authors ª 2014 The Authors The EMBO Journal Vol 33 | No 20 | 2014 2307 The EMBO Journal Gain of Hh or Yki cause niche competition Marc Amoyel et al Marc Amoyel et al Gain of Hh or Yki cause niche competition The EMBO Journal

AA′ A′′ A′′′ Figure 7. hpo mutant CySCs also skew neutral drift dynamics and outcompete GSCs. ◂ Α ex-lacZ (green, single channel A’) expression in the testis was observed in the somatic lineage (Zfh1, blue, single channel A’“) near the hub (DE-cadherin, red, single channel A”). B, C Clonal analysis, GFP (single channels, B’,C’) indicates the clone, Vasa (red) labels germ cells and Zfh1 (blue) CySCs and early cyst cells; the hub is indicated by a dotted line. GFP-labeled hpo mutant clones were generated by the MARCM technique and analyzed at 2 (B) and 14 dpci (C). Arrow (C, C’) shows displacement of wild-type GSCs by hpo mutant CySCs. D Variation of average size of hpo mutant clones as a function of time. The data points (boxes) show the mean fraction of labeled CySCs in persisting clones. The black line shows a fit of the neutral drift model, modified to have a bias in favor of the labeled cell, to the data using an induction frequency of 10%. The dashed orange line represents the predicted clonal evolution if only a single CySC clone were induced with a time-shift of 3 days with the same set of parameters. For details of the biased drift model and the notation, see Supplementary Materials and Methods. n = 69, 64, 71, 43 for 2, 7, 14, 28 dpci, respectively. Error bars denote SEM. E Distribution of clone sizes of persistent hpo mutant clones. The boxes show experimental data, and lines show the predictions of the model. Error bars denote SEM. F Graph showing the number of GSCs at 14 dpci when CySC clones of the indicated genotype were present. Asterisks denote statistically significant change for the comparisons indicated. n = 26 (control), 36 (ptc), 47 (hpo), 15 (hpo; stg/+). Error bars denote SEM. G, H YkiAct-overexpressing clones outcompeted both CySCs and GSCs at 14 dpci. Graph in (G) shows the number of GSCs when clones overexpressing YkiAct were B B′ CC′ present. Asterisks denote statistically significant change for the comparisons indicated. n = 15 (control), 19 (UAS-YkiAct). Error bars denote SEM. Clones are green (single channel H’), Vasa is red (single channel H”), and Tj is blue (single channel H”’). The hub is indicated by a dotted line.

expression of the Yki target gene ex-lacZ in ptc mutant clones replication of DNA versus rapid production of offspring) determine (Fig 8D, arrow), suggesting no direct link between these pathways which type of self-renewal strategy a stem cell adopts. in this tissue. These data establish Yki as a central regulator of somatic stem cell fate in the testis and suggest a parallel require- Characterizing the testis stem cell niche ment for the Hh and Hpo pathways in CySC proliferation, through independent or convergent control of cell cycle progression genes. Our study revealed an unexpected ratio of CySCs to GSCs, close to 1:1 and different from the 2:1 ratio described by Hardy et al. DE However we note that both studies find the same number of CySCs Discussion (approximately 13), and that the difference resides in the number of GSCs. Indeed, Hardy et al find a ratio of 1.3 CySCs:1 GSC in larval In this study we characterized the behavior of somatic CySCs in the testes which increases to 1.8:1 in young adults, due entirely to a Drosophila testis and explored the molecular mechanisms that regu- drop in the number of GSCs (Hardy et al, 1979). This may be a func- late their ability to compete with their neighbors for limited space at tion of the genetic background used by these authors, as we estab- the niche. We found that single stem cell clones bias stem cell lished our 1:1 ratio through three different experiments in distinct replacement dynamics in their favor, leading to non-neutral compe- genetic stocks. Although the analysis of the data is consistent with tition, when they had increases in Hh signaling, Yki activity or in neutral competition between 13 equipotent CySCs, by the nature of the rate of proliferation, but not when JAK/STAT signaling or adhe- the neutral competition model, we cannot rule out the possibility sion were dys-regulated. Furthermore, we found that the dynamics that the stem cell compartment is heterogeneous with cells moving of CySCs were well-described by a model in which they were contin- reversibly between states in which they become primed for duplica- FG ually and stochastically lost and replaced, leading to neutral drift tion or loss, as recently defined in the mouse intestinal crypt dynamics and a consolidation of clonal diversity. (Ritsma et al, 2014). In this case, the effective number of CySCs This observation contrasts with the dynamics of GSC offspring may be smaller than the observed figure of N = 13, while the true fate choices, where oriented divisions and mother centromere reten- loss/replacement rate, k, might be proportionately adjusted to a tion determine which cells remain as stem cells and which are thrust lower value such that the ratio N2/k remains constant. out of the niche to differentiate (Yamashita et al, 2003, 2007; Sheng & Matunis, 2011). However, careful analysis of GSC dynamics has Mechanisms of niche competition by CySCs suggested that they also undergo neutral competition, albeit at a slower loss/replacement rate than CySCs (Wallenfang et al, 2006; Our results also show that the predominant force driving niche colo- Sheng & Matunis, 2011; Salzmann et al, 2013). Thus, within the nization by CySCs is proliferation. How proliferation causes stem same stem cell niche, two markedly different strategies for self- cells to replace neighbors more efficiently is not established by this HH′ H′′ H′′′ renewal are in use, exemplified by the requirement for yki in CySC study. However, we hypothesize that in such a competitive situa- self-renewal, but not in GSC self-renewal (this study and Sun et al, tion, the rate of stem cell loss is not altered but the overproliferating 2008). This is particularly surprising as the two stem cell populations mutants simply produce more offspring, which are in the right place are by necessity linked, in that they need to produce offspring in the to fill a vacant seat at the niche. It remains possible that a mecha- correct ratio, as well as the fact that CySCs support GSC self-renewal nism of active displacement is involved in CySC dominance (i.e., through BMP production (Leatherman & Dinardo, 2010). It has been the colonizing stem cells crowd out the wild-type ones), and live- hypothesized that the careful choice of stem cell retention in the GSC imaging of competing clones might distinguish between passive pool is a requirement of their role in preserving the genetic integrity replacement and active displacement. of the species (Yuan & Yamashita, 2010). CySCs are under no such A related issue is how CySCs outcompete GSCs. We found that constraint, and moreover, need to proliferate twice as fast in order to GSC loss is only observed after most of the CySC pool is comprised produce two cyst cells for every germ cyst (Inaba et al, 2011). Thus of colonizing mutant CySCs (Supplementary Fig S4F). We therefore it may be that the functional imperatives of the tissue (e.g., careful favor the model that competition among CySCs for niche space

2308 The EMBO Journal Vol 33 | No 20 | 2014 ª 2014 The Authors ª 2014 The Authors The EMBO Journal Vol 33 | No 20 | 2014 2309 The EMBO Journal Gain of Hh or Yki cause niche competition Marc Amoyel et al Marc Amoyel et al Gain of Hh or Yki cause niche competition The EMBO Journal

AA′ A′′ A′′′ Figure 8. Hh and Yki regulate niche competition and self-renewal independently. A–C yki mutant CySCs did not self-renew (A) and loss of yki suppressed the ptc colonization phenotype (C). Testes with negatively marked yki (A), ptc (B), or ptc yki ◂ double mutant (C) clones at 7 dpci. Absence of GFP (single channel A’,B’,C’) marks the clones, Vasa (single channel A”,B”,C”) is red, and Tj (single channel A”’,B’”, C’”) is blue. Arrows point to differentiated mutant cyst cells in (A) and (C). D ptc mutant clones (arrow) did not display increases in ex-lacZ expression compared to wild type (arrowhead). Clones are labeled by GFP (single channel D’), b-gal is in red (single channel D”), and Zfh1 is in blue (single channel D’”). The hub is outlined with a dotted line in all panels. E Graph of clone recovery rates over time for the indicated genotypes. n = 61, 21, 88 for control at 2, 7, 14 dpci, respectively; n = 46, 39, 89 for ptc at 2, 7, 14 dpci, respectively; n = 50, 94, 35 for yki at 2, 7, 14 dpci, respectively; n = 40, 59, 64 for ptc yki at 2, 7, 14 dpci, respectively; n = 39, 67 for hpo at 2, 14 dpci, respectively. F yki mutant GSCs could be recovered at 2 and 7 dpci. yki mutant CySCs could be recovered at 2 dpci but not 7 dpci. n = 61, 21 for control at 2, 7 dpci, respectively. n = 50, 94 for yki at 2, 7 dpci, respectively.

BB′ B′′ B′′′ precedes that between CySCs and GSCs. It is unclear whether the Materials and Methods numerous offspring of the competitive CySCs are passively replacing GSCs that have spontaneously left a vacancy at the niche, or whether Fly stocks and genotypes are described in Supplementary Materials colonizing CySCs actively push the GSCs out of the niche. The latter and Methods. For ptc mutants, ptcS2 was used in all experiments scenario is reminiscent of competition among GSCs in the Drosophila shown, but similar results were obtained with ptcIIw. hpoKC202 ovary, where the contact area between the GSC and niche depended phenotypes were confirmed using hpo42-47. on DE-cadherin. GSCs that elevated cadherins adhered better to the Freshly eclosed adult males were aged for 1 day and then heat niche and caused the physical displacement of neighbors (Jin et al, shocked for 1 h at 37°C to induce clones and raised at 25°C until the 2008; Tian et al, 2012). We explored the contribution of integrin- appropriate time for dissection. For self-renewal assays, CySCs were and cadherin-based adhesion and found that neither affected the scored as Zfh1-positive or Tj-positive cells one cell diameter away competitiveness of CySCs. Moreover, we found that integrin binding from the hub, and GSCs as Vasa-positive cells in contact with the CC′ C′′ C′′′ was entirely dispensable for CySC self-renewal, unlike cadherin hub. For control, ptc or hpo CySCs, the method of counting is (Voog et al, 2008). Importantly, clonal gain of integrin or cadherin detailed in the text. did not lead to niche colonization, indicating that they are not Dissections and immunohistochemistry were performed as previ- instructive for CySC maintenance. Moreover, we found no role for ously described (Flaherty et al, 2010). Primary antibodies used were JAK/STAT signaling in inducing competition at the niche. The fact rabbit anti-GFP (1:500, Invitrogen), mouse anti-GFP (1:500, Invitro- that neither Stat92E nor integrin was causal to colonization in clonal gen), chicken anti-b-galactosidase (1:250, Immunology Consultants assays is surprising because both were ascribed critical roles in Lab), goat anti-Vasa (1:400, Santa Cruz), rabbit anti-Zfh1 (1:5,000, CySC-dependent niche competition (Issigonis et al, 2009). The gift of Ruth Lehmann), guinea pig anti-Zfh1 (1:1,000, gift of James reasons for the difference in results by our group and the previous Skeath), guinea pig anti-Tj (1:3,000, gift of Dorothea Godt), rabbit study are not entirely clear. However, we note that gain of Stat92E anti-Stat92E (1:1,000), mouse anti-Ptc (1:100, DSHB), rat anti-DE activity in CySCs in an otherwise wild-type background leads to cadherin (1:50, DSHB), mouse anti-bPS-integrin (1:20, DSHB), D D′ D′′ D′′′ expansion (not loss) of GSCs because JAK/STAT signaling in CySCs rabbit anti-cleaved caspase 3 (1:50, Cell Signaling). enables their extended niche function to support GSC self-renewal For 5-ethynyl-20-deoxyuridine (EdU, Invitrogen) labeling, (Leatherman & Dinardo, 2008, 2010). The latter niche role is specific samples were incubated for 30 min before fixation in Ringer’s to JAK/STAT signaling in CySCs and cannot be fulfilled by Hh signal- medium containing 10 lM EdU. Testes were fixed and processed ing, another CySC self-renewal pathway (Amoyel et al, 2013). More- normally for antibody labeling and then treated per manufacturer’s over, our clonal assays (as opposed to lineage mis-expression) are instructions. able to recapitulate the constant jostling for space at the niche that For statistical tests, we used the GraphPad Prism software. To normally occurs. Regardless, our findings establish that competition compare two samples, we used the Mann–Whitney U-test to deter- and self-renewal are two facets of the same homeostatic process mine significance; for multiple conditions, we used the Kruskal– (i.e., proliferation) and that colonizing stem cells have not acquired Wallis test and the Sidak’s multiple comparisons test for post hoc a new cellular property, but are simply better at self-renewing. analysis. Our study exemplifies how corrupting the naturally occurring The mathematical model is described in Supplementary Materials EF process of neutral competition endows a stem cell with greater and Methods. competitiveness, enabling it to gain dominance within a tissue. Such behavior may be relevant to the early steps of oncogenesis driven Supplementary information for this article is available online: by tumor-initiating cells, which have stem cell-like properties (Reya http://emboj.embopress.org et al, 2001), as in the case of carcinoma, glioma and leukemia caused by sustained Hh signaling (Clement et al, 2007; Zhao et al, Acknowledgements 2009; Youssef et al, 2012). The process described here of biasing We thank R. Lehmann, F. Schöck, L. Johnston, D. Kalderon, U. Tepass, J. neutral drift by stem cells harboring oncogenic mutations and the Treisman, S. Grewal, Y. Yamashita, T. Harris, B. Ohlstein, L. Buttitta, D. Godt, mechanism underlying it appear to be conserved (Vermeulen et al, Bloomington, and DSHB for antibodies and reagents. We thank members of 2013; Snippert et al, 2014). Taken together, these findings may the Bach laboratory for fruitful discussions. We are extremely grateful to explain observations such as field cancerization, in which a Esteban Mazzoni, Andrew Tomlinson, and Gary Struhl for their generosity in molecular lesion spreads through a tissue, causing multiple foci of the aftermath of Superstorm Sandy. BDS acknowledges the support of the the primary tumor (Vanharanta & Massague, 2012). Wellcome Trust (Grant Number 098357/Z/12/Z). Work in the Bach laboratory is

2310 The EMBO Journal Vol 33 | No 20 | 2014 ª 2014 The Authors ª 2014 The Authors The EMBO Journal Vol 33 | No 20 | 2014 2311 The EMBO Journal Gain of Hh or Yki cause niche competition Marc Amoyel et al Marc Amoyel et al Gain of Hh or Yki cause niche competition The EMBO Journal

supported by grants from the NIH (R01-GM085075-05 and R01-GM085075- Forbes AJ, Lin H, Ingham PW, Spradling AC (1996) Hedgehog is required for Moreno E, Basler K (2004) dMyc transforms cells into super-competitors. Cell Tanentzapf G, Devenport D, Godt D, Brown NH (2007) Integrin-dependent 05S1) and NYSTEM (C028132 and C024284). the proliferation and specification of ovarian somatic cells prior to egg 117: 117 – 129 anchoring of a stem-cell niche. Nat Cell Biol 9: 1413 – 1418 chamber formation in Drosophila. Development 122: 1125 – 1135 de Navascues J, Perdigoto CN, Bian Y, Schneider MH, Bardin AJ, Tapon N, Ito N, Dickson BJ, Treisman JE, Hariharan IK (2001) The Drosophila Author contributions Goberdhan DC, Paricio N, Goodman EC, Mlodzik M, Wilson C (1999) Martinez-Arias A, Simons BD (2012) Drosophila midgut homeostasis tuberous sclerosis complex gene homologs restrict cell growth and cell MA designed, carried out, and analyzed the experiments. EAB contributed to Drosophila tumor suppressor PTEN controls cell size and number by involves neutral competition between symmetrically dividing intestinal proliferation. Cell 105: 345 – 355 experimental design and analysis. BDS contributed to experimental design and antagonizing the Chico/PI3-kinase signaling pathway. Genes Dev 13: stem cells. EMBO J 31: 2473 – 2485 Thacker SA, Bonnette PC, Duronio RJ (2003) The contribution of provided the mathematical analysis. MA, BDS, and EAB wrote the manuscript. 3244 – 3258 Neufeld TP, de la Cruz AF, Johnston LA, Edgar BA (1998) Coordination of E2F-regulated transcription to Drosophila PCNA gene function. Curr Biol Hamaratoglu F, Willecke M, Kango-Singh M, Nolo R, Hyun E, Tao C, growth and cell division in the Drosophila wing. Cell 93: 1183 – 1193 13: 53 – 58 Conflict of interest Jafar-Nejad H, Halder G (2006) The tumour-suppressor genes NF2/Merlin Nystul T, Spradling A (2007) An epithelial niche in the Drosophila ovary Tian JP, Jin Z, Xie T (2012) Mathematical model for two germline stem cells The authors declare that they have no conflict of interest. and Expanded act through Hippo signalling to regulate cell proliferation undergoes long-range stem cell replacement. Cell Stem Cell 1: 277 – 285 competing for niche occupancy. Bull Math Biol 74: 1207 – 1225 and apoptosis. Nat Cell Biol 8: 27 – 36 Oh H, Irvine KD (2008) In vivo regulation of Yorkie phosphorylation and Tulina N, Matunis E (2001) Control of stem cell self-renewal in Drosophila Hardy RW, Tokuyasu KT, Lindsley DL, Garavito M (1979) The germinal localization. Development 135: 1081 – 1088 spermatogenesis by JAK-STAT signaling. Science 294: 2546 – 2549 References proliferation center in the testis of Drosophila melanogaster. J Ultrastruct Pan D (2010) The hippo signaling pathway in development and cancer. Dev Vanharanta S, Massague J (2012) Field cancerization: something new under Res 69: 180 – 190 Cell 19: 491 – 505 the sun. Cell 149: 1179 – 1181 Amoyel M, Bach EA (2014) Cell competition: how to eliminate your Inaba M, Yuan H, Yamashita YM (2011) String (Cdc25) regulates stem cell Potter CJ, Huang H, Xu T (2001) Drosophila Tsc1 functions with Tsc2 to Verdu J, Buratovich MA, Wilder EL, Birnbaum MJ (1999) Cell-autonomous neighbours. Development 141: 988 – 1000 maintenance, proliferation and aging in Drosophila testis. Development antagonize insulin signaling in regulating cell growth, cell proliferation, regulation of cell and organ growth in Drosophila by Akt/PKB. Nat Cell Biol Amoyel M, Sanny J, Burel M, Bach EA (2013) Hedgehog is required for CySC 138: 5079 – 5086 and organ size. Cell 105: 357 – 368 1: 500 – 506 self-renewal but does not contribute to the GSC niche in the Drosophila Ingham PW, Taylor AM, Nakano Y (1991) Role of the Drosophila patched gene Price MA, Kalderon D (1999) Proteolysis of cubitus interruptus in Drosophila Vermeulen L, Morrissey E, van der Heijden M, Nicholson AM, Sottoriva A, testis. Development 140: 56 – 65 in positional signalling. Nature 353: 184 – 187 requires phosphorylation by protein kinase A. Development 126: Buczacki S, Kemp R, Tavare S, Winton DJ (2013) Defining stem cell Chen C, Jack J, Garofalo RS (1996) The Drosophila insulin receptor is required Issigonis M, Tulina N, de Cuevas M, Brawley C, Sandler L, Matunis E (2009) 4331 – 4339 dynamics in models of intestinal tumor initiation. Science 342: 995 – 998 for normal growth. Endocrinology 137: 846 – 856 JAK-STAT signal inhibition regulates competition in the Drosophila testis Reya T, Morrison SJ, Clarke MF, Weissman IL (2001) Stem cells, cancer, and Voog J, D’Alterio C, Jones DL (2008) Multipotent somatic stem cells Chen Y, Struhl G (1996) Dual roles for patched in sequestering and stem cell niche. Science 326: 153 – 156 cancer stem cells. Nature 414: 105 – 111 contribute to the stem cell niche in the Drosophila testis. Nature 454: transducing Hedgehog. Cell 87: 553 – 563 Jin Z, Kirilly D, Weng C, Kawase E, Song X, Smith S, Schwartz J, Xie T (2008) Ritsma L, Ellenbroek SI, Zomer A, Snippert HJ, de Sauvage FJ, Simons BD, 1132 – 1136 Chen HW, Chen X, Oh SW, Marinissen MJ, Gutkind JS, Hou SX (2002) mom Differentiation-defective stem cells outcompete normal stem cells for Clevers H, van Rheenen J (2014) Intestinal crypt homeostasis Wallenfang MR, Nayak R, DiNardo S (2006) Dynamics of the male germline identifies a receptor for the Drosophila JAK/STAT signal transduction niche occupancy in the Drosophila ovary. Cell Stem Cell 2: 39 – 49 revealed at single-stem-cell level by in vivo live imaging. Nature 507: stem cell population during aging of Drosophila melanogaster. Aging Cell 5: pathway and encodes a protein distantly related to the mammalian Kawase E, Wong MD, Ding BC, Xie T (2004) Gbb/Bmp signaling is essential 362 – 365 297 – 304 cytokine receptor family. Genes Dev 16: 388 – 398 for maintaining germline stem cells and for repressing bam transcription Salzmann V, Inaba M, Cheng J, Yamashita YM (2013) Lineage tracing Wang L, Li Z, Cai Y (2008) The JAK/STAT pathway positively regulates DPP Cheng J, Tiyaboonchai A, Yamashita YM, Hunt AJ (2011) Asymmetric division in the Drosophila testis. Development 131: 1365 – 1375 quantification reveals symmetric stem cell division in Drosophila male signaling in the Drosophila germline stem cell niche. J Cell Biol 180: 721 – 728 of cyst stem cells in Drosophila testis is ensured by anaphase spindle Kiger AA, Jones DL, Schulz C, Rogers MB, Fuller MT (2001) Stem cell germline stem cells. Cell Mol Bioeng 6: 441 – 448 Xie T, Spradling AC (1998) Decapentaplegic is essential for the maintenance repositioning. Development 138: 831 – 837 self-renewal specified by JAK-STAT activation in response to a support cell Sarpal R, Pellikka M, Patel RR, Hui FY, Godt D, Tepass U (2012) Mutational and division of germline stem cells in the Drosophila ovary. Cell 94: 251 – 260 Clayton E, Doupe DP, Klein AM, Winton DJ, Simons BD, Jones PH (2007)A cue. Science 294: 2542 – 2545 analysis supports a core role for Drosophila alpha-catenin in adherens Xie T, Spradling AC (2000) A niche maintaining germ line stem cells in the single type of progenitor cell maintains normal epidermis. Nature 446: Klein AM, Nakagawa T, Ichikawa R, Yoshida S, Simons BD (2010) Mouse germ junction function. J Cell Sci 125: 233 – 245 Drosophila ovary. Science 290: 328 – 330 185 – 189 line stem cells undergo rapid and stochastic turnover. Cell Stem Cell 7: Sheng XR, Matunis E (2011) Live imaging of the Drosophila spermatogonial Yamashita YM, Jones DL, Fuller MT (2003) Orientation of asymmetric stem Clement V, Sanchez P, de Tribolet N, Radovanovic I, Ruiz i Altaba A (2007) 214 – 224 stem cell niche reveals novel mechanisms regulating germline stem cell cell division by the APC tumor suppressor and centrosome. Science 301: HEDGEHOG-GLI1 signaling regulates human glioma growth, cancer stem Leatherman JL, Dinardo S (2008) Zfh-1 controls somatic stem cell output. Development 138: 3367 – 3376 1547 – 1550 cell self-renewal, and tumorigenicity. Curr Biol 17: 165 – 172 self-renewal in the Drosophila testis and nonautonomously influences Shivdasani AA, Ingham PW (2003) Regulation of stem cell maintenance and Yamashita YM, Mahowald AP, Perlin JR, Fuller MT (2007) Asymmetric de la Cova C, Abril M, Bellosta P, Gallant P, Johnston LA (2004) Drosophila germline stem cell self-renewal. Cell Stem Cell 3: 44 – 54 transit amplifying cell proliferation by tgf-beta signaling in Drosophila inheritance of mother versus daughter centrosome in stem cell division. myc regulates organ size by inducing cell competition. Cell 117: 107 – 116 Leatherman JL, Dinardo S (2010) Germline self-renewal requires cyst stem spermatogenesis. Curr Biol 13: 2065 – 2072 Science 315: 518 – 521 de Cuevas M, Matunis EL (2011) The stem cell niche: lessons from the cells and stat regulates niche adhesion in Drosophila testes. Nat Cell Biol Simons BD, Clevers H (2011) Strategies for homeostatic stem cell self-renewal Youssef KK, Lapouge G, Bouvree K, Rorive S, Brohee S, Appelstein O, Larsimont Drosophila testis. Development 138: 2861 – 2869 12: 806 – 811 in adult tissues. Cell 145: 851 – 862 JC, Sukumaran V, Van de Sande B, Pucci D, Dekoninck S, Berthe JV, Aerts S, Dinardo S, Okegbe T, Wingert L, Freilich S, Terry N (2011) Lines and bowl Lee T, Luo L (1999) Mosaic analysis with a repressible cell marker for studies Singh SR, Zheng Z, Wang H, Oh SW, Chen X, Hou SX (2010) Competitiveness Salmon I, del Marmol V, Blanpain C (2012) Adult interfollicular affect the specification of cyst stem cells and niche cells in the Drosophila of gene function in neuronal morphogenesis. Neuron 22: 451 – 461 for the niche and mutual dependence of the germline and somatic stem tumour-initiating cells are reprogrammed into an embryonic hair follicle testis. Development 138: 1687 – 1696 Leevers SJ, Weinkove D, MacDougall LK, Hafen E, Waterfield MD (1996) The cells in the Drosophila testis are regulated by the JAK/STAT signaling. J Cell progenitor-like fate during basal cell carcinoma initiation. Nat Cell Biol 14: Dong J, Feldmann G, Huang J, Wu S, Zhang N, Comerford SA, Gayyed MF, Drosophila phosphoinositide 3-kinase Dp110 promotes cell growth. EMBO J Physiol 223: 500 – 510 1282 – 1294 Anders RA, Maitra A, Pan D (2007) Elucidation of a universal size-control 15: 6584 – 6594 Snippert HJ, van der Flier LG, Sato T, van Es JH, van den Born M, Yuan H, Yamashita YM (2010) Germline stem cells: stems of the next mechanism in Drosophila and mammals. Cell 130: 1120 – 1133 Lehner CF, O’Farrell PH (1990) Drosophila cdc2 homologs: a functional Kroon-Veenboer C, Barker N, Klein AM, van Rheenen J, Simons BD, generation. Curr Opin Cell Biol 22: 730 – 736 Doupe DP, Alcolea MP, Roshan A, Zhang G, Klein AM, Simons BD, Jones PH homolog is coexpressed with a cognate variant. EMBO J 9: 3573 – 3581 Clevers H (2010) Intestinal crypt homeostasis results from neutral Zhang Y, Kalderon D (2001) Hedgehog acts as a somatic stem cell factor in (2012) A single progenitor population switches behavior to maintain and Lopez-Garcia C, Klein AM, Simons BD, Winton DJ (2010) Intestinal stem cell competition between symmetrically dividing Lgr5 stem cells. Cell 143: the Drosophila ovary. Nature 410: 599 – 604 repair esophageal epithelium. Science 337: 1091 – 1093 replacement follows a pattern of neutral drift. Science 330: 822 – 825 134 – 144 Zhao C, Chen A, Jamieson CH, Fereshteh M, Abrahamsson A, Blum J, Kwon Duronio RJ, O’Farrell PH, Xie JE, Brook A, Dyson N (1995) The transcription Margolis J, Spradling A (1995) Identification and behavior of epithelial stem Snippert HJ, Schepers AG, van Es JH, Simons BD, Clevers H (2014) Biased HY, Kim J, Chute JP, Rizzieri D, Munchhof M, VanArsdale T, Beachy PA, factor E2F is required for S phase during Drosophila embryogenesis. Genes cells in the Drosophila ovary. Development 121: 3797 – 3807 competition between Lgr5 intestinal stem cells driven by oncogenic Reya T (2009) Hedgehog signalling is essential for maintenance of cancer Dev 9: 1445 – 1455 Michel M, Kupinski AP, Raabe I, Bokel C (2012) Hh signalling is essential for mutation induces clonal expansion. EMBO Rep 15: 62 – 69 stem cells in myeloid leukaemia. Nature 458: 776 – 779 Edgar BA, O’Farrell PH (1989) Genetic control of cell division patterns in the somatic stem cell maintenance in the Drosophila testis niche. Development Sun S, Zhao S, Wang Z (2008) Genes of Hippo signaling network act Drosophila embryo. Cell 57: 177 – 187 139: 2663 – 2669 unconventionally in the control of germline proliferation in Drosophila. License: This is an open access article under the Flaherty MS, Salis P, Evans CJ, Ekas LA, Marouf A, Zavadil J, Banerjee U, Bach Montagne J, Stewart MJ, Stocker H, Hafen E, Kozma SC, Thomas G (1999) Dev Dyn 237: 270 – 275 terms of the Creative Commons Attribution 4.0 EA (2010) Chinmo is a functional effector of the JAK/STAT pathway that Drosophila S6 kinase: a regulator of cell size. Science 285: 2126 – 2129 Tanentzapf G, Brown NH (2006) An interaction between integrin and the License, which permits use, distribution and reproduc- regulates eye development, tumor formation and stem cell self-renewal in Morata G, Ripoll P (1975) Minutes: mutants of drosophila autonomously talin FERM domain mediates integrin activation but not linkage to the tion in any medium, provided the original work is Drosophila. Dev Cell 18: 556 – 568 affecting cell division rate. Dev Biol 42: 211 – 221 cytoskeleton. Nat Cell Biol 8: 601 – 606 properly cited.

2312 The EMBO Journal Vol 33 | No 20 | 2014 ª 2014 The Authors ª 2014 The Authors The EMBO Journal Vol 33 | No 20 | 2014 2313 The EMBO Journal Gain of Hh or Yki cause niche competition Marc Amoyel et al Marc Amoyel et al Gain of Hh or Yki cause niche competition The EMBO Journal

2312 The EMBO Journal Vol 33 | No 20 | 2014 ª 2014 The Authors ª 2014 The Authors The EMBO Journal Vol 33 | No 20 | 2014 2313 Jung-Soon Mo et al Hippo pathway in stem cells and cancer EMBO reports Review

G-protein. For example, activation of G12/13 or Gq/11 stimulates Glossary YAP/TAZ, while Gs inhibits YAP/TAZ. Surprisingly, the LATS1/2 BMP Bone morphogenetic protein kinase may not be involved in YAP/TAZ regulation by mechanical CC Cholangiocarcinoma The Hippo signaling pathway in stem cell biology CSNK1 Casein kinase 1 cues [50], whereas the LATS1/2 kinase is involved in GPCR-medi- FGF Fibroblast growth factor cated hormonal cues to YAP/TAZ and MST1/2 is not required. The HCC Hepatocellular carcinoma precise mechanism by which the actin cytoskeleton relays upstream and cancer Hering canal cells Origin of liver stem/progenitor cells in adult cues to modulate LATS1/2 kinase activity is still not fully under- livers stood and remains a key question in the field. HIPK2 Homeodomain-interacting protein kinase2 Jung-Soon Mo, Hyun Woo Park & Kun-Liang Guan* Hippo Ste20-like kinase Hpo Beyond the main components of the Hippo pathway described Id Inhibitor of DNA binding proteins above, many other additional proteins have been reported to modu- LATS1/2 Large tumor suppressor kinase 1/2, Wts orthologs late the Hippo pathway, including TAOK1-3 (thousand and one LIF Leukemia inhibitory factor amino acid protein kinases) [57,58], MARK1-4 (MAP/microtubule Abstract bind to other DNA binding proteins including Mad, Homothorax MARK1-4 MAP/microtubule affinity-regulating kinases affinity-regulating kinases) [59,60], SIK1-3 (salt-inducible kinases) MASK Multiple ankyrin repeats single KH domain- (Hth), and teashirt to promote gene expression [18,19]. containing protein [61], RASSF (RAS association domain-containing family protein) The Hippo signaling pathway, consisting of a highly conserved The Hippo pathway is a tumor suppressor pathway because Mats Mob as tumor suppressor [61–63], MASK (multiple ankyrin repeats single KH domain-contain- kinase cascade (MST and Lats) and downstream transcription co- mutations in these regulatory pathway components result in an Mob1 MOB kinase activator 1A ing protein) [64,65], HIPK2 (homeodomain-interacting protein activators (YAP and TAZ), plays a key role in tissue homeostasis overgrowth phenotype. MST1/2 Mammalian Ste2-like kinases, Hpo orthologs kinase 2) [66,67], and CSNK1 (casein kinase 1) [23,68] (Fig 1). and organ size control by regulating tissue-specific stem cells. In mammals, the Hippo pathway consists of the serine/threonine PALS1 Membrane-associated palmitoylated protein 5 PATJ PALS-1-associated tight junction protein Moreover, this pathway plays a prominent role in tissue repair and kinases MST1/2 (mammalian Ste2-like kinases, Hpo orthologs) and PcG Polycomb group protein regeneration. Dysregulation of the Hippo pathway is associated LATS1/2 (large tumor suppressor kinase 1/2, Wts orthologs) [7,20– RASSF RAS association domain-containing family protein Hippo signaling in embryogenesis and embryonic with cancer development. Recent studies have revealed a complex 22]. Activation of the Hippo pathway results in the inactivation of Sav1 Salvador homolog 1 stem cells network of upstream inputs, including cell density, mechanical YAP (Yes-associated protein, Yki ortholog) by LATS1/2-mediated SIK1-3 Salt-inducible kinases sensation, and G-protein-coupled receptor (GPCR) signaling, that direct phosphorylation on YAP Ser127 (in humans). Phosphorylated TAOK1-3 Thousand and one amino acid protein kinases The first cell differentiation event in mammalian development occurs TAZ Transcriptional coactivator with PDZ binding modulate Hippo pathway activity. This review focuses on the role YAP is sequestered in the cytoplasm via binding to 14-3-3 and is motif during preimplantation, when the outer blastomeres of the embryo of the Hippo pathway in stem cell biology and its potential impli- degraded in a ubiquitin-proteasome-dependent manner, which TGF-b/activin Transforming growth factor B form an outer epithelial trophectoderm (TE) that envelops the remain- cations in tissue homeostasis and cancer. depends on phosphorylation of YAP Ser381 and Ser384 [23]. Warts NDR family kinase Wts ing blastomeres, the inner cell mass (ICM). The TE is necessary for Conversely, dephosphorylated YAP acts mainly through TEAD YAP Yes-associated protein, Yki ortholog implantation and later contributes to the placenta. Embryonic stem Keywords cancer; Hippo pathway; regeneration; stem cell; YAP family transcription factors to promote cell proliferation and organ cells (ESCs) are pluripotent cells, derived from the ICM of an early blas- DOI 10.15252/embr.201438638 | Received 17 February 2014 | Revised 16 April growth [24]. TAZ (transcriptional coactivator with PDZ binding tocyst, that have the potential to self-renew and differentiate into differ- 2014 | Accepted 16 April 2014 | Published online 13 May 2014 motif), a paralog of YAP in mammals, is regulated by the LATS1/2 ent cell types and tissues. This pluripotent capacity raises hope for EMBO Reports (2014) 15, 642–656 in a similar manner. YAP/TAZ are the major downstream mediators proteins, such as PATJ, PALS1, AMOT (angiomotin), ZO-1 (zona their potential application in regenerative medicine [69]. of the Hippo pathway. Besides the TEAD family transcription occludens protein 1) [38,39], E-cadherin [40], a/b-catenin [41,42], The association between Hippo signaling and stem cell-like prop- / See the Glossary for abbreviations used in this article. factors, YAP/TAZ also interacts with other transcription factors PTPN14 (protein tyrosine phosphatase non-receptor type 14) erties has been previously shown. For example, YapÀ À embryos including Smad, Runx1/2, p73, ErbB4, Pax3, and T-box transcrip- [43–45], and Ajuba/Zyxin protein [46,47], have also been identified arrest during development around E8.5 and display a yolk sac tion factor 5 (TBX5) to mediate the transcription of a diverse array as regulators or interacting partners of core Hippo pathway compo- vascular defect [70]. The YAP target transcription factors, TEADs, Introduction: The Hippo signaling pathway of genes, although the biological functions of these other transcrip- nents in mammals. Other newly characterized Hippo pathway regu- are the earliest genes expressed at high levels during embryo devel- tion factors in mediating Hippo signaling are less clear [25]. lator is the PCP (planar cell polarity) complex, composed of opment, and TEAD4 is required for specification of the TE lineage The Hippo pathway is evolutionally conserved and regulates diverse Although the core signaling cascade from Hpo (MST1/2) to Yki transmembrane cadherins Ft (Fat) and Ds (Dachsous). In Drosoph- during preimplantation of the mouse embryo [71–73]. At the blasto- cellular processes, including cell survival, proliferation, differentia- (YAP) is well understood, the upstream regulators of the Hippo ila, Ds binds to Ft, which in turn activates the Hippo pathway by cyst stage, TEAD4 promotes expression of multiple genes associated tion, and organ size. This pathway was initially characterized pathway are just beginning to be delineated. Interestingly, accumu- inhibiting the interaction between Zyxin and Wts, thus favoring its with trophoblast specification, including Cdx2 and Gata3, which are through clonal genetic screens identifying genes involved in tissue lating evidence from both Drosophila and mammals has shown that (Wts) degradation [47]. The vertebrate homolog of Ft is FAT4, but it selectively expressed only in blastomeres destined to become TE growth control in Drosophila melanogaster. In Drosophila, the core apical–basal polarity proteins may regulate the Hippo pathway by is still unclear whether FAT4 is involved in regulating the Hippo [71,74,75] (Fig 2A). Moreover, it has been shown that this activity components of the Hippo pathway include the kinase cascade of controlling YAP/TAZ localization. For instance, earlier studies in pathway in vertebrates [48,49]. In any case, further studies are of TEAD4 is dependent on YAP localization in the nucleus, which is Ste20-like kinase Hpo (Hippo) and NDR family kinase Wts (Warts) Drosophila implicated the apical membrane-associated FERM- needed to define the exact role of the PCP in regulating the Hippo modulated by cell–cell contact and LATS1/2-mediated phosphoryla- [1–7]. Hpo complexes with the scaffolding protein Sav (Salvador) to domain proteins Mer (Merlin) and Ex (Expanded), which are apical pathway. tion. This finding suggests that YAP localization is essential for phosphorylate and activate Wts, which then forms a complex with tumor suppressors, and the WW and C2 domain-containing protein Several studies have reported that YAP/TAZ is regulated by TEAD4 activity and cell fate specification [76]. Additionally, NF2 its regulatory protein Mats (Mob as tumor suppressor) [8–10]. Kibra (kidney and brain protein) as components upstream of Hpo. mechanical cues from neighboring cells and the extracellular matrix (Neurofibromin 2) and AMOT, two upstream components of the When in complex with Mats, Wts directly phosphorylates the tran- The Mer/Ex/Kibra complex recruits Hpo to the plasma membrane [50–52]. Moreover, the Hippo pathway is potently and acutely regu- Hippo pathway, facilitate YAP phosphorylation via LATS1/2 during scriptional coactivator Yki (Yorkie), sequestering it in the cytoplasm to enhance its kinase activity [26–30]. The apical transmembrane lated by a wide array of extracellular hormones, including lysophos- cell fate specification of mouse preimplantation development by promoting its interaction with 14-3-3 [11–15]. Conversely, when protein Crb (Crumbs) also interacts with Ex and modulates its (Ex) phatidic acid (LPA), sphingosine-1-phosphate (S1P), epinephrine, [77,78] (Fig 2B). Although TAZ is also highly enriched in the devel- the Hippo pathway is inactivated, unphosphorylated Yki translo- localization and stability [31–35]. Similar to Crb, the Scrib (Scribble) glucagon, and thrombin [53–55]. These mechanical and hormonal oping mouse embryo, inactivation of the gene encoding TAZ, cates into the nucleus where it associates with the TEAD/TEF family complex (Scrib/Dlg/Lgl) and Par3 polarity complex (Par3/Par6/ cues appear to be mediated through the actin cytoskeleton. Mechan- Wwtr1, results in only minor skeletal defects and the development transcription factor Sd (Scalloped) to initiate gene expression, aPKC) have been implicated in the regulation of the Hippo pathway ically, stabilization of F-actin results in YAP/TAZ activation, while of renal cysts, and these mice still grow to adulthood [79]. Alto- promoting cell survival and proliferation [16,17]. Yorkie can also activity [34,36,37]. In addition, a multitude of other cellular junction the disruption of F-actin leads to YAP/TAZ inactivation [56]. gether, these results demonstrate a critical role for YAP and TEADs Hormonally, G-protein-coupled receptors (GPCRs) transduce extra- in the process of cell fate determination in early mouse embryos cellular hormonal cues through RHO GTPases and the actin cyto- [67,73]. Department of Pharmacology and Moores Cancer Center, University of California, San Diego, La Jolla, CA, USA skeleton to modulate YAP/TAZ. GPCR signaling can either stimulate Recently, the Hippo pathway has also emerged as a crucial *Corresponding author. Tel: +1 858 822 7945; E-mail: [email protected] or inhibit YAP/TAZ activity in a manner dependent on the coupled regulator of pluripotency in vitro [80,81]. Initially, BMP and LIF

642 EMBO reports Vol 15 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO reports Vol 15 | No 6 | 2014 643 Jung-Soon Mo et al Hippo pathway in stem cells and cancer EMBO reports Review

642 EMBO reports Vol 15 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO reports Vol 15 | No 6 | 2014 643 EMBO reports Hippo pathway in stem cells and cancer Jung-Soon Mo et al Jung-Soon Mo et al Hippo pathway in stem cells and cancer EMBO reports

DROSOPHILA MAMMALS Apical domain Apical Extra- AMOT actin cellular GPCR GPCR network signal Crumbs NF2 F-actin Dachsous Apical Lats1/2 DPP actin LIN7C α-catenin Crumbs αq/11 α12/13 αs network TKV TE lineage βγ βγ βγ FRMD6 PALS AA β-catenin PATJ Athrophin FAT NF2 Kibra AMOT Rho P TJ Expanded YAP / TAZ Nanog AJ Merlin YAP Oct4 P TAZ YAP Psx1 Mad Ajub YAP YAP TAZ Expanded F-actin TAZ TAZ Hand1 ZYX PAR3 Cdx2 TEAD Eomes Kibra Merlin NF2 PAR6 TJ Gata3 PAR3 MST1/2 αPKC Merlin TJ P PAR6 FAT1– 4 Salvador Basolateral domain F-actin P αPKC Lats1/2 MOB1A/2B P SIK1–3 P P Hippo P PP2A MARK1–4 YAP Salvador Echinoid NF2 P TAZ YAP F-actin RASSF6 TAOK1–3 TAZ AMOT Lats1/2 RASSF YAP YAP TAZ -catenin YAP RASSF1A TAZ α TAZ PP2A YAP P TAZ YAP TAZ β-catenin YAP P AMOT Dlg TAZ TAZ Igl P/ YAP/TAZ Warts YAP

ZO-1 YA Scrib TAZ SJ Dlg Mats P P Dachs PTPN14 Scrib Nanog ZYX E-cadherin Oct4 α-catenin BA Inner Cell Mass lineage P 14-3-3 β-catenin AJ TEAD Cdx2 Yorkie P 14-3-3 YAP / TAZ

Figure 2. A model of TE and ICM specification regulated by Hippo-YAP pathway in preimplantation embryo. During preimplantation, the outer blastomeres of the embryo form an outer epithelial trophectoderm (TE) that envelopes the remaining blastomeres, the inner cell mass Yorkie YAP YAP YAP YAP YAP YAP Tgi P VGL4 TAZ TAZ TAZ TAZ TAZ (ICM). The Hippo pathway plays important roles in this cell fate specification. The outer cells have an outside exposed surface and are composed of plasma membranes with Scalloped Mad Hth Scalloped Tsh TEAD TEAD p73 RUNX TBX5 SMAD PAX3 apical and basolateral domains, whereas inner cells are completely surrounded by outer cells. (A) In the outer cells, the nuclear localization of YAP and TEAD4 regulates specification of the TE lineage through activation of the TE-specific genes such as Gata3, Cdx2, and Eomes. (B) In the inner cells, cell–cell adhesions influence Hippo signaling. Activated Hippo pathway impairs YAP nuclear localization in the ICM lineage, thereby limiting TEAD4 transcription and abrogating expression of other TE-specific genes. Activation of Oct4 and Nanog maintains pluripotency and generates the ICM in mouse embryos. The yellow spheres indicate phosphorylation of target proteins by kinase.

transcription factors: Sox, Oct3/4, c-Myc, and KLF4 [88]. YAP is Thus, in ESCs, YAP/TAZ promotes stemness directly, as well as Figure 1. Schematic models of the Hippo pathway in Drosophila and mammals. activated during the reprogramming of human embryonic fibro- indirectly by mediating TGF-b/BMP or LIF signaling, through regu- Cells are shown with respective cellular junctions; adherens junction (AJ), tight junction (TJ), septate junction (SJ). Hippo pathway components in Drosophila and mammals are lating the expression of genes responsible for maintaining pluripo- shown in various colors, with arrows and blunt lines indicating activation and inhibition, respectively. The yellow spheres indicate phosphorylation of target proteins by blasts into iPSCs, and the addition of YAP to Sox2, Oct4, and kinase. Continuous lines indicate known interactions, whereas dashed lines indicate unknown mechanisms. See introduction for further details. KLF4 increases iPSC’s reprogramming efficiency in mouse embry- tency both in vivo and in vitro [86,87]. These studies implicate the onic fibroblasts, further confirming a positive role of YAP in Hippo pathway with those involved in maintaining ESC pluripoten- stemness [87]. cy and controlling cell fate specification in development. In conclu- signals were shown to maintain mouse ESCs in an undifferentiated, inhibiting TEAD function resulted in differentiation toward the Moreover, it has been reported that the Hippo pathway can sion, YAP, TAZ, and TEAD proteins seem to be key regulators for pluripotent state, whereas human ESCs require FGF, BMP, and endoderm lineage [86]. Conversely, YAP protein and mRNA interact with other pathways to promote and maintain pluripo- maintaining the pluripotent properties of both ESCs and iPSCs in TGF-b/activin [82–84]. Fine-tuning these multiple signaling path- levels are significantly decreased with the loss of pluripotent tency. For example, TAZ associates with Smad2/3 to maintain mammals. The transcription coactivator activity of YAP/TAZ is ways is crucial in maintaining the balance between differentiation markers during ESC differentiation [87]. In addition, YAP is the nuclear accumulation of Smad complexes, thereby promoting similarly required for promoting stem cells as well as normal cell and self-renewal in ESCs. Supporting the role for YAP and TEADs sequestered and thereby inactivated in the cytoplasm, and conse- expression of pluripotency markers (Oct4, Nanog) in response to proliferation. In addition, TEAD is likely to be involved in both the in maintaining pluripotency, the high expression of YAP and quently, a large number of genes important for stem cell mainte- TGF-b stimulation [80]. Another piece of evidence linking the stemness and proliferation function of YAP/TAZ. However, depend- TEAD2 in ESCs, neural stem cells, and hematopoietic stem cells nance and function, including PcG, Nanog, Oct3/4, and Sox2, are Hippo and TGF-b/BMP pathways is the finding that YAP binds ing on cell context, YAP/TAZ must induce expression of different initially placed these genes into a general ‘stemness’ transcrip- repressed. Smad1 to regulate the induction of Id family members for mESC genes between stem cells and differentiated cells. It is also possible tional signature based on transcriptional profiling [85]. Tamm Additional evidence for the role of YAP in pluripotency is seen maintenance upon stimulation with BMP [81]. Finally, TAZ has that besides TEAD, different transcription factors may be used by et al found that YAP and TEAD2 could activate the expression of in induced pluripotent stem cells (iPSCs). The seminal findings been identified as a coactivator of Pax3-dependent transcription, YAP/TAZ in stem cells compared to differentiated cells to induce ESC master transcriptional regulators Oct4 and Nanog in mamma- by Yamanaka’s group demonstrated that mouse somatic cells can which influences the expression of various genes during embryo- downstream target gene expression, thereby promoting and main- lian ESCs. Furthermore, restricting YAP and TEAD2 expression or be reprogrammed into iPSCs by inducing the activity of four genesis [89]. taining stem cells.

644 EMBO reports Vol 15 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO reports Vol 15 | No 6 | 2014 645 EMBO reports Hippo pathway in stem cells and cancer Jung-Soon Mo et al Jung-Soon Mo et al Hippo pathway in stem cells and cancer EMBO reports

ZO-1 YA Scrib TAZ SJ Dlg Mats P P Dachs PTPN14 Scrib Nanog ZYX E-cadherin Oct4 α-catenin BA Inner Cell Mass lineage P 14-3-3 β-catenin AJ TEAD Cdx2 Yorkie P 14-3-3 YAP / TAZ

Liver: Liver progenitor cells and tumorigenesis to play a major role in overproliferation and tumorigenesis in this model [95]. The liver is the most important metabolic organ and has a high Intriguingly, liver-specific deletion of Sav1 enhanced proliferation Endodermal progenitors regenerative capacity and is able to regenerate after more than 70% and expansion of hepatic progenitor cells (OCs) and these mice hepatectomy. Hepatocytes are the predominant cell type in the adult eventually developed liver tumors with a mixed HCC and CC pheno- liver and are mitotically quiescent. The regenerative capacity of the type, distinct from HCC which originates from the aberrant prolifer- liver depends on hepatocyte proliferation, although the liver also ation of hepatocytes only. However, the levels of phosphorylated contains oval cells (OCs) which are capable of generating a transit YAP and phosphorylated LATS1/2 were not affected in the Sav1 KO Cholangiocytes Hering canal cells precursor compartment. Liver regeneration has been known for livers, suggesting that Sav is likely to play an essential role in OC many years, although the underlying mechanisms and how the liver expansion and tumorigenesis in this model but, surprisingly, acts Hepatocytes Oval cells senses when it has reached its original size are still poorly under- independently of LATS1/2 and YAP [96,97]. stood [90]. Studies of NF2 conditional knockout mice also support a role for Previous studies in Drosophila have implicated the Hippo path- YAP in liver tumorigenesis [93,98,99]. Inactivation of NF2 results in hepatocyte and BEC proliferation, widespread hepatocellular way as a central mechanism that restricts tissue overgrowth during HEPATOTOXINS development and it derailed under pathological conditions contrib- carcinoma, and bile duct hamartomas comprising cytokeratin- LIVER INJURY utes to tumorigenesis [91]. The Hippo pathway impinges on the positive biliary epithelial cells. Zhang et al [98] reported that NF2 transcriptional coactivator Yki to regulate the transcription of target and YAP act antagonistically to each other in the Hippo pathway to genes involved in cell growth, proliferation, and survival. Conserva- regulate liver development and physiology. Deletion of only one tion of mammalian homologs for all the known components of the copy of YAP was sufficient to reverse the expansion of liver AA BA Drosophila Hippo pathway has facilitated investigation of the physi- progenitor cells and tumorigenesis driven by the loss of NF2. MST1/2 RASSF MST1/2 ological roles of Hippo signaling in mammals. While it was already Consistent with this finding, the NF2-deficient liver showed reduced NF2 suggested based on the Drosophila data that the Hippo pathway is phosphorylation of YAP and LATS1/2 and increased YAP nuclear MST1/2 involved in mammalian tumorigenesis, Dong et al [12] provided localization, providing functional evidence that the main tumor P Salvador functional evidence that the mammalian Hippo pathway is a potent suppressive mechanism of NF2 is mediated through inactivating NF2 Lats1/2 Lats1/2 regulator of organ size and that its dysregulation leads to tumorigen- YAP. On the other hand, EGFR signaling has also been implicated MOB MOB esis in the liver. Induction of YAP overexpression using a condi- in NF2 deletion-induced tumorigenesis. Pharmacologic inhibitors of ? ? tional YAP transgenic mouse resulted in massive hepatomegaly via EGFR blocked OC expansion and tumorigenesis triggered by NF2 an increase in the number, but not the size, of the liver cells. Inter- deletion [99]. Benhamouche et al also showed that liver-specific P P estingly, the YAP-induced enlarged livers reverted back to their deletion of NF2 leads to an early and dramatic expansion of YAP YAP CYTOPLASMIC YAP YAP original size without any gross abnormalities when the expression progenitor cells without any detectable alteration in YAP localiza- TAZ Inhibition TAZ RETENTION TAZ Inhibition TAZ of transgenic YAP was repressed. These data clearly establish a tion and phosphorylation, arguing against a role for YAP in NF2 DEGRADATION predominant role of YAP in organ size control in mice [92,93]. KO-induced tumorigenesis. Future studies are necessary to clarify However, when YAP overexpression was maintained for an the discrepancy of these two reports regarding NF2 deletion- YAP YAP extended period of time, the transgenic mice develop liver tumors induced YAP activation [98,99]. However, the general consensus is TAZ PROLIFERATION PROLIFERATION TAZ similar to hepatocellular carcinoma, suggesting a role of hyper-YAP that NF2 acts upstream of YAP and that other downstream effectors TEAD EXPANSION EXPANSION TEAD activation in cancer development. of NF2 may also contribute to tumorigenesis. More recently, other components of the Hippo pathway have Collectively, these data suggest that Hippo pathway components been shown to repress proliferation and restrict liver growth. Dele- may play an important role in maintaining hepatocyte quiescence tion of both MST1 and MST2 results in embryonic lethality [94– and regulating organ size in mammals, yet their dysregulation can 96]. However, a single copy of MST1 or MST2 (mice with geno- lead to stem cell expansion, overgrowth, and tumorigenesis through Figure 3. Schematic illustration of Hippo pathway in liver. type of either MST1 / , MST2+/ or MST1+/ , MST2 / ) is multiple mechanisms. There are differences in the phenotypes Endodermal progenitors generate hepatocytes and cholangiocytes that surround the bile duct system in adult liver. The Hering canal cells give rise to bipotential oval cells, À À À À À À sufficient to support normal embryonic development. During later observed in the conditional knockout mouse models of various which are capable of generating both hepatocytes and cholangiocytes. Hepatocyte regeneration is responsible for liver growth after partial hepatectomy. The exposure of the adult liver to hepatotoxins induces the proliferation of oval cell, but hepatocytes are slow to respond or do not respond at all to toxic injury. (A) In the hepatocytes, MST1/2 stages of development, loss of MST1 or MST2 promotes prolifera- Hippo pathway components (Fig 3). Thus, further studies are is activated by proteolytic cleavage that resulted in the loss of the Sav1 interacting SARAH domain. Cleaved MST1/2 is required to phosphorylate Mob1, but SAV1 is not tion of liver stem cells/progenitor cells such as oval cells. Prolifer- needed to fully elucidate the roles of these Hippo pathway compo- required for Hippo pathway activity. This facilitates YAP phosphorylation, resulting in cytoplasmic retention by 14-3-3 binding and degradation by ubiquitin-proteasome- ation of liver progenitor cells gives rise to both hepatocytes and nents and their mechanisms of action in regulating of liver progeni- dependent manner. Lats1/2 activity is unaffected by MST1/2 inactivation in hepatocytes. But loss of NF2 decreases Lats1/2 and YAP phosphorylation, suggesting that cholangiocytes (biliary epithelial cells, BECs), which are the prom- tor cells. the existence of unknown kinase other than MST1/2. Additionally, RASSF family proteins seem to have a role in MST1/2 regulation. Lats1/2 might indirectly inhibit YAP by inent epithelial cells of the bile duct. This eventually leads to the activating unidentified kinase distinct from MST1/2. (B) In the oval cells, MST1/2-regulated phosphorylation on YAP Ser127 is unaffected by Sav1 inactivation. However, Sav1 regulates YAP protein level and localization via as yet defined mechanisms. There are no clear links between MST1/2 and Lats1/2 activation in oval cells. The mechanism development of liver tumors due to the loss of heterozygosity of underlying the inactivation of YAP inhibits oval cell proliferation. However, the role of oval cells in liver regeneration remains controversial. the remaining MST1 or MST2. These mice display characteristics Skin: Epidermal progenitor cells of hepatocellular carcinoma (HCC) and cholangiocarcinoma (CC) with expansion and transformation of a mixed population of The skin (epidermal tissue) in the human body undergoes constant tumor-associated liver progenitors [94]. Interestingly, YAP protein replenishing, completely replacing itself every 2 weeks throughout toward the skin’s surface as they terminally differentiate, eventually Recent findings have implicated the importance of the Hippo levels were increased, while YAP phosphorylation and LATS1/2 an individual’s life [100]. The epidermal stem cells are located leading to constant skin remodeling. When the skin is injured, pathway in epidermal development and homeostasis. It has been phosphorylation were significantly reduced, relative to wild-type, within the basal layer and have a high proliferative capacity to wound healing greatly accelerates this regenerating process by shown that inactivation of Sav1 (WW45) alleles leads to early in the absence of MST1/2, indicating that YAP is a downstream continuously produce new epidermis while still maintaining struc- which these inner progenitor cells migrate outwards. Epidermal embryonic lethality, and histological examination displayed a thick- effector of MST1/2 in the liver. In contrast, TAZ protein levels and tural integrity. During development, the basal epidermal cells gener- growth must be carefully balanced, because inadequate proliferation ening of the epidermal skin layer in the embryos [101]. WW45-null phosphorylation status are decreased in the MST1/2-knockout liver ate proliferative progenitor cells, which can only divide for a limited results in the thinning of skin and loss of protection, whereas exces- primary keratinocytes show hyperproliferation, progenitor expan- and tumors, suggesting that TAZ, a potential oncogene, is unlikely number of cycles; these cells then leave the basal layer, migrate sive growth leads to hyperproliferative disorders. sion, decreased apoptosis, and inhibition of terminal differentiation.

646 EMBO reports Vol 15 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO reports Vol 15 | No 6 | 2014 647 EMBO reports Hippo pathway in stem cells and cancer Jung-Soon Mo et al Jung-Soon Mo et al Hippo pathway in stem cells and cancer EMBO reports

These observations suggest that the Hippo pathway restricts the tube [106]. Both YAP gain-of-function and loss-of-function studies Cardiac progenitor cells and muscle progenitor cells Intestine: Intestinal stem cells pool of these progenitor cells. in Xenopus demonstrate that YAP is required for expansion of Through molecular and genetic studies, two groups have inde- Sox2+ neural plate progenitors and Pax3+ neural crest progenitors The fetal heart grows through the proliferation of cardiomyocytes, Intestinal stem cells (ISCs) are responsible for the constant pendently shown that YAP overexpression results in expansion of at the neural plate border and for maintaining these progenitor and following birth, postnatal cardiomyocytes undergo hypertrophy renewal and repair of the intestinal epithelia to maintain tissue the epidermal stem cells and progenitor cells in the epidermis cells in an undifferentiated state. The effects of YAP on Pax3+ to reach an optimal size. Although it was traditionally believed that homeostasis [125,126]. Recent studies have highlighted the role of [42,102]. Mice carrying the YAP transgene reveal epidermal thick- neural crest progenitors are through the direct regulation of Pax3 the adult human heart lacks adequate myocardium regenerative the Hippo pathway and its effectors YAP and Yki in intestinal ening, hyperkeratosis, and squamous cell-like carcinoma in skin transcription. YAP acts through TEAD to stimulate Pax3 expres- potential for repair, recent studies have identified endogenous stem regeneration following tissue injury in both mice and Drosophila, grafts. Conversely, deletion of YAP in the epidermis or disruption of sion. Previous studies have also suggested that mouse TEAD is cells with the regenerative capacity to repair lost or damaged heart respectively. In general, the loss of Hippo signaling and/or the the YAP–TEAD interaction during epidermal development resulted responsible for activating the Pax3 promoter and neural crest tissue during the late cardiac development of the adult heart [115]. elevated YAP activity is associated with stem cell expansion in in epidermal hypoplasia and loss of keratinocyte proliferation. This expression in the mouse as well [108]. It is well documented that Unlike other tissues such as the liver, the role of Hippo signaling various organs [125,127]. However, in the intestine, there are phenotype was attributed to the gradual loss of the epidermal stem/ the expansion of mouse neural progenitors is mediated by the acti- in the heart is less well understood. It has recently been shown that contradictory reports regarding the role of YAP in ISC expansion progenitor cells and the progenitor cells’ limited capacity for self- vation of the Notch pathway; however, in the frog embryo, YAP’s a cardiac-specific deletion of Sav1 or overexpression of a constitu- and intestinal regeneration across different species and experimen- renewal. ability to repress neural differentiation is likely independent of tively active YAP mutant in embryos results in embryos within tal settings. Surprisingly, deletion of MST1/2 did not lead to epidermal hyper- Notch signaling [107]. It has been shown that YAP is amplified or cardiomegaly due to increased cardiomyocyte proliferation. Ablation The function of the Hippo pathway and YAP in ISCs has mostly plasia, indicating that YAP is regulated through an alternative mech- up-regulated in human Shh-dependent medulloblastoma, a brain of either the MST1/2 or LATS1/2 kinases, the upstream inhibitory been studied in the context of intestinal regeneration following anism that is not dependent on canonical Hippo pathway tumor in children. Similarly, it was observed that YAP and its kinases of YAP, causes perinatal lethality resulting from an over- tissue injury in transgenic animal models (Fig 4). In the DSS- components MST1/2 in the skin [41]. Consistent with an MST1/2- target transcription factor TEAD1 are highly expressed in mouse grown heart due to elevated cardiomyocyte proliferation, similar to induced colonic regeneration model by Cai et al, YAP protein levels independent regulation of YAP, recent studies have shown that Shh-dependent medulloblastomas [109]. In addition, YAP is a the Sav cKO heart [116,117]. Genetic interaction studies have shown are elevated following tissue injury. In addition, the specific deletion MST1/2 is not required for YAP activation by G-protein-coupled target of Shh signaling in the developing cerebellum. YAP expres- that nuclear YAP interacts with b-catenin in cardiomyocytes, of YAP in the intestinal epithelium prevented DSS-induced intestinal receptor (GPCR) signaling. Cell adhesion and a-catenin have also sion and nuclear localization are induced in proliferating cerebellar directly activating b-catenin target genes to promote Wnt signaling, regeneration, suggesting that YAP is required for these processes been implicated in YAP regulation. Interestingly, skin-specific dele- granule neural precursors, which are thought to be the cells of which has already been implicated in cardiac repair and cell repro- [128]. Correlating with the function of the Hippo pathway to tion of a-catenin, a component of adherens junctions and an impor- origin for certain medulloblastomas. Additionally, it has been gramming. Loss of b-catenin in the Sav cKO hearts suppressed the suppress YAP activity, loss of Hippo signaling in Sav1-deficient tant tumor suppressor in epithelia, resulted in keratinocyte suggested that mutation of Patched1 (PTCH1), which encodes an overgrowth phenotype caused by Hippo pathway inactivation, crypts displayed accelerated regeneration upon DSS-induced injury hyperproliferation and squamous cell carcinoma that resemble the inhibitor of hedgehog pathway, leads to the activation of YAP in a suggesting that the Hippo pathway restrains cardiomyocyte prolifer- in a YAP-dependent manner [128]. Similarly, Zhou et al [129] phenotypes observed in YAP transgenic mice [41]. a-Catenin is non-cell-autonomous manner and alters hedgehog pathway in ation and heart size by inhibiting Wnt signaling [118]. Another showed that deletion of the core Hippo kinase MST1/2 in the intesti- considered a critical sensor for cell density and provides the cell medulloblastoma cells and tissue samples [110]. These studies recent study showed that YAP activates the IGF pathway during nal epithelium resulted in a marked expansion of the ISC compart- with neighborhood information through the formation of density- show a critical role for YAP and TEAD in neuronal progenitor cells heart development, resulting in the inactivation of GSK3b, which in ments due to YAP hyperactivation. Ubiquitous overexpression of dependent cell–cell junctions (adherens junctions). Similar to a-cate- and medulloblastoma development. turn inhibits b-catenin degradation [119]. More recently, Xin et al YAP-S127A, which lacks the phosphorylation site required for inac- nin, the Hippo signaling pathway has been implicated in cell contact Large-scale RNAi screens reveal that FatJ cadherin, the closest have reported that expression of constitutively active YAP promotes tivation by the Hippo pathway, also resulted in the loss of differenti- inhibition of proliferation as well as tissue growth control [103]. homolog of the Drosophila dFat, is spatially restricted to the inter- proliferation of adult cardiomyocytes and enhances adult heart ation markers and expansion of an undifferentiated cell population Notably, a-catenin can directly interact with YAP and suppress YAP mediate regions of the neural tube and acts though YAP to regu- regeneration in response to injury. YAP-expressing cardiomyocytes in the mouse intestine [92]. function, possibly by sequestering YAP at the plasma membrane late the number of neural progenitor cell pools within the dp4-vp1 behave similar to embryonic cells with regard to their regenerative On the other hand, Barry et al [130] reported that specific expres- and preventing it from entering the nucleus [41]. These findings domain [111]. Loss of NF2 also caused an overexpansion of the potential [120]. sion of YAP in the intestinal epithelium suppresses intestinal provide a mechanistic explanation for how a-catenin modulates neocortical progenitor pool by increasing YAP/TAZ protein levels, Conversely, loss of YAP leads to embryonic lethality through renewal and reduces the ISC population by restricting Wnt/b- YAP activity by translating context-dependent information to regu- enhancing nuclear localization of both these proteins, and up- myocardial hypoplasia, due to reduced cardiomyocyte proliferation catenin signaling. Intestinal regeneration after irradiation is charac- late stem cell proliferation and tissue expansion. It should be noted regulating their target genes in the mammalian dorsal telence- in the embryonic heart [119,121]. Thus, YAP connects Hippo signal- terized by hyperactivation of Wnt/b-catenin signaling. Consistently, that there is strong evidence supporting that angiomotin mediates phalon [112]. In addition, Hippo signaling had previously been ing and other growth-promoting pathways, such as IGF and Wnt deletion of YAP resulted in Wnt hypersensitivity and led to ISC cell–cell contact and tight junction signals to inhibit YAP function implicated in Ft/Ds signaling through its regulation of cell prolifer- signaling, to regulate embryonic and neonatal cardiomyocyte prolif- expansion and crypt hyperplasia after injury by irradiation. These by both increasing YAP phosphorylation and physical binding ation and differentiation in Drosophila, although there was no eration. This is mediated at least in part by its interaction with results are at odds with the role of YAP in the DSS-induced colonic [78,104]. direct evidence to implicate Ft/Ds signaling in regulating the b-catenin, directly promoting a stemness gene expression program regeneration model. vertebrate Hippo pathway [113]. Cappello et al recently suggested [117–119,121]. Another inconsistency is in the crosstalk between YAP and Wnt/ a connection of FAT4/DCHS1 and YAP in mammals. They A role for the Hippo pathway in skeletal muscle is beginning to b-catenin signaling and their role in intestinal regeneration. The Nervous system: neural progenitor cells reported that knockdown of FAT4 or DCHS1 promotes neural be delineated. YAP overexpression in C2C12 myoblasts and primary Sav1-deficient mouse colons developed polyps after DSS-induced progenitor cell proliferation and malpositioning of cells in the mouse muscle stem cells blocks the progression of myoblasts regeneration, which showed nuclear accumulation of YAP, but not YAP and TEAD2 are highly expressed in neural stem cells (NSCs), developing cerebral cortex [114]. These mouse data demonstrate through the myogenic program and preserves the progenitor-like b-catenin [128]. This is consistent with the observation by Barry which are multipotent progenitors present in the nervous system. that reduced levels of FAT4 and DCHS1 increase the activity of and proliferative properties [122,123]. High YAP expression and et al [130] that YAP-S127A expression restricts Wnt/b-catenin NSCs are capable of self-renewing and produce multiple neural unphosphorylated YAP and a YAP-responsive transcriptional activity expands the pool of activated satellite cells, the resident signaling during intestinal regeneration. In contrast, Zhou et al lineages which ultimately compose the central nervous system reporter. Together, these findings reveal a novel function of stem cells in skeletal muscle, and prevents the differentiation of this [129] reported that in the MST1/2-deficient intestinal epithelium, (CNS) [85,105]. In the vertebrate’s developing neural tube, YAP is NF2 and FAT4 signaling in inhibiting neural progenitor expansion cell population. Interestingly, overexpression of TAZ increases nuclear accumulation of YAP correlates with b-catenin activation. expressed by ventricular zone progenitor cells and co-localizes during brain development and establish YAP/TAZ as key myogenic gene expression in a MyoD-dependent manner, thereby Uncontrolled tissue regeneration after injury can become oncogenic, with Sox2, a neural progenitor marker [106,107]. Overexpression effectors. promoting myogenic differentiation [124]. Despite the high level of like in colon cancer. In this context, Barry et al underscored that of either YAP or a transcriptionally active form of TEAD in the To date, the proposed model is that YAP promotes NSC prolif- sequence identity between YAP and TAZ, their opposite effects on YAP is silenced in a subset of highly aggressive human colorectal neural tube leads to reduced neural differentiation and a marked eration by serving as an effector of the Shh pathway in the brain. muscle progenitor fate is a nice illustration of the complexity and carcinomas, whereas Zhou and co-workers showed a striking preva- increase in neural progenitor cell numbers due to accelerated cell A full understanding of the role of the Hippo pathway in NSC context specificity associated with Hippo pathway activation or inhi- lence of YAP overexpression in 95% of colonic cancer specimens cycle progression and recurring cell cycle exit. These effects are requires future studies to examine crosstalk between Hippo bition and the resulting transcriptional response. Obviously, further [129,130]. The complex nature of YAP in the context of ISC expan- associated with the induction of cyclin D1 and the down-regula- and other signaling pathways such as the MAPK, Ephrin, studies need to be carried out in vivo to conclusively determine the sion, intestinal regeneration, and its relation to Wnt/b-catenin tion of NeuroM. Conversely, loss of YAP triggers cell death and Wnt, and Notch pathways that are also thought to control brain role of Hippo signaling, particularly the opposing functions of YAP signaling certainly requires further investigation. Nevertheless, promotes premature neuronal differentiation in the chick neural development. and TAZ, in cardiac and skeletal muscle biology. these studies point to a role of YAP in ISC, either positively by

648 EMBO reports Vol 15 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO reports Vol 15 | No 6 | 2014 649 EMBO reports Hippo pathway in stem cells and cancer Jung-Soon Mo et al Jung-Soon Mo et al Hippo pathway in stem cells and cancer EMBO reports

–/– NORMAL MST1/2 Moreover, TAZ has been shown to be a key regulator of cancer NF2 is a potent upstream regulator of the Hippo pathway, and an stem cells (CSCs) in breast cancer [139]. In addition, YAP and TEAD inactivating mutation in NF2 is associated with several human are highly expressed in CSCs of medulloblastomas [109]. Increasing cancers including acoustic neuromas, meningiomas of the brain, and MST1/2 YAP/β-catenin evidence has suggested that tumor growth is dependent on CSCs, schwannomas of the dorsal roots of the spinal cord [146,147]. A high Sav activation which represent a small subset of cells within a tumor but have the frequency of NF2 mutations has also been reported in mesothelioma ability to self-renew, differentiate into other tumor cell types, and [148,149]. Recently, a TAZ and calmodulin-binding transcription P YAP STEM/PROGENITOR initiate tumor formation. CSCs are also thought to be resistant to activator 1 (CAMTA1) fusion gene has been reported in epithelioid TAZ EXPANSION β-catenin chemotherapeutic agents and are responsible for cancer recurrence hemangioendothelioma, a rare form of sarcoma [150,151]. The role and metastasis. High-grade tumors are characterized by a higher and mechanism of this fusion protein in cancer progression is still population of CSCs within the tumor. Microarray analysis of 993 unclear, but may relate to the transcriptional regulatory functions primary human breast tumors has identified a list of genes highly ascribed to both TAZ and CAMTA1. Intestinal crypt expressed in G3 (tumors that poorly differentiated tumors) Many studies have reported a high frequency of mutations in various YAP TAZ β-catenin compared to G1 (benign tumors) [139]. Interestingly, elevated YAP/ GPCRs (GPR98, GRM3, AGTRL1, LPHN3, and BAI3) and G-proteins TAZ activity is observed in G3 tumors, which are also characterized (GNAS, GNAQ, and GNAO1) across a wide range of cancers, particu- TEAD TCF TEAD TCF by the expression of embryonic and normal mammary stem cell larly in melanoma [152–154]. Notably, activating mutations in genes. Using a model for tumor progression, Cordenonsi et al GNAQ and GNA11 have been observed in approximately 50% of demonstrated a role for TAZ in breast cancer cells [139]. Upon injec- uveal melanomas. And in these uveal melanomas with activating tion in mice, MII cells, which are Ras-transformed MCF10A-T1k, mutations in GNAq or GNA11, we found that YAP is constitutively YAP –/– YAP Tg generate low-grade tumors. On the other hand, MIV cells, which are activated and its activation is pathologically critically important. malignant MCF10A-CA1a cells derived from the in vivo spontaneous Collectively, extensive studies have established a critical role for RSPO evolution of MII cells, readily formed tumors resembling G3 tumors. the Hippo pathway in human tumorigenesis. Inhibiting YAP/TAZ Wnt TAZ was highly expressed in the MIV cells, but not the MII cells, may be a new therapeutic area for treating cancers with a dysregu- Wnt hyperactivation whereas YAP levels were comparable across both cell lines. Overex- lated Hippo pathway. P Dvl YAP pression of active TAZ increases MCF10A proliferation and the Dvl TAZ formation of invasive carcinomas. These observations support an STEM/PROGENITOR P β-catenin Conclusions EXPANSION YAP important role of TAZ in breast cancer stem cells. TAZ Other studies have shown that nuclear TAZ is highly expressed in high-grade glioblastomas. Ectopic expression of TAZ leads to Although most of the Hippo pathway components were initially increased invasion, self-renewal, and tumor initiating capacity to identified in Drosophila, much research has recently been done in generate properties similar to mesenchymal-like stem cells [140]. mammalian cells and animal models, revealing this pathway’s Dvl Conversely, knockdown of TAZ expression in mesenchymal-like important contribution to tissue homeostasis, organ size control, β-catenin YAP TAZ stem cells decreases their mesenchymal properties and limits their cancer development, and stem cell biology. As the key downstream TCF TEAD TCF capacity to self-renewal and initiate in glioma. Collectively, it is effectors of the Hippo pathway, YAP/TAZ is involved in embryonic clear that TAZ enhances the self-renewal capacity and tumorigenic stem cells as well as tissue-specific stem cell self-renewal, and tissue potential contributing to both the initiation and progression of regeneration and homeostasis of the liver, intestine, pancreas, heart, breast cancer and glioma. Therefore, TAZ could be a potential skin, and central nervous system. Moreover, compelling evidence molecular target for treating aggressive tumors that have uncon- supports a role for YAP/TAZ in cancer stem cells. Therefore, compo- Figure 4. The context-dependent role of YAP in intestinal stem cell expansion. trolled TAZ activation. nents of the Hippo pathway may be good therapeutic targets in In the intestinal stem cells (ISC), the Hippo pathway inhibits YAP activity by phosphorylation and cytosolic retention of YAP. The cytosolic YAP directly binds Dysregulation of the Hippo pathway has been identified in a diseases such as degeneration and cancer. to -catenin and subsequently inhibits the canonical Wnt signaling. In Mst1/2 / intestinal epithelia, loss of Hippo pathway regulation promotes dephosphorylation b À À broad range of human cancers, including liver, lung, colorectal, Since the discovery of the Wrts kinase in Drosophila in 1995, for and nuclear translocation of YAP/b-catenin and induces their target gene expression. Activation of YAP/b-catenin results in the expansion of ISC. However, a controversial / ovarian, and prostate [12,103,137]. Studies have shown that YAP the first decade research in the Hippo pathway was largely limited role of YAP has been demonstrated in the context of Wnt-induced intestinal regeneration. In YAPÀ À intestinal epithelia, hyperactivation of Wnt/b-catenin signaling results in ISC expansion, whereas YAP overexpression represses Wnt/b-catenin signaling, which leads to the loss of ISC and epithelial self-renewal. In this context, YAP functions to activity is increased as a result of increased expression and nuclear to Drosophila. However, rapid progress, especially in the last several inhibit the nuclear translocation of disheveled (Dvl). localization in human tumor samples. This is consistent with inacti- years, has been made regarding the identification of upstream vation of the Hippo pathway which is known to inhibit YAP and components, signals, and mechanisms of regulation in both TAZ activity mainly by promoting these transcriptional coactivators’ Drosophila and mammalian systems. Cell polarity, adhesion, cytoplasmic localization and ubiquitin-mediated degradation. In mechanotransduction, as well as diffusible signals acting through directly promoting ISC or negatively by indirectly inhibiting Wnt Hippo signaling and cancer stem cells addition, YAP gene amplification (somatic mutation) has been GPCRs, have all been identified as regulators of Hippo pathway signaling. reported in various human and murine tumor models [136,141]. activity. However, many key questions remain to be addressed. The In the Drosophila midgut, the Hippo pathway and Yki facilitate As discussed above, the Hippo pathway plays a key role in regulat- Collectively, these data suggest that unrestrained YAP activity can function of YAP/TAZ has been investigated in only a few cell types. intestinal regeneration after tissue injury [131–133]. Perturbation of ing organ size and tumorigenesis by inhibiting cell proliferation, counteract classical tumor suppressor checkpoints. Further studies to uncover the physiological roles of YAP/TAZ in a Hippo signaling or overexpression of a constitutively active Yki promoting apoptosis, and regulating stem/progenitor cell expansion Compared with other well-known oncogenic signaling pathway, broad range of tissue-specific stem cells and various types of cancer mutant (Yki-S168A) induced the expression of the Upd (Out- [134,135]. Phosphorylated YAP/TAZ localizes to the cytosol, only few cancers are known to be associated with a direct mutation stem cells will likely expand our knowledge of the Hippo pathway stretched), which is a cytokine that stimulates expansion of ISC decreasing tumor growth, whereas unphosphorylated YAP/TAZ is of a Hippo pathway component. Of note, Lats2 is mutated in approxi- in regulating tissue homeostasis during development and adulthood through the JAK/STAT pathway. However, further investigation is localized mainly in the nucleus and promotes cell and tumor mately 40% of mesothelioma cases [142]. Interestingly, Mst1/2 and as well as cancer initiation and metastasis. Organ size regulation is required to address whether Upd acts in an autocrine fashion via growth. Indeed, there is considerable evidence that abnormal Hippo Lats1 are tumor suppressors in mice, and although mutation in these a fundamental question in biology, though the signals critical for Hippo-Yki signaling in the ISC [131] or whether it triggers a non- signaling is associated with tumor progression. As expected, genes have not been identified in human cancer, silencing of these sensing organ size control, with each organ presumably having its autonomous increase in ISC expansion via Hippo-Yki signaling in elevated expression and activity of YAP/TAZ correlates with various genes have been reported to data, suggesting that these genes may own specific signals, are unknown. Research into the molecular the enterocytes [132]. human cancers [103,136–138]. be inactivated by non-mutational mechanisms [94–96,129,143–146]. signals controlling organ size will be of paramount importance not

650 EMBO reports Vol 15 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO reports Vol 15 | No 6 | 2014 651 EMBO reports Hippo pathway in stem cells and cancer Jung-Soon Mo et al Jung-Soon Mo et al Hippo pathway in stem cells and cancer EMBO reports

6. Udan RS, Kango-Singh M, Nolo R, Tao C, Halder G (2003) Hippo 25. Wang K, Degerny C, Xu M, Yang XJ (2009) YAP, TAZ, and Yorkie: a acts downstream of alpha-catenin to control epidermal proliferation. Sidebar A: In need of answers promotes proliferation arrest and apoptosis in the Salvador/Warts conserved family of signal-responsive transcriptional coregulators in Cell 144: 782 – 795 (i) Does the stemness function of Hippo-YAP pathway differ from its pathway. Nat Cell Biol 5: 914 – 920 animal development and human disease. Biochem Cell Biol 87: 77 – 91 42. Silvis MR, Kreger BT, Lien WH, Klezovitch O, Rudakova GM, Camargo more classic role in cell growth regulation? Are different down- 7. Wu S, Huang J, Dong J, Pan D (2003) hippo encodes a Ste-20 family 26. McCartney BM, Kulikauskas RM, LaJeunesse DR, Fehon RG (2000) The FD, Lantz DM, Seykora JT, Vasioukhin V (2011) alpha-catenin is a tumor stream target transcription utilized in both processes? protein kinase that restricts cell proliferation and promotes apoptosis neurofibromatosis-2 homologue, Merlin, and the tumor suppressor suppressor that controls cell accumulation by regulating the localiza- (ii) What are the molecular bases for the differential functions of YAP in conjunction with salvador and warts. Cell 114: 445 – 456 expanded function together in Drosophila to regulate cell proliferation tion and activity of the transcriptional coactivator Yap1. Sci Signal 4: and TAZ in stem cells? Is this simply due to differential expression 8. Kango-Singh M, Nolo R, Tao C, Verstreken P, Hiesinger PR, Bellen HJ, and differentiation. Development 127: 1315 – 1324 ra33 of these two proteins? Halder G (2002) Shar-pei mediates cell proliferation arrest during 27. Hamaratoglu F, Willecke M, Kango-Singh M, Nolo R, Hyun E, Tao C, 43. Liu X, Yang N, Figel SA, Wilson KE, Morrison CD, Gelman IH, Zhang J (iii) How is YAP/TAZ regulated by stemness signals? Is there hippo imaginal disc growth in Drosophila. Development 129: 5719 – 5730 Jafar-Nejad H, Halder G (2006) The tumour-suppressor genes NF2/ (2013) PTPN14 interacts with and negatively regulates the oncogenic kinase cascade MST-Lats independent regulation of YAP/TAZ? 9. Tapon N, Harvey KF, Bell DW, Wahrer DC, Schiripo TA, Haber D, Hariha- Merlin and Expanded act through Hippo signalling to regulate cell function of YAP. Oncogene 32: 1266 – 1273 ran IK (2002) salvador Promotes both cell cycle exit and apoptosis in proliferation and apoptosis. Nat Cell Biol 8: 27 – 36 44. Wang W, Huang J, Wang X, Yuan J, Li X, Feng L, Park JI, Chen J (2012) (iv) YAP/TAZ clearly plays different roles in different stem cell types, Drosophila and is mutated in human cancer cell lines. Cell 110: 28. Baumgartner R, Poernbacher I, Buser N, Hafen E, Stocker H (2010) The PTPN14 is required for the density-dependent control of YAP1. Genes either being inhibitory or activating. What determines the specificity and complexity of the Hippo-YAP in tissue-specific stem cells? 467 – 478 WW domain protein Kibra acts upstream of Hippo in Drosophila. Dev Dev 26: 1959 – 1971 10. Lai ZC, Wei X, Shimizu T, Ramos E, Rohrbaugh M, Nikolaidis N, Ho LL, Cell 18: 309 – 316 45. Huang JM, Nagatomo I, Suzuki E, Mizuno T, Kumagai T, Berezov A, Li Y (2005) Control of cell proliferation and apoptosis by mob as tumor 29. Genevet A, Wehr MC, Brain R, Thompson BJ, Tapon N (2010) Kibra is a Zhang H, Karlan B, Greene MI, Wang Q (2013) YAP modifies cancer cell only for the Hippo pathway but also for the field of developmental suppressor, mats. Cell 120: 675 – 685 regulator of the Salvador/Warts/Hippo signaling network. Dev Cell 18: sensitivity to EGFR and survivin inhibitors and is negatively regulated biology. Because the Hippo pathway is regulated by a wide range of 11. Huang J, Wu S, Barrera J, Matthews K, Pan D (2005) The Hippo 300 – 308 by the non-receptor type protein tyrosine phosphatase 14. Oncogene signals, both physical and chemical, how the Hippo pathway inte- signaling pathway coordinately regulates cell proliferation and apopto- 30. Yu J, Zheng Y, Dong J, Klusza S, Deng WM, Pan D (2010) Kibra 32: 2220 – 2229 grates all of these inputs from multiple signaling pathways to gener- sis by inactivating Yorkie, the Drosophila Homolog of YAP. Cell 122: functions as a tumor suppressor protein that regulates Hippo signaling 46. Das Thakur M, Feng Y, Jagannathan R, Seppa MJ, Skeath JB, Longmore ate a concerted cellular response remains a question of high 421 – 434 in conjunction with Merlin and Expanded. Dev Cell 18: GD (2010) Ajuba LIM proteins are negative regulators of the Hippo interest. Understanding the molecular mechanisms by which the 12. Dong J, Feldmann G, Huang J, Wu S, Zhang N, Comerford SA, Gayyed 288 – 299 signaling pathway. Curr Biol 20: 657 – 662 Hippo pathway controls development, regeneration, tissue homeo- MF, Anders RA, Maitra A, Pan D (2007) Elucidation of a universal size- 31. Varelas X, Samavarchi-Tehrani P, Narimatsu M, Weiss A, Cockburn K, 47. Rauskolb C, Pan G, Reddy BV, Oh H, Irvine KD (2011) Zyxin links fat stasis, and injury/repair will require the input of researchers across control mechanism in Drosophila and mammals. Cell 130: 1120 – 1133 Larsen BG, Rossant J, Wrana JL (2010) The Crumbs complex couples cell signaling to the hippo pathway. PLoS Biol 9:e1000624 multiple disciplines, including genetic, genomic, developmental, 13. Staley BK, Irvine KD (2012) Hippo signaling in Drosophila: recent density sensing to Hippo-dependent control of the TGF-beta-SMAD 48. Mao Y, Mulvaney J, Zakaria S, Yu T, Morgan KM, Allen S, Basson MA, systems biology, cell biology, biochemistry, and cancer biology. advances and insights. Dev Dyn 241: 3 – 15 pathway. Dev Cell 19: 831 – 844 Francis-West P, Irvine KD (2011) Characterization of a Dchs1 mutant Given the increasing research interest in this pathway, continued 14. Oh H, Irvine KD (2008) In vivo regulation of Yorkie phosphorylation 32. Chen CL, Gajewski KM, Hamaratoglu F, Bossuyt W, Sansores-Garcia L, mouse reveals requirements for Dchs1-Fat4 signaling during mamma- rapid progress is eagerly anticipated. and localization. Development 135: 1081 – 1088 Tao C, Halder G (2010) The apical-basal cell polarity determinant lian development. Development 138: 947 – 957 15. Ren F, Zhang L, Jiang J (2010) Hippo signaling regulates Yorkie nuclear Crumbs regulates Hippo signaling in Drosophila. Proc Natl Acad Sci U S 49. Bossuyt W, Chen CL, Chen Q, Sudol M, McNeill H, Pan D, Kopp A, Acknowledgments localization and activity through 14-3-3 dependent and independent A 107: 15810 – 15815 Halder G (2014) An evolutionary shift in the regulation of the Hippo We apologize for the many important contributions to the Hippo pathway mechanisms. Dev Biol 337: 303 – 312 33. Robinson BS, Huang J, Hong Y, Moberg KH (2010) Crumbs regulates pathway between mice and flies. Oncogene 33: 1218 – 1228 field that we could not cite owing to space constraints. The authors would like 16. Mahoney WM Jr, Hong JH, Yaffe MB, Farrance IK (2005) The transcrip- Salvador/Warts/Hippo signaling in Drosophila via the FERM-domain 50. Dupont S, Morsut L, Aragona M, Enzo E, Giulitti S, Cordenonsi M, to thank Steven W Plouffe and Dr. Carsten G. Hansen for critical reading of this tional co-activator TAZ interacts differentially with transcriptional protein Expanded. Curr Biol 20: 582 – 590 Zanconato F, Le Digabel J, Forcato M, Bicciato S et al (2011) Role of manuscript. This work is supported by grants from the National institutes of enhancer factor-1 (TEF-1) family members. Biochem J 388: 217 – 225 34. Grzeschik NA, Parsons LM, Allott ML, Harvey KF, Richardson HE (2010) YAP/TAZ in mechanotransduction. Nature 474: 179 – 183 Health to K.L.G. (CA132809 and EYO226116). 17. Vassilev A, Kaneko KJ, Shu H, Zhao Y, DePamphilis ML (2001) TEAD/TEF Lgl, aPKC, and Crumbs regulate the Salvador/Warts/Hippo pathway 51. Wada K, Itoga K, Okano T, Yonemura S, Sasaki H (2011) Hippo pathway transcription factors utilize the activation domain of YAP65, a Src/Yes- through two distinct mechanisms. Curr Biol 20: 573 – 581 regulation by cell morphology and stress fibers. Development 138: Conflict of interest associated protein localized in the cytoplasm. Genes Dev 15: 35. Ling C, Zheng Y, Yin F, Yu J, Huang J, Hong Y, Wu S, Pan D (2010) The 3907 – 3914 The authors declare that they have no conflict of interest. 1229 – 1241 apical transmembrane protein Crumbs functions as a tumor suppressor 52. Zhao B, Li L, Wang L, Wang CY, Yu J, Guan KL (2012) Cell detachment 18. Peng HW, Slattery M, Mann RS (2009) Transcription factor choice in that regulates Hippo signaling by binding to Expanded. Proc Natl Acad activates the Hippo pathway via cytoskeleton reorganization to induce the Hippo signaling pathway: homothorax and yorkie regulation of the Sci U S A 107: 10532 – 10537 anoikis. Genes Dev 26: 54 – 68 References microRNA bantam in the progenitor domain of the Drosophila eye 36. Martin-Belmonte F, Perez-Moreno M (2012) Epithelial cell polarity, 53. Yu FX, Zhao B, Panupinthu N, Jewell JL, Lian I, Wang LH, Zhao J, Yuan imaginal disc. Genes Dev 23: 2307 – 2319 stem cells and cancer. Nat Rev Cancer 12: 23 – 38 H, Tumaneng K, Li H et al (2012) Regulation of the Hippo-YAP pathway 1. Harvey KF, Pfleger CM, Hariharan IK (2003) The Drosophila Mst ortho- 19. Oh H, Irvine KD (2011) Cooperative regulation of growth by Yorkie and 37. Menendez J, Perez-Garijo A, Calleja M, Morata G (2010) A tumor- by G-protein-coupled receptor signaling. Cell 150: 780 – 791 log, hippo, restricts growth and cell proliferation and promotes apopto- Mad through bantam. Dev Cell 20: 109 – 122 suppressing mechanism in Drosophila involving cell competition and 54 . Mo JS, Yu FX, Gong R, Brown JH, Guan KL (2012) Regulation of the sis. Cell 114: 457 – 467 20. Chan EH, Nousiainen M, Chalamalasetty RB, Schafer A, Nigg EA, Sillje the Hippo pathway. Proc Natl Acad Sci U S A 107: 14651 – 14656 Hippo-YAP pathway by protease-activated receptors (PARs). Genes Dev 2. Jia J, Zhang W, Wang B, Trinko R, Jiang J (2003) The Drosophila HH (2005) The Ste20-like kinase Mst2 activates the human large tumor 38. Remue E, Meerschaert K, Oka T, Boucherie C, Vandekerckhove J, 26: 2138 – 2143 Ste20 family kinase dMST functions as a tumor suppressor by suppressor kinase Lats1. Oncogene 24: 2076 – 2086 Sudol M, Gettemans J (2010) TAZ interacts with zonula occludens-1 55. Miller E, Yang J, DeRan M, Wu C, Su AI, Bonamy GM, Liu J, Peters EC, restricting cell proliferation and promoting apoptosis. Genes Dev 17: 21. Praskova M, Xia F, Avruch J (2008) MOBKL1A/MOBKL1B phosphorylation and -2 proteins in a PDZ-1 dependent manner. FEBS Lett 584: Wu X (2012) Identification of serum-derived sphingosine-1-phosphate 2514 – 2519 by MST1 and MST2 inhibits cell proliferation. Curr Biol 18: 311 – 321 4175 – 4180 as a small molecule regulator of YAP. Chem Biol 19: 955 – 962 3. Justice RW, Zilian O, Woods DF, Noll M, Bryant PJ (1995) The Drosophila 22. Callus BA, Verhagen AM, Vaux DL (2006) Association of mammalian 39. Oka T, Remue E, Meerschaert K, Vanloo B, Boucherie C, Gfeller D, Bader 56. Aragona M, Panciera T, Manfrin A, Giulitti S, Michielin F, Elvassore N, tumor suppressor gene warts encodes a homolog of human myotonic sterile twenty kinases, Mst1 and Mst2, with hSalvador via C-terminal GD, Sidhu SS, Vandekerckhove J, Gettemans J et al (2010) Functional Dupont S, Piccolo S (2013) A mechanical checkpoint controls multicel- dystrophy kinase and is required for the control of cell shape and coiled-coil domains, leads to its stabilization and phosphorylation. FEBS complexes between YAP2 and ZO-2 are PDZ domain-dependent, and lular growth through YAP/TAZ regulation by actin-processing factors. proliferation. Genes Dev 9: 534 – 546 J 273: 4264 – 4276 regulate YAP2 nuclear localization and signalling. Biochem J 432: Cell 154: 1047 – 1059 4. Pantalacci S, Tapon N, Leopold P (2003) The Salvador partner Hippo 23. Zhao B, Li L, Tumaneng K, Wang CY, Guan KL (2010) A coordinated 461 – 472 57. Boggiano JC, Vanderzalm PJ, Fehon RG (2011) Tao-1 phosphorylates promotes apoptosis and cell-cycle exit in Drosophila. Nat Cell Biol 5: phosphorylation by Lats and CK1 regulates YAP stability through SCF 40. Kim NG, Koh E, Chen X, Gumbiner BM (2011) E-cadherin mediates Hippo/MST kinases to regulate the Hippo-Salvador-Warts tumor 921 – 927 (beta-TRCP). Genes Dev 24: 72 – 85 contact inhibition of proliferation through Hippo signaling-pathway suppressor pathway. Dev Cell 21: 888 – 895 5. Xu T, Wang W, Zhang S, Stewart RA, Yu W (1995) Identifying tumor 24. Zhao B, Ye X, Yu J, Li L, Li W, Li S, Yu J, Lin JD, Wang CY, Chinnaiyan components. Proc Natl Acad Sci U S A 108: 11930 – 11935 58. Poon CL, Lin JI, Zhang X, Harvey KF (2011) The sterile 20-like kinase suppressors in genetic mosaics: the Drosophila lats gene encodes a AM et al (2008) TEAD mediates YAP-dependent gene induction and 41. Schlegelmilch K, Mohseni M, Kirak O, Pruszak J, Rodriguez JR, Zhou D, Tao-1 controls tissue growth by regulating the Salvador-Warts-Hippo putative protein kinase. Development 121: 1053 – 1063 growth control. Genes Dev 22: 1962 – 1971 Kreger BT, Vasioukhin V, Avruch J, Brummelkamp TR et al (2011) Yap1 pathway. Dev Cell 21: 896 – 906

652 EMBO reports Vol 15 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO reports Vol 15 | No 6 | 2014 653 EMBO reports Hippo pathway in stem cells and cancer Jung-Soon Mo et al Jung-Soon Mo et al Hippo pathway in stem cells and cancer EMBO reports

6. Udan RS, Kango-Singh M, Nolo R, Tao C, Halder G (2003) Hippo 25. Wang K, Degerny C, Xu M, Yang XJ (2009) YAP, TAZ, and Yorkie: a acts downstream of alpha-catenin to control epidermal proliferation. Sidebar A: In need of answers promotes proliferation arrest and apoptosis in the Salvador/Warts conserved family of signal-responsive transcriptional coregulators in Cell 144: 782 – 795 (i) Does the stemness function of Hippo-YAP pathway differ from its pathway. Nat Cell Biol 5: 914 – 920 animal development and human disease. Biochem Cell Biol 87: 77 – 91 42. Silvis MR, Kreger BT, Lien WH, Klezovitch O, Rudakova GM, Camargo more classic role in cell growth regulation? Are different down- 7. Wu S, Huang J, Dong J, Pan D (2003) hippo encodes a Ste-20 family 26. McCartney BM, Kulikauskas RM, LaJeunesse DR, Fehon RG (2000) The FD, Lantz DM, Seykora JT, Vasioukhin V (2011) alpha-catenin is a tumor stream target transcription utilized in both processes? protein kinase that restricts cell proliferation and promotes apoptosis neurofibromatosis-2 homologue, Merlin, and the tumor suppressor suppressor that controls cell accumulation by regulating the localiza- (ii) What are the molecular bases for the differential functions of YAP in conjunction with salvador and warts. Cell 114: 445 – 456 expanded function together in Drosophila to regulate cell proliferation tion and activity of the transcriptional coactivator Yap1. Sci Signal 4: and TAZ in stem cells? Is this simply due to differential expression 8. Kango-Singh M, Nolo R, Tao C, Verstreken P, Hiesinger PR, Bellen HJ, and differentiation. Development 127: 1315 – 1324 ra33 of these two proteins? Halder G (2002) Shar-pei mediates cell proliferation arrest during 27. Hamaratoglu F, Willecke M, Kango-Singh M, Nolo R, Hyun E, Tao C, 43. Liu X, Yang N, Figel SA, Wilson KE, Morrison CD, Gelman IH, Zhang J (iii) How is YAP/TAZ regulated by stemness signals? Is there hippo imaginal disc growth in Drosophila. Development 129: 5719 – 5730 Jafar-Nejad H, Halder G (2006) The tumour-suppressor genes NF2/ (2013) PTPN14 interacts with and negatively regulates the oncogenic kinase cascade MST-Lats independent regulation of YAP/TAZ? 9. Tapon N, Harvey KF, Bell DW, Wahrer DC, Schiripo TA, Haber D, Hariha- Merlin and Expanded act through Hippo signalling to regulate cell function of YAP. Oncogene 32: 1266 – 1273 ran IK (2002) salvador Promotes both cell cycle exit and apoptosis in proliferation and apoptosis. Nat Cell Biol 8: 27 – 36 44. Wang W, Huang J, Wang X, Yuan J, Li X, Feng L, Park JI, Chen J (2012) (iv) YAP/TAZ clearly plays different roles in different stem cell types, either being inhibitory or activating. What determines the specific- Drosophila and is mutated in human cancer cell lines. Cell 110: 28. Baumgartner R, Poernbacher I, Buser N, Hafen E, Stocker H (2010) The PTPN14 is required for the density-dependent control of YAP1. Genes ity and complexity of the Hippo-YAP in tissue-specific stem cells? 467 – 478 WW domain protein Kibra acts upstream of Hippo in Drosophila. Dev Dev 26: 1959 – 1971 10. Lai ZC, Wei X, Shimizu T, Ramos E, Rohrbaugh M, Nikolaidis N, Ho LL, Cell 18: 309 – 316 45. Huang JM, Nagatomo I, Suzuki E, Mizuno T, Kumagai T, Berezov A, Li Y (2005) Control of cell proliferation and apoptosis by mob as tumor 29. Genevet A, Wehr MC, Brain R, Thompson BJ, Tapon N (2010) Kibra is a Zhang H, Karlan B, Greene MI, Wang Q (2013) YAP modifies cancer cell only for the Hippo pathway but also for the field of developmental suppressor, mats. Cell 120: 675 – 685 regulator of the Salvador/Warts/Hippo signaling network. Dev Cell 18: sensitivity to EGFR and survivin inhibitors and is negatively regulated biology. Because the Hippo pathway is regulated by a wide range of 11. Huang J, Wu S, Barrera J, Matthews K, Pan D (2005) The Hippo 300 – 308 by the non-receptor type protein tyrosine phosphatase 14. Oncogene signals, both physical and chemical, how the Hippo pathway inte- signaling pathway coordinately regulates cell proliferation and apopto- 30. Yu J, Zheng Y, Dong J, Klusza S, Deng WM, Pan D (2010) Kibra 32: 2220 – 2229 grates all of these inputs from multiple signaling pathways to gener- sis by inactivating Yorkie, the Drosophila Homolog of YAP. Cell 122: functions as a tumor suppressor protein that regulates Hippo signaling 46. Das Thakur M, Feng Y, Jagannathan R, Seppa MJ, Skeath JB, Longmore ate a concerted cellular response remains a question of high 421 – 434 in conjunction with Merlin and Expanded. Dev Cell 18: GD (2010) Ajuba LIM proteins are negative regulators of the Hippo interest. Understanding the molecular mechanisms by which the 12. Dong J, Feldmann G, Huang J, Wu S, Zhang N, Comerford SA, Gayyed 288 – 299 signaling pathway. Curr Biol 20: 657 – 662 Hippo pathway controls development, regeneration, tissue homeo- MF, Anders RA, Maitra A, Pan D (2007) Elucidation of a universal size- 31. Varelas X, Samavarchi-Tehrani P, Narimatsu M, Weiss A, Cockburn K, 47. Rauskolb C, Pan G, Reddy BV, Oh H, Irvine KD (2011) Zyxin links fat stasis, and injury/repair will require the input of researchers across control mechanism in Drosophila and mammals. Cell 130: 1120 – 1133 Larsen BG, Rossant J, Wrana JL (2010) The Crumbs complex couples cell signaling to the hippo pathway. PLoS Biol 9:e1000624 multiple disciplines, including genetic, genomic, developmental, 13. Staley BK, Irvine KD (2012) Hippo signaling in Drosophila: recent density sensing to Hippo-dependent control of the TGF-beta-SMAD 48. Mao Y, Mulvaney J, Zakaria S, Yu T, Morgan KM, Allen S, Basson MA, systems biology, cell biology, biochemistry, and cancer biology. advances and insights. Dev Dyn 241: 3 – 15 pathway. Dev Cell 19: 831 – 844 Francis-West P, Irvine KD (2011) Characterization of a Dchs1 mutant Given the increasing research interest in this pathway, continued 14. Oh H, Irvine KD (2008) In vivo regulation of Yorkie phosphorylation 32. Chen CL, Gajewski KM, Hamaratoglu F, Bossuyt W, Sansores-Garcia L, mouse reveals requirements for Dchs1-Fat4 signaling during mamma- rapid progress is eagerly anticipated. and localization. Development 135: 1081 – 1088 Tao C, Halder G (2010) The apical-basal cell polarity determinant lian development. Development 138: 947 – 957 15. Ren F, Zhang L, Jiang J (2010) Hippo signaling regulates Yorkie nuclear Crumbs regulates Hippo signaling in Drosophila. Proc Natl Acad Sci U S 49. Bossuyt W, Chen CL, Chen Q, Sudol M, McNeill H, Pan D, Kopp A, Acknowledgments localization and activity through 14-3-3 dependent and independent A 107: 15810 – 15815 Halder G (2014) An evolutionary shift in the regulation of the Hippo We apologize for the many important contributions to the Hippo pathway mechanisms. Dev Biol 337: 303 – 312 33. Robinson BS, Huang J, Hong Y, Moberg KH (2010) Crumbs regulates pathway between mice and flies. Oncogene 33: 1218 – 1228 field that we could not cite owing to space constraints. The authors would like 16. Mahoney WM Jr, Hong JH, Yaffe MB, Farrance IK (2005) The transcrip- Salvador/Warts/Hippo signaling in Drosophila via the FERM-domain 50. Dupont S, Morsut L, Aragona M, Enzo E, Giulitti S, Cordenonsi M, to thank Steven W Plouffe and Dr. Carsten G. Hansen for critical reading of this tional co-activator TAZ interacts differentially with transcriptional protein Expanded. Curr Biol 20: 582 – 590 Zanconato F, Le Digabel J, Forcato M, Bicciato S et al (2011) Role of manuscript. This work is supported by grants from the National institutes of enhancer factor-1 (TEF-1) family members. Biochem J 388: 217 – 225 34. Grzeschik NA, Parsons LM, Allott ML, Harvey KF, Richardson HE (2010) YAP/TAZ in mechanotransduction. Nature 474: 179 – 183 Health to K.L.G. (CA132809 and EYO226116). 17. Vassilev A, Kaneko KJ, Shu H, Zhao Y, DePamphilis ML (2001) TEAD/TEF Lgl, aPKC, and Crumbs regulate the Salvador/Warts/Hippo pathway 51. Wada K, Itoga K, Okano T, Yonemura S, Sasaki H (2011) Hippo pathway transcription factors utilize the activation domain of YAP65, a Src/Yes- through two distinct mechanisms. Curr Biol 20: 573 – 581 regulation by cell morphology and stress fibers. Development 138: Conflict of interest associated protein localized in the cytoplasm. Genes Dev 15: 35. Ling C, Zheng Y, Yin F, Yu J, Huang J, Hong Y, Wu S, Pan D (2010) The 3907 – 3914 The authors declare that they have no conflict of interest. 1229 – 1241 apical transmembrane protein Crumbs functions as a tumor suppressor 52. Zhao B, Li L, Wang L, Wang CY, Yu J, Guan KL (2012) Cell detachment 18. Peng HW, Slattery M, Mann RS (2009) Transcription factor choice in that regulates Hippo signaling by binding to Expanded. Proc Natl Acad activates the Hippo pathway via cytoskeleton reorganization to induce the Hippo signaling pathway: homothorax and yorkie regulation of the Sci U S A 107: 10532 – 10537 anoikis. Genes Dev 26: 54 – 68 References microRNA bantam in the progenitor domain of the Drosophila eye 36. Martin-Belmonte F, Perez-Moreno M (2012) Epithelial cell polarity, 53. Yu FX, Zhao B, Panupinthu N, Jewell JL, Lian I, Wang LH, Zhao J, Yuan imaginal disc. Genes Dev 23: 2307 – 2319 stem cells and cancer. Nat Rev Cancer 12: 23 – 38 H, Tumaneng K, Li H et al (2012) Regulation of the Hippo-YAP pathway 1. Harvey KF, Pfleger CM, Hariharan IK (2003) The Drosophila Mst ortho- 19. Oh H, Irvine KD (2011) Cooperative regulation of growth by Yorkie and 37. Menendez J, Perez-Garijo A, Calleja M, Morata G (2010) A tumor- by G-protein-coupled receptor signaling. Cell 150: 780 – 791 log, hippo, restricts growth and cell proliferation and promotes apopto- Mad through bantam. Dev Cell 20: 109 – 122 suppressing mechanism in Drosophila involving cell competition and 54 . Mo JS, Yu FX, Gong R, Brown JH, Guan KL (2012) Regulation of the sis. Cell 114: 457 – 467 20. Chan EH, Nousiainen M, Chalamalasetty RB, Schafer A, Nigg EA, Sillje the Hippo pathway. Proc Natl Acad Sci U S A 107: 14651 – 14656 Hippo-YAP pathway by protease-activated receptors (PARs). Genes Dev 2. Jia J, Zhang W, Wang B, Trinko R, Jiang J (2003) The Drosophila HH (2005) The Ste20-like kinase Mst2 activates the human large tumor 38. Remue E, Meerschaert K, Oka T, Boucherie C, Vandekerckhove J, 26: 2138 – 2143 Ste20 family kinase dMST functions as a tumor suppressor by suppressor kinase Lats1. Oncogene 24: 2076 – 2086 Sudol M, Gettemans J (2010) TAZ interacts with zonula occludens-1 55. Miller E, Yang J, DeRan M, Wu C, Su AI, Bonamy GM, Liu J, Peters EC, restricting cell proliferation and promoting apoptosis. Genes Dev 17: 21. Praskova M, Xia F, Avruch J (2008) MOBKL1A/MOBKL1B phosphorylation and -2 proteins in a PDZ-1 dependent manner. FEBS Lett 584: Wu X (2012) Identification of serum-derived sphingosine-1-phosphate 2514 – 2519 by MST1 and MST2 inhibits cell proliferation. Curr Biol 18: 311 – 321 4175 – 4180 as a small molecule regulator of YAP. Chem Biol 19: 955 – 962 3. Justice RW, Zilian O, Woods DF, Noll M, Bryant PJ (1995) The Drosophila 22. Callus BA, Verhagen AM, Vaux DL (2006) Association of mammalian 39. Oka T, Remue E, Meerschaert K, Vanloo B, Boucherie C, Gfeller D, Bader 56. Aragona M, Panciera T, Manfrin A, Giulitti S, Michielin F, Elvassore N, tumor suppressor gene warts encodes a homolog of human myotonic sterile twenty kinases, Mst1 and Mst2, with hSalvador via C-terminal GD, Sidhu SS, Vandekerckhove J, Gettemans J et al (2010) Functional Dupont S, Piccolo S (2013) A mechanical checkpoint controls multicel- dystrophy kinase and is required for the control of cell shape and coiled-coil domains, leads to its stabilization and phosphorylation. FEBS complexes between YAP2 and ZO-2 are PDZ domain-dependent, and lular growth through YAP/TAZ regulation by actin-processing factors. proliferation. Genes Dev 9: 534 – 546 J 273: 4264 – 4276 regulate YAP2 nuclear localization and signalling. Biochem J 432: Cell 154: 1047 – 1059 4. Pantalacci S, Tapon N, Leopold P (2003) The Salvador partner Hippo 23. Zhao B, Li L, Tumaneng K, Wang CY, Guan KL (2010) A coordinated 461 – 472 57. Boggiano JC, Vanderzalm PJ, Fehon RG (2011) Tao-1 phosphorylates promotes apoptosis and cell-cycle exit in Drosophila. Nat Cell Biol 5: phosphorylation by Lats and CK1 regulates YAP stability through SCF 40. Kim NG, Koh E, Chen X, Gumbiner BM (2011) E-cadherin mediates Hippo/MST kinases to regulate the Hippo-Salvador-Warts tumor 921 – 927 (beta-TRCP). Genes Dev 24: 72 – 85 contact inhibition of proliferation through Hippo signaling-pathway suppressor pathway. Dev Cell 21: 888 – 895 5. Xu T, Wang W, Zhang S, Stewart RA, Yu W (1995) Identifying tumor 24. Zhao B, Ye X, Yu J, Li L, Li W, Li S, Yu J, Lin JD, Wang CY, Chinnaiyan components. Proc Natl Acad Sci U S A 108: 11930 – 11935 58. Poon CL, Lin JI, Zhang X, Harvey KF (2011) The sterile 20-like kinase suppressors in genetic mosaics: the Drosophila lats gene encodes a AM et al (2008) TEAD mediates YAP-dependent gene induction and 41. Schlegelmilch K, Mohseni M, Kirak O, Pruszak J, Rodriguez JR, Zhou D, Tao-1 controls tissue growth by regulating the Salvador-Warts-Hippo putative protein kinase. Development 121: 1053 – 1063 growth control. Genes Dev 22: 1962 – 1971 Kreger BT, Vasioukhin V, Avruch J, Brummelkamp TR et al (2011) Yap1 pathway. Dev Cell 21: 896 – 906

59. Huang HL, Wang S, Yin MX, Dong L, Wang C, Wu W, Lu Y, Feng M, Dai 76. Nishioka N, Inoue K, Adachi K, Kiyonari H, Ota M, Ralston A, Yabuta N, 94. Song H, Mak KK, Topol L, Yun K, Hu J, Garrett L, Chen Y, Park O, Chang 111. Van Hateren NJ, Das RM, Hautbergue GM, Borycki AG, Placzek M, C, Guo X et al (2013) Par-1 regulates tissue growth by influencing Hirahara S, Stephenson RO, Ogonuki N et al (2009) The Hippo signaling J, Simpson RM et al (2010) Mammalian Mst1 and Mst2 kinases play Wilson SA (2011) FatJ acts via the Hippo mediator Yap1 to restrict the hippo phosphorylation status and hippo-salvador association. PLoS Biol pathway components Lats and Yap pattern Tead4 activity to distin- essential roles in organ size control and tumor suppression. Proc Natl size of neural progenitor cell pools. Development 138: 1893 – 1902 11:e1001620 guish mouse trophectoderm from inner cell mass. Dev Cell 16: Acad Sci U S A 107: 1431 – 1436 112. Lavado A, He Y, Pare J, Neale G, Olson EN, Giovannini M, Cao X (2013) 60. Mohseni M, Sun J, Lau A, Curtis S, Goldsmith J, Fox VL, Wei C, Frazier 398 – 410 95. Zhou D, Conrad C, Xia F, Park JS, Payer B, Yin Y, Lauwers GY, Thasler W, Tumor suppressor Nf2 limits expansion of the neural progenitor pool M, Samson O, Wong KK et al (2014) A genetic screen identifies an 77. Cockburn K, Biechele S, Garner J, Rossant J (2013) The Hippo pathway Lee JT, Avruch J et al (2009) Mst1 and Mst2 maintain hepatocyte quies- by inhibiting Yap/Taz transcriptional coactivators. Development 140: LKB1-MARK signalling axis controlling the Hippo-YAP pathway. Nat Cell member Nf2 is required for inner cell mass specification. Curr Biol 23: cence and suppress hepatocellular carcinoma development through 3323 – 3334 Biol 16: 108 – 117 1195 – 1201 inactivation of the Yap1 oncogene. Cancer Cell 16: 425 – 438 113. Cho E, Feng Y, Rauskolb C, Maitra S, Fehon R, Irvine KD (2006) 61. Wehr MC, Holder MV, Gailite I, Saunders RE, Maile TM, Ciirdaeva E, 78. Hirate Y, Hirahara S, Inoue K, Suzuki A, Alarcon VB, Akimoto K, Hirai T, 96. Lu L, Li Y, Kim SM, Bossuyt W, Liu P, Qiu Q, Wang Y, Halder G, Finegold Delineation of a Fat tumor suppressor pathway. Nat Genet 38: Instrell R, Jiang M, Howell M, Rossner MJ et al (2013) Salt-inducible Hara T, Adachi M, Chida K et al (2013) Polarity-dependent distribution MJ, Lee JS et al (2010) Hippo signaling is a potent in vivo growth and 1142 – 1150 kinases regulate growth through the Hippo signalling pathway in of angiomotin localizes Hippo signaling in preimplantation embryos. tumor suppressor pathway in the mammalian liver. Proc Natl Acad Sci 114. Cappello S, Gray MJ, Badouel C, Lange S, Einsiedler M, Srour M, Chita- Drosophila. Nat Cell Biol 15: 61 – 71 Curr Biol 23: 1181 – 1194 USA107: 1437 – 1442 yat D, Hamdan FF, Jenkins ZA, Morgan T et al (2013) Mutations in 62. Praskova M, Khoklatchev A, Ortiz-Vega S, Avruch J (2004) Regulation 79. Hossain Z, Ali SM, Ko HL, Xu J, Ng CP, Guo K, Qi Z, Ponniah S, Hong W, 97. Lee KP, Lee JH, Kim TS, Kim TH, Park HD, Byun JS, Kim MC, Jeong WI, genes encoding the cadherin receptor-ligand pair DCHS1 and FAT4 of the MST1 kinase by autophosphorylation, by the growth inhibi- Hunziker W (2007) Glomerulocystic kidney disease in mice with a Calvisi DF, Kim JM et al (2010) The Hippo-Salvador pathway restrains disrupt cerebral cortical development. Nat Genet 45: 1300 – 1308 tory proteins, RASSF1 and NORE1, and by Ras. Biochem J 381: targeted inactivation of Wwtr1. Proc Natl Acad Sci U S A 104: hepatic oval cell proliferation, liver size, and liver tumorigenesis. Proc 115. Nadal-Ginard B, Kajstura J, Leri A, Anversa P (2003) Myocyte death, 453 – 462 1631 – 1636 Natl Acad Sci U S A 107: 8248 – 8253 growth, and regeneration in cardiac hypertrophy and failure. Circ Res 63. Ikeda M, Kawata A, Nishikawa M, Tateishi Y, Yamaguchi M, Nakagawa 80. Varelas X, Sakuma R, Samavarchi-Tehrani P, Peerani R, Rao BM, Demb- 98. Zhang N, Bai H, David KK, Dong J, Zheng Y, Cai J, Giovannini M, Liu P, 92: 139 – 150 K, Hirabayashi S, Bao Y, Hidaka S, Hirata Y et al (2009) Hippo pathway- owy J, Yaffe MB, Zandstra PW, Wrana JL (2008) TAZ controls Smad Anders RA, Pan D (2010) The Merlin/NF2 tumor suppressor functions 116. Yamamoto S, Yang G, Zablocki D, Liu J, Hong C, Kim SJ, Soler S, Odashi- dependent and -independent roles of RASSF6. Sci Signal 2: ra59 nucleocytoplasmic shuttling and regulates human embryonic stem-cell through the YAP oncoprotein to regulate tissue homeostasis in ma M, Thaisz J, Yehia G et al (2003) Activation of Mst1 causes dilated 64. Sidor CM, Brain R, Thompson BJ (2013) Mask proteins are cofactors of self-renewal. Nat Cell Biol 10: 837 – 848 mammals. Dev Cell 19: 27 – 38 cardiomyopathy by stimulating apoptosis without compensatory Yorkie/YAP in the Hippo pathway. Curr Biol 23: 223 – 228 81. Alarcon C, Zaromytidou AI, Xi Q, Gao S, Yu J, Fujisawa S, Barlas A, Miller 99. Benhamouche S, Curto M, Saotome I, Gladden AB, Liu CH, Giovannini ventricular myocyte hypertrophy. J Clin Invest 111: 1463 – 1474 65. Sansores-Garcia L, Atkins M, Moya IM, Shahmoradgoli M, Tao C, Mills AN, Manova-Todorova K, Macias MJ et al (2009) Nuclear CDKs drive M, McClatchey AI (2010) Nf2/Merlin controls progenitor homeostasis 117. Matsui Y, Nakano N, Shao D, Gao S, Luo W, Hong C, Zhai P, Holle E, Yu GB, Halder G (2013) Mask is required for the activity of the Hippo path- Smad transcriptional activation and turnover in BMP and TGF-beta and tumorigenesis in the liver. Genes Dev 24: 1718 – 1730 X, Yabuta N et al (2008) Lats2 is a negative regulator of myocyte size way effector Yki/YAP. Curr Biol 23: 229 – 235 pathways. Cell 139: 757 – 769 100. Fuchs E, Chen T (2013) A matter of life and death: self-renewal in stem in the heart. Circ Res 103: 1309 – 1318 66. Poon CL, Zhang X, Lin JI, Manning SA, Harvey KF (2012) Homeodomain- 82. Biswas A, Hutchins R (2007) Embryonic stem cells. Stem Cells Dev 16: cells. EMBO Rep 14: 39 – 48 118. Heallen T, Zhang M, Wang J, Bonilla-Claudio M, Klysik E, Johnson RL, interacting protein kinase regulates Hippo pathway-dependent tissue 213 – 222 101. Lee JH, Kim TS, Yang TH, Koo BK, Oh SP, Lee KP, Oh HJ, Lee SH, Kong Martin JF (2011) Hippo pathway inhibits Wnt signaling to restrain growth. Curr Biol 22: 1587 – 1594 83. Darr H, Benvenisty N (2006) Human embryonic stem cells: the battle YY, Kim JM et al (2008) A crucial role of WW45 in developing epithelial cardiomyocyte proliferation and heart size. Science 332: 458 – 461 67. Chen J, Verheyen EM (2012) Homeodomain-interacting protein kinase between self-renewal and differentiation. Regen Med 1: 317 – 325 tissues in the mouse. EMBO J 27: 1231 – 1242 119. Xin M, Kim Y, Sutherland LB, Qi X, McAnally J, Schwartz RJ, Richardson regulates Yorkie activity to promote tissue growth. Curr Biol 22: 84. Xiao L, Yuan X, Sharkis SJ (2006) Activin A maintains self-renewal and 102. Zhang H, Pasolli HA, Fuchs E (2011) Yes-associated protein (YAP) tran- JA, Bassel-Duby R, Olson EN (2011) Regulation of insulin-like growth 1582 – 1586 regulates fibroblast growth factor, Wnt, and bone morphogenic scriptional coactivator functions in balancing growth and differentia- factor signaling by Yap governs cardiomyocyte proliferation and 68. Liu CY, Zha ZY, Zhou X, Zhang H, Huang W, Zhao D, Li T, Chan SW, protein pathways in human embryonic stem cells. Stem Cells 24: tion in skin. Proc Natl Acad Sci U S A 108: 2270 – 2275 embryonic heart size. Sci Signal 4: ra70 Lim CJ, Hong W et al (2010) The hippo tumor pathway promotes TAZ 1476 – 1486 103. Zhao B, Wei X, Li W, Udan RS, Yang Q, Kim J, Xie J, Ikenoue T, Yu J, Li L 120. Xin M, Kim Y, Sutherland LB, Murakami M, Qi X, McAnally J, Porrello degradation by phosphorylating a phosphodegron and recruiting the 85. Ramalho-Santos M, Yoon S, Matsuzaki Y, Mulligan RC, Melton DA et al (2007) Inactivation of YAP oncoprotein by the Hippo pathway is ER, Mahmoud AI, Tan W, Shelton JM et al (2013) Hippo pathway effec- SCF{beta}-TrCP E3 ligase. J Biol Chem 285: 37159 – 37169 (2002) “Stemness”: transcriptional profiling of embryonic and adult involved in cell contact inhibition and tissue growth control. Genes Dev tor Yap promotes cardiac regeneration. Proc Natl Acad Sci U S A 110: 69. Cockburn K, Rossant J (2010) Making the blastocyst: lessons from the stem cells. Science 298: 597 – 600 21: 2747 – 2761 13839 – 13844 mouse. J Clin Invest 120: 995 – 1003 86. Tamm C, Bower N, Anneren C (2011) Regulation of mouse embryonic 104. Zhao B, Li L, Lu Q, Wang LH, Liu CY, Lei Q, Guan KL (2011) Angiomotin 121. von Gise A, Lin Z, Schlegelmilch K, Honor LB, Pan GM, Buck JN, Ma Q, 70. Morin-Kensicki EM, Boone BN, Howell M, Stonebraker JR, Teed J, Alb JG, stem cell self-renewal by a Yes-YAP-TEAD2 signaling pathway down- is a novel Hippo pathway component that inhibits YAP oncoprotein. Ishiwata T, Zhou B, Camargo FD et al (2012) YAP1, the nuclear target Magnuson TR, O’Neal W, Milgram SL (2006) Defects in yolk sac vascu- stream of LIF. J Cell Sci 124: 1136 – 1144 Genes Dev 25: 51 – 63 of Hippo signaling, stimulates heart growth through cardiomyocyte logenesis, chorioallantoic fusion, and embryonic axis elongation in 87. Lian I, Kim J, Okazawa H, Zhao J, Zhao B, Yu J, Chinnaiyan A, Israel MA, 105. Ma DK, Bonaguidi MA, Ming GL, Song H (2009) Adult neural stem cells proliferation but not hypertrophy. Proc Natl Acad Sci U S A 109: mice with targeted disruption of Yap65. Mol Cell Biol 26: 77 – 87 Goldstein LS, Abujarour R et al (2010) The role of YAP transcription in the mammalian central nervous system. Cell Res 19: 672 – 682 2394 – 2399 71. Yagi R, Kohn MJ, Karavanova I, Kaneko KJ, Vullhorst D, DePamphilis ML, coactivator in regulating stem cell self-renewal and differentiation. 106. Cao X, Pfaff SL, Gage FH (2008) YAP regulates neural progenitor cell 122 . Judson RN, Tremblay AM, Knopp P, White RB, Urcia R, De Bari C, Buonanno A (2007) Transcription factor TEAD4 specifies the trophecto- Genes Dev 24: 1106 – 1118 number via the TEA domain transcription factor. Genes Dev 22: Zammit PS, Camargo FD, Wackerhage H (2012) The Hippo pathway derm lineage at the beginning of mammalian development. Develop- 88. Takahashi K, Yamanaka S (2006) Induction of pluripotent stem cells 3320 – 3334 member Yap plays a key role in influencing fate decisions in muscle ment 134: 3827 – 3836 from mouse embryonic and adult fibroblast cultures by defined factors. 107. Gee ST, Milgram SL, Kramer KL, Conlon FL, Moody SA (2011) Yes-associ- satellite cells. J Cell Sci 125: 6009 – 6019 72. Chen Z, Friedrich GA, Soriano P (1994) Transcriptional enhancer factor Cell 126: 663 – 676 ated protein 65 (YAP) expands neural progenitors and regulates Pax3 123. Watt KI, Judson R, Medlow P, Reid K, Kurth TB, Burniston JG, Ratkevi- 1 disruption by a retroviral gene trap leads to heart defects and 89. Murakami M, Tominaga J, Makita R, Uchijima Y, Kurihara Y, Nakagawa expression in the neural plate border zone. PLoS ONE 6:e20309 cius A, De Bari C, Wackerhage H (2010) Yap is a novel regulator of embryonic lethality in mice. Genes Dev 8: 2293 – 2301 O, Asano T, Kurihara H (2006) Transcriptional activity of Pax3 is co- 108. Milewski RC, Chi NC, Li J, Brown C, Lu MM, Epstein JA (2004) Identifi- C2C12 myogenesis. Biochem Biophys Res Commun 393: 619 – 624 73. Sawada A, Kiyonari H, Ukita K, Nishioka N, Imuta Y, Sasaki H (2008) activated by TAZ. Biochem Biophys Res Commun 339: 533 – 539 cation of minimal enhancer elements sufficient for Pax3 expression in 124. Jeong H, Bae S, An SY, Byun MR, Hwang JH, Yaffe MB, Hong JH, Hwang Redundant roles of Tead1 and Tead2 in notochord development and 90. Michalopoulos GK (2007) Liver regeneration. J Cell Physiol 213: 286 – 300 neural crest and implication of Tead2 as a regulator of Pax3. Develop- ES (2010) TAZ as a novel enhancer of MyoD-mediated myogenic differ- the regulation of cell proliferation and survival. Mol Cell Biol 28: 91. Halder G, Johnson RL (2011) Hippo signaling: growth control and ment 131: 829 – 837 entiation. FASEB J 24: 3310 – 3320 3177 – 3189 beyond. Development 138: 9 – 22 109. Fernandez LA, Northcott PA, Dalton J, Fraga C, Ellison D, Angers S, 125. Barry ER, Camargo FD (2013) The Hippo superhighway: signaling cross- 74. Nishioka N, Yamamoto S, Kiyonari H, Sato H, Sawada A, Ota M, Nakao 92. Camargo FD, Gokhale S, Johnnidis JB, Fu D, Bell GW, Jaenisch R, Taylor MD, Kenney AM (2009) YAP1 is amplified and up-regulated in roads converging on the Hippo/Yap pathway in stem cells and develop- K, Sasaki H (2008) Tead4 is required for specification of trophectoderm Brummelkamp TR (2007) YAP1 increases organ size and expands undif- hedgehog-associated medulloblastomas and mediates Sonic ment. Curr Opin Cell Biol 25: 247 – 253 in pre-implantation mouse embryos. Mech Dev 125: 270 – 283 ferentiated progenitor cells. Curr Biol 17: 2054 – 2060 hedgehog-driven neural precursor proliferation. Genes Dev 23: 126. Barker N (2014) Adult intestinal stem cells: critical drivers of epithelial 75. Ralston A, Cox BJ, Nishioka N, Sasaki H, Chea E, Rugg-Gunn P, Guo G, 93. Liu-Chittenden Y, Huang B, Shim JS, Chen Q, Lee SJ, Anders RA, Liu JO, 2729 – 2741 homeostasis and regeneration. Nat Rev Mol Cell Biol 15: 19 – 33 Robson P, Draper JS, Rossant J (2010) Gata3 regulates trophoblast Pan D (2012) Genetic and pharmacological disruption of the TEAD-YAP 110. Kagey JD, Brown JA, Moberg KH (2012) Regulation of Yorkie activity in 127. Johnson R, Halder G (2014) The two faces of Hippo: targeting the development downstream of Tead4 and in parallel to Cdx2. Develop- complex suppresses the oncogenic activity of YAP. Genes Dev 26: Drosophila imaginal discs by the Hedgehog receptor gene patched. Hippo pathway for regenerative medicine and cancer treatment. Nat ment 137: 395 – 403 1300 – 1305 Mech Dev 129: 339 – 349 Rev Drug Discov 13: 63 – 79

654 EMBO reports Vol 15 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO reports Vol 15 | No 6 | 2014 655 EMBO reports Hippo pathway in stem cells and cancer Jung-Soon Mo et al Jung-Soon Mo et al Hippo pathway in stem cells and cancer EMBO reports

654 EMBO reports Vol 15 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO reports Vol 15 | No 6 | 2014 655 EMBO reports Hippo pathway in stem cells and cancer Jung-Soon Mo et al

128. Cai J, Zhang N, Zheng Y, de Wilde RF, Maitra A, Pan D (2010) The 142. Murakami H, Mizuno T, Taniguchi T, Fujii M, Ishiguro F, Fukui T, Akat- Hippo signaling pathway restricts the oncogenic potential of an intes- suka S, Horio Y, Hida T, Kondo Y et al (2011) LATS2 is a tumor suppres- tinal regeneration program. Genes Dev 24: 2383 – 2388 sor gene of malignant mesothelioma. Cancer Res 71: 873 – 883 129. Zhou D, Zhang Y, Wu H, Barry E, Yin Y, Lawrence E, Dawson D, Willis 143. St John MA, Tao W, Fei X, Fukumoto R, Carcangiu ML, Brownstein DG, JE, Markowitz SD, Camargo FD et al (2011) Mst1 and Mst2 protein Parlow AF, McGrath J, Xu T (1999) Mice deficient of Lats1 develop kinases restrain intestinal stem cell proliferation and colonic tumori- soft-tissue sarcomas, ovarian tumours and pituitary dysfunction. Nat genesis by inhibition of Yes-associated protein (Yap) overabundance. Genet 21: 182 – 186 Proc Natl Acad Sci U S A 108:E1312 – E1320 144. Takahashi Y, Miyoshi Y, Takahata C, Irahara N, Taguchi T, Tamaki Y, 130. Barry ER, Morikawa T, Butler BL, Shrestha K, de la Rosa R, Yan KS, Noguchi S (2005) Down-regulation of LATS1 and LATS2 mRNA expres- Fuchs CS, Magness ST, Smits R, Ogino S et al (2013) Restriction of sion by promoter hypermethylation and its association with biologi- intestinal stem cell expansion and the regenerative response by YAP. cally aggressive phenotype in human breast cancers. Clin Cancer Res Nature 493: 106 – 110 11: 1380 – 1385 131. Karpowicz P, Perez J, Perrimon N (2010) The Hippo tumor suppressor 145. Jiang Z, Li X, Hu J, Zhou W, Jiang Y, Li G, Lu D (2006) Promoter hyper- pathway regulates intestinal stem cell regeneration. Development 137: methylation-mediated down-regulation of LATS1 and LATS2 in human 4135 – 4145 astrocytoma. Neurosci Res 56: 450 – 458 132. Staley BK, Irvine KD (2010) Warts and Yorkie mediate intestinal regen- 146. Seidel C, Schagdarsurengin U, Blumke K, Wurl P, Pfeifer GP, Haupt- eration by influencing stem cell proliferation. Curr Biol 20: 1580 – 1587 mann S, Taubert H, Dammann R (2007) Frequent hypermethylation of 133. Ren F, Wang B, Yue T, Yun EY, Ip YT, Jiang J (2010) Hippo signaling MST1 and MST2 in soft tissue sarcoma. Mol Carcinog 46: 865 – 871 regulates Drosophila intestine stem cell proliferation through multiple 147. Evans DG (2009) Neurofibromatosis 2 [Bilateral acoustic neurofibroma- pathways. Proc Natl Acad Sci U S A 107: 21064 – 21069 tosis, central neurofibromatosis, NF2, neurofibromatosis type II]. Genet 134. Pan D (2010) The hippo signaling pathway in development and cancer. Med 11: 599 – 610 Dev Cell 19: 491 – 505 148. Thurneysen C, Opitz I, Kurtz S, Weder W, Stahel RA, Felley-Bosco E 135. Zhao B, Lei QY, Guan KL (2008) The Hippo-YAP pathway: new connec- (2009) Functional inactivation of NF2/merlin in human mesothelioma. tions between regulation of organ size and cancer. Curr Opin Cell Biol Lung Cancer 64: 140 – 147 20: 638 – 646 149. Bianchi AB, Mitsunaga SI, Cheng JQ, Klein WM, Jhanwar SC, Seizinger 136. Zender L, Spector MS, Xue W, Flemming P, Cordon-Cardo C, Silke J, Fan B, Kley N, Klein-Szanto AJ, Testa JR (1995) High frequency of inactivat- ST, Luk JM, Wigler M, Hannon GJ et al (2006) Identification and valida- ing mutations in the neurofibromatosis type 2 gene (NF2) in primary tion of oncogenes in liver cancer using an integrative oncogenomic malignant mesotheliomas. Proc Natl Acad Sci U S A 92: 10854 – 10858 approach. Cell 125: 1253 – 1267 150. Tanas MR, Sboner A, Oliveira AM, Erickson-Johnson MR, Hespelt J, 137. Steinhardt AA, Gayyed MF, Klein AP, Dong J, Maitra A, Pan D, Mont- Hanwright PJ, Flanagan J, Luo Y, Fenwick K, Natrajan R et al (2011) gomery EA, Anders RA (2008) Expression of Yes-associated protein in Identification of a disease-defining gene fusion in epithelioid heman- common solid tumors. Hum Pathol 39: 1582 – 1589 gioendothelioma. Sci Transl Med 3: 98ra82 138. Chan SW, Lim CJ, Guo K, Ng CP, Lee I, Hunziker W, Zeng Q, Hong W 151. Errani C, Zhang L, Sung YS, Hajdu M, Singer S, Maki RG, Healey JH, (2008) A role for TAZ in migration, invasion, and tumorigenesis of Antonescu CR (2011) A novel WWTR1-CAMTA1 gene fusion is a consis- breast cancer cells. Cancer Res 68: 2592 – 2598 tent abnormality in epithelioid hemangioendothelioma of different 139. Cordenonsi M, Zanconato F, Azzolin L, Forcato M, Rosato A, Frasson C, anatomic sites. Genes Chromosom Cancer 50: 644 – 653 Inui M, Montagner M, Parenti AR, Poletti A et al (2011) The Hippo 152. Onken MD, Worley LA, Long MD, Duan S, Council ML, Bowcock AM, transducer TAZ confers cancer stem cell-related traits on breast cancer Harbour JW (2008) Oncogenic mutations in GNAQ occur early in uveal cells. Cell 147: 759 – 772 melanoma. Invest Ophthalmol Vis Sci 49: 5230 – 5234 140. Bhat KP, Salazar KL, Balasubramaniyan V, Wani K, Heathcock L, 153. Prickett TD, Wei X, Cardenas-Navia I, Teer JK, Lin JC, Walia V, Gartner J, Hollingsworth F, James JD, Gumin J, Diefes KL, Kim SH et al (2011) The Jiang J, Cherukuri PF, Molinolo A et al (2011) Exon capture analysis of transcriptional coactivator TAZ regulates mesenchymal differentiation G protein-coupled receptors identifies activating mutations in GRM3 in in malignant glioma. Genes Dev 25: 2594 – 2609 melanoma. Nat Genet 43: 1119 – 1126 141. Overholtzer M, Zhang J, Smolen GA, Muir B, Li W, Sgroi DC, Deng CX, 154. Kan Z, Jaiswal BS, Stinson J, Janakiraman V, Bhatt D, Stern HM, Yue P, Brugge JS, Haber DA (2006) Transforming properties of YAP, a candi- Haverty PM, Bourgon R, Zheng J et al (2010) Diverse somatic mutation date oncogene on the chromosome 11q22 amplicon. Proc Natl Acad Sci patterns and pathway alterations in human cancers. Nature 466: USA103: 12405 – 12410 869 – 873

656 EMBO reports Vol 15 | No 6 | 2014 ª 2014 The Authors Scientific Report

Heterotrimeric G proteins control stem cell proliferation through CLAVATA signaling in Arabidopsis

Takashi Ishida1,†, Ryo Tabata1,†, Masashi Yamada2,3,†, Mitsuhiro Aida4, Kanako Mitsumasu1, Masayuki Fujiwara4, Katsushi Yamaguchi5, Shuji Shigenobu5, Masayuki Higuchi4, Hiroyuki Tsuji4, Ko Shimamoto4, Mitsuyasu Hasebe6,7, Hiroo Fukuda2 & Shinichiro Sawa1,*

Abstract hormone and the extracellular leucine-rich repeat (LRR) domain- containing receptors CLV1, CLV2-CORYNE (CRN)/SUPPRESSOR OF Cell-to-cell communication is a fundamental mechanism for coor- LLP1 2 (SOL2), and RECEPTOR-LIKE PROTEIN KINASE 2 (RPK2) dinating developmental and physiological events in multicellular [1,2]. In the shoot apical meristem (SAM) of Arabidopsis thaliana organisms. Heterotrimeric G proteins are key molecules that trans- (Arabidopsis), CLV signaling restricts the expression of the homeobox- mit extracellular signals; similarly, CLAVATA signaling is a crucial containing transcription factor WUSCHEL (WUS) [3–5]. Conversely, regulator in plant development. Here, we show that Arabidopsis WUS promotes the expression of CLV3, forming a negative feedback thaliana Gb mutants exhibit an enlarged stem cell region, which is loop that controls the number of stem cells [3–5]. While the similar to that of clavata mutants. Our genetic and cell biological peptide-binding plasma membrane components have been well analyses suggest that the G protein beta-subunit1 AGB1 and RPK2, studied, the molecules that mediate intracellular signaling by these one of the major CLV3 peptide hormone receptors, work synergisti- receptors are largely unknown. The protein phosphatase KAPP and cally in stem cell homeostasis through their physical interactions. a Rho GTPase-related protein have been shown to physically inter- We propose that AGB1 and RPK2 compose a signaling module to act with the CLV1 receptor [6], and the protein phosphatase 2Cs facilitate meristem development. POLTERGEIST (POL) and POL-LIKE 1 (PLL1) are also known to be signaling mediators [7]. However, further analyses are needed to Keywords Arabidopsis thaliana; heterotrimeric G protein; peptide hormone; trace the signaling pathway from the receptor to cellular processes. RECEPTOR-LIKE PROTEIN KINASE 2; stem cell homeostasis On the other hand, heterotrimeric G proteins, composed of alpha Subject Categories Plant Biology; Stem Cells (Ga), beta (Gb), and gamma (Gc) subunits, are important signaling DOI 10.15252/embr.201438660 | Received 21 February 2014 | Revised 13 molecules that link extracellular signals to intracellular mechanisms August 2014 | Accepted 14 August 2014 | Published online 26 September 2014 in eukaryotes [8,9]. The basic components and mechanisms of G EMBO Reports (2014) 15: 1202–1209 protein signaling have been studied extensively in mammalian cells: G protein-coupled receptors (GPCRs) sense extracellular ligands and stimulate G proteins, whereupon Ga and Gbc dissociate and Introduction provoke variable cellular events [8]. In resting cells, GDP-bound Ga associates with Gbc, and ligand-stimulated GPCRs promote the Coordinated cell proliferation and cell differentiation are essential exchange of GDP for GTP, causing Ga and Gbc to dissociate. processes in multicellular organisms. To achieve these functions, Although land plants express similar G protein components, they organisms have developed scrupulously designed cell-to-cell are controlled by slightly different systems compared with canonical communication systems over the course of evolution. Plants have G proteins. In plants, Ga can spontaneously exchange GDP for GTP, established unique ligand-receptor-based signaling modules, such as while the seven-pass transmembrane domain-containing protein the CLAVATA (CLV) pathway, which comprises the CLV3 peptide RGS1 inhibits G signaling through the formation of an inactive

1 Graduate School of Science and Technology, Kumamoto University, Kumamoto, Japan 2 Department of Biological Sciences, Graduate School of Science, The University of Tokyo, Tokyo, Japan 3 Department of Biology and Institute for Genome Science and Policy Center for Systems Biology, Duke University, Durham, NC, USA 4 Graduate School of Biological Sciences, Nara Institute of Science and Technology, Ikoma, Japan 5 Functional Genomics Facility, National Institute for Basic Biology, Okazaki, Japan 6 Division of Evolutionary Biology, National Institute for Basic Biology, Okazaki, Japan 7 School of Life Science, The Graduate University for Advanced Studies, Okazaki, Japan *Corresponding author. Tel: +81 96 342 3439; E-mail: [email protected] †These authors are contributed equally to this work

1202 EMBO reports Vol 15 | No 11 | 2014 ª 2014 The Authors Takashi Ishida et al Gbeta and CLAVATA control stem cells in plants EMBO reports EMBO reports Gbeta and CLAVATA control stem cells in plants Takashi Ishida et al

complex [10–12]. Extracellular ligands bound to RGS1 stimulate the The nonsense mutation in the AGB1 gene is expected to be G A WT B clv2 C clv2 clen1 90 release of Ga to activate signaling. This self-activating ability helps responsible for the observed clen1 phenotypes. We therefore exam- N.S. explain the absence of clear GPCR homologues in plant genomes ined SAM size and carpel number in the previously isolated agb1-2 80 * * * * [13]. Several transmembrane proteins have been annotated as plant and clv2 agb1-2 mutants [17]. agb1-2 produced 1.4-fold larger SAMs 70 * GPCRs based on their sequence; however, clear evidence that these than wild-type plants, while the carpel number was still 2 (Fig 1D, m) 60 candidates function as GPCRs has not been reported [9,13]. These G, K and N), suggesting that the single mutation in AGB1 is suffi- facts provide possibilities for the mode of G signaling which is cient to affect SAM height. Moreover, the clv2 agb1-2 double mutant 50 the presence of alternative, non-canonical GPCRs and GPCR- exhibited a clv2 clen1-like phenotype, showing 6.5-fold larger SAMs 40 independent function of G proteins. Despite their unique regulatory and 1.9 times the number of carpels compared with wild-type plants 30 * mechanisms, plant G proteins are involved in various aspects (Fig 1E, G, L and N). Although overexpression of AGB1 did not 20 of morphological and physiological processes, much like their affect plant architecture in the wild type (Supplementary Fig S2), it D agb1-2 E clv2 agb1-2 F clv2 clen1 SAM height ( µ 10 mammalian counterparts [12,14–18]. suppressed the enhanced abnormalities of the clv2 clen1 mutant, + 35S:AGB1 Recently, Bommert et al [19] reported genetic evidence that resulting in a clv2-like phenotype (Fig 1F, G, M and N). These plants 0 maize Ga modulates CLV signaling in the control of shoot meristem also resembled a clv2 clen1 mutant that harbors a genomic fragment WT clv2 clv2 agb1-2 clv2 clv2 size. However, the biochemical and cell biological processes under- of the AGB1 gene (Supplementary Fig S3). These results show that a clen1 agb1-2 clen1 lying the cross-talk between the CLV pathway and G proteins mutation in AGB1 enhances the abnormalities of the clv2 mutant N + 35S:AGB1 6 remain unclear, as these extraordinary phenotypes have only been and suggest that AGB1 regulates SAM height and carpel number. N.S. reported for Ga mutants in maize [19]. Here, we show that the N.S. 5 * Arabidopsis G protein beta-subunit1 (agb1) mutant exhibits an AGB1 is involved in SAM maintenance through the CLV3 * * enlarged SAM, similar to that of clv mutants. Genetic analysis signaling pathway * 4 suggests that AGB1 works together with RPK2, a leucine-rich HI J KL M repeat-receptor-like kinase (LRR-RLK), in stem cell homeostasis. CLV3 restricts cell proliferation in the SAM, and synthetic CLV3 * 3 Bimolecular fluorescence complementation (BiFC) assays and peptide treatment induces SAM consumption due to diminished cell co-immunoprecipitation (co-IP) analyses indicate that AGB1 associ- proliferation [5,21]. To investigate whether the enlarged SAM 2 ates with RPK2. These results establish the involvement of AGB1 in phenotype observed in agb1 is a consequence of a disturbance of carpel number carpel meristem development in the RPK2-dependent signaling pathway CLV3 signaling, we examined the sensitivity of the agb1-2 mutant to 1 and indicate the diversity of CLV signaling in plants. the CLV3 peptide (Fig 2). Wild-type seedlings grown on MS media containing 5 lM CLV3 did not develop stems under these condi- 0 tions, even at 20 days after germination (Fig 2E, M and Q). Conver- WT clv2 clv2 agb1-2 clv2 clv2 Results and discussion sely, 10% of agb1-2 mutants developed a stem at the same stage WT clv2 clv2 agb1-2 clv2 clv2 clen1 clen1 agb1-2 clen1 clen1 agb1-2 + 35S:AGB1 (Fig 2G, O and, Q). Furthermore, clv2 agb1-2 double mutants + 35S:AGB1 Identification of mutations in a gene encoding a heterotrimeric G showed strong resistance compared with clv2 or agb1-2 mutants protein b subunit in clv2 enhancer 1 mutants (Fig 2H, P and Q). Figure 1. The G protein b-subunit regulates SAM maintenance. Next, we examined the genetic relationship between AGB1 and A–F DIC images of SAMs. Wild-type (A), clv2 (B), clv2 clen1 (C), agb1-2 (D), clv2 agb1-2 (E), and clv2 clen1 plants transformed with 35S:AGB1 (F) are shown. Brackets indicate height of SAM. To decipher the molecular mechanisms underlying the CLV signal- WUS, which is known to function downstream of CLV signaling G Quantitative analysis of SAM height. ing pathway, we conducted a genetic screen to search for mutations [4,5]. Similar to the wus-101 single mutant, the SAM was terminated H–M Dissecting microscope images of mature siliques. Wild-type (H), clv2 (I), clv2 clen1 (J), agb1-2 (K), clv2 agb1-2 (L), and clv2 clen1 plants transformed with 35S:AGB1 that enhance the phenotypes of clv2 mutants. As a result, we in the wus-101 agb1-2 double mutant (Supplementary Fig S6), indi- (M) are shown. isolated 48 mutants with obviously enlarged SAMs, which have cating that WUS is epistatic to AGB1. Taken together, these findings N Quantitative analysis of carpel number. been designated clv2 enhancer (clen) mutants. From these mutants, suggest that AGB1 regulates the SAM activities through a CLV3- Data information: Scale bars = 25 lm in (A–F), 1 mm in (H–M). Error bars in (G) and (N) represent SD. The histograms and complete data are shown in Supplementary clv2 enhancer 1 (clen1) was selected for further study. clv2-101,a related pathway. Figs S4 and S5 and Supplementary Table S1. An asterisk indicates a statistically significant difference from the neighboring value (*P < 0.05). N.S. indicates not significant. null allele of clv2 mutation, displays approximately twofold (21.25 lm) larger SAMs than wild-type plants, in which SAM height Heterotrimeric G proteins are expressed in the is 10.18 lm on average, and the clv2 clen1 double mutant exhibits inflorescence meristem sevenfold (71.73 lm) larger SAMs (Fig 1A–C and G). Similarly, the The heterotrimeric G protein c subunit, but not Ga, is also and 10% of the Gc mutants maintained a SAM even on CLV3- pistils of wild-type flowers have 2 carpels, whereas clv2-101 and Given that AGB1 is a heterotrimeric G protein subunit, the involve- involved in CLV3 signaling containing media (Supplementary Table S1, Supplementary Fig double-mutant plants present 2.5 and 3.8 carpels on average, ment of other G protein components was predicted. To examine S8A), similar to what was observed in the agb1-2 mutants. Further- respectively (Fig 1H–J and N). The enlarged SAMs and increased whether G proteins are expressed in SAMs, we performed in situ To investigate the possibility that the Ga and Gc subunits are more, clv2 agg1-1c agg2-1 triple mutants showed enhanced abnor- carpel numbers observed in clv2 clen1 mutants relative to wild-type mRNA hybridization experiments. Expression of both GPA1 and involved in CLV signaling, we examined both SAM height and CLV3 malities and SAMs were maintained in the triple mutant at the or clv2 mutant plants suggest decreased CLV3 signaling activity [3]. AGB1 was observed in the inflorescence meristem, floral meristem, peptide sensitivity in Ga and Gc null mutants, designated gpa1-4 similar frequency as in the clv2 agb1 double mutants when treated Using a positional cloning approach, the clen1 mutation was and floral organ primordium (Supplementary Fig S7A and B). and agg1-1c agg2-1, respectively [22,23]. A recent report showed with CLV3 (Supplementary Table S1, Supplementary Fig S8A). roughly mapped to near the nga1139 marker (33/34 chromosomes) Conversely, weak AGG1 expression signals were detected, whereas that a maize Ga mutant exhibited a very large meristem phenotype In mammals and plants, Gb and Gc are known to form a hetero- on chromosome 4, and the genomic DNA sequence was analyzed AGG2 was not (Supplementary Fig S7C and D). The expression of [19]. In contrast, gpa1-4 plants did not show obvious SAM or carpel dimer [8,9,24]. Arabidopsis Gc appears to act with Gb during CLV via the SOLiD system to identify mutations [20]. Thus, we detected these genes in vegetative SAMs and inflorescences was supported by abnormalities, and the mutation did not affect the degree of resis- signaling to regulate SAM height and carpel number. This idea is a nucleotide substitution in the AGB1 gene that converts a Trp resi- the Arabidopsis eFP Browser database (Supplementary Fig S7E–G). tance to CLV3 compared with the wild type (Supplementary Table also supported by the results of the examination of carpel number due into a stop codon (Supplementary Fig S1). In addition to the Thus, the fact that not only AGB1 but also other G protein compo- S1, Supplementary Fig S8A). Furthermore, the additional mutation phenotypes at higher temperatures. Morphological abnormalities in clen1 mutant, we have identified this mutation in 8 clen mutants nents were expressed in SAMs highlighted the possibility that the G of GPA1 did not affect the clv2 mutant phenotype (Supplementary flowers are occasionally strengthened at higher temperatures [25]. and have found other four point mutations in five additional clen protein signaling complex is involved in the CLV signaling path- Table S1). Conversely, the agg1-1c agg2-1 double mutant produced Accordingly, the agb1-2 and agg1-1c agg2-1 mutants both exhibited mutants (Supplementary Fig S1). ways. 1.4-fold larger SAMs than the wild type as well as 2-carpel siliques, 3-carpel pistils, whereas wild-type and gpa1-4 plants all presented 2

ª 2014 The Authors EMBO reports Vol 15 | No 11 | 2014 1203 1204 EMBO reports Vol 15 | No 11 | 2014 ª 2014 The Authors Takashi Ishida et al Gbeta and CLAVATA control stem cells in plants EMBO reports EMBO reports Gbeta and CLAVATA control stem cells in plants Takashi Ishida et al

WT clv2 agb1-2 clv2 agb1-2 carpels (Supplementary Fig S8B). The difference between the Ga We next tested whether AGB1 associates with CLV1, RPK2, and mutant phenotypes and those of the Gb and Gc mutants is not unex- CLV2 using BiFC assays. Protoplasts transformed with AGB1 and ABCD pected, as Ga mutations are often reported as the exception among RPK2 exhibited a positive BiFC signal when a CLV3-expressing G protein mutants [13]. vector was co-transformed, and the signal was localized to the plasma membrane (Fig 4A and C). However, we did not detect an Gbc controls SAM maintenance in the RPK2 pathway interaction between AGB1 and either CLV1 or CLV2 (Fig 4A and C). AGB1 is therefore predicted to receive CLV signals through RPK2, Mock The fact that the clv2-101 agb1-2 double mutant showed detectable though it is unclear how CLV3 facilitates this interaction. Further- additive phenotypes suggests that AGB1 mediates CLV3 signaling in more, we performed a co-IP assay to confirm the physical interac- a CLV2-independent manner. In contrast to clv2 agb1-2 mutants, tion between AGB1 and RPK2. FLAG-tagged AGB1 was pulled down which exhibit 3.1-fold larger SAMs than clv2 mutants, the clv1 agb1-2 with Venus-tagged RPK2c, which contains the C-terminal intracellu- EFGH and rpk2 agb1-2 mutants exhibited 2-fold and 1.5-fold larger SAMs lar domain, whereas most of the AGB1-FLAG disappeared after IP than the corresponding single mutants (Fig 3A). Furthermore, when expressed alone or with mCherry-Venus (Fig 4D). These µ M comparing rpk2 and rpk2 agb1-2 plants revealed similar carpel results suggest that RPK2 is capable of interacting with AGB1. numbers, whereas the clv2 agb1-2 and clv1 agb1-2 mutants showed a Taking these results together with the genetic data, we propose that significantly increased carpel number relative to the single mutants AGB1 functions preferentially with RPK2 on the CLV signaling to (Fig 3B). All of the double mutants showed a significantly enhanced regulate cell proliferation activities in SAMs. phenotype with the corresponding single mutants. However, the + CLV3 5 + CLV3 degree of enhancement was smaller in rpk2 agb1 for SAM height and only the rpk2 agb1 produced a similar number of carpels when Conclusion compared with the rpk2 single mutant. These results lead us to I JKL hypothesize that AGB1 is at least partially involved in the RPK2- Heterotrimeric G proteins are evolutionarily conserved signaling dependent CLV signaling pathway. molecules that mediate the transduction of extracellular cues into intracellular signals, in combination with transmembrane GPCRs. AGB1 associates with RPK2 in planta In plants, several transmembrane proteins have been reported as GPCRs. In this study, we have shown that an LRR-RLK receptor, Mock Based on the results of the genetic analyses, AGB1 is expected to RPK2, is able to interact with G proteins. Surprisingly, among the interact with CLV signaling components either directly or indirectly. examined G protein mutants, only the Ga mutant did not exhibit AGB1 has been observed to localize to the plasma membrane, any abnormalities in CLV signaling-related processes, suggesting nucleus, cytoplasm, and Golgi apparatus [24,26–28]. Our expression that mutations in GPA1 did not disrupt CLV signaling. However, MNOP analysis confirmed the presence of AGB1-GFP signals at the plasma we cannot exclude the possibility that Ga or related proteins serve membrane (Supplementary Fig S9). Because LRR-containing recep- as a bridge between RPK2 and Gbc dimers, as in canonical GPCR µ M tor complexes also localize to the plasma membrane, any complex and G protein interactions. The Arabidopsis genome encodes three that these proteins form is likely to be present here. extra-large G proteins (XLGs), which contain a Ga-like domain

AB 80 5 + CLV3 5 + CLV3 70 * 4 60 Q 100% * 50 3 * 80% N.S. 40 * N.S. 60% 30 * 2 20 40% 1 * number carpel

SAM height ( µ m) 10

percentage 20% Terminated SAM 0 0 Normal 0% WT clv2 agb1-2 clv2 WT clv2 agb1-2 clv2 agb1-2 agb1-2 MS MS + CLV3 5µM Figure 3. Genetic relationship between the AGB1 and CLV3 receptors. A, B Quantitative analyses of SAM height (A) and carpel number (B) are shown. White bars correspond to the wild-type plants or single mutants for the indicated Figure 2. AGB1 is involved in SAM maintenance through CLV3 signaling. receptors, whereas gray bars represent agb1-2 single mutants or double mutants for AGB1 and the indicated receptors. Error bars represent SD. Note that the A–P Eighteen-day-old seedlings of wild-type (A, E, I, and M), clv2 (B, F, J, and N), agb1-2 (C, G, K, and O), and clv2 agb1-2 (D, H, L, and P) plants. The plants were grown carpel number observed in all wild-type and agb1-2 plants was 2, and the corresponding SD was therefore 0. An asterisk indicates a statistically significant on agar medium with (E–F, M–P) or without (A–D, I–L) 5 lM CLV3 peptide. (I–P) represent closer views of (A–H). Scale bars = 1 cm in (A–H), 1 mm in (I–P). difference from the neighboring value. N.S. indicates not significant (*P < 0.05). The histograms and complete data are shown in Supplementary Figs S4 and S5 and Q Quantification of the seedlings showing terminated SAMs observed 20 days after germination. Supplementary Table S1.

ª 2014 The Authors EMBO reports Vol 15 | No 11 | 2014 1205 1206 EMBO reports Vol 15 | No 11 | 2014 ª 2014 The Authors Takashi Ishida et al Gbeta and CLAVATA control stem cells in plants EMBO reports EMBO reports Gbeta and CLAVATA control stem cells in plants Takashi Ishida et al

AB 80 5 + CLV3 5 + CLV3 70 * 4 60 Q 100% * 50 3 * 80% N.S. 40 * N.S. 60% 30 * 2 20 40% 1 * number carpel

SAM height ( µ m) 10 percentage 20% Terminated SAM 0 0 Normal 0% WT clv2 agb1-2 clv2 WT clv2 agb1-2 clv2 agb1-2 agb1-2 MS MS + CLV3 5µM Figure 3. Genetic relationship between the AGB1 and CLV3 receptors. A, B Quantitative analyses of SAM height (A) and carpel number (B) are shown. White bars correspond to the wild-type plants or single mutants for the indicated Figure 2. AGB1 is involved in SAM maintenance through CLV3 signaling. receptors, whereas gray bars represent agb1-2 single mutants or double mutants for AGB1 and the indicated receptors. Error bars represent SD. Note that the A–P Eighteen-day-old seedlings of wild-type (A, E, I, and M), clv2 (B, F, J, and N), agb1-2 (C, G, K, and O), and clv2 agb1-2 (D, H, L, and P) plants. The plants were grown carpel number observed in all wild-type and agb1-2 plants was 2, and the corresponding SD was therefore 0. An asterisk indicates a statistically significant on agar medium with (E–F, M–P) or without (A–D, I–L) 5 lM CLV3 peptide. (I–P) represent closer views of (A–H). Scale bars = 1 cm in (A–H), 1 mm in (I–P). difference from the neighboring value. N.S. indicates not significant (*P < 0.05). The histograms and complete data are shown in Supplementary Figs S4 and S5 and Q Quantification of the seedlings showing terminated SAMs observed 20 days after germination. Supplementary Table S1.

Arabidopsis and maize utilize common G proteins for meristem 23012034, 24114001, 24114009, 24370024, 24657035, and 24658032 to SS; maintenance, different systems consisting of various combinations 25119713 and 25440134 to TI) and the NIBB cooperative research program of receptors and G protein subunits are employed. Therefore, these (12-103) to SS. differences might contribute to the diversity of the signaling pathways in plant development. Author contributions MH, HT, KS, MH, HF, and SS conceived or designed the experiments. TI, RT, MY, MA, KM, MF, KY, and SS performed the experiments. KY, SS, and MH analyzed Materials and Methods the data. TI, RT, MY, MA, and SS wrote the manuscript. KS was deceased on September 28, 2013. Plant materials Conflict of interest The following Arabidopsis wild-type and mutant lines were The authors declare that they have no conflict of interest. obtained: wild-type Columbia-0 (Col-0); clv1-101 (CS858348) in the Col-2 background; clv3-8 ER in an unknown background (CS3604) [32]; rpk2-2 [5], clv2-101 (GK686A09), gpa1-4 (SALK_001846), References agb1-2 (CS6536), and wus-101 (GK870H12) in the Col-0 background; and agg1-1c agg2-1 (kindly provided by Jimmy Botella) in a 1. Betsuyaku S, Sawa S, Yamada M (2011) The function of the CLE peptides mix of Col-0 and Wassilewskija (Ws). in plant development and plant-microbe interactions. Arabidopsis Book 9:e0149 SAM measurement 2. Yamada M, Sawa S (2013) The roles of peptide hormones during plant root development. Curr Opin Plant Biol 16: 56 – 61 Seven-day-old seedlings were fixed with 70% ethanol, cleared in a 3. Clark SE (1995) CLAVATA3 is a specific regulator of shoot and floral meri- mixture of chloral hydrate, glycerol, and water (8:1:2; w/v/v), and stem development affecting the same processes as CLAVATA1. Develop- observed using a microscope (ZEISS AXIO Imager M1) that was ment 121: 2057 equipped with Nomarski optics. The base of the SAM was defined 4. Schoof H, Lenhard M, Haecker A, Mayer KF, Jurgens G, Laux T (2000) as the location of the leaf primordium, and the height was measured The stem cell population of Arabidopsis shoot meristems in maintained between the top and base of the SAM, as described [5]. by a regulatory loop between the CLAVATA and WUSCHEL genes. Cell 100: 635 – 644 Peptide assay 5. Kinoshita A, Betsuyaku S, Osakabe Y, Mizuno S, Nagawa S, Stahl Y, Simon R, Yamaguchi-Shinozaki K, Fukuda H, Sawa S (2010) RPK2 is an The CLV3 peptide was synthesized as described previously [33]. essential receptor-like kinase that transmits the CLV3 signal in Arabidop- Seedlings were grown on MS plates containing CLV3 peptides until sis. Development 137: 3911 – 3920 18–20 days after germination. 6. Trotochaud AE, Hao T, Wu G, Yang Z, Clark SE (1999) The CLAVATA1 Figure 4. AGB1 is able to interact with RPK2. receptor-like kinase requires CLAVATA3 for its assembly into a signaling A Protoplasts expressing the indicated proteins tagged with the N- or C-terminal halves of Venus. Protoplast transformation complex that includes KAPP and a Rho-related protein. Plant Cell 11: B Positive and negative controls for the BiFC analysis. BiFC signals (upper), mCherry fluorescence (middle), and merged (bottom) images are shown. 393 – 406 C Quantification of the BiFC assays. The results for the positive control and the experiments for CLV1, CLV2, RPK2, and AGB1 are shown. BiFC signals were measured as described in Materials and Methods. The percentages of cells with BiFC signals are indicated by yellow bars (n = 20). Arabidopsis leaf mesophyll protoplast transformation was 7. Song SK, Lee MM, Clark SE (2006) POL and PLL1 phosphatases are CLAV- D Co-IP assay showing the physical interaction between AGB1 and RPK2. AGB1-FLAG alone or with mCherry-Venus or RPK2c-Venus was transiently co-expressed in performed as described previously [34]. True leaves of 3-week-old ATA1 signaling intermediates required for Arabidopsis shoot and floral protoplasts. Total protein extracts were subjected to IP with an anti-GFP antibody. The presence of AGB1-FLAG (upper) and Venus-tagged proteins (bottom) was seedlings were collected and chopped in an enzyme solution stem cells. Development 133: 4691 – 4698 determined by Western blotting. Note that the AGB1-FLAG co-expressed with RPK2c-Venus was condensed after IP. The co-IP experiments were repeated three times, containing 0.6% Cellulase ‘ONOZUKA’ RS (Yakult Pharmaceutical 8. Gilman AG (1987) G proteins: transducers of receptor-generated signals. with similar results. Industry) and 0.6% Macerozyme R10 (Yakult Pharmaceutical Indus- Annu Rev Biochem 56: 615 – 649 Data information: Scale bars = 10 lm in (A) and (B). try). Isolated protoplasts were washed and re-suspended at a 9. Assmann SM (2005) G proteins Go green: a plant G protein signaling concentration of 2 × 107 protoplasts per ml for polyethylene glycol FAQ sheet. Science 310: 71 – 73 (PEG)-mediated transformation. Vectors for transient expression 10. Johnston CA, Taylor JP, Gao Y, Kimple AJ, Grigston JC, Chen JG, Siderovski [9,29]. As any potential overlapping functions of XLGs were not Although Arabidopsis, rice, and other species, including dicots, were mixed with the protoplasts in transformation buffer [0.4 M DP, Jones AM, Willard FS (2007) GTPase acceleration as the rate-limiting addressed in this study, further analyses will be needed to evalu- gymnosperms, and animals, harbor the conserved amino acid in mannitol, 0.1 M Ca(NO3)2, and 40% PEG (w/v) (Sigma)]. After step in Arabidopsis G protein-coupled sugar signaling. Proc Natl Acad Sci ate the biological relevance of XLGs not only in CLV signaling but the Ga subunit, maize and some other monocots do not exhibit washing, the protoplasts were incubated in liquid culture medium USA 104: 17317 – 17322 also in G protein function. In fact, a recent report showed that this residue [12]. This evolutionarily distinct background of containing 0.4 M mannitol for 12–24 h at 23°C. 11. Jones JC, Duffy JW, Machius M, Temple BR, Dohlman HG, Jones AM maize Ga mutants exhibit enlarged meristem phenotype leading heterotrimeric G proteins could be another explanation for the Further experimental details are provided in Supplementary (2011) The crystal structure of a self-activating G protein alpha subunit the authors to infer a function of Ga in SAM maintenance [19]. differing meristem phenotypes observed in maize compared with Methods. reveals its distinct mechanism of signal initiation. Sci Signal 4: ra8 Despite clear evidence of the involvement of Ga in the maize Arabidopsis (Fig 1 and [12,19]). 12. Urano D, Jones JC, Wang H, Matthews M, Bradford W, Bennetzen JL, CLV-like pathway, further research is required before any general- Taken together, our results suggest the hypothesis that CLV3- Supplementary information for this article is available online: Jones AM (2012) G protein activation without a GEF in the plant king- izations can be made because severe phenotypes, such as those RPK2 signaling activates a heterotrimeric G protein through an http://embor.embopress.org dom. PLoS Genet 8:e1002756 observed in the maize Ga mutant, have not been reported in interaction, at least in Arabidopsis. Thus, these results support 13. Urano D, Chen JG, Botella JR, Jones AM (2013) Heterotrimeric G protein either Arabidopsis or rice Ga mutants [13,19,30]. The critical the notion that LRR-RLK-type receptor RPK2 acts as an alterna- Acknowledgements signalling in the plant kingdom. Open Biol 3: 120186 amino acids for Ga function have been reported. In particular, the tive GPCR, similar to canonical GPCR-G protein systems. This situa- We thank A. Miyawaki (RIKEN), S. Takayama (NAIST), and M. Kakita (Nagoya 14. Ullah H, Chen JG, Young JC, Im KH, Sussman MR, Jones AM (2001) Thr residue in the switch I region of GPA1 is important for the tion contrasts with that in maize, where FEA2, an LRR type University) for the BiFC vectors and N. Inada (NAIST) for confocal microscopy. Modulation of cell proliferation by heterotrimeric G protein in Arabidop- interaction between regulatory proteins for activation [31]. receptor, mediates CLV-like signaling and G proteins [19]. Although This work was supported by grants from KAKENHI (221S0002, 23119517, sis. Science 292: 2066 – 2069

ª 2014 The Authors EMBO reports Vol 15 | No 11 | 2014 1207 1208 EMBO reports Vol 15 | No 11 | 2014 ª 2014 The Authors Takashi Ishida et al Gbeta and CLAVATA control stem cells in plants EMBO reports EMBO reports Gbeta and CLAVATA control stem cells in plants Takashi Ishida et al

15. Wang XQ, Ullah H, Jones AM, Assmann SM (2001) G protein regulation 24. Wang S, Assmann SM, Fedoroff NV (2008) Characterization of the of ion channels and abscisic acid signaling in Arabidopsis guard cells. Arabidopsis heterotrimeric G protein. J Biol Chem 283: 13913 – 13922 Science 292: 2070 – 2072 25. Ishiguro S, Watanabe Y, Ito N, Nonaka H, Takeda N, Sakai T, Kanaya H, 16. Trusov Y, Rookes JE, Chakravorty D, Armour D, Schenk PM, Botella JR Okada K (2002) SHEPHERD is the Arabidopsis GRP94 responsible for the (2006) Heterotrimeric G proteins facilitate Arabidopsis resistance to formation of functional CLAVATA proteins. EMBO J 21: 898 – 908 necrotrophic pathogens and are involved in jasmonate signaling. Plant 26. Chen JG, Gao Y, Jones AM (2006) Differential roles of Arabidopsis hetero- Physiol 140: 210 – 220 trimeric G-protein subunits in modulating cell division in roots. Plant 17. Ullah H, Chen JG, Temple B, Boyes DC, Alonso JM, Davis KR, Ecker JR, Physiol 141: 887 – 897 Jones AM (2003) The beta-subunit of the Arabidopsis G protein nega- 27. Anderson DJ, Botella JR (2007) Expression analysis and subcellular locali- tively regulates auxin-induced cell division and affects multiple develop- zation of the Arabidopsis thaliana G-protein beta-subunit AGB1. Plant mental processes. Plant Cell 15: 393 – 409 Cell Rep 26: 1469 – 1480 18. Mudgil Y, Uhrig JF, Zhou J, Temple B, Jiang K, Jones AM (2009) Arabidop- 28. Adjobo-Hermans MJ, Goedhart J, Gadella TW Jr (2006) Plant G protein sis N-MYC DOWNREGULATED-LIKE1, a positive regulator of auxin trans- heterotrimers require dual lipidation motifs of Galpha and Ggamma and port in a G protein-mediated pathway. Plant Cell 21: 3591 – 3609 do not dissociate upon activation. J Cell Sci 119: 5087 – 5097 19. Bommert P, Je BI, Goldshmidt A, Jackson D (2013) The maize Galpha 29. Ding L, Pandey S, Assmann SM (2008) Arabidopsis extra-large G proteins gene COMPACT PLANT2 functions in CLAVATA signalling to control shoot (XLGs) regulate root morphogenesis. Plant J 53: 248 – 263 meristem size. Nature 502: 555 – 558 30. Fujisawa Y, Kato T, Ohki S, Ishikawa A, Kitano H, Sasaki T, Asahi T, 20. Tabata R, Kamiya T, Shigenobu S, Yamaguchi K, Yamada M, Hasebe M, Iwasaki Y (1999) Suppression of the heterotrimeric G protein causes Fujiwara T, Sawa S (2013) Identification of an EMS-induced causal abnormal morphology, including dwarfism, in rice. Proc Natl Acad Sci mutation in a gene required for boron-mediated root development by USA 96: 7575 – 7580 low-coverage genome re-sequencing in Arabidopsis. Plant Signal Behav 31. Tesmer JJ, Berman DM, Gilman AG, Sprang SR (1997) Structure of RGS4 8:e22534 bound to AlF4–activated G(i alpha1): stabilization of the transition state 21. Kinoshita A, Nakamura Y, Sasaki E, Kyozuka J, Fukuda H, Sawa S (2007) for GTP hydrolysis. Cell 89: 251 – 261 Gain-of-function phenotypes of chemically synthetic CLAVATA3/ESR-re- 32. Dievart A, Dalal M, Tax FE, Lacey AD, Huttly A, Li J, Clark SE (2003) CLAV- lated (CLE) peptides in Arabidopsis thaliana and Oryza sativa. Plant Cell ATA1 dominant-negative alleles reveal functional overlap between Physiol 48: 1821 – 1825 multiple receptor kinases that regulate meristem and organ develop- 22. Trusov Y, Rookes JE, Tilbrook K, Chakravorty D, Mason MG, Anderson D, ment. Plant Cell 15: 1198 – 1211 Chen JG, Jones AM, Botella JR (2007) Heterotrimeric G protein gamma 33. Kondo T, Sawa S, Kinoshita A, Mizuno S, Kakimoto T, Fukuda H, Saka- subunits provide functional selectivity in Gbetagamma dimer signaling gami Y (2006) A plant peptide encoded by CLV3 identified by in situ in Arabidopsis. Plant Cell 19: 1235 – 1250 MALDI-TOF MS analysis. Science 313: 845 – 848 23. Jones AM, Ecker JR, Chen JG (2003) A reevaluation of the role of the 34. Abel S, Theologis A (1994) Transient transformation of Arabidopsis leaf heterotrimeric G protein in coupling light responses in Arabidopsis. Plant protoplasts: a versatile experimental system to study gene expression. Physiol 131: 1623 – 1627 Plant J 5: 421 – 427

ª 2014 The Authors EMBO reports Vol 15 | No 11 | 2014 1209 Research Article

Targeted gene therapy and cell reprogramming in Fanconi anemia

Paula Rio1,2,†, Rocio Baños1,2,†, Angelo Lombardo3,†, Oscar Quintana-Bustamante1,2, Lara Alvarez1,2, Zita Garate1,2, Pietro Genovese3, Elena Almarza1,2, Antonio Valeri1,2, Begoña Díez1,2, Susana Navarro1,2, Yaima Torres4, Juan P Trujillo2,5, Rodolfo Murillas6, Jose C Segovia1,2, Enrique Samper4, Jordi Surralles5, Philip D Gregory7, Michael C Holmes7, Luigi Naldini3,8,** & Juan A Bueren1,2,*

Abstract Introduction

Gene targeting is progressively becoming a realistic therapeutic The progressive development of engineered nucleases has markedly alternative in clinics. It is unknown, however, whether this tech- improved the efficacy and specificity of targeted gene therapy, open- nology will be suitable for the treatment of DNA repair deficiency ing new possibilities for the treatment of inherited and acquired syndromes such as Fanconi anemia (FA), with defects in homology- diseases in the clinics (Tebas et al, 2014). In contrast to conven- directed DNA repair. In this study, we used zinc finger nucleases tional gene therapy with integrative vectors, targeted gene therapy and integrase-defective lentiviral vectors to demonstrate for the enables the insertion of foreign sequences (i.e., therapeutic genes or first time that FANCA can be efficiently and specifically targeted small oligonucleotides) in specific sites of the cell genome. Thus, into the AAVS1 safe harbor locus in fibroblasts from FA-A patients. depending on the genetic etiology of the disease, the gene-targeting Strikingly, up to 40% of FA fibroblasts showed gene targeting approach may pursue the correction of a specific mutation or, alter- 42 days after gene editing. Given the low number of hematopoi- natively, the insertion of the therapeutic transgene into safe loci of etic precursors in the bone marrow of FA patients, gene-edited FA the genome, often referred to as ‘safe harbors’ (Naldini, 2011). fibroblasts were then reprogrammed and re-differentiated toward In spite of the advances in the field, the question of whether or the hematopoietic lineage. Analyses of gene-edited FA-iPSCs not targeted gene therapy will be applicable to diseases where confirmed the specific integration of FANCA in the AAVS1 locus in homology-directed repair (HDR) is affected has never been explored. all tested clones. Moreover, the hematopoietic differentiation of Taking into account that Fanconi anemia (FA) proteins participate in these iPSCs efficiently generated disease-free hematopoietic HDR (Taniguchi et al, 2002; Yamamoto et al, 2003; Niedzwiedz progenitors. Taken together, our results demonstrate for the first et al, 2004; Yang et al, 2005; Nakanishi et al, 2011) and coordinate time the feasibility of correcting the phenotype of a DNA repair the action of multiple DNA repair processes, including the action of deficiency syndrome using gene-targeting and cell reprogramming different nucleases and homologous recombination (see reviews in strategies. Kee & D’Andrea, 2010; Kottemann & Smogorzewska, 2013; Moldo- van & D’Andrea, 2009), we aimed to investigate for the first time the Keywords cell reprogramming; Fanconi anemia; gene-targeting; iPSCs; zinc possibility of conducting a targeted gene therapy strategy in FA cells. finger nucleases Genetically, FA is a complex disease where mutations in sixteen Subject Categories Genetics, Gene Therapy & Genetic Disease; different genes (FANC-A,-B,-C,-D1/BRCA2,-D2,-E,-F,-G,-I,-J/ Haematology; Stem Cells BRIP1,-L –M, –N/PALB2, -O/RAD51C; -P/SLX4; -Q/ERCC4/XPF) have DOI 10.15252/emmm.201303374 | Received 9 August 2013 | Revised 16 April been reported (Bogliolo et al, 2013). Among all these genes, muta- 2014 | Accepted 17 April 2014 | Published online 23 May 2014 tions in FANCA account for about 60% of total FA patients (Casado EMBO Mol Med (2014) 6: 835–848 et al, 2007; Auerbach, 2009). Importantly, while few recurrent mutations (i.e., truncation of exon 4 in Spanish gypsies or mutations

1 Division of Hematopoietic Innovative Therapies, CIEMAT/CIBERER, Madrid, Spain 2 Instituto de Investigación Sanitaria Fundación Jiménez Díaz (IIS-FJD, UAM), Madrid, Spain 3 San Raffaele Telethon Institute for Gene Therapy, San Raffaele Scientific Institute, Milan, Italy 4 NIMGenetics SL, Madrid, Spain 5 Universidad Autónoma Barcelona/CIBERER, Barcelona, Spain 6 Division of Epithelial Biomedicine, CIEMAT/CIBERER, Madrid, Spain 7 Sangamo BioSciences Inc., Richmond, CA, USA 8 Vita Salute San Raffaele University, Milan, Italy *Corresponding author. Tel: +34 913 466 518; Fax: +34 913 466 484; E-mail: [email protected] **Corresponding author. Tel: +02 2643 4681; Fax: +02 2643 4621; E-mail: [email protected] †These authors contributed equally to this work.

ª 2014 The Authors. Published under the terms of the CC BY 4.0 license EMBO Molecular Medicine Vol 6 | No 6 | 2014 835 EMBO Molecular Medicine Targeted gene therapy in Fanconi anemia Paula Rio et al Paula Rio et al Targeted gene therapy in Fanconi anemia EMBO Molecular Medicine

in exons 13, 36, and 38) have been observed in FA-A patients, the EGFP and FANCA transgenes flanked by AAVS1 homology arms A 10.6 Kb FANCA mutations are generally private mutations, which include (donor IDLV) was generated (Fig 1A top). In this donor IDLV, 6.3 Kb Donor IDLV HR Arm HR Arm point mutations, microinsertions, microdeletions, splicing mutations FANCA is under the transcriptional control of the human PGK 2A EGFP PGK FANCA and large intragenic deletions (Castella et al, 2011). Thus, consider- promoter. In addition, a promoterless EGFP cDNA preceded by a BGHpA SA SV40pA U5-R-ΔU3 ing the large number of genes and mutations that can account for splice acceptor (SA) site and a translational self-cleaving 2A ZFNs the FA disease, the insertion of a functional FA gene in a ‘safe sequence was also included upstream of the FANCA cassette. Upon AAVS1 locus harbor’ locus would lead to the generation of a targeted gene addi- targeted-mediated insertion into AAVS1, the EGFP cassette will be 1 2 tion platform with a broad application in FA, regardless of the placed under the transcriptional control of the promoter of the ubiq- Exon Intron 1 Exon complementation group and mutation type of each patient. uitously expressed PPP1R12C gene, thus allowing the FACSorting of Recent studies by our group and others aiming at the identifica- gene-targeted cells (Fig 1A). Besides the donor IDLV, an adenoviral Targeted AAVS1 locus 1 2A EGFP PGK FANCA 2 tion of ‘safe harbor sites’ in the human genome have shown robust vector expressing a ZFN pair (AdV5/35-ZFN), designed to induce a BGHpA and stable expression of transgenes integrated in the human DNA double-strand break in the AAVS1 locus, was used to enhance SD SA SV40pA PPP1R12C gene, a locus also known as AAVS1, across different cell the efficiency of gene targeting in this locus (Hockemeyer et al, 5’ TI ( 1,195 bp) types (Smith et al, 2008; Lombardo et al, 2011). Additionally, no 2009). 3’ TI ( 1,314 bp) detectable transcriptional perturbations of the PPP1R12C and its To investigate the feasibility of performing gene targeting in FA- flanking genes were observed after integration of transgenes in this A cells, skin fibroblasts from four FA-A patients with different muta- B C locus, indicating that AAVS1 may represent a safe landing path for tions in FANCA were transduced either with the donor IDLV alone, days 1428 42 therapeutic transgene insertion in the human genome (Lombardo or with the donor IDLV and the AdV5/35-ZFNs simultaneously. et al, 2011). These observations, together with the development of Fourteen days after transduction, cells were analyzed by flow artificial zinc finger nucleases (ZFNs) that efficiently and selectively cytometry to measure the proportion of EGFP+ fibroblasts. While 14 target the AAVS1 locus, have facilitated gene editing strategies <0.05% of the cells transduced with the donor IDLV alone were 12 5 - hTERT aiming at inserting therapeutic transgenes in this locus, not only in positive for EGFP, 0.2–1.1% of FA fibroblasts that had been co- 10 4 + hTERT immortalized cell lines but also in several primary human cell types, transduced with the donor IDLV and the ZFNs-AdV were EGFP+ 8 3 including induced pluripotent stem cells (hiPSCs; Hockemeyer et al, (See Fig 1B and representative analyses in Supplementary Fig S1). 6 + 2 2009; DeKelver et al, 2010; Lombardo et al, 2011; Zou et al, 2011b; Strikingly, the percentage of EGFP cells markedly increased during 4 % EGFP % EGFP Chang & Bouhassira, 2012). the in vitro culture of these cells, reaching levels between 5.5 and 2 1 Because a defective FA pathway not only predisposes FA patients 13.4% (Fig 1B), showing the proliferation advantage of gene-edited 0 0 FA-52 FA-123 FA-664 to cancer (Rosenberg et al, 2008) but also to the early development of FA-A fibroblasts. FA-5 FA-123 FA-664 FA-52 bone marrow failure due to the progressive extinction of the HSCs Because the prolonged in vitro culture of FA fibroblasts results in (Larghero et al, 2002; Jacome et al, 2006), our final aim in these stud- increased rates of cell senescence (Muller et al, 2012), in a new set D E ies was the generation of gene-edited, disease-free FA-HSCs, obtained of experiments, fibroblasts from three FA patients (FA-52, FA-123 from non-hematopoietic tissues of the patient. Thus, in our current and FA-644) were transduced with an excisable hTERT-expressing Days 14 28 42 studies, we firstly pursued the specific insertion of the therapeutic LV (Salmon et al, 2000) prior to performing the gene-targeting 40 5’TI FANCA gene in the AAVS1 locus of FA-A patients’ fibroblasts. There- procedure. Transduction of FA fibroblasts with hTERT-LVs resulted 35 Ladder Z+D D H20 after, gene-edited FA cells were reprogrammed to generate self- in a marked increase in telomerase activity (see representative data - renewing disease-free iPSCs and finally re-differentiated toward the in Supplementary Fig S2). Significantly, the proportion of EGFP+ 30 1,195 bp hematopoietic lineage, as previously described with FA cells corrected cells was markedly increased (3–4-fold) in hTERT-transduced versus 25 by conventional LV-mediated gene therapy (Raya et al, 2009). untransduced FA fibroblasts from FA patients (Fig 1C), indicating 20 Our goal of conducting a combined approach of gene editing and that hTERT improved the efficacy of gene targeting in FA-A fibro- EGFP % 15 3’TI cell reprogramming in FA cells was particularly challenging taking into blasts. Consistent with data obtained with non-immortalized fibro- Ladder - Z+D D H20 account the relevance of the FA pathway both in HDR (Taniguchi blasts, when immortalized gene-edited FA fibroblasts were 10 et al, 2002; Yamamoto et al, 2003; Niedzwiedz et al, 2004; Yang et al, maintained in culture, a progressive increase in the proportion of 5 1,314 bp + 2005; Moldovan & D’Andrea, 2009; Kee & D’Andrea, 2010; EGFP cells was also observed (see data from geFA-52T in Fig 1D). 0 Nakanishi et al, 2011; Kottemann & Smogorzewska, 2013) and cell Strikingly, around 40% of treated FA-A fibroblasts were EGFP+ FA-52T reprogramming (Raya et al, 2009; Muller et al, 2012; Yung et al, after 42 days in culture in the absence of any selectable drug 2013). In spite of these hurdles, the strong selective growth advan- (Fig 1D). tage characteristic of corrected FA cells allowed us to establish a new PCR analyses with two pairs of primers that amplify, respec- Figure 1. Efficacy of gene targeting of FANCA in the AAVS1 locus of primary hFA-A fibroblasts. A Top: schematic representation of the donor integrase-defective lentiviral vector (IDLV) used to promote insertion of the EGFP/FANCA cassette into the AAVS1 locus. approach for the efficient generation of FA HPCs harboring specific tively, the 50 and the 30 integration junctions between the EGFP/ Middle: AAVS1 locus with the zinc finger nucleases (ZFNs) target site. Bottom: AAVS1 locus upon ZFN-mediated targeted insertion of the EGFP/PGK-FANCA cassette. integrations of the therapeutic FANCA gene in a safe harbor locus. FANCA cassette and the endogenous AAVS1 locus evidenced the Black arrow shows transcription of the EGFP from the endogenous PPP1R12C promoter. HA, homology arm; SD, splice donor; SA, splice acceptor; BGHpA, bovine insertion of the EGFP/FANCA cassette into the AAVS1 locus of growth hormone polyadenylation signal; SV40pA, simian virus 40 polyadenylation signal. Constituents of the LTR (U5-R-DU3) are also indicated. sorted EGFP+ geFA-52T fibroblasts (Fig 1E). In these gene-edited B Proliferation advantage of targeted Fanconi anemia (FA) fibroblasts (EGFP+ cells) during in vitro incubation. Results FA fibroblasts, the activity of hTERT was also confirmed (Supple- C Comparative analysis of gene targeting in FA-A fibroblasts, untransduced or transduced with a lentiviral vector expressing hTERT. Analyses were performed 14 days after gene targeting. mentary Fig S2). D In vitro proliferation advantage of targeted FA fibroblasts (EGFP+) previously transduced with hTERT (FA-52T fibroblasts). Efficient gene-targeting-mediated complementation of To investigate whether the insertion of the therapeutic hFANCA E Targeted integration analysis of the EGFP/PGK-FANCA cassette into the AAVS1 site by PCR using primers specific for the 50 or 30 integration junctions (red arrows in the fibroblasts from FA-A patients cassette in the AAVS1 locus of FA-A fibroblasts corrected the cellu- top schematic) defined as 50 TI or 30 TI, respectively. lar phenotype of the disease, the functionality of the FA pathway in To promote insertion of a FANCA expression cassette into the FA-52T fibroblasts was tested both before (negative control) and AAVS1 locus, an integrase-defective lentiviral vector (IDLV) harboring after the gene-targeting procedure. As a positive control, healthy

836 EMBO Molecular Medicine Vol 6 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO Molecular Medicine Vol 6 | No 6 | 2014 837 EMBO Molecular Medicine Targeted gene therapy in Fanconi anemia Paula Rio et al Paula Rio et al Targeted gene therapy in Fanconi anemia EMBO Molecular Medicine

donor fibroblasts (H.D. Fib) were analyzed in parallel. The presence mimicking the response of H.D. fibroblasts (Fig 2A). Because the radial chromosomes, typically found in FA patients0 cells—the same promoters were hypomethylated in gene-corrected FA-iPSC clones, of nuclear FANCD2 foci, fully dependent on the expression of all the main characteristic of FA cells is the increased chromosomal DEB treatment did not induce any increase in the number of chro- in clear contrast to the high level of methylation observed in H.D. FA core complex proteins, including FANCA (Garcia-Higuera et al, instability upon exposure to DNA inter-strand cross-linking (ICL) mosomal aberrations in geFA-52T fibroblasts (Fig 2B). fibroblasts (Supplementary Fig S4C). To further demonstrate the 2001), was determined in these samples after DNA damage induced drugs, we also investigated the response of both uncorrected and Taken together, these results show the feasibility of correcting pluripotency of geFA-iPSC16 cells in vivo, cells were subcutane- by mitomycin C (MMC). In contrast to uncorrected FA-52T fibro- gene-edited FA-A fibroblasts to diepoxybutane (DEB). While in the phenotype of FA cells using gene targeting strategies, in particu- ously inoculated in NSG mice. Characteristic teratomas containing blasts (FA-52T Fib.), which did not generate FANCD2 foci even after FA-52T fibroblasts DEB induced a significant increase in the number lar by promoting the insertion and expression of FANCA in the complex structures representing the three embryonic germ layers MMC exposure, a significant proportion of the geFA-52T fibroblasts of chromosomal aberrations per cell (from 0.05 0.05 to AAVS1 safe harbor locus of fibroblasts from FA-A patients. were observed 8–10 weeks after implantation. Immunofluores- Æ generated FANCD2 foci, mainly after treatment with MMC, thus 1.7 0.46 aberrations/cell)— including chromatid breaks and cence staining confirmed the expression of definitive endoderm Æ Efficient generation of disease-free iPSCs from FA fibroblasts markers (Fox2A), neural structures that expressed neuroectoder- corrected by gene targeting mal markers (ß-III-tubulin) and the generation of mesoderm (Brachyury) and mesoderm derivatives tissue such as muscle A p= 0.87 To generate disease-free FA-iPSCs, FA fibroblasts subjected to gene (a-SMA; Fig 3B). p= 0.0001 p= 0.003 editing (geFA-123, geFA-52 and geFA-52T) were first enriched for To confirm the insertion of the FANCA cassette into the AAVS1 35 -- MMC EGFP+ cells by cell sorting and then reprogrammed using a poly- locus in the gene-corrected FA-iPSC clones, Southern blot analyses 30 + MMC cistronic excisable LV expressing the human SOX2, OCT4, KLF4, were performed on genomic DNA extracted from gene-edited geFA- 25 and cMYC transgenes from the EF1A promoter (STEMCCA vector; iPSC clones 16, 26, and 31. Blots hybridized with probes for the 20 Somers et al, 2010). Consistent with previous observations (Raya exogenous EGFP and the endogenous AAVS1 genes confirmed the 15 et al, 2009), uncorrected FA fibroblasts did not generate iPSCs after monoallelic integration of the EGFP/FANCA cassette into the AAVS1 10 reprogramming, even after transduction with the TERT-LV (data locus and the absence of random integration in any of the three 5 not shown). Although several iPSC-like colonies were generated tested clones (Fig 3C,D). + % % of cells with FANCD2 foci 0 from gene-edited FA-123 fibroblasts (115 AP cells/100,000 fibro- Once demonstrated the generation of bona fide gene-edited FA- H.D.Fib. H.DFib. Fib. FA-52T FA-52T Fib. geFib. geFA-52T FA-52T Fib. blasts), no stable iPSC lines could be generated from FA fibroblasts iPSCs, in the next set of experiments, we aimed to verify whether simply subjected to gene editing, most probably because of the these geFA-iPSCs were disease free, as shown for their parental H.D. Fib. FA-52T Fib. geFA-52T Fib. pro-senescence nature of these cells. In marked contrast to these gene-edited FA fibroblasts (Fig 2). First, we verified by qRT-PCR observations, the reprogramming of FA fibroblasts that were first that hFANCA mRNA levels corresponding to the three tested geFA- FANCD2 FANCD2 FANCD2 transduced with the hTERT-LV and then subjected to gene editing iPSC clones were very similar to levels observed in the control ES generated 230 iPSC-like clones, most of which could be maintained cell line and markedly higher when compared to uncorrected FA- after serial in vitro passages (Supplementary Fig S3). Twelve iPSC 52T fibroblasts (Fig 4A). Western blot analysis confirmed the

+ MMC clones generated from geFA-52T fibroblasts were further expanded expression of FANCA in all the three tested clones (Fig 4B). Even and differentiated into fibroblasts to perform additional studies to more, since FANCA is necessary for the relocation of FANCD2 to confirm the integration site of the EGFP/FANCA construct. First, damaged DNA sites, we investigated the presence of nuclear qPCR analyses were conducted to determine the mean copy FANCD2 foci in three geFA-iPSC clones exposed to MMC. As shown number per cell of the EGFP/FANCA cassette. As shown in Supple- in Fig 4C, these analyses further confirmed the expression and func- B FA-52T Fib. geFA-52T Fib. mentary Table S1, 11 out of the 12 geFA-iPSC clones analyzed tionality of FANCA in the three tested geFA-iPSC clones. Consistent were positive for EGFP integration and contained an average of with the restored FA pathway of gene-edited FA-iPSCs, DEB did not p=0.001 0.98 0.44 EGFP copies per cell. The only iPSC clone that did not induce a significant increase in the number of chromosomal aberra- 2.5 � harbor any EGFP copy (clone 5) did not progress more than six tions in FA-corrected cells. Remarkably, the number of chromo- 2 passages in culture. somal aberrations in geFA-iPSCs (0.2 0.1 aberrations/cell;

of aberrations / of p=0.29 gration of the cassette were also positive for the PCR band corre- FA-52 (c.710-5T>C and c.3558insG) was investigated by Sanger

er 0.5 N sponding to the specific insertion in the AAVS1 locus. sequencing both on FA-52T fibroblasts and geFA-iPSC clones 16, 26, Three geFA-iPSC clones (clones 16, 26 and 31) were selected and 31 (Supplementary Fig S5). The confirmation of both patho- 0 DEB -+-+ for further characterization. The pluripotency of these gene- genic mutations in the three tested geFA-iPSCs, together with our + DEB FA-52T Fib. geFA-52T Fib. corrected clones was first analyzed both by alkaline phosphatase observations showing that all stable iPSC clones contained the (AP) staining and immunohistochemistry staining of different AAVS1-targeted FANCA gene (Supplementary Table S1) and had a pluripotency genes. Representative pictures in Fig 3A and Supple- functional FA pathway, demonstrates that the disease-free nature of mentary Fig S4A showed that all tested geFA-iPSCs clones were gene-edited FA-iPSCs is a consequence of the functional insertion of Figure 2. Phenotypic correction of the gene-edited FA-A fibroblasts. highly positive for AP, NANOG, TRA-1-60, OCT4, and SSEA-4 FANCA within the AAVS1 safe harbor site of these reprogrammed A Top: histogram showing the percentage of FA-A fibroblasts, unstransduced or co-transduced with the donor integrase-defective lentiviral vector (IDLV) and the AdV5/ 35-ZFNs (geFA-52T Fib), showing FANCD2 foci in the absence or the presence of mitomycin C (MMC). Bottom: representative images of FANCD2 foci (red) in cells expression. RT-qPCR analyses of the expression of endogenous FA cells. shown in the top histogram, after MMC treatment. pluripotency genes NANOG, OCT4, SOX2, KLF4, and cMYC were Aiming to excise the STEMCCA vector from the genome of B Chromosomal instability induced by diepoxybutane (DEB) in untreated (FA-52T) and gene-edited FA fibroblasts (geFA-52T Fib). Left: representative FISH analysis was consistent with the pluripotent nature of these clones (Supplemen- geFA-iPSCs, cells from clone 16 were transduced with an IDLV performed by staining telomeres (in green), centromeres (in pink) and chromosomes (in blue). Right: histogram showing the number of chromosomal aberrations per tary Fig S4B). In all cases, a very low expression of the ectopic co-expressing the Cre recombinase and the Cherry fluorescence cell. reprogramming transgenes was found, indicating substantial marker (Papapetrou et al, 2011). Thereafter, individual colonies Data information: Values are shown as mean s.e. from three independent experiments (A) or analysis of twenty different metaphases per group (B). All P-values were Æ inactivation of the EF1A promoter present in the reprogramming were isolated to select those clones with a lower number of copies calculated using two-tailed unpaired Student’s t-test. vector. As expected for bona fide iPSC clones, OCT4 and NANOG of the STEMCCA provirus. Two clones were selected: Excised clones

838 EMBO Molecular Medicine Vol 6 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO Molecular Medicine Vol 6 | No 6 | 2014 839 EMBO Molecular Medicine Targeted gene therapy in Fanconi anemia Paula Rio et al Paula Rio et al Targeted gene therapy in Fanconi anemia EMBO Molecular Medicine

cell � To investigate whether the EGFP/FANCA cassette was specifically Fig 4D) was ten times lower to the number observed in their paren- - DEB 1.5 integrated in the AAVS1 locus of all these iPSC clones, 30 primers tal uncorrected fibroblasts (see Fig 2B). previously used in analyses of Fig 1E were used. As shown in To assure the identity of the different geFA-iPSC clones, the 1 Supplementary Table S1, all iPSC clones that were positive for inte- presence of the original pathogenic mutations described in patient of aberrations / of p=0.29 gration of the cassette were also positive for the PCR band corre- FA-52 (c.710-5T>C and c.3558insG) was investigated by Sanger er 0.5 N sponding to the specific insertion in the AAVS1 locus. sequencing both on FA-52T fibroblasts and geFA-iPSC clones 16, 26, Three geFA-iPSC clones (clones 16, 26 and 31) were selected and 31 (Supplementary Fig S5). The confirmation of both patho- 0 DEB -+-+ for further characterization. The pluripotency of these gene- genic mutations in the three tested geFA-iPSCs, together with our + DEB FA-52T Fib. geFA-52T Fib. corrected clones was first analyzed both by alkaline phosphatase observations showing that all stable iPSC clones contained the (AP) staining and immunohistochemistry staining of different AAVS1-targeted FANCA gene (Supplementary Table S1) and had a pluripotency genes. Representative pictures in Fig 3A and Supple- functional FA pathway, demonstrates that the disease-free nature of mentary Fig S4A showed that all tested geFA-iPSCs clones were gene-edited FA-iPSCs is a consequence of the functional insertion of Figure 2. Phenotypic correction of the gene-edited FA-A fibroblasts. highly positive for AP, NANOG, TRA-1-60, OCT4, and SSEA-4 FANCA within the AAVS1 safe harbor site of these reprogrammed A Top: histogram showing the percentage of FA-A fibroblasts, unstransduced or co-transduced with the donor integrase-defective lentiviral vector (IDLV) and the AdV5/ 35-ZFNs (geFA-52T Fib), showing FANCD2 foci in the absence or the presence of mitomycin C (MMC). Bottom: representative images of FANCD2 foci (red) in cells expression. RT-qPCR analyses of the expression of endogenous FA cells. shown in the top histogram, after MMC treatment. pluripotency genes NANOG, OCT4, SOX2, KLF4, and cMYC were Aiming to excise the STEMCCA vector from the genome of B Chromosomal instability induced by diepoxybutane (DEB) in untreated (FA-52T) and gene-edited FA fibroblasts (geFA-52T Fib). Left: representative FISH analysis was consistent with the pluripotent nature of these clones (Supplemen- geFA-iPSCs, cells from clone 16 were transduced with an IDLV performed by staining telomeres (in green), centromeres (in pink) and chromosomes (in blue). Right: histogram showing the number of chromosomal aberrations per tary Fig S4B). In all cases, a very low expression of the ectopic co-expressing the Cre recombinase and the Cherry fluorescence cell. reprogramming transgenes was found, indicating substantial marker (Papapetrou et al, 2011). Thereafter, individual colonies Data information: Values are shown as mean s.e. from three independent experiments (A) or analysis of twenty different metaphases per group (B). All P-values were Æ inactivation of the EF1A promoter present in the reprogramming were isolated to select those clones with a lower number of copies calculated using two-tailed unpaired Student’s t-test. vector. As expected for bona fide iPSC clones, OCT4 and NANOG of the STEMCCA provirus. Two clones were selected: Excised clones

A geFA-iPSC 16 A 100

NANOG TRA 1-60 OCT4 SSEA-4 DAPI AP

mRNA hFANCA mRNA 1 (relative expression) (relative FA-52T Clone 16 Clone 26 Clone 31 H9 Fibroblasts geFA-iPSCs Control ES B geFA-iPSC 16

ECTODERM ENDODERM MESODERM B Fibroblasts geFA-iPSCs

H.D. FA-A 16 26 31 FANCA

ß-Actin

ß-III Tubulin Fox2A α-SMA Brachyury

C geFA-iPSC 16 geFA-iPSC 26 geFA-iPSC 31 C geFA-IPSCs Fibroblasts FANCD2 FANCD2 FANCD2 16 26 31 geFA FA

EGFP hPGK hFANCA

9.6 Kb MMC + P Bgl I Targeted Locus Bgl I

Non Targeted D geFA-iPSC 16 Locus - DEB + DEB P 2.5 3.3 Kb Bgl I Bgl I 2

D 1.5 geFA-IPSCs Fibroblasts 1 p=0.1

16 26 31 geFA FA of aberra Ɵ ons/cell

er 0.5 N 5.1 Kb EGFP hPGK hFANCA Targeted Locus 0 DEB - + BstXI P BstXI geFA-iPSCs 16

Figure 4. Disease-free Fanconi anemia phenotype of corrected geFA-iPSCs. A Histogram showing the levels of hFANCA expression in gene-edited FA-iPSC clones and human ES (H9) relative to untreated FA-52T fibroblasts. Data are shown as mean s.e. of three different analyses. Æ B Western blot analysis showing FANCA expression in geFA-iPSC clones in comparison with fibroblasts from HD and a FA-A patient. C Representative immunofluorescence analysis of FANCD2 foci in geFA-iPSCs after DNA damage with mitomycin C (MMC). D Chromosomal instability induced by diepoxybutane (DEB) was also tested in geFA-iPSC 16. FISH analysis was performed using probes to detect telomeres (green), Figure 3. Pluripotency characterization and insertion site analyses of gene-edited FA-A iPSCs. centromeres (pink) and chromosomes (blue). Right: histogram showing the number of chromosomal aberrations per cell. A Expression of TRA1-60, SSEA-4, OCT4, and NANOG pluripotency markers by immunofluorescence staining of gene-edited FA-iPSCs (geFA-iPSCs; clone 16). Data information: Data are shown as mean s.e. from three different experiments (A) or analysis of twenty different metaphases per group (D). All P-values were B Immunofluorescence analysis of ectoderm (b-II-tubulin), endoderm (Fox2A), and mesoderm (a-SMA and Brachyury) in teratomas generated from geFA-iPSCs Æ (clone 16). calculated using two-tailed unpaired Student’s t-test. C Southern blot analysis of genomic DNA extracted from the indicated gene-corrected FA iPSC clones (geFA-IPSCs) and from parental fibroblasts, either unmanipulated (FA) or after gene editing (ge-FA iPSCs, clones 16, 26 and 31). Genomic DNA was digested with BglI and hybridized with a probe for PPP1R12C. The band of 9.6 kb corresponds to the targeted integration in PPP1R12C, while the 3.3 kb correspond to the untargeted allele. D Southern blot analysis of samples shown in (C) digested with BstXI and hybridized with a probe (P) for EGFP. One single band of 5.1 kb is expected for specific 16.1 and 16.2, with a number of 0.35 0.10 and <0.05 copies/cell, OCT4, KLF4, NANOG, and cMYC) and the absence of ectopic transg- integrations in PPP1R12C. Æ respectively. In clone 16.2, the excision of the hTERT provirus was enes expression (Supplementary Fig S6A). As expected from bona fide also confirmed (<0.05 copies as deduced from q-PCR analyses). pluripotent iPSC clones, these two clones generated teratomas with RT-qPCR analysis performed in these two subclones showed the structures characteristics of the three germ layers (Supplementary persistent expression of endogenous pluripotency genes (SOX2, Fig S6B).

840 EMBO Molecular Medicine Vol 6 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO Molecular Medicine Vol 6 | No 6 | 2014 841 EMBO Molecular Medicine Targeted gene therapy in Fanconi anemia Paula Rio et al Paula Rio et al Targeted gene therapy in Fanconi anemia EMBO Molecular Medicine

A geFA-iPSC 16 A 100

NANOG TRA 1-60 OCT4 SSEA-4 DAPI AP

mRNA hFANCA mRNA 1 (relative expression) (relative FA-52T Clone 16 Clone 26 Clone 31 H9 Fibroblasts geFA-iPSCs Control ES B geFA-iPSC 16

ECTODERM ENDODERM MESODERM B Fibroblasts geFA-iPSCs

H.D. FA-A 16 26 31 FANCA

ß-Actin

ß-III Tubulin Fox2A α-SMA Brachyury

C geFA-iPSC 16 geFA-iPSC 26 geFA-iPSC 31 C geFA-IPSCs Fibroblasts FANCD2 FANCD2 FANCD2 16 26 31 geFA FA

EGFP hPGK hFANCA

9.6 Kb MMC + P Bgl I Targeted Locus Bgl I

Non Targeted D geFA-iPSC 16 Locus - DEB + DEB P 2.5 3.3 Kb Bgl I Bgl I 2

D 1.5 geFA-IPSCs Fibroblasts 1 p=0.1

16 26 31 geFA FA of aberraƟ ons/cell

er 0.5 N 5.1 Kb EGFP hPGK hFANCA Targeted Locus 0 DEB - + BstXI P BstXI geFA-iPSCs 16

A Analysis of the genetic stability of gene-edited FA fibroblasts of the excised vs the non-excised clones (see data from two indepen- 16 and iPSCs dent experiments in Fig 5A and Supplementary Fig S8). Consistent 14 with the flow cytometry data, granulo-macrophage and erythroid geFA-iPSC16 12 Because of the chromosomal instability of FA cells, we investigated colonies were generated by geFA-iPSC-differentiated cells in methyl- geFA-iPSC 16.1 Ex by means of karyotype analyses and aCGH analyses whether the cellulose. As it was observed in the flow cytometry studies, higher 10 different manipulations of FA-52 fibroblasts and their corresponding numbers of hematopoietic progenitors were generated by excised iPSCs induced chromosomal instability. As shown in Table 1, no versus non-excised geFA-iPSC (Fig 5B). In all instances, colonies 8 evident karyotype or aCGH abnormalities were observed in derived from geFA-iPSC were almost as resistant to MMC as healthy 6 expanded FA-52 parental fibroblasts when compared with a cord blood progenitor cells, in contrast to the MMC hypersensitivity % % positive cells reference human DNA sample. Even more, the transduction with observed in BM progenitors from FA patients (Fig 5C). 4 hTERT-LV and the gene-editing process did not induce evident Finally, to investigate whether gene-edited FA-iPSCs were also 2 chromosomal abnormalities in these cells. Reprogrammed geFA-52 able to differentiate toward the hematopoietic lineage in vivo, one of iPSCs also had a normal karyotype, although a deletion in the the teratomas generated by the excised geFA-52 iPSCs (clone 16.2) 0 16p12.2p12.1 locus was noted in the aCGH analysis. After excision was analyzed for the presence of human hematopoietic markers. As CD34+CD43+ CD34+CD45+ CD45+ with the Cre recombinase, in addition to the 16p deletion, a mosaic shown in Supplementary Fig S9, 3% of the cells present in this tera- + B trisomy in chromosome 5 was observed (See Table 1 and Supple- toma consisted on hCD45 /mCD45À cells. Within this population, mentary Fig S7). 3.5% corresponded to hCD34+ cells, thus revealing the in vivo p= 0.019 differentiation potential of this clone. 1000 p= 0.2 BFU-ES Generation of disease-free hematopoietic progenitors from p= 0.04

gene-edited FA-A iPSCs cells Discussion 5 CFU-GMs 100 In experiments corresponding to Fig 5 and Supplementary Figs S8 and S9, we investigated whether hematopoietic progenitor cells Thanks to the development of artificial nucleases capable of generat- derived from gene-edited FA-iPSCs were disease-free. To conduct ing DNA double-strand breaks (DSBs) in pre-determined sequences 10

these experiments, embryoid bodies from geFA-iPSCs were incu- of the genome (Porteus & Baltimore, 2003; Urnov et al, 2010; Cong CFCs/1x10 of er

bated with hematopoietic cytokines as described in Materials and et al, 2013; Joung & Sander, 2013), targeted gene therapy is entering N methods. As shown in representative analyses from Supplementary into the clinics (Tebas et al, 2014). Whether these approaches will 1 Fig S8A, the hematopoietic differentiation of geFA-iPSCs after be amenable to the treatment of DNA repair deficiency syndromes geFA-IPSC16 geFA-iPSC16 Ex H.D. CB 21 days of in vitro stimulation was demonstrated by the presence of such as FA is, however, uncertain. In this respect, it is currently hematopoietic precursors (CD43+/CD34+), committed hematopoi- known that FA proteins participate in maintaining the genomic etic progenitors (CD34+/CD45+) and also mature hematopoietic stability of the cell and coordinate the actions of multiple repair + cells (CD34À/CD45 ). When the hematopoietic differentiation of processes, including HDR (Kottemann & Smogorzewska, 2013), C - MMC excised and non-excised iPSC clones was compared, the proportion making these cells particularly appropriate for investigating the 100 + MMC of CD45+ and CD34+/CD45+ was consistently increased in the case feasibility of performing targeted gene therapy in syndromes associated with DNA repair defects and genome instability. Although the 80 mechanisms explaining how the FA pathway promotes HDR are still unclear, most evidence suggests that the monoubiquitination of 60 Table 1. aCGH analysis in FA-52 fibroblasts prior to and after gene FANCD2—which is critically dependent on the presence of all the editing and in gene-edited IPSCs-derived clones FA core complex proteins, including FANCA—is essential for the 40 aCGH result recruitment of several HDR factors (such as BRCA1, BRCA2, and

RAD51) to damaged chromatin (see review in Kee & D’Andrea, CFCs survival (%) OMIM 20 Cells Alteration Locus GENES Karyotype 2010). To investigate whether gene targeting was feasible in FA cells we FA-52 –– –46 XY 0 a focused on the most frequent FA complementation group, FA-A fibroblasts geFA-iPSC16 H.D. CB FA-664 BM FA-82 BM (Casado et al, 2007; Auerbach, 2009), and investigated the possibil- geFA-52T fibr.b –– –46 XY ity of inserting the therapeutic transgene in a safe harbor locus of Figure 5. Hematopoietic differentiation of gene-edited FA-IPSCs. geFA-52T iPSC clones the human genome—the AAVS1 locus (Lombardo et al, 2011). + + + + + c A Analysis of the percentage of CD43 CD34 , CD45 CD34 , and CD45 cells generated by unexcised and excised geFA-iPSCs (clones 16 and Ex 16.1). Clone 16 Deletion 16p12.2p12.1 * 46 XY Strikingly, our first results in Fig 1 clearly demonstrate the B Left: Representative pictures of hematopoietic colonies generated by geFA-iPSCs. Right: Analysis of the clonogenic potential of unexcised and excised ge-FAiPSCs c Clone 16 Ex Deletion 16p12.2p12.1 * 46 XY feasibility of performing gene targeting in FA-A cells with signifi- (clones 16 and Ex 16.1) in comparison with H.D. cord blood cells. Mosaic 5 – 46 XY cant efficacies (up to 4%), comparable with efficacies reported in C Survival to mitomycin C (MMC) of CFCs obtained from geFA-IPSCs (clone 16) in comparison with BM CFCs from two different FA patients (FA-664 BM and FA-82 BM) trisomy primary cells competent for DNA repair (DeKelver et al, 2010; and with CFCs from a healthy cord blood (H.D. CB). Lombardo et al, 2011; Sebastiano et al, 2011; Soldner et al, 2011; Data information: Values are shown as mean s.e. of three experiments. All P-values were calculated using two-tailed unpaired Student’s t-test. *EEF2K, CDR2, HS3ST2, SCNN1G, SCNN1B, COG7, GGA2, EARS2, NDUFAB1, Æ PALB2, DCTN5, PLK1, ERN2, PRKCB, CACNG3, RBBP6. Zou et al, 2011a). The feasibility of performing gene targeting in aComparison analyses between expanded fibroblasts from patient FA-52 (FA- FA-A cells could be explained by different hypotheses. First, as 52 fibroblasts) and a reference male DNA sample. previously described in other systems (Matrai et al, 2011; Peluffo this hypothesis, we should contemplate the possibility that the tion of the therapeutic cassette in the AAVS1 locus might have bComparison analyses between expanded, TERT-transduced, and gene-edited FA-52 fibroblasts (geFA-52T fibr.) with respect to FA-52 fibroblasts. et al, 2013), a transient though early expression of FANCA may be limited HDR activity of FA-A cells (Nakanishi et al, 2005, 2011) occurred through an HDR-independent process, as reported in cComparison analyses between geFA-52T iPSCs clone 16 and clone 16 Ex induced by the donor IDLV, thus facilitating the insertion of the could be sufficient to facilitate the ZFN-mediated integration of other models (Anguela et al, 2014), PCR and Southern blot analy- (after excision of the reprogramming cassette) and FA-52 fibroblasts. exogenous therapeutic cassette through a HDR process. Besides our donor IDLV in the AAVS1 site. Finally, although the integra- ses showed the expected amplicons and band length for targeted

842 EMBO Molecular Medicine Vol 6 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO Molecular Medicine Vol 6 | No 6 | 2014 843 EMBO Molecular Medicine Targeted gene therapy in Fanconi anemia Paula Rio et al Paula Rio et al Targeted gene therapy in Fanconi anemia EMBO Molecular Medicine

editing (Fig 1). Whether or not this effect is specific for FA cells or approaches (Raya et al, 2009; Muller et al, 2012), our new study the EF1a STEMCCA lentiviral vector kindly provided by isotypes were used as controls. 40,6-Diamidino-2-phenylindole whether it is simply mediated by the enhanced proliferation rate of shows the efficient hematopoietic differentiation of gene-edited FA- Dr Mostoslavsky was used (Sommer et al, 2010). This vector (DAPI; Roche)-positive cells were excluded from the analysis. TERT-transduced FA cells is currently unknown. Nevertheless, to iPSCs. Moreover in the current study, we observed the generation of contains the cDNAs for OCT4, SOX2, c-MYC, and KLF4 flanked by Analysis was performed using FlowJo software. the best of our knowledge, the improved gene targeting mediated by increased numbers of hematopoietic progenitors from geFA-iPSCs loxP sequences for their subsequent excision. ZFNs targeting hTERT observed in our experiments constitutes a new finding that subjected to excision of the reprogramming cassette, confirming intron 1 of the PPP1R12C gene were expressed from an Adenoviral Inmunofluorescence and Western blot of Fanconi has not been previously reported in any other experimental model. previous observations showing that the residual expression of repro- Vector (AdV5/35) under the control of the CMV promoter (Lom- anemia proteins The observation that transduction with hTERT also facilitates the gramming genes limits the iPSC differentiation potential (Ramos- bardo et al, 2011). generation of gene-edited FA-iPSCs is consistent with previous data Mejia et al, 2012). The hematopoietic differentiation observed in Analyses of FANCD2 foci were performed by immunofluorescence showing the relevance of hTERT in cell reprogramming (Batista these experiments and the robust expression of FANCA targeted into Cell transduction of primary fibroblasts or iPSCs treated for 16 h with 200 nM of et al, 2011; Pomp et al, 2011; Winkler et al, 2013). In safety terms, the safe harbor AAVS1 locus should account for the generation of a MMC. After MMC treatment, cells were stained with rabbit poly- even though the hTERT provirus could be efficiently excised from high number of hematopoietic progenitors with normalized For gene editing experiments, fibroblasts from FA-A patients were clonal anti-FANCD2 (Abcam, ab2187-50) as previously described transduced cells with the Cre recombinase, further approaches response to MMC. transduced either with donor IDLV alone (150 ng HIV Gag p24/ml) (Hotta & Ellis, 2008; Raya et al, 2009). Cells with more than ten based on the transient expression of hTERT during gene editing In summary, our study demonstrates for the first time the possi- or together with AdV5/35-ZFNs (multiplicity of infection (MOI) foci were scored as positive. FANCA expression was analyzed by and/or cell reprogramming would constitute safer approaches to bility of conducting efficient and precise targeted-mediated gene 200). Fourteen days post-transduction, the proportion of EGFP+ Western blot (Raya et al, 2009) using the following antibodies: limit potential genomic insults during the ex vivo manipulation of therapy in HDR-deficient cells. Moreover, we show the feasibility of cells was determined by flow cytometry (BD LSRFortessa cell hFANCA (ab5063 Abcam) and anti-beta Actin to mouse antibody the samples. reprogramming these cells to generate iPSC-derived gene-edited analyzer, Becton Dickinson Pharmingen). To immortalize fibro- (ab6276, Abcam) as control. Goat polyclonal antibody to rabbit Interestingly, EGFP analyses in gene-edited FA fibroblasts hematopoietic progenitors characterized by a disease-free pheno- blasts from FA-52 to FA-123 patients, 105 cells were transduced at IgG (HRP; ab6721-1; Abcam) and sheep polyclonal antibody to showed that in the absence of any artificial selection process, a type. Our approach thus constitutes a new proof-of-concept with a MOI 1 with the pLox.TERT.ires.TK LV (Salmon et al, 2000) for mouse IgG—H&L (HRP; ab 6808, Abcam) were used as secondary progressive increase in the proportion of targeted cells (up to 40% potential future clinical impact to optimize the generation of gene- 24 h. To excise the reprogramming cassette and hTERT from estab- antibodies. Protein quantification was done with Image J soft- after 42 days in culture) was observed, mimicking the improved corrected HSCs from non-hematopoietic tissues of patients with lished hiPSCs, single cell suspensions were generated by incubation ware. growth proliferation properties of FA precursor cells in mosaic inherited diseases, including DNA repair deficiency and genetic with accutase (Gibco) and transduced for 10 h with the IDLV patients (Waisfisz et al, 1999; Gregory et al, 2001; Gross et al, instability syndromes, like FA. pLM.CMV.Cherry.2A.Cre. Immediately after transduction, FANCA expression by qRT-PCR 2002) or in experimental models of FA gene therapy (Rio et al, 2 × 104 cells/10 cm2 dish, expressing Cherry protein, were sorted 2008). Consistent with previous observations in FA cells corrected and new subclones of the parental geFA-IPSCs were generated. The expression of human FANCA mRNA was analyzed in the differ- by LV-mediated gene therapy (Raya et al, 2009), this proliferation Materials and Methods ent clones of geFA-iPSCs by real-time quantitative reverse transcrip- competence of FA-corrected cells was particularly remarkable when Hematopoietic differentiation tase-polymerase chain reaction (qRT-PCR; Gonzalez-Murillo et al, samples were subjected to cell reprogramming, confirming the rele- Cell lines and primary fibroblasts from FA-A patients 2010) using primers described in Supplementary Methods. Parental vance of the FA pathway during the process of iPSC generation. iPSC colonies were detached using colagenase type IV (Gibco) for fibroblasts from FA-52 and ES H9 were used as controls. Similar conclusions were obtained in two additional studies (Muller 293T and HT1080 cells (ATCC: CRL-11268 and ATCC: CCL-121) 30 min at 37°C, washed and centrifuged at 200× g, resuspended in et al, 2012; Yung et al, 2013), although these studies showed that were used for the production and titration of the LVs, respec- differentiation media composed by KO-DMEM (Gibco) supple- Gene targeting analysis: PCR and Southern blots reprogramming of FA cells can occur, albeit with a very low effi- tively. Cells were grown in Dulbecco’s modified medium mented with 20% non-heat-inactivated FBS (Biowhitaker), 1% ciency compared to gene-complemented FA cells. GlutaMAXTM (DMEM; Gibco) supplemented with 10% fetal bovine NEAA (Lonza; Biowhitaker), L-Glu (1 mM; Invitrogen), b-mercap- For PCR analysis, genomic DNA was extracted with DNeasy Blood Studies in Figs 2 and 4 showing the generation of nuclear serum (FBS, Biowhitaker) and 0.5% penicillin/streptomycin solu- toethanol (0.1 mM; Gibco) and hrBMP4 (0.5 ng/ml; Prepotech) & Tissue Kit (Qiagen). To detect the targeted integration of the HDR

completely corrected the phenotype of FA-A fibroblasts and bona cillin/streptomycin solution (Gibco) at 37°C under hypoxic (40 mM; Sigma), ascorbic acid (50 lg/ml; Invitrogen), hrSCF, 2 min at 94°C, 40 cycles of 30 s at 94°C, 30 s at 58°C (50 TI) and

fide iPSCs. Although transduction of FA fibroblasts with the conditions (5% of O2) and 5% of CO2. Patients were classified as hrFlt3 ligand and TPO (100 ng/ml; EuroBioSciences), hrIL3 59°C (30 TI), 1 min at 72°C and one final step for 5 min at 72°C. The hTERT-LV might have had consequences upon the genetic insta- FA-A patients as previously described (Casado et al, 2007). The (10 ng/ml; Biosource), hrIL6 (10 ng/ml; Prepotech), hrBMP4 proper target integration amplified a 1195 pb amplicon for the 50 TI bility of FA cells, our karyotype and aCGH studies indicate that ES4 and H9 (NIH Human Embryonic Stem Cell Registry, http:// (50 ng/ml; Prepotech), Wnt11 (200 ng/ml; R&D), and rhVEGF and a 1314 pb fragment for the 30 TI that were resolved in agarose neither the expansion nor the transduction with hTERT-LV or the stemcells.nih.gov/research/registry/) lines of hES cells were main- (5 ng/ml; Prepotech). Media were changed every 3–4 days. At gel at 2%. For Southern blot analyses, genomic DNA was extracted gene-editing processes induced evident chromosomal abnormali- tained as originally described (Raya et al, 2008). FA patients and day 7, media were replaced by fresh media where rhWnt-11 was and digested either with BstXI enzyme or with BglI (both from New ties in FA fibroblasts. In contrast to these results, data in Table 1 healthy donors were encoded to protect their confidentiality, and substituted by rhWnt-3a (200 ng/ml; R&D). Media were changed England Biolabs). Matched DNA amounts were separated on 0.8% and Supplementary Fig S7 showed the presence of chromosomal informed consents were obtained in all cases according to Institu- every 3–4 days. At day 14 and 21, immunophenotypic analysis of agarose gel, transferred to a nylon membrane (Hybond XL, GE abnormalities in reprogrammed and excised geFA-iPSCs. Impor- tional regulations of the CIEMAT. All studies conformed the prin- the differentiated cells was performed by flow cytometry, and Healthcare) and probed either with the 32P-radiolabeled sequence of tantly, different genetic defects have also been reported in non- ciples set out in the World Medical Association Declaration of colony-forming unit assays were conducted (See Supplementary a fragment of EGFP to detect specific (5.1 kb) and non-specific inte- FA-iPSCs (Mayshar et al, 2010; Gore et al, 2011; Laurent et al, Helsinki. Methods). grations or with a probe of AAVS1 gene located outside of the

844 EMBO Molecular Medicine Vol 6 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO Molecular Medicine Vol 6 | No 6 | 2014 845 EMBO Molecular Medicine Targeted gene therapy in Fanconi anemia Paula Rio et al Paula Rio et al Targeted gene therapy in Fanconi anemia EMBO Molecular Medicine

integration of the cassette, strongly suggesting that AAVS1 target- 2011; Cheng et al, 2012; Ruiz et al, 2013) that were associated Vectors Flow cytometry ing took place through a HDR mechanism. In this respect, while with the generation of the iPSCs (Mayshar et al, 2010; Gore et al, the specificity of gene targeting might be reduced in FA cells, our 2011; Hussein et al, 2011; Laurent et al, 2011) and/or with muta- pCCL.sin.cPPT.AAVS1.loxP.SA.2A.GFP.pA.loxP.PGK.FANCA.pA.Wpre Transduction with the AdV5/35-ZFNs and the donor IDLV, was data clearly show that all the FA-iPSC clones harbored one single tions that pre-existed in the somatic population of origin (Young donor transfer LV (donor IDLV) was generated using elements analyzed by flow cytometry analysis (FACSCalibur; Becton copy of FANCA specifically integrated in the PPPR12C target gene et al, 2012). This indicates that the presence of chromosomal from the backbones pCCL.PGK.FANCA.Wpre* (Gonzalez-Murillo Dickinson Pharmingen). Immunophenotypic analysis of the hemato- (Table 1). Consequently, this result further supports the efficacy abnormalities in our iPSCs is not exclusive of their FA genetic et al, 2010) and pCCLsin.cPPT.AAVS1.2A.GFP.pA (Lombardo poietic differentiated cells was performed using the following anti- and the specificity of our gene targeting approach. background and that the different mechanisms accounting for et al, 2011). The integrase-defective third-generation packaging bodies according to the manufacturer’s instructions: phycoerythrin With the main objective of preventing the predisposition to mutations in non-FA-iPSCs would be applicable to our geFA- plasmid pMD.Lg/pRRE.D64Vint was used to produce IDLV parti- (PE)-Cy7-conjugated anti-human CD34 (BD Pharmingen), PE-conju- senescence of FA cells (Muller et al, 2012), the transduction of iPSCs. cles (Lombardo et al, 2007). pLM.CMV.Cherry.2A.Cre (Papapetrou gated anti-human CD31 (eBiosciences), allophycocyanin (APC)- hTERT-LV in FA-A fibroblasts induced an unexpected effect in these Consistent with the previous study showing the generation of et al, 2011) and pLox.TERT.ires.TK vectors (Salmon et al, 2000) conjugated anti-human CD45 (BD), and fluorescein isothiocyanate cells, which consisted of a significant increase in the efficacy of gene disease-free FA-iPSCs through conventional gene therapy were provided by Addgene. For reprogramming experiments, (FITC)-conjugated anti-human CD43 (BD). Fluorochrome-matched editing (Fig 1). Whether or not this effect is specific for FA cells or approaches (Raya et al, 2009; Muller et al, 2012), our new study the EF1a STEMCCA lentiviral vector kindly provided by isotypes were used as controls. 40,6-Diamidino-2-phenylindole whether it is simply mediated by the enhanced proliferation rate of shows the efficient hematopoietic differentiation of gene-edited FA- Dr Mostoslavsky was used (Sommer et al, 2010). This vector (DAPI; Roche)-positive cells were excluded from the analysis. TERT-transduced FA cells is currently unknown. Nevertheless, to iPSCs. Moreover in the current study, we observed the generation of contains the cDNAs for OCT4, SOX2, c-MYC, and KLF4 flanked by Analysis was performed using FlowJo software. the best of our knowledge, the improved gene targeting mediated by increased numbers of hematopoietic progenitors from geFA-iPSCs loxP sequences for their subsequent excision. ZFNs targeting hTERT observed in our experiments constitutes a new finding that subjected to excision of the reprogramming cassette, confirming intron 1 of the PPP1R12C gene were expressed from an Adenoviral Inmunofluorescence and Western blot of Fanconi has not been previously reported in any other experimental model. previous observations showing that the residual expression of repro- Vector (AdV5/35) under the control of the CMV promoter (Lom- anemia proteins The observation that transduction with hTERT also facilitates the gramming genes limits the iPSC differentiation potential (Ramos- bardo et al, 2011). generation of gene-edited FA-iPSCs is consistent with previous data Mejia et al, 2012). The hematopoietic differentiation observed in Analyses of FANCD2 foci were performed by immunofluorescence showing the relevance of hTERT in cell reprogramming (Batista these experiments and the robust expression of FANCA targeted into Cell transduction of primary fibroblasts or iPSCs treated for 16 h with 200 nM of et al, 2011; Pomp et al, 2011; Winkler et al, 2013). In safety terms, the safe harbor AAVS1 locus should account for the generation of a MMC. After MMC treatment, cells were stained with rabbit poly- even though the hTERT provirus could be efficiently excised from high number of hematopoietic progenitors with normalized For gene editing experiments, fibroblasts from FA-A patients were clonal anti-FANCD2 (Abcam, ab2187-50) as previously described transduced cells with the Cre recombinase, further approaches response to MMC. transduced either with donor IDLV alone (150 ng HIV Gag p24/ml) (Hotta & Ellis, 2008; Raya et al, 2009). Cells with more than ten based on the transient expression of hTERT during gene editing In summary, our study demonstrates for the first time the possi- or together with AdV5/35-ZFNs (multiplicity of infection (MOI) foci were scored as positive. FANCA expression was analyzed by and/or cell reprogramming would constitute safer approaches to bility of conducting efficient and precise targeted-mediated gene 200). Fourteen days post-transduction, the proportion of EGFP+ Western blot (Raya et al, 2009) using the following antibodies: limit potential genomic insults during the ex vivo manipulation of therapy in HDR-deficient cells. Moreover, we show the feasibility of cells was determined by flow cytometry (BD LSRFortessa cell hFANCA (ab5063 Abcam) and anti-beta Actin to mouse antibody the samples. reprogramming these cells to generate iPSC-derived gene-edited analyzer, Becton Dickinson Pharmingen). To immortalize fibro- (ab6276, Abcam) as control. Goat polyclonal antibody to rabbit Interestingly, EGFP analyses in gene-edited FA fibroblasts hematopoietic progenitors characterized by a disease-free pheno- blasts from FA-52 to FA-123 patients, 105 cells were transduced at IgG (HRP; ab6721-1; Abcam) and sheep polyclonal antibody to showed that in the absence of any artificial selection process, a type. Our approach thus constitutes a new proof-of-concept with a MOI 1 with the pLox.TERT.ires.TK LV (Salmon et al, 2000) for mouse IgG—H&L (HRP; ab 6808, Abcam) were used as secondary progressive increase in the proportion of targeted cells (up to 40% potential future clinical impact to optimize the generation of gene- 24 h. To excise the reprogramming cassette and hTERT from estab- antibodies. Protein quantification was done with Image J soft- after 42 days in culture) was observed, mimicking the improved corrected HSCs from non-hematopoietic tissues of patients with lished hiPSCs, single cell suspensions were generated by incubation ware. growth proliferation properties of FA precursor cells in mosaic inherited diseases, including DNA repair deficiency and genetic with accutase (Gibco) and transduced for 10 h with the IDLV patients (Waisfisz et al, 1999; Gregory et al, 2001; Gross et al, instability syndromes, like FA. pLM.CMV.Cherry.2A.Cre. Immediately after transduction, FANCA expression by qRT-PCR 2002) or in experimental models of FA gene therapy (Rio et al, 2 × 104 cells/10 cm2 dish, expressing Cherry protein, were sorted 2008). Consistent with previous observations in FA cells corrected and new subclones of the parental geFA-IPSCs were generated. The expression of human FANCA mRNA was analyzed in the differ- by LV-mediated gene therapy (Raya et al, 2009), this proliferation Materials and Methods ent clones of geFA-iPSCs by real-time quantitative reverse transcrip- competence of FA-corrected cells was particularly remarkable when Hematopoietic differentiation tase-polymerase chain reaction (qRT-PCR; Gonzalez-Murillo et al, samples were subjected to cell reprogramming, confirming the rele- Cell lines and primary fibroblasts from FA-A patients 2010) using primers described in Supplementary Methods. Parental vance of the FA pathway during the process of iPSC generation. iPSC colonies were detached using colagenase type IV (Gibco) for fibroblasts from FA-52 and ES H9 were used as controls. Similar conclusions were obtained in two additional studies (Muller 293T and HT1080 cells (ATCC: CRL-11268 and ATCC: CCL-121) 30 min at 37°C, washed and centrifuged at 200× g, resuspended in et al, 2012; Yung et al, 2013), although these studies showed that were used for the production and titration of the LVs, respec- differentiation media composed by KO-DMEM (Gibco) supple- Gene targeting analysis: PCR and Southern blots reprogramming of FA cells can occur, albeit with a very low effi- tively. Cells were grown in Dulbecco’s modified medium mented with 20% non-heat-inactivated FBS (Biowhitaker), 1% ciency compared to gene-complemented FA cells. GlutaMAXTM (DMEM; Gibco) supplemented with 10% fetal bovine NEAA (Lonza; Biowhitaker), L-Glu (1 mM; Invitrogen), b-mercap- For PCR analysis, genomic DNA was extracted with DNeasy Blood Studies in Figs 2 and 4 showing the generation of nuclear serum (FBS, Biowhitaker) and 0.5% penicillin/streptomycin solu- toethanol (0.1 mM; Gibco) and hrBMP4 (0.5 ng/ml; Prepotech) & Tissue Kit (Qiagen). To detect the targeted integration of the HDR

FANCD2 foci and the chromosomal stability of gene-edited FA tion (Gibco). Skin fibroblasts were obtained from FA-5, FA-123, and plated in ultra-low attachment plates (Costar). After 2 days, cassette in the AAVS1 locus, two different pair of primers for the 30 fibroblasts and iPSCs upon exposure to ICL drugs demonstrate FA-664, and FA-52 patients and were maintained in DMEM (Invi- media were replaced by Stempro 34 (Invitrogen) supplemented or the 50 integration junction (50 TI and 30 TI, respectively) were that the specific targeting of FANCA in the AAVS1 locus has trogen) supplemented with 20% FBS (Biowhitaker) and 1% peni- with 0.5% pen/streptomicin, L-Glu (2 mM; Invitrogen), MTG used (Supplementary Table S2). PCR was conducted as follows: completely corrected the phenotype of FA-A fibroblasts and bona cillin/streptomycin solution (Gibco) at 37°C under hypoxic (40 mM; Sigma), ascorbic acid (50 lg/ml; Invitrogen), hrSCF, 2 min at 94°C, 40 cycles of 30 s at 94°C, 30 s at 58°C (50 TI) and fide iPSCs. Although transduction of FA fibroblasts with the conditions (5% of O2) and 5% of CO2. Patients were classified as hrFlt3 ligand and TPO (100 ng/ml; EuroBioSciences), hrIL3 59°C (30 TI), 1 min at 72°C and one final step for 5 min at 72°C. The hTERT-LV might have had consequences upon the genetic insta- FA-A patients as previously described (Casado et al, 2007). The (10 ng/ml; Biosource), hrIL6 (10 ng/ml; Prepotech), hrBMP4 proper target integration amplified a 1195 pb amplicon for the 50 TI bility of FA cells, our karyotype and aCGH studies indicate that ES4 and H9 (NIH Human Embryonic Stem Cell Registry, http:// (50 ng/ml; Prepotech), Wnt11 (200 ng/ml; R&D), and rhVEGF and a 1314 pb fragment for the 30 TI that were resolved in agarose neither the expansion nor the transduction with hTERT-LV or the stemcells.nih.gov/research/registry/) lines of hES cells were main- (5 ng/ml; Prepotech). Media were changed every 3–4 days. At gel at 2%. For Southern blot analyses, genomic DNA was extracted gene-editing processes induced evident chromosomal abnormali- tained as originally described (Raya et al, 2008). FA patients and day 7, media were replaced by fresh media where rhWnt-11 was and digested either with BstXI enzyme or with BglI (both from New ties in FA fibroblasts. In contrast to these results, data in Table 1 healthy donors were encoded to protect their confidentiality, and substituted by rhWnt-3a (200 ng/ml; R&D). Media were changed England Biolabs). Matched DNA amounts were separated on 0.8% and Supplementary Fig S7 showed the presence of chromosomal informed consents were obtained in all cases according to Institu- every 3–4 days. At day 14 and 21, immunophenotypic analysis of agarose gel, transferred to a nylon membrane (Hybond XL, GE abnormalities in reprogrammed and excised geFA-iPSCs. Impor- tional regulations of the CIEMAT. All studies conformed the prin- the differentiated cells was performed by flow cytometry, and Healthcare) and probed either with the 32P-radiolabeled sequence of tantly, different genetic defects have also been reported in non- ciples set out in the World Medical Association Declaration of colony-forming unit assays were conducted (See Supplementary a fragment of EGFP to detect specific (5.1 kb) and non-specific inte- FA-iPSCs (Mayshar et al, 2010; Gore et al, 2011; Laurent et al, Helsinki. Methods). grations or with a probe of AAVS1 gene located outside of the

homology arm (in the 30 region) to detect specific integration in the Hum Genet 92: 800 – 806 targeted genome editing. Nat Rev Mol Cell Biol 14: 49 – 55 Porteus MH, Baltimore D (2003) Chimeric nucleases stimulate gene targeting proper target locus (9.6 kb) and the unmodified AAVS1 locus Casado JA, Callen E, Jacome A, Rio P, Castella M, Lobitz S, Ferro T, Munoz A, Kee Y, D’Andrea AD (2010) Expanded roles of the Fanconi anemia pathway in in human cells. Science 300: 763 (3.3 kb). To detect the radiolabel signal, auto-radiographic films Sevilla J, Cantalejo A et al (2007) A comprehensive strategy for the preserving genomic stability. Genes Dev 24: 1680 – 1694 Ramos-Mejia V, Montes R, Bueno C, Ayllon V, Real PJ, Rodriguez R, Menendez were used (Amershan Hyperfilm ECL, GE Healthcare) and they were subtyping of Fanconi Anemia patients: conclusions from the Spanish Kottemann MC, Smogorzewska A (2013) Fanconi anaemia and the repair of P(2012) Residual expression of the reprogramming factors prevents exposed in an automatic reveal machine Curix60 (AGFA). Fanconi Anemia research network. J Med Genet 44: 241 – 249 Watson and Crick DNA crosslinks. Nature 493: 356 – 363 differentiation of iPSC generated from human fibroblasts and cord blood Castella M, Pujol R, Callen E, Trujillo JP, Casado JA, Gille H, Lach FP, Larghero J, Marolleau JP, Soulier J, Filion A, Rocha V, Benbunan M, Gluckman CD34+ progenitors. PLoS ONE 7:e35824 Supplementary information for this article is available online: Auerbach AD, Schindler D, Benitez J et al (2011) Origin, functional role E(2002) Hematopoietic progenitor cell harvest and functionality in Raya A, Rodriguez-Piza I, Aran B, Consiglio A, Barri PN, Veiga A, Izpisua http://embomolmed.embopress.org and clinical impact of Fanconi anemia FANCA mutations. Blood 117: Fanconi anemia patients. Blood 100: 3051 Belmonte JC (2008) Generation of cardiomyocytes from new human 3759 – 3769 Laurent LC, Ulitsky I, Slavin I, Tran H, Schork A, Morey R, Lynch C, Harness JV, embryonic stem cell lines derived from poor-quality blastocysts. Cold Acknowledgements Chang CJ, Bouhassira EE (2012) Zinc-finger nuclease mediated correction of Lee S, Barrero MJ et al (2011) Dynamic changes in the copy number of Spring Harb Symp Quant Biol 73: 127 – 135 The authors would like to thank Prof. Juan C. Izpisua-Belmonte and Dr Guiller- alpha-thalassemia in iPS cells. Blood 120: 3906 – 3914 pluripotency and cell proliferation genes in human ESCs and iPSCs during Raya A, Rodriguez-Piza I, Guenechea G, Vassena R, Navarro S, Barrero MJ, mo Guenechea for helpful discussions; Laura Cerrato for technical assistance Cheng L, Hansen NF, Zhao L, Du Y, Zou C, Donovan FX, Chou BK, Zhou G, Li reprogramming and time in culture. Cell Stem Cell 8: 106 – 118 Consiglio A, Castella M, Rio P, Sleep E et al (2009) Disease-corrected with iPSCs; and Aurora de la Cal for coordination with the FA Network. We are S, Dowey SN et al (2012) Low incidence of DNA sequence variation in Lombardo A, Genovese P, Beausejour CM, Colleoni S, Lee YL, Kim KA, Ando D, haematopoietic progenitors from Fanconi anaemia induced pluripotent also indebted to the FA patients, families, and clinicians from the FA network. human induced pluripotent stem cells generated by nonintegrating Urnov FD, Galli C, Gregory PD et al (2007) Gene editing in human stem stem cells. Nature 460: 53 – 59 This work was supported by grants to J.A.B. from the European Union (FP7 GA plasmid expression. Cell Stem Cell 10: 337 – 344 cells using zinc finger nucleases and integrase-defective lentiviral vector Rio P, Meza NW, Gonzalez-Murillo A, Navarro S, Alvarez L, Surralles J, Castella 222878 PERSIST), Spanish Ministry of Economy and Competitiveness (Interna- Cong L, Ran FA, Cox D, Lin S, Barretto R, Habib N, Hsu PD, Wu X, Jiang W, delivery. Nat Biotechnol 25: 1298 – 1306 M, Guenechea G, Segovia JC, Hanenberg H et al (2008) In vivo tional Cooperation on Stem Cell Research Plan E; Ref PLE 2009/0100; SAF Marraffini LA et al (2013) Multiplex genome engineering using CRISPR/Cas Lombardo A, Cesana D, Genovese P, Di Stefano B, Provasi E, Colombo DF, Neri proliferation advantage of genetically corrected hematopoietic stem cells 2009-07164 and SAF 2012-39834), Fondo de Investigaciones Sanitarias, Institu- systems. Science 339: 819 – 823 M, Magnani Z, Cantore A, Lo Riso P et al (2011) Site-specific integration in a mouse model of Fanconi anemia FA-D1. Blood 112: 4853 – 4861 to de Salud Carlos III (RETICS-RD06/0010/0015 and RD12/0019/0023), Direc- DeKelver RC, Choi VM, Moehle EA, Paschon DE, Hockemeyer D, Meijsing SH, and tailoring of cassette design for sustainable gene transfer. Nat Methods Rosenberg PS, Alter BP, Ebell W (2008) Cancer risks in Fanconi anemia: ción General de Investigación de la Comunidad de Madrid (CellCAM; Ref Sancak Y, Cui X, Steine EJ, Miller JC et al (2010) Functional genomics, 8: 861 – 869 findings from the German Fanconi Anemia Registry. Haematologica 93: S2010/BMD-2420), and La Fundació Privada La Marató de TV3, 121430/31/32; proteomics, and regulatory DNA analysis in isogenic settings using zinc Matrai J, Cantore A, Bartholomae CC, Annoni A, Wang W, Acosta-Sanchez A, 511 – 517 to J.S. from the Generalitat de Catalunya (SGR0489-2009), the ICREA-Academia finger nuclease-driven transgenesis into a safe harbor locus in the human Samara-Kuko E, De Waele L, Ma L, Genovese P et al (2011) Ruiz S, Gore A, Li Z, Panopoulos AD, Montserrat N, Fung HL, Giorgetti A, Bilic program, the Marató de TV3 (464/C/2012), the Spanish Ministry of Science and genome. Genome Res 20: 1133 – 1142 Hepatocyte-targeted expression by integrase-defective lentiviral vectors J, Batchelder EM, Zaehres H et al (2013) Analysis of protein-coding Innovation (SAF2012-31881), the European Commission (HEALTH-F5-2012- Garcia-Higuera I, Taniguchi T, Ganesan S, Meyn MS, Timmers C, Hejna J, induces antigen-specific tolerance in mice with low genotoxic risk. mutations in hiPSCs and their possible role during somatic cell 305421), and the European Regional Development FEDER Funds; to L.N. from Grompe M, D’Andrea AD (2001) Interaction of the Fanconi anemia Hepatology 53: 1696 – 1707 reprogramming. Nat Commun 4: 1382 Telethon (TIGET grant D2), European Union (FP7 GA 222878 PERSIST, ERC proteins and BRCA1 in a common pathway. Mol Cell 7: Mayshar Y, Ben-David U, Lavon N, Biancotti JC, Yakir B, Clark AT, Plath K, Salmon P, Kindler V, Ducrey O, Chapuis B, Zubler RH, Trono D (2000) Advanced Grant 249845 TARGETINGGENETHERAPY) and the Italian Ministry of 249 – 262 Lowry WE, Benvenisty N (2010) Identification and classification of High-level transgene expression in human hematopoietic progenitors and

846 EMBO Molecular Medicine Vol 6 | No 6 | 2014 ª 2014 The Authors ª 2014 The Authors EMBO Molecular Medicine Vol 6 | No 6 | 2014 847 EMBO Molecular Medicine Targeted gene therapy in Fanconi anemia Paula Rio et al Paula Rio et al Targeted gene therapy in Fanconi anemia EMBO Molecular Medicine

Health. The authors also thank the Fundación Marcelino Botín for promoting Gonzalez-Murillo A, Lozano ML, Alvarez L, Jacome A, Almarza E, Navarro S, chromosomal aberrations in human induced pluripotent stem cells. Cell The paper explained translational research at the Hematopoietic Innovative Therapies Division of Segovia JC, Hanenberg H, Guenechea G, Bueren JA et al (2010) Stem Cell 7: 521 – 531 the CIEMAT. CIBERER is an initiative of the Instituto de Salud Carlos III, Spain. Development of lentiviral vectors with optimized transcriptional activity Moldovan GL, D’Andrea AD (2009) How the Fanconi anemia pathway guards Problem Gene targeting is becoming a true alternative to conventional gene for the gene therapy of patients with Fanconi anemia. Hum Gene Ther 21: the genome. Annu Rev Genet 43: 223 – 249 therapy with integrative gammaretroviral or lentiviral vectors. It is Author contributions 623 – 630 Muller LU, Milsom MD, Harris CE, Vyas R, Brumme KM, Parmar K, Moreau LA, however unknown whether these approaches would be applicable to Contribution: PR, RB, AL, LN, and JAB conceived and designed the experiments. Gore A, Li Z, Fung HL, Young JE, Agarwal S, Antosiewicz-Bourget J, Canto I, Schambach A, Park IH, London WB et al (2012) Overcoming inherited syndromes like FA, characterized by homology-directed DNA PR, RB, AL, OQ-B, LA, ZG, PG, EA, AV, BD, SN, YT, JPT, and RM conducted experi- Giorgetti A, Israel MA, Kiskinis E et al (2011) Somatic coding mutations in reprogramming resistance of Fanconi anemia cells. Blood 119: 5449 – 5457 repair (HDR) defects. Additionally, the existence of 16 different FA ments. JCS, ES, JS, PDG, and MCH provided reagents, tools, and ideas. PR, RB, human induced pluripotent stem cells. Nature 471: 63 – 67 Nakanishi K, Yang YG, Pierce AJ, Taniguchi T, Digweed M, D’Andrea AD, Wang genes, each of them with multiple mutations potentially accounting for the disease, would imply the necessity of developing individualized AL, LN, and JAB wrote the paper. Gregory JJ Jr, Wagner JE, Verlander PC, Levran O, Batish SD, Eide CR, ZQ, Jasin M (2005) Human Fanconi anemia monoubiquitination pathway targeted gene therapy strategies in FA patients. Steffenhagen A, Hirsch B, Auerbach AD (2001) Somatic mosaicism in promotes homologous DNA repair. Proc Natl Acad Sci USA 102: 1110 – 1115 Conflict of interest Fanconi anemia: evidence of genotypic reversion in lymphohematopoietic Nakanishi K, Cavallo F, Perrouault L, Giovannangeli C, Moynahan ME, Barchi Results P.D.G. and M.C.H. are current or former employees of Sangamo BioSciences, stem cells. Proc Natl Acad Sci USA 98: 2532 – 2537 M, Brunet E, Jasin M (2011) Homology-directed Fanconi anemia pathway We have demonstrated for the first time an efficient and specific Inc. The rest of the authors declare that they have no conflict of interest. Gross M, Hanenberg H, Lobitz S, Friedl R, Herterich S, Dietrich R, Gruhn B, cross-link repair is dependent on DNA replication. Nat Struct Mol Biol 18: targeting of FANCA in the AAVS1 safe harbor locus of FA-A patients’ Schindler D, Hoehn H (2002) Reverse mosaicism in Fanconi anemia: 500 – 503 fibroblasts. This approach allowed us to develop a gene-editing platform applicable to all FA subtypes and FA gene mutations based on For more information natural gene therapy via molecular self-correction. Cytogenet Genome Res Naldini L (2011) Ex vivo gene transfer and correction for cell-based therapies. the insertion of the therapeutic FA gene in a safe harbor locus. More- Fanconi Anemia Research Foundation: www.fanconi.org. 98: 126 – 135 Nat Rev Genet 12: 301 – 315 over, gene-edited FA-A fibroblasts were reprogrammed to generate Hockemeyer D, Soldner F, Beard C, Gao Q, Mitalipova M, DeKelver RC, Katibah Niedzwiedz W, Mosedale G, Johnson M, Ong CY, Pace P, Patel KJ (2004) The disease-free iPSCs, which could be re-differentiated toward the hema- GE, Amora R, Boydston EA, Zeitler B et al (2009) Efficient targeting of Fanconi anaemia gene FANCC promotes homologous recombination and topoietic lineage in a process that resulted in the generation of gene- References expressed and silent genes in human ESCs and iPSCs using zinc-finger error-prone DNA repair. Mol Cell 15: 607 – 620 edited, disease-free, hematopoietic progenitor cells. nucleases. Nat Biotechnol 27: 851 – 857 Papapetrou EP, Lee G, Malani N, Setty M, Riviere I, Tirunagari LM, Kadota K, Anguela XM, Sharma R, Doyon Y, Miller JC, Li H, Haurigot V, Rohde ME, Wong Roth SL, Giardina P, Viale A et al (2011) Genomic safe harbors permit high Impact Hotta A, Ellis J (2008) Retroviral vector silencing during iPS cell induction: an Our data showing that gene targeting is feasible in FA opens the SY, Davidson RJ, Zhou S et al (2014) Robust ZFN-mediated genome epigenetic beacon that signals distinct pluripotent states. J Cell Biochem beta-globin transgene expression in thalassemia induced pluripotent stem possibility of using similar strategies in different inherited syndromes editing in adult hemophilic mice. Blood 122: 3283 – 3287 105: 940 – 948 cells. Nat Biotechnol 29: 73 – 78 characterized by defects in HDR and genome instability. The genera- Auerbach AD (2009) Fanconi anemia and its diagnosis. Mutat Res 668: 4 – 10 Hussein SM, Batada NN, Vuoristo S, Ching RW, Autio R, Narva E, Ng S, Peluffo H, Foster E, Ahmed SG, Lago N, Hutson TH, Moon L, Yip P, Wanisch K, tion of disease-free HSCs through the specific insertion of therapeutic Batista LF, Pech MF, Zhong FL, Nguyen HN, Xie KT, Zaug AJ, Crary SM, Choi J, Sourour M, Hamalainen R, Olsson C et al (2011) Copy number variation Caraballo-Miralles V, Olmos G et al (2013) Efficient gene expression from transgenes in a safe harbor locus of non-hematopoietic cell tissues, Sebastiano V, Cherry A et al (2011) Telomere shortening and loss of and selection during reprogramming to pluripotency. Nature 471: 58 – 62 integration-deficient lentiviral vectors in the spinal cord. Gene Ther 20: additionally constitutes an implemented approach to overcome HSC defects characteristic of many DNA repair deficiency syndromes, like self-renewal in dyskeratosis congenita induced pluripotent stem cells. Jacome A, Navarro S, Casado JA, Rio P, Madero L, Estella J, Sevilla J, Badell I, 645 – 657 Fanconi anemia. Nature 474: 399 – 402 Ortega JJ, Olive T et al (2006) A simplified approach to improve the Pomp O, Dreesen O, Leong DF, Meller-Pomp O, Tan TT, Zhou F, Colman A Bogliolo M, Schuster B, Stoepker C, Derkunt B, Su Y, Raams A, Trujillo JP, efficiency and safety of ex vivo hematopoietic gene therapy in fanconi (2011) Unexpected X chromosome skewing during culture and Minguillon J, Ramirez MJ, Pujol R et al (2013) Mutations in ERCC4, anemia patients. Hum Gene Ther 17: 245 – 250 reprogramming of human somatic cells can be alleviated by exogenous encoding the DNA-repair endonuclease XPF, cause Fanconi anemia. Am J Joung JK, Sander JD (2013) TALENs: a widely applicable technology for telomerase. Cell Stem Cell 9: 156 – 165 homology arm (in the 30 region) to detect specific integration in the Hum Genet 92: 800 – 806 targeted genome editing. Nat Rev Mol Cell Biol 14: 49 – 55 Porteus MH, Baltimore D (2003) Chimeric nucleases stimulate gene targeting proper target locus (9.6 kb) and the unmodified AAVS1 locus Casado JA, Callen E, Jacome A, Rio P, Castella M, Lobitz S, Ferro T, Munoz A, Kee Y, D’Andrea AD (2010) Expanded roles of the Fanconi anemia pathway in in human cells. Science 300: 763 (3.3 kb). To detect the radiolabel signal, auto-radiographic films Sevilla J, Cantalejo A et al (2007) A comprehensive strategy for the preserving genomic stability. Genes Dev 24: 1680 – 1694 Ramos-Mejia V, Montes R, Bueno C, Ayllon V, Real PJ, Rodriguez R, Menendez were used (Amershan Hyperfilm ECL, GE Healthcare) and they were subtyping of Fanconi Anemia patients: conclusions from the Spanish Kottemann MC, Smogorzewska A (2013) Fanconi anaemia and the repair of P(2012) Residual expression of the reprogramming factors prevents exposed in an automatic reveal machine Curix60 (AGFA). Fanconi Anemia research network. J Med Genet 44: 241 – 249 Watson and Crick DNA crosslinks. Nature 493: 356 – 363 differentiation of iPSC generated from human fibroblasts and cord blood Castella M, Pujol R, Callen E, Trujillo JP, Casado JA, Gille H, Lach FP, Larghero J, Marolleau JP, Soulier J, Filion A, Rocha V, Benbunan M, Gluckman CD34+ progenitors. PLoS ONE 7:e35824 Supplementary information for this article is available online: Auerbach AD, Schindler D, Benitez J et al (2011) Origin, functional role E(2002) Hematopoietic progenitor cell harvest and functionality in Raya A, Rodriguez-Piza I, Aran B, Consiglio A, Barri PN, Veiga A, Izpisua http://embomolmed.embopress.org and clinical impact of Fanconi anemia FANCA mutations. Blood 117: Fanconi anemia patients. Blood 100: 3051 Belmonte JC (2008) Generation of cardiomyocytes from new human 3759 – 3769 Laurent LC, Ulitsky I, Slavin I, Tran H, Schork A, Morey R, Lynch C, Harness JV, embryonic stem cell lines derived from poor-quality blastocysts. Cold Acknowledgements Chang CJ, Bouhassira EE (2012) Zinc-finger nuclease mediated correction of Lee S, Barrero MJ et al (2011) Dynamic changes in the copy number of Spring Harb Symp Quant Biol 73: 127 – 135 The authors would like to thank Prof. Juan C. Izpisua-Belmonte and Dr Guiller- alpha-thalassemia in iPS cells. Blood 120: 3906 – 3914 pluripotency and cell proliferation genes in human ESCs and iPSCs during Raya A, Rodriguez-Piza I, Guenechea G, Vassena R, Navarro S, Barrero MJ, mo Guenechea for helpful discussions; Laura Cerrato for technical assistance Cheng L, Hansen NF, Zhao L, Du Y, Zou C, Donovan FX, Chou BK, Zhou G, Li reprogramming and time in culture. Cell Stem Cell 8: 106 – 118 Consiglio A, Castella M, Rio P, Sleep E et al (2009) Disease-corrected with iPSCs; and Aurora de la Cal for coordination with the FA Network. We are S, Dowey SN et al (2012) Low incidence of DNA sequence variation in Lombardo A, Genovese P, Beausejour CM, Colleoni S, Lee YL, Kim KA, Ando D, haematopoietic progenitors from Fanconi anaemia induced pluripotent also indebted to the FA patients, families, and clinicians from the FA network. human induced pluripotent stem cells generated by nonintegrating Urnov FD, Galli C, Gregory PD et al (2007) Gene editing in human stem stem cells. Nature 460: 53 – 59 This work was supported by grants to J.A.B. from the European Union (FP7 GA plasmid expression. Cell Stem Cell 10: 337 – 344 cells using zinc finger nucleases and integrase-defective lentiviral vector Rio P, Meza NW, Gonzalez-Murillo A, Navarro S, Alvarez L, Surralles J, Castella 222878 PERSIST), Spanish Ministry of Economy and Competitiveness (Interna- Cong L, Ran FA, Cox D, Lin S, Barretto R, Habib N, Hsu PD, Wu X, Jiang W, delivery. Nat Biotechnol 25: 1298 – 1306 M, Guenechea G, Segovia JC, Hanenberg H et al (2008) In vivo tional Cooperation on Stem Cell Research Plan E; Ref PLE 2009/0100; SAF Marraffini LA et al (2013) Multiplex genome engineering using CRISPR/Cas Lombardo A, Cesana D, Genovese P, Di Stefano B, Provasi E, Colombo DF, Neri proliferation advantage of genetically corrected hematopoietic stem cells 2009-07164 and SAF 2012-39834), Fondo de Investigaciones Sanitarias, Institu- systems. Science 339: 819 – 823 M, Magnani Z, Cantore A, Lo Riso P et al (2011) Site-specific integration in a mouse model of Fanconi anemia FA-D1. Blood 112: 4853 – 4861 to de Salud Carlos III (RETICS-RD06/0010/0015 and RD12/0019/0023), Direc- DeKelver RC, Choi VM, Moehle EA, Paschon DE, Hockemeyer D, Meijsing SH, and tailoring of cassette design for sustainable gene transfer. Nat Methods Rosenberg PS, Alter BP, Ebell W (2008) Cancer risks in Fanconi anemia: ción General de Investigación de la Comunidad de Madrid (CellCAM; Ref Sancak Y, Cui X, Steine EJ, Miller JC et al (2010) Functional genomics, 8: 861 – 869 findings from the German Fanconi Anemia Registry. Haematologica 93: S2010/BMD-2420), and La Fundació Privada La Marató de TV3, 121430/31/32; proteomics, and regulatory DNA analysis in isogenic settings using zinc Matrai J, Cantore A, Bartholomae CC, Annoni A, Wang W, Acosta-Sanchez A, 511 – 517 to J.S. from the Generalitat de Catalunya (SGR0489-2009), the ICREA-Academia finger nuclease-driven transgenesis into a safe harbor locus in the human Samara-Kuko E, De Waele L, Ma L, Genovese P et al (2011) Ruiz S, Gore A, Li Z, Panopoulos AD, Montserrat N, Fung HL, Giorgetti A, Bilic program, the Marató de TV3 (464/C/2012), the Spanish Ministry of Science and genome. Genome Res 20: 1133 – 1142 Hepatocyte-targeted expression by integrase-defective lentiviral vectors J, Batchelder EM, Zaehres H et al (2013) Analysis of protein-coding Innovation (SAF2012-31881), the European Commission (HEALTH-F5-2012- Garcia-Higuera I, Taniguchi T, Ganesan S, Meyn MS, Timmers C, Hejna J, induces antigen-specific tolerance in mice with low genotoxic risk. mutations in hiPSCs and their possible role during somatic cell 305421), and the European Regional Development FEDER Funds; to L.N. from Grompe M, D’Andrea AD (2001) Interaction of the Fanconi anemia Hepatology 53: 1696 – 1707 reprogramming. Nat Commun 4: 1382 Telethon (TIGET grant D2), European Union (FP7 GA 222878 PERSIST, ERC proteins and BRCA1 in a common pathway. Mol Cell 7: Mayshar Y, Ben-David U, Lavon N, Biancotti JC, Yakir B, Clark AT, Plath K, Salmon P, Kindler V, Ducrey O, Chapuis B, Zubler RH, Trono D (2000) Advanced Grant 249845 TARGETINGGENETHERAPY) and the Italian Ministry of 249 – 262 Lowry WE, Benvenisty N (2010) Identification and classification of High-level transgene expression in human hematopoietic progenitors and

differentiated blood lineages after transduction with improved lentiviral Winkler T, Hong SG, Decker JE, Morgan MJ, Wu C, Hughes WM 5th, Yang Y, vectors. Blood 96: 3392 – 3398 Wangsa D, Padilla-Nash HM, Ried T et al (2013) Defective telomere Sebastiano V, Maeder ML, Angstman JF, Haddad B, Khayter C, Yeo DT, elongation and hematopoiesis from telomerase-mutant aplastic anemia Goodwin MJ, Hawkins JS, Ramirez CL, Batista LF et al (2011) In situ iPSCs. J Clin Invest 123: 1952 – 1963 genetic correction of the sickle cell anemia mutation in human induced Yamamoto K, Ishiai M, Matsushita N, Arakawa H, Lamerdin JE, Buerstedde pluripotent stem cells using engineered zinc finger nucleases. Stem Cells JM, Tanimoto M, Harada M, Thompson LH, Takata M (2003) Fanconi 29: 1717 – 1726 anemia FANCG protein in mitigating radiation- and enzyme-induced DNA Smith JR, Maguire S, Davis LA, Alexander M, Yang F, Chandran S, double-strand breaks by homologous recombination in vertebrate cells. ffrench-Constant C, Pedersen RA (2008) Robust, persistent transgene Mol Cell Biol 23: 5421 – 5430 expression in human embryonic stem cells is achieved with Yang YG, Herceg Z, Nakanishi K, Demuth I, Piccoli C, Michelon J, Hildebrand AAVS1-targeted integration. Stem Cells 26: 496 – 504 G, Jasin M, Digweed M, Wang ZQ (2005) The Fanconi anemia group A Soldner F, Laganiere J, Cheng AW, Hockemeyer D, Gao Q, Alagappan R, protein modulates homologous repair of DNA double-strand breaks in Khurana V, Golbe LI, Myers RH, Lindquist S et al (2011) Generation of mammalian cells. Carcinogenesis 26: 1731 – 1740 isogenic pluripotent stem cells differing exclusively at two early onset Young MA, Larson DE, Sun CW, George DR, Ding L, Miller CA, Lin L, Pawlik Parkinson point mutations. Cell 146: 318 – 331 KM, Chen K, Fan X et al (2012) Background mutations in parental cells Somers A, Jean JC, Sommer CA, Omari A, Ford CC, Mills JA, Ying L, Sommer AG, account for most of the genetic heterogeneity of induced pluripotent Jean JM, Smith BW et al (2010) Generation of transgene-free lung stem cells. Cell Stem Cell 10: 570 – 582 disease-specific human induced pluripotent stem cells using a single Yung SK, Tilgner K, Ledran MH, Habibollah S, Neganova I, Singhapol C, excisable lentiviral stem cell cassette. Stem Cells 28: 1728 – 1740 Saretzki GC, Stojkovic M, Armstrong L, Przyborski SA et al (2013) Human Sommer CA, Sommer AG, Longmire TA, Christodoulou C, Thomas DD, Gostissa pluripotent stem cell models of Fanconi Anaemia deficiency reveal an M, Alt FW, Murphy GJ, Kotton DN, Mostoslavsky G (2010) Excision of important role for Fanconi Anaemia proteins in cellular reprogramming reprogramming transgenes improves the differentiation potential of iPS and survival of haematopoietic progenitors. Stem Cells 31: 1022 – 1029 cells generated with a single excisable vector. Stem Cells 28: 64 – 74 Zou J, Mali P, Huang X, Dowey SN, Cheng L (2011a) Site-specific gene Taniguchi T, Garcia-Higuera I, Andreassen PR, Gregory RC, Grompe M, correction of a point mutation in human iPS cells derived from an adult D’Andrea AD (2002) S-phase-specific interaction of the Fanconi anemia patient with sickle cell disease. Blood 118: 4599 – 4608 protein, FANCD2, with BRCA1 and RAD51. Blood 100: 2414 – 2420 Zou J, Sweeney CL, Chou BK, Choi U, Pan J, Wang H, Dowey SN, Cheng L, Tebas P, Stein D, Tang WW, Frank I, Wang SQ, Lee G, Spratt SK, Surosky RT, Malech HL (2011b) Oxidase-deficient neutrophils from X-linked chronic Giedlin MA, Nichol G et al (2014) Gene editing of CCR5 in autologous CD4 granulomatous disease iPS cells: functional correction by zinc finger T cells of persons infected with HIV. N Engl J Med 370: 901 – 910 nuclease-mediated safe harbor targeting. Blood 117: 5561 – 5572 Urnov FD, Rebar EJ, Holmes MC, Zhang HS, Gregory PD (2010) Genome editing with engineered zinc finger nucleases. Nat Rev Genet 11: 636 – 646 License: This is an open access article under the Waisfisz Q, Morgan NV, Savino M, de Winter JP, van Berkel CG, Hoatlin ME, terms of the Creative Commons Attribution 4.0 Ianzano L, Gibson RA, Arwert F, Savoia A et al (1999) Spontaneous License, which permits use, distribution and reproduc- functional correction of homozygous fanconi anaemia alleles reveals novel tion in any medium, provided the original work is mechanistic basis for reverse mosaicism. Nat Genet 22: 379 – 383 properly cited.

848 EMBO Molecular Medicine Vol 6 | No 6 | 2014 ª 2014 The Authors Article

Intercellular network structure and regulatory motifs in the human hematopoietic system

Wenlian Qiao1, Weijia Wang1, Elisa Laurenti2,3, Andrei L Turinsky4, Shoshana J Wodak4,5, Gary D Bader3,6,7, John E Dick2,3 & Peter W Zandstra1,7,8,9,10,*

Abstract stem cells (HSCs), at the apex of the hematopoietic developmental hierarchy, populate and sustain the system through highly coordi- The hematopoietic system is a distributed tissue that consists of nated self-renewal and differentiation processes. Increasing functionally distinct cell types continuously produced through evidence suggests that HSC fate decisions are regulated in part via hematopoietic stem cell (HSC) differentiation. Combining genomic feedback mechanisms including HSC autocrine signaling and para- and phenotypic data with high-content experiments, we have built crine signaling from differentiated hematopoietic cells (Csaszar a directional cell–cell communication network between 12 cell et al, 2012; Heazlewood et al, 2013). However, the key signaling types isolated from human umbilical cord blood. Network structure molecules and cell types involved and how multiple often compet- analysis revealed that ligand production is cell type dependent, ing feedback signals act to regulate HSC fate in a coordinated whereas ligand binding is promiscuous. Consequently, additional manner are poorly understood. control strategies such as cell frequency modulation and compart- We previously used mathematical modeling and bioinformatic mentalization were needed to achieve specificity in HSC fate regu- strategies to systematically characterize the role of feedback signal- lation. Incorporating the in vitro effects (quiescence, self-renewal, ing in regulating human umbilical cord blood (UCB) HSC fate in proliferation, or differentiation) of 27 HSC binding ligands into the vitro (Kirouac et al, 2009, 2010). We identified lineage-dependent topology of the cell–cell communication network allowed coding of stimulatory and inhibitory signals that constitute a dynamic and cell type-dependent feedback regulation of HSC fate. Pathway complex feedback signaling network for hematopoietic stem and enrichment analysis identified intracellular regulatory motifs progenitor cell (HSPC) proliferation. This led to the development of enriched in these cell type- and ligand-coupled responses. This an effective culture system capable of expanding human UCB HSC study uncovers cellular mechanisms of hematopoietic cell feedback by globally diluting inhibitory feedback signals (Csaszar et al, in HSC fate regulation, provides insight into the design principles of 2012), pointing to the relevance of the network that our modeling the human hematopoietic system, and serves as a foundation for approach uncovered. However, how the feedback signaling network the analysis of intercellular regulation in multicellular systems. is organized and how HSCs sense and interpret the signals produced by different cell types remains to be elucidated. Keywords feedback regulation; hematopoietic stem cell; intercellular signaling Network analysis is a powerful approach to detect the design Subject Categories Network Biology; Stem Cells principles of many types of distributed systems. This strategy has DOI 10.15252/msb.20145141 | Received 20 January 2014 | Revised 9 June been used to interpret ecological (Olesen et al, 2007), social 2014 | Accepted 17 June 2014 (Apicella et al, 2012), financial (Vitali et al, 2011), and molecular Mol Syst Biol. (2014) 10: 741 (Jeong et al, 2001) systems, but has never been applied to cell–cell communication (CCC) networks. We hypothesized that mapping the hierarchical hematopoietic signaling network would provide insight Introduction into its regulatory structure and function, in particular how feedback mechanisms control HSC fate decisions. From a network struc- The hematopoietic system is a distributed tissue consisting of multiple ture perspective, we were particularly interested in understanding phenotypically and functionally distinct cell types. Hematopoietic how network structures including modular (network division into

1 Institute of Biomaterials and Biomedical Engineering, University of Toronto, Toronto, ON, Canada 2 Princess Margaret Cancer Centre, University Health Network, Toronto, ON, Canada 3 Department of Molecular Genetics, University of Toronto, Toronto, ON, Canada 4 The Hospital for Sick Children, Toronto, ON, Canada 5 Department of Biochemistry, University of Toronto, Toronto, ON, Canada 6 Department of Computer Science, University of Toronto, Toronto, ON, Canada 7 The Donnelly Centre, University of Toronto, Toronto, ON, Canada 8 Department of Chemical Engineering and Applied Chemistry, University of Toronto, Toronto, ON, Canada 9 McEwen Centre for Regenerative Medicine, University of Health Network, Toronto, ON, Canada 10 Heart & Stroke/Richard Lewar Centre of Excellence, Toronto, ON, Canada *Corresponding author. Tel: +1 416 978 8888; E-mail: [email protected]

ª 2014 The Authors. Published under the terms of the CC BY 4.0 license Molecular Systems Biology 10: 741 | 2014 1 2 orltdrcpo xrsina ihrcniec (average confidence higher exhibited same at the cells of expression primitive cells receptor the in genes cell Specifically, correlated receptor simi- stage. and indicating mature 2C), ligand developmental (Fig of versus types expression cell compartment lar 12 the cell between compartment) receptor developmental (primitive the the recapitulated genes) of values relationship ligand expression clustering gene (or ligand Hierarchical ligands and Methods). 253 to and and genes) equivalent (Materials receptor ligands (or cognate receptors 341 their of S2) Table (Supplementary PreB to related genes properties; over-expressed biological (PreB) (Neut) differentiation. cells genes neutrophil B precursor over-expressed and (Mega) and leukocyte (Mono) megakaryocyte to erythroblasts monocytes related and and genes; MEP erythrocyte differentiation genes; over-expressed differentiation (EryB) and for and except cells proliferation cells), S1 primitive progenitor the Table and as megakaryocyte stem Supplementary to by referred example, intuitive also collectively defined For network (hereafter the see as Methods). for captured and types 2B; (Fig Materials data cell resource ontology corresponding The the gene of 1a). were properties step 2A) biological (Fig 1; (Fig the types construction throughout cell to refer topoietic will we al CCC that et (Novershtern 1 hematopoietic data Fig Transcriptomic analyzing in manuscript. and shown is constructing networks for strategy Our data transcriptomic from cell hematopoietic A Results cells between interactions theoretical on (Frankenstein based constructed been regulation. fate HSC and hema- and impact behavior connectivity system strategies topoietic components) (overlapping network of promiscuous subspecialization and sub-networks) Biology Systems Molecular nlzdtepoete ftesse n aiae predictions computationally validated and We system samples. the of UCB using properties the human types analyzed cell uncultured mature and phenotypi- progenitor, and from stem, of resolved populations highly defined represent 12 cally to between network CCC cells signaling a hematopoietic intercellular created we of profiling, sorting transcriptome high-resolution and of (Kirouac advantage populations taking cell heterogeneous in actions ute,orapoc rvdsafnaetlynwsrtg for strategy new systems. fate. multicellular HSC fundamentally in of regulation a intercellular regulation provides analyzing feedback approach the focusing in our system CCC Further, of hematopoietic mechanisms human into local the the insight on provide of and findings principles our design frequency Overall, the regulation. cell fate HSC specific- in impose relative ity to opportunities as are (niches) differentiated such signals. compartmentalization feedback whereby parameters type-dependent cell model Control through intracellu- nodes key a regulating regulatory by lar fates support HSC influence cells results hematopoietic Our ligands. ocntutteCCntok ecmie database a compiled we network, CCC the construct To have networks signaling intercellular hematopoietic Existing oeua ytm Biology Systems Molecular 03 f1 hntpclydfnd ihyerce hema- enriched highly defined, phenotypically 12 of 2013) , nvitro in S aersosst ewr-rdce HSC-targeting network-predicted to responses fate HSC rtri rgntr MP,oe-xrse HSC over-expressed (MEP), progenitors –erythroid tal et 06 rcrto fligand of curation or 2006) , elcmuiainntoki constructed is network communication –cell 10 : 741 | 2014 tal et 01 Laurenti 2011; , eetrinter- –receptor tal et 00.By 2010). , rcse Fg2) itiuino h eltpsbsdo the linear on a based by approximated types was cell ligands the two produced the their of in of ligands Distribution numbers interacting 2E). their (Fig of and different number processes in the production ranked on were based ligand types orders cell representing The respectively. networks binding, ligand two into network new a sought next paves we network properties. and CCC its hierarchy, ligands analyze hematopoietic This to 117 the 1c). the and depicting step interest mediated of constructed 1; way of ligands was (Fig nodes HSCe 178 cell nodes targeting 12 which cell the of 13 between connection S4), and CCC Table nodes a (Supplementary ligand relationships rules, that production-binding 249 these types ligand between on cell 1,344 containing Based additional network population hematopoiesis. of “Others” impact number hypothetical potentially unknown or a an ligands to Such representing about. connected information were have not receptors did system our which P S-nihdcls(Se ua C Lin UCB on human expressed KIT example, (HSCe: For types. cells cell analyzed part- HSC-enriched interaction differentially the have Some in not conclu- below. did ners our receptors shown and affect as ligands the structures not over-expressed to cognate network did its binds on and practice ligand sions ligand this each the independently; of that interactions receptors for probability the expresses considered the cell We that a higher cell. that assumed the species we ligand, receptor of a perspective, gene number modeling the from greater constructed a cascades the was from hetero- network signaling data, our with intracellular expression that interaction Given initiate require 2007). OSM to (Robb, IL11, and IL9, receptors IL7, LIF, IL6, multimeric IL5, IL21, IL4, (Materials IL15, cytokines IL3, 2C class-1 IL2, IL13, Fig CTF1, Sixteen CSF2, in construction). CNTF, ligands network including on 253 for based and Methods interest, receptors and of 341 types the cell involving the over-expressed ligand of differentially genes 933 receptor between and connections ligand of composed 1b). S3B) step 1; used Table (Fig were analyses threshold (Supplementary subsequent the this in receptors to type- (Supple- according and ligands cell over-expressed the differentially S3A) known thus Table the and S1), mentary detected and Fig sensitivity 10% (Supplementary of combination specificity of optimal the FDR with Materials receptors associated A (see a receptors to Methods). and type-associated threshold 25% and each cell and at known 20% identified of receptors 10%, benchmark of 5%, set 1%, the of compared over- then FDRs differential tested define to we determine (FDR) expression, rate To discovery properties. false biological appropriate type’s an cell the of representative (average CD45RA eiaclrclsi h oemro ih (Ding niche marrow bone the in cells perivascular eltp’ rti xrsin(Schwanhausser the expression of predictive protein are type type’s cell each cell of genes over-expressed progression entially during cells blood in differentiation. the expression through in ligand changes suspected and we receptor Thus, respectively. 0.0570, and 0.0900 P eaooei cell Hematopoietic vle o eetrepeso n iadepeso were expression ligand and expression receptor for -values = sasatn on o u nlss esprtdteCCC the separated we analysis, our for point starting a As C ewr sadrcinlbpriegah(i 2D) (Fig graph bipartite directional a is network CCC A ntecntuto fCCntok,w sue httediffer- the that assumed we networks, CCC of construction the In .0)adcreae iadepeso tlwrconfidence lower at expression ligand correlated and 0.005) À CD49f P eetritrcinpis(upeetr al S2) Table (Supplementary pairs interaction –receptor = .7)ta h aueclsi hc average which in cells mature the than 0.175) + CD90 elcmuiainnetwork communication –cell +/ À id oSF iadpoue by produced ligand a SCF, to binds ) tal et ela ioe al et Qiao Wenlian ª À 2014 CD34 01,and 2011), , tal et h Authors The + 2012), , CD38 À ela ioe al et Qiao Wenlian ª 2014 Figure h td sdvddit ewr osrcin nlss n aiainsae.Scesv tp ihnec tg eeapaeial aee.HC:hmnUBHSC- UCB human HSCe: labeled. alphabetically were stage each within steps Successive stages. validation and analysis, construction, (Lin network enriched into divided is study The h Authors The 1 opttoa n xeietlwrfo ftestudy. the of workflow experimental and Computational . À CD 3. Validation 2. Analysis 1. Construction 34 + CD38 eaooei cell Hematopoietic A production A Yes À Ligand CD45 Receptor Ligand

No genes (R) genes (L) Hematopoiesis RA production network À 11111 111 1 11 111 1 11111 1 1 1 1 1 1 111 1 1 1 1 111 1 11 11 1 1 11 11 111 1 Analysis ofligand hierarchy 6 CD49 12 cell types 1111 111 11 1 111 111 1 123 1 111 1 111 111 11 1 11 1 1 111 11 1 1 11 1 111 1 111 11111111 1 4 Ligand genes f 7 + 111 1111 1 11 CD90 1 11 111 elcmuiainnetwork communication –cell 12 celltypes 11 11 +/ 8 À … ... 2 cells. ) 11 9 B n 5 1 6 10 4 HSCe C 7 11 111 1 A cell typex Cell-cell communication 3 HSCe-targeting ligands 12 8 2 In vitroscreensfor 1 H D C E B A G F 9 network Transcriptomic data HSCe 5 10 B cell typey

11 R L Protein-protein interactiondatabses 3 12 BioMart

Cell-cell communication 933 ligand-receptor

network for HSCe fate interactions B Number of per ligand regulation receptors oeua ytm Biology Systems Molecular Analysis ofligand iRefWeb

12 cell types binding network 123 1 4 11 4 33 1 1 4 4 Ligand genes 2 3 3 1 3 2 3 2 oeua ytm Biology Systems Molecular 2 2 3 2 3 2 1 1 1 4 4 44 4 1 4 1 4 22 4 333 1 … ... 1 2 2 1 44 33 2 11 2 3 3 3 1 44 4 1 4 44 1 4 33 1 3 10 n 1 4 1 3 4 1 Prior knowledge : 741 | 2014 3 2 orltdrcpo xrsina ihrcniec (average confidence higher exhibited same at the cells of expression primitive cells receptor the in genes cell Specifically, correlated receptor simi- stage. and indicating mature 2C), ligand developmental (Fig of versus types expression cell compartment lar 12 the cell between compartment) receptor developmental (primitive the the recapitulated genes) of values relationship ligand expression clustering gene (or ligand Hierarchical ligands and Methods). 253 to and and genes) equivalent (Materials receptor ligands (or cognate receptors 341 their of S2) Table (Supplementary PreB to related genes properties; over-expressed biological (PreB) (Neut) differentiation. cells genes neutrophil B precursor over-expressed and (Mega) and leukocyte (Mono) megakaryocyte to erythroblasts monocytes related and and genes; MEP erythrocyte differentiation genes; over-expressed differentiation (EryB) and for and except cells proliferation cells), S1 primitive progenitor the Table and as megakaryocyte stem Supplementary to by referred example, intuitive also collectively defined For network (hereafter the see as Methods). for captured and types 2B; (Fig Materials data cell resource ontology corresponding The the gene of 1a). were properties step 2A) biological (Fig 1; (Fig the types construction throughout cell to refer topoietic will we al CCC that et (Novershtern 1 hematopoietic data Fig Transcriptomic analyzing in manuscript. and shown is constructing networks for strategy Our data transcriptomic from cell hematopoietic A Results cells between interactions theoretical on (Frankenstein based constructed been regulation. fate HSC and hema- and impact behavior connectivity system strategies topoietic components) (overlapping network of promiscuous subspecialization and sub-networks) Biology Systems Molecular nlzdtepoete ftesse n aiae predictions computationally validated and We system samples. the of UCB using properties the human types analyzed cell uncultured mature and phenotypi- progenitor, and from stem, of resolved populations highly defined represent 12 cally to between network CCC cells signaling a hematopoietic intercellular created we of profiling, sorting transcriptome high-resolution and of (Kirouac advantage populations taking cell heterogeneous in actions ute,orapoc rvdsafnaetlynwsrtg for strategy new systems. fate. multicellular HSC fundamentally in of regulation a intercellular regulation provides analyzing feedback approach the focusing in our system CCC Further, of hematopoietic mechanisms human into local the the insight on provide of and findings principles our design frequency Overall, the regulation. cell fate HSC specific- in impose relative ity to opportunities as are (niches) differentiated such signals. compartmentalization feedback whereby parameters type-dependent cell model Control through intracellu- nodes key a regulating regulatory by lar fates support HSC influence cells results hematopoietic Our ligands. ocntutteCCntok ecmie database a compiled we network, CCC the construct To have networks signaling intercellular hematopoietic Existing oeua ytm Biology Systems Molecular 03 f1 hntpclydfnd ihyerce hema- enriched highly defined, phenotypically 12 of 2013) , nvitro in S aersosst ewr-rdce HSC-targeting network-predicted to responses fate HSC rtri rgntr MP,oe-xrse HSC over-expressed (MEP), progenitors –erythroid tal et 06 rcrto fligand of curation or 2006) , elcmuiainntoki constructed is network communication –cell 10 : 741 | 2014 tal et 01 Laurenti 2011; , eetrinter- –receptor tal et 00.By 2010). , rcse Fg2) itiuino h eltpsbsdo the linear on a based by approximated types was cell ligands the two produced the their of in of ligands Distribution numbers interacting 2E). their (Fig of and different number processes in the production ranked on were based ligand types orders cell representing The respectively. networks binding, ligand two into network new a sought next paves we network properties. and CCC its hierarchy, ligands analyze hematopoietic This to 117 the 1c). the and depicting step interest mediated of constructed 1; way of ligands was (Fig nodes HSCe 178 cell nodes targeting 12 which cell the of 13 between connection S4), and CCC Table nodes a (Supplementary ligand relationships rules, that production-binding 249 these types ligand between on cell 1,344 containing Based additional network population hematopoiesis. of “Others” impact number hypothetical potentially unknown or a an ligands to Such representing about. connected information were have not receptors did system our which (average P S-nihdcls(Se ua C Lin UCB on human expressed KIT example, (HSCe: For types. cells cell analyzed part- HSC-enriched interaction differentially the have Some in not conclu- below. did ners our receptors shown and affect as ligands the structures not over-expressed to cognate network did its binds on and practice ligand sions ligand this each the independently; of that interactions receptors for probability the expresses considered the cell We that a higher cell. that assumed the species we ligand, receptor of a perspective, gene number modeling the from greater constructed a cascades the was from hetero- network signaling data, our with intracellular expression that interaction Given initiate require 2007). OSM to (Robb, IL11, and IL9, receptors IL7, LIF, IL6, multimeric IL5, IL21, IL4, (Materials IL15, cytokines IL3, 2C class-1 IL2, IL13, Fig CTF1, Sixteen CSF2, in construction). CNTF, ligands network including on 253 for based and Methods interest, receptors and of 341 types the cell involving the over-expressed ligand of differentially genes 933 receptor between and connections ligand of composed 1b). S3B) step 1; used Table (Fig were analyses threshold (Supplementary subsequent the this in receptors to type- (Supple- according and ligands cell over-expressed the differentially S3A) known thus Table the and S1), mentary detected and Fig sensitivity 10% (Supplementary of combination specificity of optimal the FDR with Materials receptors associated A (see a receptors to Methods). and type-associated threshold 25% and each cell and at known 20% identified of receptors 10%, benchmark of 5%, set 1%, the of compared over- then FDRs differential tested define to we determine (FDR) expression, rate To discovery properties. false biological appropriate type’s an cell the of representative CD45RA eiaclrclsi h oemro ih (Ding niche marrow bone the in cells perivascular eltp’ rti xrsin(Schwanhausser the expression of predictive protein are type type’s cell each cell of genes over-expressed progression entially during cells blood in differentiation. the expression through in ligand changes suspected and we receptor Thus, respectively. 0.0570, and 0.0900 P eaooei cell Hematopoietic vle o eetrepeso n iadepeso were expression ligand and expression receptor for -values = sasatn on o u nlss esprtdteCCC the separated we analysis, our for point starting a As C ewr sadrcinlbpriegah(i 2D) (Fig graph bipartite directional a is network CCC A ntecntuto fCCntok,w sue httediffer- the that assumed we networks, CCC of construction the In .0)adcreae iadepeso tlwrconfidence lower at expression ligand correlated and 0.005) À CD49f P eetritrcinpis(upeetr al S2) Table (Supplementary pairs interaction –receptor = .7)ta h aueclsi hc average which in cells mature the than 0.175) + CD90 elcmuiainnetwork communication –cell +/ À id oSF iadpoue by produced ligand a SCF, to binds ) tal et ela ioe al et Qiao Wenlian ª À 2014 CD34 01,and 2011), , tal et h Authors The + 2012), , CD38 À ela ioe al et Qiao Wenlian ª 2014 Figure h td sdvddit ewr osrcin nlss n aiainsae.Scesv tp ihnec tg eeapaeial aee.HC:hmnUBHSC- UCB human HSCe: labeled. alphabetically were stage each within steps Successive stages. validation and analysis, construction, (Lin network enriched into divided is study The h Authors The 1 opttoa n xeietlwrfo ftestudy. the of workflow experimental and Computational . À CD 3. Validation 2. Analysis 1. Construction 34 + CD38 eaooei cell Hematopoietic A production A Yes À Ligand CD45 Receptor Ligand

11 R L Protein-protein interactiondatabses 3 12 BioMart

Cell-cell communication 933 ligand-receptor

network for HSCe fate interactions B Number of per ligand regulation receptors oeua ytm Biology Systems Molecular Analysis ofligand iRefWeb

12 cell types binding network 123 1 4 11 4 33 1 1 4 4 Ligand genes 2 3 3 1 3 2 3 2 oeua ytm Biology Systems Molecular 2 2 3 2 3 2 1 1 1 4 4 44 4 1 4 1 4 22 4 333 1 … ... 1 2 2 11 1 44 33 2 2 3 3 3 44 1 4 1 4 44 1 1 4 33 3 10 n 1 4 1 3 4 1 Prior knowledge : 741 | 2014 3 Molecular Systems Biology Hematopoietic cell–cell communication network Wenlian Qiao et al Wenlian Qiao et al Hematopoietic cell–cell communication network Molecular Systems Biology

A B Figure 2. Construction of cell–cell communication networks. ◂ A Transcriptomic profiles of 12 phenotypically defined hematopoietic cell types isolated from human UCB were used. CMP, common myeloid progenitors; MEP, megakaryocyte–erythroid progenitors; GMP, granulocyte–monocyte progenitors; EryB, erythroblasts; Mega, megakaryocytes; Neut, neutrophils; Baso, basophils; Eos, eosinophils; Mono, monocytes; MLP, multilymphoid progenitors; PreB, precursor B cells. Stem cell B Hematopoietic gene ontology enrichment analysis. Shown is the enriched gene ontology with hypergeometric (HG) Z-scores > 1.15. compartment HSCe C Hierarchical relationships between the 12 cell types based on their ligand and receptor gene expression profiles. Hierarchical clusters for (i) 253 ligand genes and (ii) 341 receptor genes. Bootstrapped P-values (or approximately unbiased P-values) on the dendrograms score the uncertainty of the clusters. Dendrograms of gene clusters are not shown. CMP MLP D Concepts of cell–cell communication network constructed from differentially over-expressed ligand and receptor genes of each cell type. Progenitor cell E Ranks of the 13 cell types including “Others” based on the numbers of their produced ligands and the numbers of their bound ligands. compartment HSC proliferationHSC differentiationCMP ErythrocyteproliferationMEGAGranulocyte differentiationdifferentiationNEUTNEUT degranulationdifferentiationNEUT aggregationEOS homeostasis migrationLeukocytePREB chemotaxisM differentiation phase of meiotic cell cycle See also Supplementary Fig S1. MEP GMP HSCe MLP HG Z-score CMP > 1.15 GMP Interaction between ligands and blood cells in the ligand binding Mature cell function, whereas that based on the numbers of bound ligands was MEP approximated by a step-like function—on average, EryB, Neut, and process is promiscuous compartment PreB EryB HSCe bound three times as many ligands as the other cell types. EryB Mega Mono This difference posed the hypothesis that cells and ligands possess We next sought to determine whether the ligand binding network Mega Baso Eos Neut Mono Neut distinct interaction patterns in ligand production and binding had a similar structure to the ligand production network. A Eos processes, a hypothesis we explored by analyzing the structure of ligand-to-cell interaction, Bji, in the ligand binding network was Baso the two networks independently. defined if cell i expressed receptor(s) for ligand j. Interrogation of PreB the network (Fig 4A) using spectral co-clustering (Dhillon, 2001) Interaction between blood cells and ligands in the ligand suggested a significantly less modular interaction structure than in C i Low ii production process is modular the ligand production network (Fig 3A) (t-test P < 0.001), with MLP P-value 0.01 MLP P-value 0.34 HSCe HSCe ubiquitously shared ligand binding among the 12 cell types due to GMP 0.00 GMP 0.00 A cell-to-ligand interaction, Aij, in the ligand production network non-specific ligand–receptor interactions (Supplementary Fig S3A). MEP 0.00 MEP 0.01 was defined if cell i produced ligand j. Simultaneously, clustering The promiscuous network structure is robust to the choice of CMP 0.01 CMP 0.35 the cell types and the ligands suggested that groups of ligands FDR threshold for differential gene over-expression (Supplementary Mono Mono High Neut 0.00 Neut 0.01 were associated with subsets of cells in the network (Fig 3A). Fig S3B) and the incorporation of hetero-multimeric receptor EryB 0.17 EryB 0.01 Silhouette widths (Rousseeuw, 1987) measuring the relatedness of expression in network construction (Supplementary Fig S3C). Mega 0.07 Mega 0.01 the cell types’ ligand production supported the existence of 4 Interestingly, HSCe which normally reside in the bone morrow niche Eos Eos 0.00 0.12 ligand–cell modules (Fig 3B, Supplementary Fig S2): the primitive with progenitor and maturing cells (Fig 4B) interacted with ligands PreB 0.19 Baso 0.01 Baso 0.11 PreB 0.18 cell module (HSCe + MLP + CMP + MEP + GMP), neutrophil– of the greatest diversity. This raised the question of how HSCe fate monocyte module (Neut + Mono), erythroid module (EryB), and a can be specifically regulated in response to physiological demand. 341 receptor genes Manhattan distance 253 ligand genes Manhattan distance module of all the other cell types (Boso + Eos + Mega + PreB) We hypothesized two different mechanisms: relative cell frequency (Fig 1; step 2a). A priori biological processes of 190 ligands that allows more abundant cell types skew the ligand species and D E (Supplementary Table S5) suggested that each blood cell module resources available to HSCe, and cell compartmentalization that Cell A CellCell AB produced ligands with biased biological functions. For instance, limits the access of HSCe to locally available ligands. We then Cell A Cell B EryB Neut ligands of the neutrophil–monocyte module enriched in exoge- explored, computationally, the effects of the two mechanisms on the HSCe neous signals that inhibit cell survival (HG Z-scores were 1.63 and quantity and identity of HSCe-targeting ligands (Fig 1; step 2b). Cell C 2.98 for Mono and Neut, respectively) and signals that mediate cell To explore the role of cell frequency in skewing HSCe- Cell C Others survival via NF-jB (HG Z-scores were 2.15 and 1.43 for Mono and targeting ligands, we compared ligand binding in two scenarios Neut, respectively); ligands of Baso, Eos, and PreB within the by assuming that the probability of binding a ligand is a function Mono EryB (Boso + Eos + Mega + PreB) module enriched in signals that direct of cell frequency given non-regulated receptor ligand affinities. In differentiation cell fates of T helper cells (HG Z-scores were 1.17, the first scenario, we modeled ligand binding in the system of Differential ligand Others Mega 2.65, and 3.18 for Baso, Eos, and PreB, respectively); and ligands mono-nucleated cells (MNC) isolated from fresh human UCB over-expression MLP Ligand production Ligand binding Baso GMP of EryB enriched in signals that regulate G1-S cell cycle transition samples. Based on flow cytometry analysis, Neut was the most Differential receptor PreB CMP (HG Z-score = 1.41) (Fig 3C). See Supplementary Table S6 for the abundant cell type in the system (Fig 4Ci) according to the over-expression Cell A Cell A CMP Mega other HG enrichment Z-scores. phenotypic definition we used; consequently, the cell type was Neut Eos In summary, our analysis suggested that blood cell ligand the major ligand sink that significantly influenced ligand accessi- HSCe Cell B Cell B Baso GMP production is peculiar to blood cell identities, and a modular inter- bility of the other cell types (Fig 4Cii). In contrast, HSCe, a Others Mono action structure exists in the ligand production network. This quantitatively underrepresented cell type in the MNC system, had MEP Cell C Cell C MLP conclusion is robust to the choice of FDR threshold for differential negligible ligand access despite the large number ligands targeting PreB

Number of produced or bound ligands MEP gene over-expression (Supplementary Fig S2B) and the incorpora- the cell type (Fig 4A). In the second scenario, we modeled ligand Eos Others Others tion of hetero-multimeric receptor expression in network construc- binding using cell frequencies from progenitor cell-enriched UCB Produced Bound tion (Supplementary Fig S2C). Furthermore, ligand production of samples (Fig 4Di), in which cell composition is reminiscent of ligands ligands hematopoietic cell modules indicated characteristic biological prop- the progenitor enrichment seen during development or in the erties. Considering HSC feedback regulation, this raised the possibil- bone marrow niche (Nombela-Arrieta et al, 2013). Increased Figure 2. ity of HSC feedback control by cell module- or cell type-specific frequency of HSCe elevated their access to the available ligand signaling. resources (Fig 4Dii). This analysis indicates that controlling

4 Molecular Systems Biology 10: 741 | 2014 ª 2014 The Authors ª 2014 The Authors Molecular Systems Biology 10: 741 | 2014 5 Molecular Systems Biology Hematopoietic cell–cell communication network Wenlian Qiao et al Wenlian Qiao et al Hematopoietic cell–cell communication network Molecular Systems Biology

A B AB Stem cell Peripheral MEP 01234567 niche tissues GMP Primitive Ligand Cell Number of interactions Cell Ligand MLP cells CMP 178 ligands MLP HSCe MLP HSCe HSCe HSCeHSC GMP Mono Neut + Mono GMP MEP Neut MEP CMP CMP Mono Baso Mono Baso + Eos + Neut Mega Neut Progenitor EryB PreB Mega + PreB Eos cells (PC) Mature cells Eos Eos Mega in peripheral PreB EryBEry (MCP) Baso EryB EryB Baso Maturing cells PreB in niche (MCN) Mega 0 0.1 0.2 178 ligands Silhouette width C Mono-nucleated cell compartment D Stem and progenitor cell compartment i ii PreB i ii PreB Neut Baso Neut MEP Baso Neut C Mono Eos 0 05 Eos Mega PreB CMP Mono EryB CMP Mono Cell adhesionCell cycleDevelopmentImmuneInflammationProliferationSurvival Mega HSCe % of each a priori ligand set EryB Mono GMP Neut Proliferation stimulation MEP Eos MLP MLP Eos MLP Survival mediated via PI3K-AKT Baso Baso CMP CMP Survival mediated via MAPK and JAK-STAT HSCe PreB Leucocyte chemotaxis MLP EryB MEP GMP EryBEry MEP Survival mediated via NF-kB ** 0 5 10 15 0 10 20 Angiogenesis regulation GMP Mega GMP Mega * Cell frequencies (%) HSCe Cell frequencies (%) HSCe Proliferation inhibition ** Innate inflammatory response EFInsignificant Signal strength ** * Survival inhibition by external signals Inflammation / G1-S phase cell cycle regulation by growth factor HSCe PC angiogenesis / * Innate inflammatory mediation G2-M phase cell cycle regulation by growth facto Angiogenesis regulation G1-S / HSCe via growth factors Survival inhibition / G1-S regulation by interleukin G1-S regulation * ** via interleukin Survival mediation/ * ** * T helper cell differentiation G2-M regulation Survival inhibition Proliferation stimulation B Survival via NF-κB Eos via PI3K-AKT CMP MEP MLP Neut PreB EryB HSCe GMP Mono Baso Mega OR mediation via MAPK & JAK-STAT Proliferation stimulation Proliferation inhibition Figure 3. Modular ligand–cell interaction structure in the ligand production network. MCN P P P P MCP A Hierarchical clustering based on Jaccard distances identifies four cell modules separated by the blue lines. HSCe PC MCN MCP Innate inflammatory mediation Angiogenesis regulation Proliferation B Silhouette widths for the four cell modules in (A). via growth factors inhibtion G1-S regulation C Expression of a priori biological function-associated ligands by each cell module in (B). Asterisks (*) indicate the enriched ligand sets defined as HG Z-score > 1.15. via interleukin G2-M regulation See also Supplementary Table S5 and Supplementary Figure S2. HSCe CMP EryB Baso Survival inhibition Survival via NF-κB GMP Mega Eos via PI3K-AKT MEP Mono Neut mediation via MAPK & JAK-STAT MLP PreB Proliferation stimulation G2-M regulation hematopoietic cell relative frequency can modulate ligand expo- in the peripheral blood or tissues (MCP = Baso + Eos + Neut) Proliferation inhibition sure to HSCe. (Fig 4E). The spatial relationship between each compartment and 0.1 1 0.1 1 Then, we explored the role of cell compartmentalization. While an HSCe was modeled by the probability of the ligands produced by the Communication to HSCe (PHSCe, PPC, PMCN, PMCP) increasing number of hematopoietic cell types such as erythroblasts compartment reaching HSCe (Materials and Methods). Specifically, we (Soni et al, 2008), megakaryocytes (Huang & Cantor, 2009), mono- assumed that (i) there is no diffusion for HSCe autocrine ligands, so G Innate G1-S cell G2-M cell Angiogenesis Survival Survival Proliferation Proliferation cytes (Chow et al, 2011), and B cell progenitors (Nagasawa, 2007) are the probability of HSCe autocrine binding PHSCe is 1; (ii) PC reside inflammation cycle phase cycle phase regulation inhibition mediation stimulation inhibition found in the stem cell niche within the bone marrow environment, the close to HSCe, so PPC is 0.8; (iii) MCN reside further away from HSCe mediation regulation regulation exact location and direct feedback role of these cell types on HSC fate than PC, so P is 0.7; (iv) physical barriers between the stem cell Mega MCN 20 Mono decisions is not clear. We used OR gates to model the feedback effect niche and the peripheral tissues prevent MCP ligands from reaching 15 EryB PreB of these cell types on HSCe as a function of their localization based HSCe, so PMCP is 0.1. We found that HSCe expressed a broad spectrum 10 on the extant knowledge of 190 ligands (Supplementary Table S5). of autocrine signals including those thought to be important for HSC 5 Contribution to 0 The model consisted of four compartments to represent cells of self-maintenance, whereas PC and MCN were the major producers of 0 0.5 1 0 0.5 1 0 0.5 1 0 0.5 1 0 0.5 1 0 0.5 1 0 0.5 1 0 0.5 1 different developmental stages: HSCe themselves, progenitor cells non-HSC supportive signals (Fig 4F). HSCe-targeting signals PMCN-x (PC = CMP + GMP + MEP + MLP), mature cells in the stem cell niche In vivo monocytes, megakaryocytes, erythroblasts, and pre-B cells Figure 4. (MCN = EryB + Mega + Mono + PreB), and granulocytic mature cells are primed to transit from the bone marrow to the peripheral blood.

6 Molecular Systems Biology 10: 741 | 2014 ª 2014 The Authors ª 2014 The Authors Molecular Systems Biology 10: 741 | 2014 7 Molecular Systems Biology Hematopoietic cell–cell communication network Wenlian Qiao et al Wenlian Qiao et al Hematopoietic cell–cell communication network Molecular Systems Biology

A This cell movement potentially alters the HSC microenvironment. We 2005; Csaszar et al, 2012). On day 7, the numbers of Cell culture Flow cytometry Test versus control analysis next sought to predict the spatial effect of Mono, Mega, EryB, and CD34+CD133+CD90+ cells (defined as HSC-enriched cells) (Mayani 1 test ligand + PreB on HSCe feedback regulation. Our simulation results (Fig 4G) & Lansdorp, 1994; Dorrell et al, 2000; Danet et al, 2001; Ito et al, Control + revealed the importance of Mega-produced HSCe-targeting ligands 2010), CD34 cells that were CD133À or CD90À (defined as progeni- x40 (SCF + THPO + FLT3LG)*

in innate inflammatory response terms and the importance of tor cells; see Supplementary Fig S5 for functional quantification CD133 A Cell count Test ligand x Mono-produced HSCe-targeting ligands in regulating angiogenesis- using the colony-forming cell assay), and CD34À cells (defined as B Culture C CD34 CD90 associated terms. Strikingly, it was evident that EryB-produced mature cells) were quantified. The BC cocktail-supplemented D for 7 days HSCe-targeting ligands are associated with regulating cell cycle culture output 704 425 (mean s.d. from 33 biological repli- E Test ligand y Æ Æ F progression, cell survival and proliferation, which warrants future cates) cells consisted of 6.35 3.21% HSC-enriched cells, G HSC- Progenitor Mature Circle size is proportional Æ H HSC-e * BC control experimental validation. This analysis indicates that regulation of cell 27.75 6.86% progenitor cells, and 65.90 10.04% mature cells. enriched cell cell to cell number Æ Æ identities in HSCe microenvironment or niche can modulate ligand This established a reference for detecting the effects of test ligands B exposure to HSCe. on HSC-e fate decisions (Supplementary Fig S6). In addition to the i TGFB1 + (SCF + THPO + FLT3LG) ii SR1 + (SCF + THPO + FLT3LG) In summary, our analysis uncovered promiscuous ligand-to-cell BC cocktail, TGFB1 (10 ng/ml) (Batard et al, 2000) and StemRege- 1 interactions in the ligand binding network. HSCe were found to nin 1 (SR1, 0.75 lM) (Boitano et al, 2010) were used as the negative 6 express receptors for a broad range of ligands, implying the exis- and positive control for HSC-e expansion, respectively (Fig 5B). tence of physical parameters such as relative cell frequency and In vitro effect of the 33 ligands was quantified by signed one-tail 4 compartmentalization in HSC fate regulation. Our subsequent simu- P-values from the nested ANOVA detailed in the Materials and 0.5 lation revealed a potential importance of Mega, Mono, and EryB Methods (Supplementary Fig S7A). P-values of the 35 ligands 2 Test / Control Test ligands in HSC fate regulation. To explore how hematopoietic cell (including TGFB1 and SR1) at their most effective dose on human / Control Test 0 type-dependent signals feedback to HSCe, we next performed high- UCB HSC-e are shown in Fig 5C. For ligands that did not have any 0 content in vitro experiments for HSCe-targeting ligands. significant effect, results of the highest working concentrations were HSC- HSC- reported. See Supplementary Fig S8 for cell number comparison Mature Mature −enriched −enriched Progenitor Validation of HSCe-targeting ligands using a high-content in vitro between the tested conditions and the BC control. See Supplemen- enrichedProgenitor enriched SC SC phenotypic assay tary Tables S8 and S9 for results of all the testing conditions. These in vitro data allowed us to examine the impact of the cell types of C ANOVA P-value High-content in vitro experiments were performed by following the interest on HSC fate regulation in the CCC network. HSC-enriched cells Test > Control protocol in Fig 5A. HSCe-targeting ligands in the CCC network < 0.01 (Supplementary Table S4) were ranked according to the molecular Provisional feedback signaling networks for cell type-associable Progenitor 0.01 ~ 0.02 interaction confidence scores (Ceol et al, 2010) for ligand–receptor HSC fate modulation cells interactions (Supplementary Table S2) and the receptor gene expres- 0.02 ~ 0.05 sion levels in HSCe from the Transcriptomic data. Thirty-three Measurement of the in vitro effect of the 33 ligands on HSC-e Mature Test < Control ligands were prioritized for experimental tests (Materials and Meth- allowed creation of a directional CCC network. First, we categorized cells < 0.01 ods, Supplementary Table S7). We examined the phenotypic impact each ligand into one of the five functional categories [inducing TGFB1 [10] SR1 [0.75] CSF2 [100] CSF3 [100] FGF1 [200] IL11 [150] TNFSF12 [100] WNT4 [100] BMP4 [25] FGF2 [50] IL12A [150] BMP2 [1] NGF [10] ANGPT1 [100] IL17A [50] ANGPT2 [1] ANGPTL3 [50] SPP1 [200] NPPC [50] BDNF [50] IL16 [0.2] MDK [50] TNF [100] LTA [200] TNFSF10 [100] BMP6 [100] INHBA [1] GDF11 [100] IL18 [1] BTC [200] CALCA [1000] NRG1 [100] VEGFA [150] VIP [1000] WNT7A [250] + of each ligand on 40 HSC-enriched cells (HSC-e: LinÀCD34 quiescence, inducing self-renewal, inducing differentiation, inducing 0.01 ~ 0.02 low + Rho CD38ÀCD45RAÀCD49f ) isolated from human UCB samples; proliferation (self-renewal + differentiation), and inhibiting prolifer- 0.02 ~ 0.05 / this population contains approximately one NOD-scid-IL2RgcÀ À ation] in terms of their manipulation in HSC-e fate decisions using repopulating cell per 13 cells (combination of 1:10 for the P-values in Supplementary Table S9 and the classifier in > 0.05 + + + CD49f CD90 and 1:20 for CD49f CD90À HSC-enriched cells) Table 1. A representative ligand is given for each category in (Notta et al, 2011). Each ligand was tested in a short-term assay at Supplementary Fig S7B. The ligands, at the working concentrations three doses in the presence of three basal cytokines (BC)—SCF, shown in Fig 5C, were categorized with different confidences Figure 5. HSC-e respond to exogenously added HSCe-targeting ligands. low + + THPO, and FLT3LG (Petzer et al, 1996; Madlambayan et al, (Fig 6A). Collectively, 27 out of the 33 ligands of interest were A The experimental and analytical protocol. HSC-e: human UCB LinÀRho CD34 CD38ÀCD45RAÀCD49f . BC basal cocktail consisted of 100 ng/ml SCF, 50 ng/ml THPO, and 100 ng/ml FLT3LG. B Fold changes between the results of (i) negative control (TGFB1)/(ii) positive control (SR1) and that of the cell culture supplemented with BC only. HSC-enriched cells: + + + + Figure 4. Promiscuous ligand–cell interaction structure in the ligand binding network. CD34 CD133 CD90 ; progenitors: CD34 cells that are CD133À or CD90À; mature cells: CD34À. Data are from 33 biological replicates. C Signed one-tail P-values from the nested ANOVA when comparing the cell counts of testing conditions to the BC control. Positive P-values indicate that effect of a A Spectral co-clustered adjacency matrix of ligand-to-cell interactions. The gray scale indicates the number of receptor genes expressed by a cell type for each of the test ligand was greater than that of the BC control, and negative P-values indicate the effect of a test ligand was less than that of the BC control. Ligand ◂ 178 ligands. concentration is in ng/ml, except for SR1 that is in lM. B Schematic in vivo HSCe feedback signaling network. C Cell frequency-dependent ligand binding network in the mono-nucleated cell compartment. (i) Composition of mono-nucleated cells isolated from fresh human UCB See also Supplementary Figs S4,S5 and S6. samples (n = 3). (ii) Potential of apparent competition (PAC) computed from the network weighted by the cell composition shown in (i). Along the edge connecting node i and node j, the width at node i indicates the competitiveness of node i to node j in terms of ligand binding. D Cell frequency-dependent ligand binding network in the stem and progenitor cell compartment. (i) Cell frequencies in lineage-depleted cells isolated from uncultured found to direct HSC-e fate decisions (Fig 1; step 3a), indicating a (ANOVA P = 0.0036), so it induced HSC-e self-renewal. At a work- human UCB samples (n = 3). (ii) PAC computed from the network weighted by the cell composition shown in (i). E Logic gates used to model in vivo HSCe feedback signaling. The probability (P) of a cell compartment feeding signals to HSCe is inversely proportional to the distance significant enrichment of prediction capacity in this analysis ing concentration of 100 ng/ml, however, the ligand led to a signifi- between the cell compartment and HSCe. (Binomial P = 0.0001, Materials and Methods). cant decrease in the number of HSC-enriched cells (ANOVA F Simulated functional effect of HSCe, PC, MCN, and MCP on HSCe as a function of feedback probability P. The color map indicates average signaling strength from 500 Intriguingly, dose-dependent HSC-e fate regulation was observed P = 0.0007), progenitor cells (ANOVA P = 0.0094), and mature cells simulations. Insignificant cell–cell communication is colored in gray. for some ligands. For example, TNFSF10, at a working concentra- (ANOVA P = 0.0207) (Supplementary Fig S6Bii), so it inhibited G Simulated functional contribution of MCN cell type x (Mega, Mono, EryB, or PreB) to HSCe-targeting ligands as a function of the distance between MCN cell type x tion of 1 ng/ml, did not affect the number of HSC-enriched cells, HSC-e proliferation, which may be due to the pro-apoptotic effect and HSCe. The simulation was performed at PHSCe = 1, PPC = 0.8, PMCN-not x = 0.7, and PMCP = 0.1. The magnitudes of contribution are with respect to PMCN-x = 0, which is set to 0. progenitor cells, or mature cells (ANOVA P-values were 0.2747, of the ligand (Zamai et al, 2000). Dose-dependent HSC-e fate See also Supplementary Figure S3. 0.2642, and 0.3721, respectively). When the ligand was used at regulation was also observed for FGF1, FGF2, IL11, and TNFSF12 10 ng/ml, it led to an increase in the number of HSC-enriched cells (Supplementary Table S9). This result is reminiscent of differential

8 Molecular Systems Biology 10: 741 | 2014 ª 2014 The Authors ª 2014 The Authors Molecular Systems Biology 10: 741 | 2014 9 Molecular Systems Biology Hematopoietic cell–cell communication network Wenlian Qiao et al Wenlian Qiao et al Hematopoietic cell–cell communication network Molecular Systems Biology

Table 1. Functional definition of ligands for HSC-e fate regulation activation of pathways that are involved in diverse biological due to low cell frequency (Fig 4Ci). Collectively, we propose that orphan signals entering the network). We found that the CCC network based on a cell number comparison between the conditions having the processes (Kale, 2004). Furthermore, categorization of some ligands both the progenitor cells and the mature cells regulate HSC-e fate deci- can be depicted in two formats based on signal directionality— ligands of interest and the basal cytokine control. such as FGF2 (working concentration, WC = 50 ng/ml) and BMP6 sions via feedback signaling yet through different mechanisms—the ligand production and ligand binding, and each format was HSC-enriched Progenitor Mature (WC = 100 ng/ml) was sensitive to the statistical significance progenitor cells feed back HSC-e self-renewal and quiescence signals, analyzed as an individual network. The high degree of modularity cells cells cells threshold, suggesting their indeterminate role in regulating HSC-e whereas the more mature cells feed back HSC-e predominantly in the ligand production network pointed to cell type-specific Neutral – ––fate decisions may be context dependent. The ligands were excluded proliferation and differentiation signals (Fig 1; step 3b). production of ligands for HSC-e cell fate regulation. In contrast, the Quiescence induction ––↓ accordingly in the subsequent analyses. ligand-to-cell interactions in the ligand binding network were – ↓ – We explored how ligands produced by different cell types influ- Pathway enrichment analysis suggested intracellular regulatory promiscuous, and HSCe were one of the cell types that bound the enced HSC-e fate decisions by performing a functional enrichment motifs for HSC-e fate decision-making most ligands, suggesting that HSCe have broad environment sensing Self-renewal ↑ –– induction analysis for the ligands expressed by each of the 12 cell types in the capacity (Takizawa et al, 2012). Our analysis raised important ques- CCC network using the ligand function categorization (Fig 6A) as a The association between HSCe-targeting ligands and different cell tions about how feedback specificity is achieved in HSC fate regula- Differentiation – ↑ – induction reference. To ensure that there were sufficient data to draw qualita- types allowed us to construct a qualitative CCC network focusing on tion. In silico simulation posed the hypothesis that additional ––↑ tive conclusions, the analysis was performed based on the categori- HSC-e fate regulation (Fig 7A). A database survey on the intracellular control mechanisms including those observed in vivo (cell type – ↑↑zation at the intermediate confidence level while excluding BMP6 in signaling pathways of the HSCe-targeting ligands suggested that frequency control and HSC niche localization or compartmentaliza- ↓ ↓↑which categorization was indeterminate at that confidence level. intracellular regulatory motifs are associable with the ligands tion) are required to confer specificity in hematopoietic cell- Assuming each ligand acts independently in HSC-e fate regulation, responsible for directive effects on HSC-e cell fate decisions in vitro mediated feedback regulation of HSC fate decisions. To test the Proliferation ↑↑– induction this analysis allowed us, for the first time, to predict the role of each (Fig 7B, Supplementary Fig S9, Materials and Methods). Specifically, hypothesis, we prioritized 33 HSCe-targeted ligands in the CCC ↑ ↑↓cell type in the HSC-e feedback regulation. As shown in Fig 6B, signaling activity of the HSC-e quiescence-inducing ligands (such network for in vitro experiments. We anticipated the roles of the 33 ↑ ↑↑progenitor cells such as CMP, MEP, GMP, and MLP predominantly as BMP6 and IHNBA), self-renewal-inducing ligands (such as ligands in directing HSC-e fate decision using a cell surface marker ↑ – ↑ expressed ligands that induced HSC-e quiescence and self-renewal; ANGPT1, ANGPT2, NGF, and TNFSF12), proliferation-inducing expression-based phenotypic assay. The in vitro data allowed us to ↑ ↓↑EryB expressed ligands of diverse functions as expected from the ligands (such as CSF2, CSF3, and IL11), and proliferation inhibitory uncover what signals each of the 12 cell types feeds back to HSC-e. results shown in Fig 3C. In contrast to a majority of the cell types, ligands (such as TGFB1, TNFSF10, and TNF) were attributable to For instance, the mature cells, particularly Mono and granulocytes Proliferation ↓ ↓↓ inhibition which expressed at most three types of directive signals for HSC-e fate SMAD (permutation P = 0.044, Supplementary Fig S9A), NF-jB (Neut, Baso, and Eos), were found to express mainly inhibitory ↓↓– decisions, HSCe expressed ligands inducing self-renewal, quiescence, (permutation P = 0.122, Supplementary Fig S9C), STAT (permutation signals for HSC-e proliferation and inducing signals for HSC-e differ- ↓ – ↓ and differentiation, and inhibiting proliferation. This is reminiscent of P = 0.04, Supplementary Fig S9C), and caspase cascade (permutation entiation, which in combination can exhaust the HSC population ↓ ––self-sufficient autocrine signaling of HSC (Kirito et al, 2005) possibly P = 0.059, Supplementary Fig S9D) pathways, respectively. because of the extensive cell cycling and division involved in the to compensate for their disadvantage in accessing exogenous signals Our qualitative CCC network can be depicted in three ways: a proliferation and differentiation processes (Hock et al, 2004; Zhang Dash “–” indicates no change from the basal cytokine control. directional network weighted by receptor frequency (Fig 7C), a et al, 2006). However, under a normal in vivo condition, monocytes directional network weighted by cell frequencies in the MNC and granulocytes mainly circulate in the peripheral tissues; their A B 0 12 compartment (Fig 7D), and a weighted directional network with secreted ligands have limited access to HSC in the bone marrow LTA BMP4 Function categories compartmentalization (Fig 7E) overlaid to illustrate the roles of compartment because of the blood–bone marrow barrier. The identi- TNF FGF2 Self-renewal HG z-score cellular dynamics and spatial distribution in HSC fate regulation fied importance of cell compartmentalization in protecting HSC from TGFB1 HSCe through feedback signaling. For example, Neut was the largest cell exogenous signals is consistent with our observation that global TNFSF10 Quiescence CMP SPP1 Proliferation MEP population in the MNC isolated from human UCB (Fig 4C), so media dilution enhances in vitro HSC production when physical ANGPTL3 Proliferation GMP TNFSF10 and TNF from Neut were potentially the major signals to barriers between HSC and the mature cells are absent (Csaszar et al, MDK Differentiation MLP inhibit HSC-e proliferation. However, the stem cell niche-peripheral 2012). We also found that progenitor cell types—CMP, MEP, GMP, BMP6 Mega barrier would typically protect HSC-e from the inhibitory signals. and MLP—that typically co-localized with HSC in the bone marrow EryB In summary, we combined the topology of the CCC network, the niche tend to function as a unit, enriched for ligands for HSC mainte- Confidence levels Neut in vitro effect of 33 ligands on HSC-e fate decisions, and pathway nance by inducing HSC quiescence and self-renewal. This finding BDNF High Baso IL18 IL16 Eos information of the ligands. Our results support a model whereby supports the use of periodic primitive cell selection to increase in vitro IL12A Intermediate BMP6 NPPC Mono hematopoietic cells influence HSC toward certain cell fates by regu- HSC production (Madlambayan et al, 2005) and suggests technologies CSF2 Low PreB FGF2 CSF3 lating the key intracellular regulatory motifs through cell type- that target the HSC niche composition to control HSC fate in vivo. ANGPT1 FGF1 Others INHBA specific feedback signals. The pathway enrichment analysis pointed to specific intracellular ANGPT2 IL11 regulatory motifs associated with ligands of different in vitro effects GDF11 BMP2 SR1 IL17A on HSC-e fate. Specifically, HSC-e quiescence-inducing ligands such NGF Discussion as BMP6 (Holien et al, 2012) and INHBA (Burdette et al, 2005) regulate the expression of SMADs to arrest cell growth. The HSC-e While it is accepted that feedback regulation of HSC fate decisions self-renewal-inducing ligands such as angiopoietins (Hughes et al, Neutral Ligands TNFSF12 is important to stable hematopoiesis (Csaszar et al, 2012; 2003), NGF (Descamps et al, 2001), and TNFSF12 (Kawakita et al, BTC / CALCA / NRG1 / WNT4 Quiescence inducing Self-renewalProliferation inducing inducingProliferation inhibiting Heazlewood et al, 2013), it has been unclear how the feedback 2004) were found to regulate the activity of NF-jB in which deletion VEGFA / VIP / WNT7A Differentiation inducing system operates. Extensive effort has been made to understand how in the mouse hematopoietic system compromised the self-renewal stromal cells in the bone marrow microenvironment regulate HSC and long-term hematopoietic repopulation ability of HSC (Zhao Figure 6. In vitro experiments lead to functional categorization of HSCe-targeting ligands. fate decisions (Zhang et al, 2003; Nakamura et al, 2010; Kunisaki et al, 2012; Stein & Baldwin, 2013). The HSC-e proliferation-inducing A Functional categorization for the 35 HSCe-targeting ligands, including the negative control TGFB1 and the positive control SR1. The ligands were categorized at et al, 2013). In addition, we propose a hematopoietic cell-driven ligands such as CSF2 (Carter, 2001; Gu et al, 2007), CSF3 (Harel- different confidence levels. High, intermediate, and low confidence levels refer to ANOVA P-value significance thresholds 0.01, 0.02 and 0.05, respectively. See Table 1 feedback system that regulates HSC fate decisions through inter- Bellan & Farrar, 1987), and IL11 (Yoshizaki et al, 2006) were found for definition of the functional categories. B Functional enrichment was performed for the HSCe-targeting ligands produced by each cell type. The color scale indicates the HG enrichment Z-scores. cellular signaling. to induce the expression of STATs for cell proliferation. Finally, the We constructed a bipartite graph to represent the CCC network HSC-e proliferation inhibitory ligands such as TGFB1 (Shima et al, See also Supplementary Figs S7 and S8. between 12 hematopoietic cell types isolated from human UCB (and 1999), TNF (Mallick et al, 2012), and TNFSF10 (Kischkel et al,

10 Molecular Systems Biology 10: 741 | 2014 ª 2014 The Authors ª 2014 The Authors Molecular Systems Biology 10: 741 | 2014 11 Molecular Systems Biology Hematopoietic cell–cell communication network Wenlian Qiao et al Wenlian Qiao et al Hematopoietic cell–cell communication network Molecular Systems Biology

A Cell-cell communication network for HSC-e fate regulation the key intracellular regulatory motifs through cell type-specific Materials and Methods feedback signals. Further, control parameters such as relative cell Directive signals for HSC-e frequency and spatial compartmentalization (niches) are opportuni- Microarray datasets ANGPT1 IL18 BMP6 SPP1 TNFSF10 fate decisions ties to impose specificity in HSC fate regulation. A particularly inter-

Self-renewal signals esting extension of our current work is to analyze how defects in Illumina data of primitive cells and progenitor B cells (ProB: Quiescence HSC-e HSC niche composition and physical structure or defects in HSC CD34+CD10+CD19+; three biological replicates) were obtained HSC-e autocrine Proliferation intracellular regulatory motifs affect feedback regulation of HSC fate from the authors of Laurenti et al (2013). The primitive cells are + + +/ Proliferation decisions in vivo and consequently causes hematopoietic disorders HSCe (LinÀCD34 CD38ÀCD49f CD45RAÀCD90 À; 10 biological + + + Differentiation such as leukemogenesis (Schepers et al, 2013). replicates), CMP (LinÀCD34 CD38 CD135 CD45RAÀCD7ÀCD10À; + + One limitation of this study is that we used only transcriptomic five biological replicates), MLP (LinÀCD34 CD38ÀCD90ÀCD45RA ; Effect on in vitro HSC-e + + ANGPT1 ANGPT2 NGF WNT4 TNFSF12 BMP6 IL18 INHBA GDF11 CSF3 IL11 FGF2 FGF1 ANGPTL3 BMP4 data rather than proteomic data to construct the CCC network. five biological replicates), MEP (LinÀCD34 CD38 CD135À production TGFB1 TNFSF10TNF Although there is a general agreement between mRNA and protein CD45RAÀCD7ÀCD10À; five biological replicates), and GMP + + + + Positive (+) expression levels of ligands and receptors in mammalian cells (De (LinÀCD34 CD38 CD135 CD45RA CD7ÀCD10À; five biological Negative (-) Haan et al, 2003; Madlambayan et al, 2005; Schwanhausser et al, replicates). The data are accessible at Gene Expression Omnibus 2011), gaining better understanding of the dynamics of mRNA (GEO) (Edgar et al, 2002) through accession number GSE42414. Degree of feedback effect expression and the corresponding protein expression can be impor- Quantile signals of the Illumina data were calculated using the Weak tant in understanding context-specific network structures and their normalizeQuantile() function in the limma package (v3.16.3) of + ++ + + + - + - + HSC-e paracrine signals Strong -+ -+ - -+ -+ - - dynamic evolution. The newly developed mass cytometry (Bendall BioConductor. + ------et al, 2011) offers a novel single cell proteomic approach to achieve Affymetrix CEL files of mature cells and ProB MLP CMP MEP GMP EryB Mega Mono PreB Baso Eos Neut this goal. A second limitation of this study is that we defined the (CD34+CD10+CD19+; five biological replicates) were downloaded exogenous effects of 33 ligands on HSC-e fate decision according to from GEO (accession number GSE24759 (Novershtern et al, 2011), B + Exogeneous signals Exogeneous signals Exogenous signals Exogeneous signals in vitro measurements of a cell surface marker expression-based accessed on 2011-11-20). The mature cells are Mega (CD34ÀCD41 + + BMP6 ANGPT1, ANGPT2 CSF2 TGFB1 phenotypic assay. Discrepancy between our observation about the CD61 CD45À; six biological replicates), EryB (CD34ÀCD71ÀGlyA ; hi hi + + INHBA NGF, TNFSF12 CSF3 TNFSF10 in vitro effects of the tested ligands and their documented effects in six biological replicates), Neut (FSC SSC CD16 CD11b ; four hi lo + + +/ IL11 TNF literature may be attributable to the differences in experimenting cell biological replicates), Baso (FSC SSC CD22 CD123 CD33 À populations and culture conditions. Further functional validation of CD45dim; six biological replicates), Eos (FSChiSSCloCD123+ dim hi lo + SMADs NF-κB STATs Caspase cascade the surface markers to cell function fidelity would certainly strengthen CD33 ; five biological replicates), Mono (FSC SSC CD14 dim + + our analysis of network directionality; ultimately, our network should CD45 ; five biological replicates), and PreB (CD34ÀCD10 CD19 ; guide the selection of potentially novel HSC-e-regulating cell types, three biological replicates). Quality of the Affymetrix arrays was Quiescence Self-renewal Proliferation Proliferation ligands, and their key intracellular signaling nodes for in-depth assessed using the simpleaffy (v2.32.0) and AnnotationDbi in vivo characterization. A final limitation of this study is that we (v1.18.4) packages of BioConductor. The arrays with average back- C Receptor number DEMono-nucleated Compart. 1 Cell frequency and used a static (human UCB) network to predict potentially dynamic ground more than 2 s.d. from the mean background level of all A A A weighted network cell frequency compartmentalization feedback relationships between HSC-e and the other cell types. arrays and the arrays with present percent is less than 1.5 s.d. from overlaid network overlaid network Exploring how the network connections change during culture the mean present% of all arrays were not used for this study. B C B B C B B C Compart. 2 evolution (Qiao et al, 2012) is an important next step. The assump- Robust multi-array average (RMA) signals of the selected arrays tion of our static network is direct (as opposed to indirect) feedback were computed using the justRMA() function in the limma package from each cell type to HSC-e. Although our in vitro study was specifi- (v3.16.3) of BioConductor. Affymetrix annotation for GeneChip cally designed to enrich for direct effects of ligands on HSC-e by U133AAofAv2 (GEO accession number: GPL4686) was used. using the HSCe receptor expression information as a criterion for To combine the Illumina and the Affymetrix datasets, each selecting test ligands and using a short culture time (7 days) (Csaszar dataset was normalized by the averaged gene expression signal et al, 2014), further analysis of multi-step and adaptive feedback is of the respective ProB arrays. An averaged signal was calculated needed to strengthen links to in vivo hematopoiesis. for probes of the same gene according to Entrez gene identifiers. Collectively, cell–cell communication is fundamental to biologic The post-processed datasets were merged by Entrez gene Figure 7. HSC-e feedback signaling network points to intracellular regulatory motifs for HSC-e fate regulation. tissues. However, it has not been extensively explored as a network identifiers. A Cell–cell communication network for HSC-e fate regulation. The hematopoietic cell-driven network for HSC-e fate regulation. The positive and negative feedback signals are in respect to in vitro expansion of CD34+CD133+CD90+ cells. because a large number of underpinning variables need to be B Intracellular regulatory motifs associated with ligands of different directive effects on in vitro HSC-e fate. considered. Here, we provide a framework to systematically depict Ligand functional enrichment analysis C Interactions between ligand-producing cells and ligands are weighted by the number of corresponding receptors (in terms of species) expressed in HSC-e. The thicker cell–cell communication as a network while exploring the roles of the edge, the higher the weight. cell frequency and spatial distribution in the system. As a next step, For the gene set enrichment analysis (GSEA) in Fig 2B, 13 hemato- D Interactions in (C) are weighted by cell frequencies obtained from fresh human UCB mono-nucleated cell samples shown in Fig 4Ci. connecting the CCC network with more widely studied protein– poietic gene sets (Supplementary Table S1) were compiled from the E Interactions in (D) are weighted by spatial compartmentalization, where 10% of the ligands from peripheral compartment (Baso, Eos and Neut) reach HSC-e. The expressed ligands of the “Others” population, such as BMP2, LTA CSF2, and IL17, are not shown due to the lack of cell frequency information. protein interaction (Kirouac et al, 2010) and gene regulatory GeneGO database on 2012-11-15. GSEA was performed using See also Supplementary Fig S9. (McKinney-Freeman et al, 2012) networks through mechanistic the GSEA software (v2, http://www.broadinstitute.org) with the models of intracellular signaling activity and the resulting cellular minimum gene set size equal to 1, and the other settings as defaults. responses (Janes et al, 2005) will allow us to understand how HSCs See Supplementary Table S1 for GSEA Z-scores. 2000) initiated caspase cascade to cause cell death. Although many activity of specific cell fate decision-associated intracellular regula- integrate exogenous signals to make fateful decisions. The outcome For the biological process enrichment analysis in Figs 3C and 4F, connections between exogeneous ligand stimulation, pathway node tory motifs, which opens opportunities for future study. will not only contribute to the development of more effective gene sets in Supplementary Table S5 were curated from the Meta- activity, and cell phenotype changes were established in cancer cell In summary, our results demonstrate the importance of cell-to- methods for HSC production, but also further our knowledge about Core pathway database (http://thomsonreuters.com/metacore/, lines, these connections led us to the anticipation that exogeneous cell communication in human UCB stem cell fate control. Hemato- HSC (niche) biology and cell–cell communication as a layer of accessed on 2014-03-05). The material is reproduced under a license ligands direct HSC-e toward different cell fate by regulating the poietic cells influence HSC toward certain cell fates by regulating biological regulation. from Thomson Reuters; it may not be copied or re-distributed in

12 Molecular Systems Biology 10: 741 | 2014 ª 2014 The Authors ª 2014 The Authors Molecular Systems Biology 10: 741 | 2014 13 Molecular Systems Biology Hematopoietic cell–cell communication network Wenlian Qiao et al Wenlian Qiao et al Hematopoietic cell–cell communication network Molecular Systems Biology

compiled using the iRefWeb (Turner et al, 2010) resource (accessed ces (Gower & Legendre, 1986). Clustering for the ligand binding ligands MHSCe was simulated using an OR gate model: At the P-value threshold of 0.02, 5 ligands were found to be on 2012-03-05). Additional 38 ligand–receptor interaction pairs from networks was performed using the spectral co-clustering algorithm neutral to HSC-e and 27 were categorized into five functional cate- literatures (as on 2013-02-04) were included. See Supplementary (downloaded from http://adios.tau.ac.il/SpectralCoClustering/ on MHSCe xHSCe LHSCe xPC LPC xMCN LMCN xMCP LMCP ; gories (inducing HSC-e quiescence, self-renewal, differentiation and ¼ð � Þ[ð � Þ[ð � Þ[ð � Þ Table S2 for the resulting 933 ligand–receptor interaction pairs. 2013-06-01) appropriate for weighted graph adjacency matrices proliferation, and inhibiting HSC-e proliferation). Assuming the

i to cell type j, Pij, was computed as and MLP), mature cells in the stem cell niche (MCN), and mature the ligand selection process was modeled as a binomial process with The hierarchical clusters in Fig 2C were obtained using the Ward cells in the peripheral tissues (MCP). Randomly generated logic distribution X~B(33, 0.5), where 33 is the number of test ligands. agglomeration method with the Manhattan distance matrix. Confi- f R f R vectors xHSCe, xPC, xMCN, and xMCP represented the probability The expected number of effective ligands was 33*0.5 16. The P i ik j jk ; � dence of the clusters was quantified by approximately unbiased ij (PHSCe, PPC, PMCN, and PMCP) of the ligands of each compartment probability of having 27 effective ligands is ¼ K 0 fiRil fmRmk1 (AU) P-values (Shimodaira, 2002, 2004), a type of bootstrap X I M to reach HSCe. Enrichment (E) of HSCe-targeting ligands MHSCe in @ A P-values, computed using the pvclust package (v1.2-2) in R (v3.0.0). P P a biological process mediated by ligand set B was quantified as 33 6 P X 27 0:527 1 0:5 0:0001 where f is the normalized cell frequency of cell type i by the total ð ¼ Þ¼ 27 ð � Þ � i following: �� Identification of differentially over-expressed genes cell frequency of the analyzed cell types, thus fi is between 0 and 1; R is the number of receptors that cell type i expressed for n MHSCe B Prior to the in vitro experiments for testing the activity of HSCe- ik E ð ^ Þ ; For the cell type of interest, one-way pairwise Wilcoxon test (R, ligand k; K is the total number of ligands that cell type i binds; I is ¼ n B targeting ligands on HSC-e, we sought to prioritize ligands for exper- ð Þ v2.15.1) was performed between the gene expression profiles of the the total number of ligands that cell type i binds; and M is the total iments. To do that, we performed a literature survey on ligands that

interested cell type and the profiles of each of the other cell types. number of cell types that ligand k binds. The figures were drawn where n(MHSCe ^ B) is the number of HSCe-targeting ligands in had been used in in vitro cell culture of human cord blood-derived P-values were adjusted using the Benjamini & Hochberg method (or by modifying the plotPAC() function in the bipartite package biological process B, and n(B) is the number of ligands in biologi- cells; 11 ligands fell in this category (Supplementary Table S7).

false discovery rates, FDR). At a given threshold, the ligand and (v1.18) in R (v.3.0.0). cal process B. For each test condition (i.e., combination of PHSCe, Ligands such as ANGPT1, ANGPT2, ANGPTL3, and BMP2 had been

receptor genes that differentially over-expressed comparing to six PPC, PMNC, and PMCP), enriched scores from 500 simulations were used in mice or human bone marrow cells (Supplementary Table other cell types (the threshold was set arbitrarily) were defined as Network comparison averaged. Content of 11 manually curated ligand sets of biological S7), so they were also prioritized for experiments in our study. the differentially over-expressed ligands and receptors of the cell processes are tabulated in Supplementary Table S5. Excluding these ligands from our analysis, 15 ligands out of 18 type. The identified receptors of each cell type were compared to To compare interaction patterns between the network of ligand tested ligands were effective. The corresponding probability is hematopoietic cell type-specific receptors using receiver operating source and the network of ligand sink, for each network, the In vitro experiments characteristic (Supplementary Fig S1). The cell type-specific recep- numbers of overlapped ligands between one module and the other 18 P X 27 0:515 1 0:5 3 0:003 tors are (1) ACVRL1 (for TGFB1), ENG (for TGFB1), EPOR (for modules were obtained. The overlap of ligands between modules in � ð ¼ Þ¼ 15 ð � Þ � Human Lin cells were isolated from UCB samples collected from �� KIT), FKBP1A (for TGFB1), IL2RG (for IL7), IL7R (for IL7), ITGAV the network of ligand source S ={9, 13, 10, 12, 12, 17}, and the consenting donors according to ethically approved procedures at (for TGFB1), ITGB6 (for TGFB1), ITGN8 (for TGFB1), KIT (for overlap of ligands between modules in the network of ligand sink Mt. Sinai Hospital (Toronto, ON, Canada). Forty Lin� To dictate the respective regulatory effects of HSCe, CMP, GMP, low + + KITLG), LTBP1 (for TGFB1), LTBP4 (for TGFB1), MPL (for THPO), T ={75, 75, 69}. Two-sample t-test was performed for S and T in R Rho CD34 CD38�CD45RA�CD49f cells were sorted and MEP, MLP, Mega, EryB, Mono, Neut, Eos, Baso, PreB, and Others TGFBR1 (for TGFB1), TGFBR2 (for TGFB1), TGFBR3 (for TGFB1), (v3.0.0). dispensed per well in a 96-well V-bottom plate with a MoFloXDP on HSC-e cell fates, the tested ligands of each cell type were VTN (for TGFB1), CD34 and ITGA6 (CD49f) for HSCe; (2) IL3RA flow cytometer (Beckman Coulter). The cells were cultured in a extracted from the CCC network in Supplementary Table S4. Func- (for IL3), CSF2RA (for CSF2), CSF2RB (for CSF2), CSF3R (for CSF2), Flow cytometry analysis serum-free condition supplemented with 100 ng/ml SCF, 100 ng/ml tional enrichment analysis was performed for each cell type using EPOR (for KIT), KIT (for KIT), MPL (for THPO), CD34, CD38, FLT3 FLT3LG, 50 ng/ml THPO, and a test ligand at specific concentration. hypergeometric Z-scores,

(CD135) for CMP; (3) MPL (for THPO), EPOR (for EPO), CD34 and Human UCB samples were collected from consenting donors accord- On day 7, cells were stained. Total cell counts (NTotal), m + + + k n CD38 for MEP; (4) CSF3R (for CSF3), CD34, CD38, FLT3, PTPRC ing to ethically approved procedures at Mt. Sinai Hospital (Toronto, CD34 CD133 CD90 cell counts (NHSC-enriched), and CD34� cell Z � N ; ¼ m N m N n (CD45RA) for GMP; (4) IL2RG (for IL7), IL7R (for IL7), CD34, ON, Canada). Mono-nucleated cells were obtained by depleting red counts (NMature) were obtained using an LSRFortessa flow cytometer n N �N N�1 PTPRC (CD45RA) for MLP; (5) MPL (for THPO), ITGA2B (CD41), blood cells (RBC) using RBC lysis buffer (0.15 M NH Cl, 0.01 M (BD Bioscience). Progenitor cell counts were calculated as � 4 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi�� ITGB3 (CD61) for Mega; (6) EPOR (for EPO), GYPA (CD235a) for KHCO3, 0.1 mM EDTA) as previously described (Kirouac et al, N N N . See also “optimization of in vitro where N = 117 is the number of HSCe-targeting ligands expressed Total � HSC-enriched � Mature EryB; (7) CD14 for Mono; (8) CD22 and IL3RA (CD123) for Baso; 2009). Lineage-negative (Lin�) cells were isolated from the experiments” in the Supplementary Information S1. by the 13 cell types, m is the number of ligands in a given function (9) IL3RA (CD123) for Eos; (10) FCGR3A (CD16) and ITGAM mono-nucleated cell fraction using the StemSep system or the group, n is the number of expressed ligands of the cell type of (CD11b) for Neut; and (11) IL2RG (for IL4), IL4R (for IL4), IL13RA1 EasySep system for human progenitor cell enrichment (StemCell Statistical analysis interest, and k is the number of expressed ligands in the function (for IL4), MME (CD10), and CD19 for PreB. Technologies, Inc., Vancouver, BC, Canada), according to the manu- group of interest. facturer’s protocol. Cell frequencies shown in Fig 4Ci and 4Di were To assess the effects of each test ligand (in addition to SCF, THPO Network construction obtained from mono-nucleated cells of fresh UCB samples and and FLT3LG) on in vitro HSC-e fate decisions, a mixed-linear model Functional HSC-e feedback signaling network thawed Lin� cell samples, respectively. The cells were stained using was constructed with the experiment identifier as the random effect Directionality of the CCC network was defined by the expression of the following antibodies in 1:100 unless stated otherwise: CD90 to account for the variability from experiment to experiment. In Fig 7C, strength of the produced signals of function group k from ligand and receptor genes on the cell types of interest, and the (FITC, 1:50), CD38 (PE, PECy5, APC), CD45RA (1:50, APC), CD34 The analysis was performed using the lme() function of the nlme cell type i to HSC-e was modeled as

14 Molecular Systems Biology 10: 741 | 2014 ª 2014 The Authors ª 2014 The Authors Molecular Systems Biology 10: 741 | 2014 15 Molecular Systems Biology Hematopoietic cell–cell communication network Wenlian Qiao et al Wenlian Qiao et al Hematopoietic cell–cell communication network Molecular Systems Biology

whole or in part without the written consent of the scientific busi- ligand–receptor pairs in Supplementary Table S2. If “Cell A” (PE-Cy7), CD49f (PE-Cy5, 1:50), CD7 (FITC), CD10 (FITC), CD135 package (v3.1-113) in R (v2.15.1). The source code is provided as ness of Thomson Reuter. expresses a receptor for ligand x and “Cell B” expresses ligand x, an (1:50, PE), CD45RA (1:50, APC), CD71 (FITC), CD235a (PE), CD61 Supplementary Information S1. arrow is drawn from “Cell B” to “Cell A.” Networks were built in R (FITC), CD41 (PE), CD45 (PE-Cy7), CD14 (PE), CD16 (PE), CD11b Since we were mostly concerned with not missing any effective Ligand/receptor database (v2.15.1) and visualized in Cytoscape (v2.8.3). The R code is avail- (PE-Cy7), CD22 (FITC), CD33 (PE), CD123 (PE-Cy5), CD19 (FITC), ligands (type II error) that will inform future research, nominal able upon request. and CD10 (PE). All the antibodies were from BD Biosciences, Missis- P-values of the mixed model were reported without correction for Using gene ontology terms “cytokine activity,” “growth factor activ- sauga, ON, Canada. multiple tests. The ligands were categorized using definition in ity,” “hormone activity,” and “receptor activity,” 417 genes with Bipartite network analysis Table 1. Ligand categorization was performed for significance ligand activity and 1,723 genes with receptor activity were compiled Logic modeling P-value thresholds of 0.01, 0.02 and 0.05 (Supplementary Table S9). from BioMart (Kasprzyk, 2011) (accessed on 2012-02-29). Ligand– Clustering for the ligand production networks was performed based See also “statistical analysis for in vitro experiments” in the receptor interaction pairs documented in public domains were on Jaccard distances appropriate for binary graph adjacency matri- The effect of cell localization on the identity of HSCe-targeting Supplementary Information S1. compiled using the iRefWeb (Turner et al, 2010) resource (accessed ces (Gower & Legendre, 1986). Clustering for the ligand binding ligands MHSCe was simulated using an OR gate model: At the P-value threshold of 0.02, 5 ligands were found to be on 2012-03-05). Additional 38 ligand–receptor interaction pairs from networks was performed using the spectral co-clustering algorithm neutral to HSC-e and 27 were categorized into five functional cate- literatures (as on 2013-02-04) were included. See Supplementary (downloaded from http://adios.tau.ac.il/SpectralCoClustering/ on MHSCe xHSCe LHSCe xPC LPC xMCN LMCN xMCP LMCP ; gories (inducing HSC-e quiescence, self-renewal, differentiation and ¼ð � Þ[ð � Þ[ð � Þ[ð � Þ Table S2 for the resulting 933 ligand–receptor interaction pairs. 2013-06-01) appropriate for weighted graph adjacency matrices proliferation, and inhibiting HSC-e proliferation). Assuming the

(Dhillon, 2001). where LHSCe, LPC, LMCN, and LMCP are the differentially over- probability that a selected ligand is functional is 0.5 and that Hierarchical clustering Potential of apparent competition (Muller et al, 1999) of cell type expressed ligands of HSCe, progenitor cells (CMP, GMP, MEP, the effectiveness of test ligands was independent from each other, i to cell type j, Pij, was computed as and MLP), mature cells in the stem cell niche (MCN), and mature the ligand selection process was modeled as a binomial process with The hierarchical clusters in Fig 2C were obtained using the Ward cells in the peripheral tissues (MCP). Randomly generated logic distribution X~B(33, 0.5), where 33 is the number of test ligands. agglomeration method with the Manhattan distance matrix. Confi- f R f R vectors xHSCe, xPC, xMCN, and xMCP represented the probability The expected number of effective ligands was 33*0.5 16. The P i ik j jk ; � dence of the clusters was quantified by approximately unbiased ij (PHSCe, PPC, PMCN, and PMCP) of the ligands of each compartment probability of having 27 effective ligands is ¼ K 0 fiRil fmRmk1 (AU) P-values (Shimodaira, 2002, 2004), a type of bootstrap X I M to reach HSCe. Enrichment (E) of HSCe-targeting ligands MHSCe in @ A P-values, computed using the pvclust package (v1.2-2) in R (v3.0.0). P P a biological process mediated by ligand set B was quantified as 33 6 P X 27 0:527 1 0:5 0:0001 where f is the normalized cell frequency of cell type i by the total ð ¼ Þ¼ 27 ð � Þ � i following: �� Identification of differentially over-expressed genes cell frequency of the analyzed cell types, thus fi is between 0 and 1; R is the number of receptors that cell type i expressed for n MHSCe B Prior to the in vitro experiments for testing the activity of HSCe- ik E ð ^ Þ ; For the cell type of interest, one-way pairwise Wilcoxon test (R, ligand k; K is the total number of ligands that cell type i binds; I is ¼ n B targeting ligands on HSC-e, we sought to prioritize ligands for exper- ð Þ v2.15.1) was performed between the gene expression profiles of the the total number of ligands that cell type i binds; and M is the total iments. To do that, we performed a literature survey on ligands that interested cell type and the profiles of each of the other cell types. number of cell types that ligand k binds. The figures were drawn where n(MHSCe ^ B) is the number of HSCe-targeting ligands in had been used in in vitro cell culture of human cord blood-derived P-values were adjusted using the Benjamini & Hochberg method (or by modifying the plotPAC() function in the bipartite package biological process B, and n(B) is the number of ligands in biologi- cells; 11 ligands fell in this category (Supplementary Table S7). false discovery rates, FDR). At a given threshold, the ligand and (v1.18) in R (v.3.0.0). cal process B. For each test condition (i.e., combination of PHSCe, Ligands such as ANGPT1, ANGPT2, ANGPTL3, and BMP2 had been receptor genes that differentially over-expressed comparing to six PPC, PMNC, and PMCP), enriched scores from 500 simulations were used in mice or human bone marrow cells (Supplementary Table other cell types (the threshold was set arbitrarily) were defined as Network comparison averaged. Content of 11 manually curated ligand sets of biological S7), so they were also prioritized for experiments in our study. the differentially over-expressed ligands and receptors of the cell processes are tabulated in Supplementary Table S5. Excluding these ligands from our analysis, 15 ligands out of 18 type. The identified receptors of each cell type were compared to To compare interaction patterns between the network of ligand tested ligands were effective. The corresponding probability is hematopoietic cell type-specific receptors using receiver operating source and the network of ligand sink, for each network, the In vitro experiments characteristic (Supplementary Fig S1). The cell type-specific recep- numbers of overlapped ligands between one module and the other 18 P X 27 0:515 1 0:5 3 0:003 tors are (1) ACVRL1 (for TGFB1), ENG (for TGFB1), EPOR (for modules were obtained. The overlap of ligands between modules in � ð ¼ Þ¼ 15 ð � Þ � Human Lin cells were isolated from UCB samples collected from �� KIT), FKBP1A (for TGFB1), IL2RG (for IL7), IL7R (for IL7), ITGAV the network of ligand source S ={9, 13, 10, 12, 12, 17}, and the consenting donors according to ethically approved procedures at (for TGFB1), ITGB6 (for TGFB1), ITGN8 (for TGFB1), KIT (for overlap of ligands between modules in the network of ligand sink Mt. Sinai Hospital (Toronto, ON, Canada). Forty Lin� To dictate the respective regulatory effects of HSCe, CMP, GMP, low + + KITLG), LTBP1 (for TGFB1), LTBP4 (for TGFB1), MPL (for THPO), T ={75, 75, 69}. Two-sample t-test was performed for S and T in R Rho CD34 CD38�CD45RA�CD49f cells were sorted and MEP, MLP, Mega, EryB, Mono, Neut, Eos, Baso, PreB, and Others TGFBR1 (for TGFB1), TGFBR2 (for TGFB1), TGFBR3 (for TGFB1), (v3.0.0). dispensed per well in a 96-well V-bottom plate with a MoFloXDP on HSC-e cell fates, the tested ligands of each cell type were VTN (for TGFB1), CD34 and ITGA6 (CD49f) for HSCe; (2) IL3RA flow cytometer (Beckman Coulter). The cells were cultured in a extracted from the CCC network in Supplementary Table S4. Func- (for IL3), CSF2RA (for CSF2), CSF2RB (for CSF2), CSF3R (for CSF2), Flow cytometry analysis serum-free condition supplemented with 100 ng/ml SCF, 100 ng/ml tional enrichment analysis was performed for each cell type using EPOR (for KIT), KIT (for KIT), MPL (for THPO), CD34, CD38, FLT3 FLT3LG, 50 ng/ml THPO, and a test ligand at specific concentration. hypergeometric Z-scores,

N Si;k fi RHSC e;k;n; and JED contributed reagents/materials/analysis tools. All the authors reviewed Conference on Knowledge Discovery and Data Mining, pp 269 – 274. New Kirito K, Fox N, Komatsu N, Kaushansky K (2005) Thrombopoietin enhances ¼ � n 1 the manuscript. York, NY, USA: ACM expression of vascular endothelial growth factor (VEGF) in primitive X¼ where fi is the frequency of cell type i, n is the number of Ding L, Saunders TL, Enikolopov G, Morrison SJ (2012) Endothelial and hematopoietic cells through induction of HIF-1 alpha. Blood 105: expressed ligands of function group k by cell type i, and R is the Conflict of interest perivascular cells maintain haematopoietic stem cells. Nature 481: 457 – 462 4258 – 4263 expression level of receptor gene n. Cell frequencies are from The authors declare that they have no conflict of interest. Dorrell C, Gan OI, Pereira DS, Hawley RG, Dick JE (2000) Expansion of human Kirouac DC, Madlambayan GJ, Yu M, Sykes EA, Ito C, Zandstra PW (2009) Fig 4Ci. cord blood CD34(+)CD38( ) cells in ex vivo culture during retroviral Cell-cell interaction networks regulate blood stem and progenitor cell fate. À transduction without a corresponding increase in SCID repopulating cell Mol Syst Biol 5: 293 Pathway analysis References (SRC) frequency: dissociation of SRC phenotype and function. Blood 95: Kirouac DC, Ito C, Csaszar E, Roch A, Yu M, Sykes EA, Bader GD, Zandstra PW 102 – 110 (2010) Dynamic interaction networks in a hierarchically organized tissue. Intracellular regulatory factors downstream of 16 out of the 19 Apicella CL, Marlowe FW, Fowler JH, Christakis NA (2012) Social networks and Edgar R, Domrachev M, Lash AE (2002) Gene expression Omnibus: NCBI gene Mol Syst Biol 6: 417 ligands shown in Fig 7A are available in the MetaCore database cooperation in hunter-gatherers. Nature 481: 497 – 501 expression and hybridization array data repository. Nucleic Acids Res 30: Kischkel FC, Lawrence DA, Chuntharapai A, Schow P, Kim KJ, Ashkenazi A (http://thomsonreuters.com/metacore/, accessed on 2014-04-01). Batard P, Monier MN, Fortunel N, Ducos K, Sansilvestri-Morel P, Phan T, 207 – 210 (2000) Apo2L/TRAIL-dependent recruitment of endogenous FADD and The regulatory factors of each ligand were compiled and compared Hatzfeld A, Hatzfeld JA (2000) TGF-(beta)1 maintains hematopoietic Frankenstein Z, Alon U, Cohen IR (2006) The immune-body cytokine network caspase-8 to death receptors 4 and 5. Immunity 12: 611 – 620 to the regulatory factors of the other ligands of the same functional immaturity by a reversible negative control of cell cycle and induces CD34 defines a social architecture of cell interactions. Biol Direct 1: 32 Kunisaki Y, Bruns I, Scheiermann C, Ahmed J, Pinho S, Zhang D, Mizoguchi T, group. Enrichment of ligands of the same functional group to each antigen up-modulation. J Cell Sci 113(Pt 3): 383 – 390 Gower JC, Legendre P (1986) Metric and Euclidean properties of dissimilarity Wei Q, Lucas D, Ito K, Mar JC, Bergman A, Frenette PS (2013) Arteriolar regulatory factor was calculated by a permutation test. For each Bendall SC, Simonds EF, Qiu P, el Amir AD, Krutzik PO, Finck R, Bruggner RV, coefficients. J Classif 3: 5 – 48 niches maintain haematopoietic stem cell quiescence. Nature 502: regulatory factor, random functional categorization (quiescence Melamed R, Trejo A, Ornatsky OI, Balderas RS, Plevritis SK, Sachs K, Pe’er Gu L, Chiang K-Y, Zhu N, Findley HW, Zhou M (2007) Contribution of STAT3 637 – 643 induction, self-renewal induction, proliferation induction, and D, Tanner SD, Nolan GP (2011) Single-cell mass cytometry of differential to the activation of survivin by GM-CSF in CD34+ cell lines. Exp Hematol Laurenti E, Doulatov S, Zandi S, Plumb I, Chen J, April C, Fan J-B, Dick JE proliferation inhibition) was performed for 100,000 times. The ratio immune and drug responses across a human hematopoietic continuum. 35: 957 – 966 (2013) The transcriptional architecture of early human hematopoiesis between the number of times that a regulatory factor randomly fell Science 332: 687 – 696 Harel-Bellan A, Farrar W (1987) Modulation of proto-oncogene expression identifies multilevel control of lymphoid commitment. Nat Immunol 14: in a functional category more frequent than the actual categoriza- Boitano AE, Wang J, Romeo R, Bouchez LC, Parker AE, Sutton SE, Walker JR, by colony stimulating factors. Biochem Biophys Res Commun 148: 756 – 763 tion and 100,000 is defined as the permutation P-value. The results Flaveny CA, Perdew GH, Denison MS, Schultz PG, Cooke MP (2010) Aryl 1001 – 1008 Madlambayan GJ, Rogers I, Kirouac DC, Yamanaka N, Mazurier F, Doedens of pathway analysis for HSC-e differentiation-inducing ligands are hydrocarbon receptor antagonists promote the expansion of human Heazlewood SY, Neaves RJ, Williams B, Haylock DN, Adams TE, Nilsson SK M, Casper RF, Dick JE, Zandstra PW (2005) Dynamic changes in not presented because pathway information was only found for one hematopoietic stem cells. Science 329: 1345 – 1348 (2013) Megakaryocytes co-localise with hemopoietic stem cells and release cellular and microenvironmental composition can be controlled to elicit differentiation-inducing ligand BMP4, and the data are not sufficient Burdette JE, Jeruss JS, Kurley SJ, Lee EJ, Woodruff TK (2005) Activin A mediates cytokines that up-regulate stem cell proliferation. Stem Cell Res 11: in vitro human hematopoietic stem cell expansion. Exp Hematol 33: for an enrichment analysis. The material from the MetaCore path- growth inhibition and cell cycle arrest through Smads in human breast 782 – 792 1229 – 1239 way database is reproduced under a license from Thomson Reuters. cancer cells. Cancer Res 65: 7968 – 7975 Hock H, Hamblen MJ, Rooke HM, Schindler JW, Saleque S, Fujiwara Y, Orkin Mallick S, Pal BC, Vedasiromoni JR, Kumar D, Saha KD (2012) Corchorusin-D Carter BZ (2001) Cytokine-regulated expression of survivin in myeloid SH (2004) Gfi-1 restricts proliferation and preserves functional integrity of directed apoptosis of K562 cells occurs through activation of Supplementary information for this article is available online: leukemia. Blood 97: 2784 – 2790 haematopoietic stem cells. Nature 431: 1002 – 1007 mitochondrial and death receptor pathways and suppression of AKT/PKB http://msb.embopress.org Ceol A, Chatr Aryamontri A, Licata L, Peluso D, Briganti L, Perfetto L, Holien T, Våtsveen TK, Hella H, Rampa C, Brede G, Grøseth LAG, Rekvig M, pathway. Cell Physiol Biochem 30: 915 – 926 Castagnoli L, Cesareni G (2010) MINT, the molecular interaction database: Børset M, Standal T, Waage A, Sundan A (2012) Bone morphogenetic Mayani BH, Lansdorp PM (1994) Thy-1 expression is linked to functional Acknowledgements 2009 update. Nucleic Acids Res 38:D532 – D539 proteins induce apoptosis in multiple myeloma cells by Smad-dependent properties of primitive hematopoietic progenitor cells from human WQ was supported by Ontario Graduate Scholarships and a National Science Chow A, Lucas D, Hidalgo A, Mendez-Ferrer S, Hashimoto D, Scheiermann C, repression of MYC. Leukemia 26: 1073 – 1080 umbilical cord blood. Blood 83: 2410 – 2417 and Engineering Research Council postgraduate scholarship. WW was Battista M, Leboeuf M, Prophete C, van Rooijen N, Tanaka M, Merad M, Huang H, Cantor AB (2009) Common features of megakaryocytes and McKinney-Freeman S, Cahan P, Li H, Lacadie SA, Huang H-T, Curran M, supported by an Ontario Stem Cell Initiative post-doctoral fellowship. EL was Frenette PS (2011) Bone marrow CD169+ macrophages promote the hematopoietic stem cells: what’s the connection? J Cell Biochem 107: Loewer S, Naveiras O, Kathrein KL, Konantz M, Langdon EM, Lengerke C, supported by the Swiss National Science Foundation, Roche, and the FSBMB retention of hematopoietic stem and progenitor cells in the mesenchymal 857 – 864 Zon LI, Collins JJ, Daley GQ (2012) The transcriptional landscape of (Foundation Suisse pour les Bourses en Médecine et Biologie). This work was stem cell niche. J Exp Med 208: 261 – 271 Hughes DP, Marron MB, Brindle NPJ (2003) The antiinflammatory endothelial hematopoietic stem cell ontogeny. Cell Stem Cell 11: 701 – 714 supported by grants to SJW from the Canadian Institutes of Health Research, Csaszar E, Kirouac DC, Yu M, Wang W, Qiao W, Cooke MP, Boitano AE, Ito C, tyrosine kinase Tie2 interacts with a novel nuclear factor-kappaB inhibitor Muller CB, Adriaanse ICT, Belshaw R, Godfray HCJ (1999) The structure of an the Ontario Research Fund, and the SickKids Foundation; grants to GDB from Zandstra PW (2012) Rapid expansion of human hematopoietic stem cells ABIN-2. Circ Res 92: 630 – 636 aphid-parasitoid community. J Anim Ecol 68: 346 – 370 NRNB (U.S. National Institutes of Health, National Center for Research by automated control of inhibitory feedback signaling. Cell Stem Cell 10: Ito CY, Kirouac DC, Madlambayan GJ, Yu M, Rogers I, Zandstra PW (2010) The Nagasawa T (2007) The chemokine CXCL12 and regulation of HSC and B + Resources Grant Number P41 GM103504); grants to JED from Genome Canada 218 – 229 AC133 CD38À, but not the rhodamine-low, phenotype tracks LTC-IC and lymphocyte development in the bone marrow niche. Adv Exp Med Biol through the Ontario Genomics Institute, Ontario Institute for Cancer Research, Csaszar E, Wang W, Usenko T, Qiao W, Delaney C, Bernstein ID, Zandstra PW SRC function in human cord blood ex vivo expansion cultures. Blood 115: 602: 69 – 75 and a Summit Award with funds from the province of Ontario, the Canadian (2014) Blood stem cell fate regulation by Delta-1-mediated rewiring of 257 – 260 Nakamura Y, Arai F, Iwasaki H, Hosokawa K, Kobayashi I, Gomei Y, Institutes for Health Research, a Canada Research Chair, the Princess Margaret IL-6 paracrine signaling. Blood 123: 650 – 658 Janes KA, Albeck JG, Gaudet S, Sorger PK, Lauffenburger DA, Yaffe MB (2005)A Matsumoto Y, Yoshihara H, Suda T (2010) Isolation and characterization Hospital Foundation, the Terry Fox Research Institute, Canadian Cancer Soci- Danet GH, Lee HW, Luongo JL, Simon MC, Bonnet DA (2001) Dissociation systems model of signaling identifies a molecular basis set for of endosteal niche cell populations that regulate hematopoietic stem cells. ety Research Institute, and in part by the Ontario Ministry of Health and Long between stem cell phenotype and NOD/SCID repopulating activity in cytokine-induced apoptosis. Science 310: 1646 – 1653 Blood 116: 1422 – 1432 Term Care (OMOHLTC, the views expressed do not necessarily reflect those of human peripheral blood CD34(+) cells after ex vivo expansion. Exp Jeong H, Mason SP, Barabasi AL, Oltvai ZN (2001) Lethality and centrality in Nombela-Arrieta C, Pivarnik G, Winkel B, Canty KJ, Harley B, Mahoney JE, Park the OMOHLTC); and grants to PWZ from the Human Frontier Science Program, Hematol 29: 1465 – 1473 protein networks. Nature 411: 41 – 42 S-Y, Lu J, Protopopov A, Silberstein LE (2013) Quantitative imaging of the Leukemia and Lymphoma Society of Canada, the Canadian Stem Cell De Haan G, Weersing E, Dontje B, van Os R, Bystrykh LV, Vellenga E, Miller G Kale VP (2004) Differential activation of MAPK signaling pathways by haematopoietic stem and progenitor cell localization and hypoxic status Network, and the Ministry of Research and Innovation of Ontario. PWZ is the (2003) In vitro generation of long-term repopulating hematopoietic stem TGF-beta1 forms the molecular mechanism behind its dose-dependent in the bone marrow microenvironment. Nat Cell Biol 15: 533 – 543 Canada Research Chair in Stem Cell Bioengineering. The authors would like to cells by fibroblast growth factor-1. Dev Cell 4: 241 – 251 bidirectional effects on hematopoiesis. Stem Cells Dev 13: 27 – 38 Notta F, Doulatov S, Laurenti E, Poeppl A, Jurisica I, Dick JE (2011) Isolation of thank the members of the PWZ laboratory and Dr. Daniel Kirouac for their Descamps S, Toillon RA, Adriaenssens E, Pawlowski V, Cool SM, Nurcombe Kasprzyk A (2011) BioMart: driving a paradigm change in biological data single human hematopoietic stem cells capable of long-term multilineage helpful discussion. V, Le Bourhis X, Boilly B, Peyrat JP, Hondermarck H (2001) Nerve management. Database (Oxford) 2011: 1 – 3 engraftment. Science 333: 218 – 221 growth factor stimulates proliferation and survival of human breast Kawakita T, Shiraki K, Yamanaka Y, Yamaguchi Y, Saitou Y, Enokimura N, Novershtern N, Subramanian A, Lawton LN, Mak RH, Haining WN, McConkey Author contributions cancer cells through two distinct signaling pathways. J Biol Chem 276: Yamamoto N, Okano H, Sugimoto K, Murata K, Nakano T (2004) ME, Habib N, Yosef N, Chang CY, Shay T, Frampton GM, Drake AC, Leskov WQ and PWZ conceived and designed the study and wrote the manuscript. 17864 – 17870 Functional expression of TWEAK in human hepatocellular carcinoma: I, Nilsson B, Preffer F, Dombkowski D, Evans JW, Liefeld T, Smutko JS, Chen WQ performed in silico studies and analyzed in vitro data. WW performed in Dhillon IS (2001) Co-clustering documents and words using bipartite spectral possible implication in cell proliferation and tumor angiogenesis. Biochem J et al (2011) Densely interconnected transcriptional circuits control cell vitro experiments and contributed to drafting the manuscript. EL, ALT, SJW, GB graph partitioning. In Proceedings of the Seventh ACM SIGKDD International Biophys Res Commun 318: 726 – 733 States in human hematopoiesis. Cell 144: 296 – 309

16 Molecular Systems Biology 10: 741 | 2014 ª 2014 The Authors ª 2014 The Authors Molecular Systems Biology 10: 741 | 2014 17 Molecular Systems Biology Hematopoietic cell–cell communication network Wenlian Qiao et al Wenlian Qiao et al Hematopoietic cell–cell communication network Molecular Systems Biology

Olesen JM, Bascompte J, Dupont YL, Jordano P (2007) The modularity of Takizawa H, Boettcher S, Manz MG (2012) Demand-adapted regulation of pollination networks. Proc Natl Acad Sci USA 104: 19891 – 19896 early hematopoiesis in infection and inflammation. Blood 119: 2991 – 3002 Petzer AL, Zandstra PW, Piret JM, Eaves CJ (1996) Differential cytokine effects Turner B, Razick S, Turinsky AL, Vlasblom J, Crowdy EK, Cho E, Morrison K, on primitive (CD34+CD38 ) human hematopoietic cells: novel responses Donaldson IM, Wodak SJ (2010) iRefWeb: interactive analysis of À to Flt3-ligand and thrombopoietin. J Exp Med 183: 2551 – 2558 consolidated protein interaction data and their supporting evidence. Qiao W, Quon G, Csaszar E, Yu M, Morris Q, Zandstra PW (2012) PERT: a Database (Oxford) 2010: baq023 method for expression deconvolution of human blood samples from Vitali S, Glattfelder JB, Battiston S (2011) The network of global corporate varied microenvironmental and developmental conditions. PLoS Comput control. PLoS ONE 6:e25995 Biol 8:e1002838 Yoshizaki A, Nakayama T, Yamazumi K, Yakata Y, Taba M, Sekine I (2006) Robb L (2007) Cytokine receptors and hematopoietic differentiation. Oncogene Expression of interleukin (IL)-11 and IL-11 receptor in human 26: 6715 – 6723 colorectal adenocarcinoma: IL-11 up-regulation of the invasive and Rousseeuw PJ (1987) Silhouettes: a graphical aid to the interpretation and proliferative activity of human colorectal carcinoma cells. Int J Oncol 29: validation of cluster analysis. J Comput Appl Math 20: 53 – 65 869 – 876 Schepers K, Pietras EM, Reynaud D, Flach J, Binnewies M, Garg T, Wagers AJ, Zamai L, Secchiero P, Pierpaoli S, Bassini A, Papa S, Alnemri ES, Guidotti L, Hsiao EC, Passegue E (2013) Myeloproliferative neoplasia remodels the Vitale M, Zauli G (2000) TNF-related apoptosis-inducing ligand (TRAIL) as endosteal bone marrow niche into a self-reinforcing leukemic niche. Cell a negative regulator of normal human erythropoiesis. Blood 95: Stem Cell 13: 285 – 299 3716 – 3724 Shimodaira H (2002) An approximately unbiased test of phylogenetic tree Zhang J, Niu C, Ye L, Huang H, He X, Tong WG, Ross J, Haug J, Johnson T, selection. Syst Biol 51: 492 – 508 Feng JQ, Harris S, Wiedemann LM, Mishina Y, Li L (2003) Identification of Shimodaira H (2004) Approximately unbiased tests of regions using the haematopoietic stem cell niche and control of the niche size. Nature multistep-multiscale boot-strap resamping. Ann Stat 32: 2616 – 2641 425: 836 – 841 Schwanhausser B, Busse D, Li N, Dittmar G, Schuchhardt J, Wolf J, Chen W, Zhang J, Grindley JC, Yin T, Jayasinghe S, He XC, Ross JT, Haug JS, Rupp D, Selbach M (2011) Global quantification of mammalian gene expression Porter-Westpfahl KS, Wiedemann LM, Wu H, Li L (2006) PTEN maintains control. Nature 473: 337 – 342 haematopoietic stem cells and acts in lineage choice and leukaemia Shima Y, Nakao K, Nakashima T, Kawakami A, Nakata K, Hamasaki K, Kato Y, prevention. Nature 441: 518 – 522 Eguchi K, Ishii N (1999) Activation of caspase-8 in transforming growth Zhao C, Xiu Y, Ashton J, Xing L, Morita Y, Jordan CT, Boyce BF (2012) factor-beta-induced apoptosis of human hepatoma cells. Hepatology 30: Noncanonical NF-jB signaling regulates hematopoietic stem cell 1215 – 1222 self-renewal and microenvironment interactions. Stem Cells 30: 709 – 718 Soni S, Bala S, Hanspal M (2008) Requirement for erythroblast-macrophage protein (Emp) in definitive erythropoiesis. Blood Cells. Mol Dis 41: License: This is an open access article under the 141 – 147 terms of the Creative Commons Attribution 4.0 Stein SJ, Baldwin AS (2013) Deletion of the NF-jB subunit p65/RelA in the License, which permits use, distribution and reproduc- hematopoietic compartment leads to defects in hematopoietic stem cell tion in any medium, provided the original work is function. Blood 121: 5015 – 5024 properly cited.

18 Molecular Systems Biology 10: 741 | 2014 ª 2014 The Authors further reading

The EMBO Journal EMBO Reports Degradation of IF1 controls Prolyl-isomerase Pin1 controls energy metabolism during normal and cancer stem cells of Induction of autophagy Reviews osteogenic differentiation of the breast. supports the bioenergetic Cancer heterogeneity--a stem cells. Rustighi A, Zannini A, Tiberi L, demands of quiescent muscle multifaceted view. Sánchez-Aragó M, García-Bermúdez Sommaggio R, Piazza S, Sorrentino G, stem cell activation. Nuzzo S, Tuscano A, Eterno V, Benvenuti De Sousa E Melo F, Vermeulen L, Fessler J, Martínez-Reyes I, Santacatterina F, Tang AH, Rando TA. F, Santarpia L, Aifantis I, Rosato A, E, Medema JP. Cuezva JM. DOI: 10.15252/embj.201488278 Bicciato S, Zambelli A, Del Sal G. DOI: 10.1038/embor.2013.92 DOI: 10.1038/embor.2013.72 Published 14.10.2014 DOI: 10.1002/emmm.201302909 Published 12.07.2013 Published 31.05.2013 Published 12.12.2013 Escargot maintains stemness A matter of life and death: self- Prdm14 promotes germline and suppresses differentiation Cardiomyocyte proliferation renewal in stem cells. fate and naive pluripotency by in Drosophila intestinal stem repressing FGF signalling and DNA and progenitor cell recruitment Fuchs E, Chen T. cells. methylation. underlie therapeutic DOI: 10.1038/embor.2012.197 Korzelius J, Naumann SK, Loza-Coll MA, regeneration after myocardial Published 11.12.2012 Grabole N, Tischler J, Hackett JA, Kim S, Chan JS, Dutta D, Oberheim J, Gläßer Tang F, Leitch HG, Magnúsdóttir E, Surani infarction in the adult mouse C, Southall TD, Brand AH, Jones DL, MA. heart. Edgar BA. Articles DOI: 10.1038/embor.2013.67 Malliaras K, Zhang Y, Seinfeld J, DOI: 10.15252/embj.201489072 Galang G, Tseliou E, Cheng K, Sun B, MicroRNA-15b promotes Published 14.05.2013 Published 08.10.2014 Aminzadeh M, Marbán E. neurogenesis and inhibits neural DOI: 10.1002/emmm.201201737 progenitor proliferation by embor.embopress.org Discrete Notch signaling Published 29.01.2013 directly repressing TET3 during requirements the specification early neocortical development. of hematopoietic stem cells. EMBO Molecular Medicine Lv X, Jiang H, Liu Y, Lei X, Jiao J. embomolmed.embopress.org Kim AD, Melick CH, Clements WK, DOI: 10.15252/embr.201438923 RARRES3 suppresses breast Stachura DL, Distel M, Panáková D, cancer lung metastasis by Molecular Systems Biology MacRae C, Mork LA, Crump JG, Traver D. Published 24.10.2014 regulating adhesion and Rapid neurogenesis through DOI: 10.15252/embj.201488784 differentiation. Published 17.09.2014 Regulation of stem cell function transcriptional activation in by protein ubiquitylation. Morales M, Arenas EJ, Urosevic J, Guiu human stem cells. M, Fernández E, Planet E, Fenwick RB, Strikoudis A, Guillamot M, Aifantis I. Dynamics of genomic Fernández-Ruiz S, Salvatella X, Reverter Busskamp V, Lewis N, Guye P, Ng AHM, H3K27me3 domains and role DOI: 10.1002/embr.201338373 D, Carracedo A, Massagué J, Gomis RR. Shipman SL, Byrne SM, Sanjana NE, Li Y, Published 20.03.2014 Weiss R, Murn J, Li S, Stadler M, of EZH2 during pancreatic DOI: 10.15252/emmm.201303675 Church GM. endocrine specification. Published 27.05.2014 DOI: 10.15252/msb.20145508 Xu CR, Li LC, Donahue G, Ying L, Zhang MYC proteins promote neuronal in press YW, Gadue P, Zaret KS. differentiation by controlling the Stimulation of endogenous DOI: 10.15252/embj.201488671 mode of progenitor cell division. cardioblasts by exogenous A competitive protein Published 08.08.2014 Zinin N, Adameyko I, Wilhelm M, Fritz N, cell therapy after myocardial interaction network buffers Uhlén P, Ernfors P, Henriksson MA. infarction. Oct4-mediated differentiation Mapping early fate DOI: 10.1002/embr.201337424 Malliaras K, Ibrahim A, Tseliou E, Liu W, to promote pluripotency in determination in Lgr5+ crypt Published 05.03.2014 Sun B, Middleton RC, Seinfeld J, Wang L, embryonic stem cells. stem cells using a novel Ki67- Sharifi BG, Marbán E. Muñoz Descalzo S, Rué P, Faunes F, RFP allele. The SCFSlimb E3 ligase complex DOI: 10.1002/emmm.201303626 Hayward P, Jakt LM, Balayo T, Garcia- regulates asymmetric division to Published 05.05.2014 Basak O, van de Born M, Korving J, Ojalvo J, Martinez Arias A. Beumer J, van der Elst S, van Es JH, inhibit neuroblast overgrowth. DOI: 10.1038/msb.2013.49 Clevers H. Li S, Wang C, Sandanaraj E, Aw SS, Koe CT, Delayed transplantation Published 08.10.2013 DOI: 10.15252/embj.201488017 Wong JJ, Yu F, Ang BT, Tang C, Wang H. of precursor cell-derived Published 04.08.2014 DOI: 10.1002/embr.201337966 astrocytes provides multiple Published 10.01.2014 benefits in a rat model of msb.embopress.org Fip1 regulates mRNA alternative Parkinsons. polyadenylation to promote Biased competition between Proschel C, Stripay JL, Shih CH, Munger stem cell self-renewal. Lgr5 intestinal stem cells driven JC, Noble MD. Lackford B, Yao C, Charles GM, Weng by oncogenic mutation induces DOI: 10.1002/emmm.201302878 L, Zheng X, Choi EA, Xie X, Wan J, Xing clonal expansion. Published 29.01.2014 Y, Freudenberg JM, Yang P, Jothi R, Hu Snippert HJ, Schepers AG, van Es JH, G, Shi Y. Simons BD, Clevers H. Prostaglandin E₂ promotes DOI: 10.1002/embj.201386537 DOI: 10.1002/embr.201337799 post-infarction cardiomyocyte Published 04.03.2014 Published 12.12.2013 replenishment by endogenous stem cells. emboj.embopress.org STAT3 promotes corticospinal Hsueh YC, Wu JM, Yu CK, Wu KK, Hsieh remodelling and functional PC. recovery after spinal cord injury. DOI: 10.1002/emmm.201303687 Lang C, Bradley PM, Jacobi A, Published 21.01.2014 Kerschensteiner M, Bareyre FM. DOI: 10.1038/embor.2013.117 Published 09.08.2013 EMBO Molecular Medicine

embor.embopress.org | emboj.embopress.org | embomolmed.embopress.org | msb.embopress.org

58403 / MITM027487