Gemc1 Governs Multiciliogenesis Through Direct Interaction with And

RESEARCH ARTICLE GemC1 governs multiciliogenesis through direct interaction with and transcriptional regulation of p73 Maria-Eleni Lalioti1, Marina Arbi2, Ioannis Loukas2, Konstantina Kaplani1, Argyro Kalogeropoulou1, Georgia Lokka1, Christina Kyrousi1, Athanasia Mizi3,4, Theodore Georgomanolis3, Natasa Josipovic3,4, Dimitrios Gkikas5, Vladimir Benes6, Panagiotis K. Politis5, Argyris Papantonis3,4, Zoi Lygerou2 and Stavros Taraviras1,*

ABSTRACT therefore, been speculated that formation of multiprotein complexes A distinct combination of transcription factors elicits the acquisition of a is the crucial event providing specificity during these processes. specific fate and the initiation of a differentiation program. Multiciliated Multiciliated cells (MCCs) comprise a specialized population of cells (MCCs) are a specialized type of epithelial cells that possess post-mitotic epithelial cells harboring dozens of motile cilia that dozens of motile cilia on their apical surface. Defects in cilia function have generate fluid flow in a variety of different tissues, including the been associated with ciliopathies that affect many organs, including brain and the airway epithelium. Through the synchronized beating brain and airway epithelium. Here we show that the geminin coiled-coil of their cilia, MCCs in the brain generate a laminar flow of – − domain-containing protein 1 GemC1 (also known as Lynkeas) regulates cerebrospinal fluid termed ependymal flow which is essential for the transcriptional activation of p73, a transcription factor central to brain homeostasis and neurogenesis (Spector et al., 2015; Stolp and multiciliogenesis. Moreover, we show that GemC1 acts in a trimeric Molnár, 2015). MCCs are also essential for development, complex with transcription factor E2F5 and tumor protein p73 (officially homeostasis and regeneration of the airway epithelium, and the known as TP73), and that this complex is important for the activation of synchronized beating of their cilia is crucial for mucus clearance. the p73 promoter. We also provide in vivo evidence that GemC1 is Defects in the generation or function of cilia cause a wide variety of necessary for p73 expression in different multiciliated epithelia. We diseases, known as ciliopathies, which include primary ciliary further show that GemC1 regulates multiciliogenesis through the control dyskinesia (PCD), mucociliary clearance-associated defects and of chromatin organization, and the epigenetic marks/tags of p73 and infertility. Moreover, a link between ependymal cilia malfunction Foxj1. Our results highlight novel signaling cues involved in the and hydrocephalus has become apparent in different animal models commitment program of MCCs across species and tissues. (Fliegauf et al., 2007; Spassky and Meunier, 2017). Recent findings have emphasized the role of the geminin family This article has an associated First Person interview with the first author members McIdas (also known as Idas) (Pefani et al., 2011) and of the paper. geminin coiled-coil domain-containing protein 1 (GMNC; also known as Lynkeas; hereafter referred to as GemC1) (Balestrini KEY WORDS: GemC1, Lynkeas, GMNC, Epigenetic regulation, et al., 2010), to multiciliate cell fate acquisition and differentiation Multiciliated cells, Transcriptional regulation, Multiciliogenesis (reviewed in Arbi et al., 2018; Kyrousi et al., 2016, 2017). GemC1 had initially been identified as a cell cycle regulator, essential for the INTRODUCTION initiation of DNA replication during the transition from G1 to S The acquisition of a functionally specialized cellular phenotype is phase (Balestrini et al., 2010) but was later shown to constitute a key an organized process that requires restriction of stem and progenitor molecule in the early steps of multiciliate cell differentiation. More fate competence. To accomplish this, the chromatin landscape of specifically, GemC1 is essential for the commitment and generation individual stem and progenitor cells and expression of a distinct of MCCs in the mouse brain (Kyrousi et al., 2015) and airway transcription factor set orchestrate gene expression to specifically epithelium (Arbi et al., 2016; Terré et al., 2016) as well as in the instruct the acquisition of a specific cell fate. General transcription oviduct, affecting female fertility (Terré et al., 2016). The role of factors act in concert with tissue-specific proteins to regulate cell GemC1 in multiciliate cell differentiation is well conserved – it has fate decisions and differentiation in multiple cell types. It has, been shown to be essential for multiciliogenesis in both zebrafish and Xenopus skin (Zhou et al., 2015). McIdas has been shown to act in a complex with E2F4 and/or

1Department of Physiology, School of Medicine, University of Patras, 26504 Patras, E2F5 and to activate the transcription of genes involved in centriole Greece. 2Department of General Biology, School of Medicine, University of Patras, amplification (Ma et al., 2014; Stubbs et al., 2012). Mutations in the Patras 26504, Greece. 3Center for Molecular Medicine Cologne, University of C-terminal domain of human McIdas, responsible for the interaction Cologne, Robert-Koch-Str. 21, 50931 Cologne, Germany. 4Department of Pathology, University Medical Center Göttingen, Robert-Koch-Str. 40, 37075 with the E2F family members, results in mucociliary clearance Göttingen, Germany. 5Department of Center for Basic Research, Biomedical disorders (Boon et al., 2014). Similarly, GemC1 has been shown to Research Foundation of the Academy of Athens, 4 Soranou Efesiou Street, regulate transcription by interacting with E2F4 and E2F5 through a 115 27 Athens, Greece. 6European Molecular Biology Laboratory (EMBL), Core Facilities and Services, Meyerhofstraße 1, Heidelberg 69117, Germany. C-terminal domain homologous to the McIdas C-terminus (Arbi et al., 2016; Kyrousi et al., 2015; Terré et al., 2016). McIdas and *Author for correspondence ([email protected]) GemC1 regulate the transcriptional activation of Myb and Foxj1, S.T., 0000-0002-7455-647X genes essential for centriole amplification and basal body docking, respectively, as well as the regulation of a broad gene expression

Received 17 January 2019; Accepted 16 April 2019 program linked with the fully differentiated phenotype of MCCs. Journal of Cell Science

1 RESEARCH ARTICLE Journal of Cell Science (2018) 132, jcs228684. doi:10.1242/jcs.228684

The tumor protein p73 (officially known as TP73) has recently been the additional level of specificity needed for the regulation of the identified as a central regulator of multiciliogenesis; it is expressed distinct gene targets required for MCC differentiation. in MCCs of various tissues, and is required for MCC differentiation First, we examined whether GemC1 interacts with p73. and for the direct regulation of a plethora of transcriptional modulators Immunoprecipitation experiments performed in HEK293T cells of multiciliogenesis (Marshall et al., 2016; Nemajerova et al., revealed that pull down of HA-tagged p73 specifically precipitated 2016). At the same time, p73 was found to be an essential regulator GFP-tagged GemC1 (GFP-GemC1; see Fig. 1A). To further for the generation of mature ependymal cells in the adult mouse brain, determine the domain of GemC1 responsible for this interaction, we indicating a global role in the regulation of multiciliogenesis performed immunoprecipitations by using truncated GemC1 forms (Gonzalez-Cano et al., 2016; Medina-Bolívar et al., 2014). lacking the coiled-coil region (GemC1Δcc), previously shown to be Moreover, p73 has a central role in orchestrating multiciliogenesis required for interactions with geminin family members (namely through the transcriptional activation of a network of key ciliogenic geminin and McIdas) (Caillat et al., 2015) or the C-terminal domain of factors (namely Foxj1, Rfx2 and Rfx3) and through post- GemC1 (GemC1Δct), as well as a combination of both deletions transcriptional processing by miR-34b and/or miR-34c (Marshall (GemC1ΔΔ). When both regions were deleted, the observed et al., 2016; Nemajerova et al., 2016). Finally, ChIP-seq analysis in interaction was completely lost, showing that these two regions are murine tracheal cells revealed a plethora of p73-target genes, including essential for GemC1 interactions with p73 (Fig. 1B). Notably, only a Foxj1 and Myb (Marshall et al., 2016; Nemajerova et al., 2016), that small decrease in interaction affinity was observed with the are also transcriptionally regulated by GemC1 in cooperation with GemC1Δcc and GemC1Δct single mutants, indicating that both of E2F5 (Arbi et al., 2016; Kyrousi et al., 2015; Terré et al., 2016). these regions can mediate interactions with p73. Even though several factors have been implicated in the activation Second, we asked whether GemC1, p73 or E2F5 could be of the multiciliogenesis program, it remains unclear how the detected as part of the same complex. We first verified that we can combination of general E2F family transcription factors and of detect specific interactions of GemC1 with E2F5, as conflicting data tissue-specific factors, like GemC1 and p73, cooperate to instruct had been reported in the literature (Terré et al., 2016; Zhou et al., acquisition of a MCC fate. Here, we show that GemC1 physically 2015). To that end, hemagglutinin (HA)-tagged E2F5, FLAG interacts with p73 and that it is essential for the recruitment of p73 to (DYK)-tagged DP1 (an E2F5 cofactor) and GFP-GemC1 were co- the complex between GemC1 and E2F5 bound to its co-factor DP1 transfected in HEK293T cells followed by HA pull-down assays. (hereafter referred to as E2F5-containing complexes). Moreover, the Indeed, E2F5 specifically co-precipitated full length GemC1, as GemC1–E2F5–p73 complex transcriptionally regulates p73, well as the truncated GemC1 form GemC1Δcc lacking the coiled- suggesting a role of this complex in the transcriptional regulation of coil domain, but not GemC1Δct, which lacks the C-terminal domain MCC cascade genes. Crucially, we also provide in vivo evidence that (Fig. S1), verifying that GemC1 specifically interacts with E2F5 GemC1 acts upstream of p73, as its deletion blocks the expression of through its C-terminal region. To detect a complex comprising p73 both in the brain and in the airway epithelium, whereas GemC1 E2F5, GemC1 and p73, HA-E2F5, DYK-DP1, GFP-GemC1 and overexpression is sufficient to induce expression of p73.Finally,we p73-GFP were co-expressed in HEK293T cells. To stabilize show that deletion of GemC1 affects the spatial configuration and the putative transient interactions, these co-immunoprecipitations epigenetic landscape of key MCC loci, such as of p73 and Foxj1, were carried out following reversible crosslinking (see Materials underscoring its vital role in the MCC differentiation pathway. and Methods). Pull down of HA-E2F5 precipitated both GFP- GemC1 and p73-GFP (Fig. 1C). To prove the simultaneous RESULTS presence of all three factors in the complex, we performed the GemC1 binds to p73 and recruits it to E2F5-containing same IP experiments with GemC1Δcc and GemC1Δct. Only in the complexes presence of GemC1Δcc, capable of binding both p73 and E2F5, was E2F family members regulate transcription of a large set of genes E2F5 able to precipitate p73. This interaction is lost in the presence of involved in cell cycle, apoptosis, DNA damage and cell fate decisions GemC1Δct, which binds p73 but not E2F5, showing that p73 is (Julian and Blais, 2015). Therefore, specific recognition of the recruited to E2F5-containing complexes through GemC1 (Fig. 1D,E). different gene targets in each process could be determined Our results demonstrate that GemC1 not only interacts with p73 but by differential interactions and the formation of multiprotein also directly recruits it to E2F5-containing complexes. complexes that can provide cellular specificity. GemC1 and McIdas are key factors regulating the generation of MCCs (Arbi et al., 2018; GemC1–E2F5–p73 regulates transcription of p73 Kyrousi et al., 2016, 2017; Ma et al., 2014; Stubbs et al., 2012; Our data suggest that GemC1, p73 and E2F5 are parts of a single Terré et al., 2016; Zhou et al., 2015). It has been shown that complex. To gain insight into the functional significance of such they interact with E2F-family members and activate transcription a complex, we examined its ability to activate transcription. We through promoter-located E2F-binding sites in genes important for performed luciferase reporter assays in HEK293T cells using the p73 multiciliogenesis, including those of transcription factors ΜYB, promoter region cloned upstream of a luciferase reporter gene FOXJ1 and MCIDAS (Arbi et al., 2016; Kyrousi et al., 2015; Ma et al., (pGL3-p73). Cells were co-transfected with GemC1, p73 and E2F5 2014; Terré et al., 2016). However, it remains unclear how GemC1 constructs, alone or in combination, before luciferase activity was and McIdas target specifically the subset of E2F-regulated genes that measured. We found that GemC1 is able to activate transcription are involved in multiciliogenesis, thereby establishing a tissue- from the p73 promoter (12-fold, Fig. 2Α), while the addition of E2F5 specific transcriptional program. Recent findings suggest that p73 is a further increases its transcriptional activity (34-fold, Fig. 2Α). crucial regulator of the MCC transcriptional program (Marshall et al., Strikingly, co-expression of p73 with GemC1 and E2F5 2016; Nemajerova et al., 2016). Given that several p73 transcriptional contributes to even higher transcriptional activation (54-fold, targets are also regulated by GemC1–E2F5 complexes, and that p53 Fig. 2Α). These results were confirmed when mouse neural stem and E2F-family members have been functionally associated in cells (NSCs) were used in a similar setup (Fig. S2). To further different cellular contexts (Engeland, 2018), we examined whether examine how GemC1 regulates transcriptional activation of the p73 E2F-Mut p73 interacts with the GemC1–E2F5 complex, as this would explain promoter, we mutated its E2F-binding motifs (P73 )and Journal of Cell Science

2 RESEARCH ARTICLE Journal of Cell Science (2018) 132, jcs228684. doi:10.1242/jcs.228684

Fig. 1. GemC1 forms a complex with p73 and E2F5. (A,B) GemC1 interacts with p73. GFP-GemC1 or truncated forms lacking the coiled coil, the C-terminus or both regions (GFP-GemC1Δcc, GFP-GemC1Δct and GFP-GemC1ΔΔ, respectively) were co-transfected into HEK293T cells together with HA-p73 or empty HA vector as a negative control. Immunoprecipitation was performed against HA (a-HA) and against GFP (a-GFP). Complete loss of interaction is observed in the presence of GFP-GemC1ΔΔ. (C,D) GemC1 recruits p73 to E2F5-containing complexes. GFP-GemC1 and its deletion mutants GFP-GemC1Δcc and GFP-GemC1Δct were co-transfected into HEK293T cells together with p73-GFP, HA-E2F5 and TfDP1-DYK. Immunoprecipitation was performed against HA. HA-E2F5 pulldown specifically precipitates p73 and GemC1wt or GemC1Δcc. In the presence of GemC1Δct, which binds to p73 (Fig. 1B) but not E2F5 (Fig. S1), association of p73 with the E2F5-containing complex is lost. #, longer exposure compared to GFP-GemC1. (E) Schematic representation of GemC1 deletion mutants used in (B,D) and their effect on p73 and E2F5 binding. TCL, Total cell lysate; IP, Immuno-precipitation, nd, not determined. Asterisks indicate IgGs (heavy chain). Journal of Cell Science

3 RESEARCH ARTICLE Journal of Cell Science (2018) 132, jcs228684. doi:10.1242/jcs.228684

Fig. 2. p73 activates its own promoter in a complex with GemC1 and E2F5. (A,B) p73 co-operates with GemC1–E2F5 to transcriptionally activate its promoter. Measurement of relative fold-induction of luciferase (RLF). For luciferase reporter gene assay, p73 regulatory elements cloned upstream of a luciferase reporter, were co-transfected into HEK293T cells, by using GemC1, E2F5 and p73 expression plasmids as indicated or empty plasmid vector as a control. (B) Luciferase reporter assays were conducted using a P73E2F-Mut promoter, which contains the same DNA region with the p73 promoter but lacks the E2F-binding motifs. GemC1, E2F5 and p73 expression vectors or an empty vector as a control, were co-transfected into HEK293T cells with the mutant P73E2F-Mut promoter, cloned upstream from a luciferase reporter. All luciferase experiments were normalized for transfection efficiency with an expression vector for Renilla luciferase. Fold induction is the ratio of the normalized luciferase activity induced by the expression constructs and that induced by the empty expression vector. Data are represented as mean±s.e.m. Statistical significance was determined by two-tailed Mann–Whitney U test from n=3 independent transfections, *P<0.05, **P<0.005. (C) p73 is part of the complex that binds to the p73 promoter. Chromatin immunoprecipitation of an E2F-containing fragment from the p73 promoter. HEK293T cells were co-transfected with vectors expressing p73-GFP and GemC1 (p73GFP/GemC1), p73-GFP alone (p73GFP) or GFP and GemC1 (GFP/GemC1) as a control. In every condition, cells were co-transfected with vectors expressing p73 regulatory elements together with E2F5 and DP1. DNA fragments within the immunoprecipitates were quantified by qPCR. Data from a representative experiment are shown as fold enrichment of the p73 promoter fragment in immunoprecipitates of GFP versus control IgG immunoprecipitates. repeated the reporter assays. This time, p73 promoter responsiveness own promoter region directly or through forming a complex with to GemC1 was markedly decreased (Fig. 2B). A similar reduction GemC1–E2F5, cells were transfected with p73-GFP, E2F5 and DP1 was observed when GemC1 was co-expressed with E2F5 and/or p73, but not GemC1 (hereafter referred to as p73GFP). In the absence of indicating that activation of p73 is mediated by E2F binding to its GemC1, the E2F-containing promoter fragment was not enriched cognate sites (Fig. 2B). following GFP immunoprecipitation, indicating that GemC1 is Given that the GemC1–E2F5 complex directly binds to the essential for p73 binding to the promoter region (Fig. 2C). Thus, we promoter regions of genes that are transcriptionally regulated by conclude that GemC1 mediates the association of p73 with the E2F5-containing complex in MCCs (Arbi et al., 2016; promoter-bound E2F5-containing complexes. Terré et al., 2016), we next examined whether p73 is part of Collectively, our data suggest that the GemC1–E2F5–p73 this complex. To this end, chromatin immunoprecipitation was complex is coopted to activate p73 transcription, suggesting a carried out in HEK293T cells transfected with p73-GFP (or GFP role for this complex as a regulator of the transcriptional program alone as a control), together with E2F5, DP1 and GemC1 in MCCs. expression constructs. We assessed the co-precipitation of a fragment containing E2F-binding sites by qPCR following p73 gene expression is regulated by GemC1 in vivo immunoprecipitation of GFP (Fig. 2C), and found that it was To provide in vivo evidence for the transcriptional regulation of enriched in p73-GFP precipitates (defined as p73GFP/GemC1 in p73 by GemC1, we overexpressed GemC1 through in utero Fig. 2C) as compared to those of the control (defined as GFP/ electroporation (IUE) in the developing cortex of E14.5 days post GemC1 in Fig. 2C), indicating that p73 is part of the complex that coitum (dpc) wild-type mouse embryos. Plasmids expressing binds the p73 promoter. To determine whether the p73 binds to its the internal ribosome entry site (IRES)-GFP (referred to as GFP) Journal of Cell Science

4 RESEARCH ARTICLE Journal of Cell Science (2018) 132, jcs228684. doi:10.1242/jcs.228684

Fig. 3. See next page for legend. or GemC1-IRES-GFP (referred to as GemC1) were injected into Mouse tracheal epithelial cells (MTECs) were further used to the telencephalic ventricles of E14.5dpc embryos. Brains were assess the effect of GemC1 overexpression on p73 regulation ex analyzed 2 days after electroporation by using immunofluorescence vivo. A lentiviral-based delivery system was used to express either a and an anti-p73 specific antibody (Fig. 3A). Upon GemC1 GFP-GemC1 fusion protein or GFP only (as a control). Following overexpression, 45% of cells displayed nuclear p73 transduction, an air liquid interface (ALI) culture setup was immunostaining, whereas control GFP-electroporated cells established to initiate ciliogenesis. GemC1-overexpressing cells completely lacked p73 expression at this developmental stage were analyzed for p73 expression at different time points during

(Fig. 3B). MCC differentiation and showed notable increase compared to Journal of Cell Science

5 RESEARCH ARTICLE Journal of Cell Science (2018) 132, jcs228684. doi:10.1242/jcs.228684

Fig. 3. GemC1 controls the expression of p73 both in embryonic radial GemC1KO/KO mice. Collectively, our data suggest that GemC1 is a glial cells and in airway epithelial cells. (A,B) Overexpression of GemC1 in sufficient and necessary activator of p73 expression in vivo. embryonic radial glial cells induces p73 protein expression. (A) In utero electroporation was performed in the developing cortex of E14.5 dpc of mouse embryos, using GFP- or GemC1-encoding plasmids. Embryos were sacrificed Lack of GemC1 triggers epigenetic reorganization at the 2 days later, and coronal sections were stained with antibodies against GFP p73 and Foxj1 loci (green) and p73 (red). DNA was stained with Draq-5 (blue). Boxed regions in Having established that GemC1 knockout abrogates p73 expression, left-most images are shown in higher magnifications in images in the middle we asked whether it also affects the epigenetic configuration of its and right. (B) Quantification of electroporation experiments. The percentage of target loci. First, we applied ‘intrinsic 4C’ (i4C), a chromosome GemC1-GFP-positive radial glial cells that co-express p73 was markedly conformation capture (3C) variant that does not require cell fixation increased, compared to cells electroporated with the GFP-only plasmid. (Brant et al., 2016), to pRGCs from GemC1WT/WT or GemC1KO/KO Data are represented as mean±s.e.m. and were obtained from n=3 mice. p73 Foxj1 Statistical analysis was assessed with two-tailed Mann–Whitney, ***P<0.001. mice in proliferating conditions using the and promoters (C,D) GemC1 induces the expression of p73 in the airway epithelial cells. as viewpoints. Following next-generation sequencing and mapping (C) Mouse tracheal epithelial cells were infected with lentiviruses expressing of i4C data to the mouse reference genome, the spatial interaction either GFP-GemC1 or GFP alone, and 2 days later transitioned to ALI culture profiles (Fig. 4A) were analyzed using the fourSig method to conditions. Cells analyzed at different time points during differentiation using prioritize interactions (Williams et al., 2014). A large fraction of antibodies against GFP (marks infected cells, green) and p73 (red). (D) For significant i4C contacts formed by either promoter (>70% for p73 each time point the percentage of infected cells expressing p73 was quantified. ∼ Foxj1 GemC1 Data are presented as the mean values of at least ten independent fields and 40% for ) remained unchanged upon knockout obtained from one representative experiment for each condition and n number (Fig. 4B). However, upon correlation of significant i4C interactions of cells analyzed, as shown. Error bars indicate±s.e.m. **P<0.01, ***P<0.001. with DNase I hypersensitive (DHS) data from wild-type mouse P-values were calculated by the nonparametric two-tailed Mann–Whitney U cerebellum (from the ENCODE project; https://genome.ucsc.edu/ test. (E,F) p73 expression is blocked in the absence of GemC1. (E) Coronal ENCODE/index.html) a considerable subset of p73 and Foxj1 sections at the level of the subventricular zone (SVZ) from newborn (P0) or contacts from GemC1KO/KO cells did not overlap such sites wild-type mice (GemC1WT/WT), or mice homozygous for deletion of the GemC1 gene (GemC1KO/KO) were immunostained with a specific antibody against p73 (Fig. 4C). This redirection of spatial interactions away from (green). DNA was stained with Draq-5 (blue). Cells expressing p73 were accessible chromatin is indicative of transcriptional repression at not detected in the GemC1KO/KO brains, as opposed to the p73-positive the p73 and Foxj1 loci, and in agreement with a reduction in the cells in the control littermates. Boxed images (bottom-left in each image) are level of modified histone H3 (H3K27ac) − an epigenetic mark magnifications of the areas indicated by asterisks. LV, lateral ventricle. associated with an increased level of transcription – at p73 and WT/WT (F) Transverse sections from trachea isolated from GemC1 or Foxj1 promoters in the absence of GemC1 (Fig. 4D). GemC1KO/KO P3 mice immunostained with antibodies against p73 (green), acetylated α-tubulin (ACT, cilia marker, red) and E-cadherin (marker for cell Finally, we generated poly(A)+ RNA-sequence data from the WT/WT boundaries, blue). Scale bars: 10 μm (C,F); 40 μm (A,E). walls of the lateral ventricles isolated from GemC1 and GemC1KO/KO brains (three biological replicates), which revealed 180 genes that are ∼4-times more down- than up-regulated upon control GFP-overexpressing cells (Fig. 3C,D). These data indicate GemC1 knockout (Fig. 4E), in accordance with its proposed role as that ectopic overexpression of GemC1 is sufficient to promote a transcriptional activator. Importantly, these 180 downregulated endogenous p73 expression in vivo and ex vivo. genes, based on the gene ontology term enrichment analysis we Furthermore, we investigated whether GemC1 is an essential performed, were relevant to multiciliated cells and included cilium activator of p73 transcription. To this end, we employed mice movement, cilium assembly, MCC differentiation or ventricular bearing a homozygous deletion of the GemC1 gene (Arbi et al., development (Fig. 4E). Taken together, our data support the idea 2016). We assessed p73 protein expression in isolated brains that GemC1 depletion drives epigenetic reorganization of key p73 derived from newborn mice (P0). Brain coronal sections at the and Foxj1 loci, and we expect additional target loci to be controlled level of the subventricular zone (SVZ), where ependymal in a similar manner, again with an impact on gene expression. progenitor cells reside, were immunostained for p73 (green, Fig. 3E). p73 expression was detected in progenitor cells near the DISCUSSION lateral wall of wild-type mice, whereas the corresponding area in MCCs are specialized epithelial cells that carry hundreds of motile homozygous GemC1 knockout mice (GemC1KO/KO) was devoid cilia to drive fluid flow across lumenal surfaces (Brooks and of p73-expressing cells. To exclude any intrinsic factors that Wallingford, 2014). Members of the E2F and geminin families, as might influence p73 expression, we used a primary cell culture well as protein p73 have been implicated in controlling early events assay of postnatal radial glial cells (pRGCs) that, under of fate commitment and differentiation in MCCs. However, how physiological conditions, differentiate into ependymal cells ex these factors cooperate to determine specificity in MCC fate is not vivo (Kyrousi et al., 2015; Paez-Gonzalez et al., 2011). pRGCs understood. derived from GemC1KO/KO mice also did not upregulate p73 GemC1, a member of the geminin family, has been shown to protein expression when induced to differentiate towards the promote the generation of MCCs across different cell tissues and ependymal lineage, which is in contrast to pRGCs from control animal models (Arbi et al., 2016, 2018; Kyrousi et al., 2015, 2016, mice (Fig. S3). 2017; Terré et al., 2016; Zhou et al., 2015). Its role was mainly To reinforce our results, we examined whether p73 expression is attributed to its ability to activate transcription of Mcidas, Foxj1 also controlled by GemC1 in other ciliated tissues. Tracheal sections and Myb genes encoding key transcription factors for early events from newborn wild-type and GemC1KO/KO mice were stained using in commitment and differentiation of MCCs. On the other hand, p73, antibodies specific to p73, as well as against acetylated α-tubulin a p53-family member, was recently highlighted for its role as a and E-cadherin that, respectively, label ciliary axonemes and cell central regulator of the multiciliogenesis transcription program boundaries (Fig. 3F). Expression of p73 was detected in ciliated (Marshall et al., 2016; Nemajerova et al., 2016). Our findings tracheal cells isolated from GemC1WT/WT mice, whereas suggest that interactions between GemC1 and p73 are crucial p73-expressing cells could not be found in tracheas derived from for multiciliogenesis. Journal of Cell Science

6 RESEARCH ARTICLE Journal of Cell Science (2018) 132, jcs228684. doi:10.1242/jcs.228684

Fig. 4. GemC1 deletion results in epigenetic reorganization at the p73 and Foxj1 loci. (A) Spatial interactomes between the p73 and Foxj1 promoters. The average of two i4C replicates generated in wild-type (WT) and GemC1 knockout (KO) cells are shown for the 0.75 Mbp around the p73 and Foxj1 promoters, i.e. ‘viewpoints’. Colored boxes denote significantly enriched contacts identified by using the fourSig algorithm. Aligned ENCODE ChIP-data from mouse cerebellum are shown below. (B) Venn diagrams showing the number of intrachromosomal (cis) fourSig contacts shared between i4C in Foxj1 and p73 WT and KO cells. (C) Heatmaps showing enrichment of DNase I-hypersensitive (DHS) sites in the 6 kbp around Foxj1 and p73 i4C cis-contacts from both viewpoints. A substantial cluster of cis-contacts enriched for DHS sites is lost in KO cells. The mean tag density of DHS data under the two conditions is shown on the right. (D) Bar plot showing H3K27ac de-enrichment (±s.e.m. from two replicates) at the p73 and Foxj1 promoters in KO cells. *P<0.05; unpaired Student’s t-test. (E) Scatter plot (left) showing up- and downregulated genes (orange and blue, respectively) upon mRNA-seq from GemC1 WT and KO in radial glial cells. Bar plot (right) showing GO terms enriched for the 180 downregulated genes in GemC1KO/KO cells. Data derived from n=3 mice.

It has been proposed that GemC1 can activate transcription – by transcriptional activation of p73 in vivo. When ectopically expressed, interacting with the general transcription factors E2F4 and E2F5 GemC1 is sufficient to induce p73 expression in neural progenitor through a motif in its C-terminal domain (Terré et al., 2016 and cells of the developing cortex and ex vivo in tracheal epithelial cells. present study) – to control activation of the ciliogenesis program. GemC1-deficient mice lack p73 expression both in the brain and in However, it was unknown how GemC1 might impart specificity to the airway epithelium, thus, providing further insight in the molecular E2F complexes, as to direct them to the subset of genes involved in mechanism through which GemC1 deletion abrogates the generation multiciliogenesis. Here, we report that GemC1 physically interacts of MCCs in different tissues (Arbi et al., 2016; Kyrousi et al., 2015; with p73 through interaction sites at both its C-terminal and coiled- Terré et al., 2016). These results show that GemC1 is both sufficient coil domains. These domains are also crucial for GemC1 and necessary for p73 expression in lung and brain tissues that interactions with other binding partners (Caillat et al., 2013, 2015; harbor MCCs, thus, implying that p73 activation by GemC1 is Terré et al., 2016). Moreover, we show that GemC1 forms a global. Both GemC1 (Arbi et al., 2016; Terré et al., 2016) and p73 complex with E2F5 and that p73 is then recruited to this complex (Marshall et al., 2016) have been implicated in cell fate decisions, through its interaction with GemC1. We suggest that formation of affecting the balance between the ciliated and the secretory cell this multiprotein complex confers specificity by restricting population in the airway epithelium. Our results, taken together transcriptional activation to a subset of E2F-responsive genes. with earlier work, establish GemC1 as the most-upstream factor Indeed, we show that such a cooperation between E2F5, GemC1 controlling the activation of the multiciliogenesis program in stem and p73 operates on the promoter of p73 itself. We show that and progenitor cells. p73 binds to its own promoter when associated with the E2F5- Given that one of the key mechanisms for gene regulation involves containing complex, thereby enhancing transcriptional activation of promoter–enhancer interactions, it is tempting to speculate that p73. Consistent with the above, we show that GemC1 regulates the GemC1 and p73 complexes are important to bring these elements into Journal of Cell Science

7 RESEARCH ARTICLE Journal of Cell Science (2018) 132, jcs228684. doi:10.1242/jcs.228684 close proximity, thus, regulating transcription of a specific subset of were conducted (n=3; GemC1–E2F5 n=2; GemC1 mutants with p73, n=2; genes required for fate commitment and differentiation of MCCs. Our GemC1-p73-E2F5, n=2; GemC1 mutants-p73-E2F5, n=1). Plasmid vectors findings, indeed, suggest that GemC1 acts as a bridging factor expressing the desired proteins in the Co-IP experiments are listed below: α between E2F complexes and other transcriptional activators, and pCMVHA-E2F5 (human, Addgene plasmid #24213), HA-p73 -pcDNA3 between chromatin remodeling activities, to promote ciliogenesis. We (human, Addgene plasmid #22102), pLVDest-GFP-GemC1 (mouse, as previously described in Arbi et al., 2016) and the corresponding mutants provide evidence that the epigenetic landscape of the promoter regions Δ Δ Δ Δ ΔΔ p73 Foxj1 mGemC1 cc ( 86-137aa), mGemC1 ct ( 300-339aa),mGemC1 of and is altered upon GemC1 deletion. Moreover, GemC1 (Δ86-137;300-339); pcDNA3.1+/C-(K)DYK-TFDP1 (human, GenScript triggers reorganization at the p73 and Foxj1 loci, as spatial contacts #OHu23102) and pEGFP-mP73. Briefly, to generate pEGFP-mP73, the were markedly rearranged in GemC1-ablated neural progenitors. This cloning strategy was the following: mp73 ORF from pcDNA3.1 (GenScript would be consistent with GemC1 stabilizing distal enhancers with #OMu22724) was cloned into the XhoI and AgeI restriction sites of the proximal promoters and bringing key regulatory domains bound by pEGFP-N1 vector (GenBank #U55762). Foxj1 into close physical proximity. It is conceivable that GemC1 will bridge different regulators to E2F complexes in different loci, HEK293T and mouse neural stem cells − culture and luciferase directing the timing of the ciliogenesis program. Positive feedback reporter assays loops, such as the autoregulation of p73 described here, would provide Luciferase assays were performed in cell lysates from human HEK293T additional levels of control. It would be interesting to further cells and mouse neural stem cells (NSCs) by using the Dual-Luciferase Reporter Assay Kit (Promega). A 1-kb genomic region upstream of the investigate whether GemC1 has a role in the coordination of the transcription start site of human p73 was cloned upstream of the coding multiciliogenesis program in terms of spatial genome organization. sequence for firefly luciferase in the pGL3 basic vector (Promega) to create In summary, we show that GemC1 is essential for the formation pGL3-p73. Specifically, upstream regulatory sequences for the p73 of E2F- and p73-containing complexes that could play a crucial role promoter region are located at chr1:3651640-3652640 (GRCh38; catalog in transcriptional activation of key genes for the commitment and no. hg38). For the pGL3-P73E2F-Mut construct, the same region was cloned differentiation into MCCs. We also propose that GemC1 modifies upstream of the firefly luciferase sequence but without the E2F4 and/or the transcriptional network of multiciliogenesis through the control E2F5-binding motifs, as described by Nemajerova et al., 2016. Expression of spatial contacts and epigenetic status at key MCC loci. Our results vectors used for co-transfections were as follows: mouse p73 was cloned in the pCDNA3.1 expression vector, mouse GemC1 was cloned in the put forward the idea that GemC1, through interactions with multiple E2F5 partners, acts as a pioneering factor for the well-conserved pCAGGS vector and human was cloned in the pCMV vector. HEK293T cells (ATCC CRL-3216) were cultured at 37°C under 5% CO2 multiciliogenesis pathway across different species and organs. in DMEM supplemented with 10% FBS (both Gibco). For transfection, This premise can be important for delineating the mechanisms cells (3.5×104 cells) were grown in 96-well plates and, using Turbofect disrupted in cilium-associated diseases. (Fermentas), co-transfected with a total of 130 ng DNA containing 25 ng of pGL3-construct with 50 ng of the expression constructs (empty vector, pCDNA3.1-p73, pCAGGS-GemC1, pCMV-E2F5) and 5 ng of Renilla MATERIALS AND METHODS luciferase phRL-SV40 vector (Promega) as a transfection efficiency control. Co-immunoprecipitation and western blot analysis Firefly and Renilla luciferase activity was measured in cell lysates 48 h after Co-immunoprecipitation experiments were conducted as previously transfection with the Dual-Luciferase Reporter Assay Kit (Promega, described (Symeonidou et al., 2013). HEK293T cells (ATCC CRL-3216), E1980), in a Victor Light Luminescence plate reader. The transcriptional tested for contaminations on a regular basis, were transiently transfected with activity of the regulatory elements was measured as the ratio of Firefly to equal amounts of the indicated plasmids using the TurboFect Transfection Renilla luciferase signal, normalized to the values obtained when the empty Reagent (Thermo Fisher Scientific; catalog no. R0531) according to the vector was co-transfected. Results from at least n=3 different wells are manufacturer’s instructions. Cells were collected 48 h post transfection, presented as the mean fold-induction relative to the control expressing washed twice with ice cold PBS and incubated for 10 min in lysis buffer vector ± the standard error of the mean (±s.e.m.). Statistical analysis was – U (50 mM Tris-HCl pH 8.2, 150 mM NaCl, 5 mM EDTA, 5 mM MgCl2,0.5% performed with the nonparametric two-tailed Mann Whitney test from Triton X-100) supplemented with fresh protease and phosphatase inhibitors data derived from n=3 independent transfections. (1 mM PMSF and 0.1 mM NaVO4). For reversible crosslinking, cells Mouse NSCs were isolated from cortex derived from E14.5 dcp C57/B6 were incubated with 2 mM Lomant’s Reagent [dithiobis (succinimidyl mice. NSC isolation experiments were performed according to the local propionate), (DSP); Thermo Fisher Scientific; catalog no. 22586] for 20 min ethics committee (Athens Prefecture Veterinarian Service; ϰ/3237/11-05- prior to incubation with lysis buffer. Next cell lysates were passed through a 2012) and took place in the animal facilities of the Center for Experimental 1-ml syringe (mechanical disruption) and centrifuged for 10 min at 15,000 g. Surgery of the Biomedical Research Foundation of the Academy of Athens. HA immunoprecipitations were performed using anti-HA antibody F-7 NSCs were plated after the third passage in 12-well plates and maintained as (Santa Cruz Biotechnology; catalog no. sc-7392, 1.5 µg per IP experiment). neurosphere cultures in a serum-free medium [1:1 mixture of Dulbecco’s For each IP reaction, 50 µl of Protein A agarose suspension (Merck Millipore; modified Eagle’s medium (DMEM) and F-12 supplemented with penicillin catalog no. IP02) was incubated with 1.5 µg of HA-antibody for 2 h on a (100 units/ml; Invitrogen) and streptomycin (100 µg/ml; Invitrogen)], rotating wheel at 4°C. After three washes with lysis buffer, the beads were containing B-27 supplement (1 ml/50 ml; Invitrogen), insulin (20 µg/ml; incubated with cell lysates (cells from two 35-mm dishes were pulled together Sigma), recombinant human basic fibroblast growth factor (bFGF 20 ng/ml; per IP) for 3 h on a rotating wheel at 4°C. Beads were washed three times with R&D Systems), and epidermal growth factor (EGF 20 ng/ml; R&D lysis buffer and proteins were eluted upon incubation with Laemmli buffer. Systems). When spheres reached the appropriate size, they were Protein samples were then analyzed by SDS-PAGE, transferred on PVDF transiently transfected with Lipofectamine 2000 (2 μl per well) with membrane (Merck Millipore; catalog no. IPVH00010) followed by western 0.5 μg of each reporter construct (pGL3-p73 promoter-Luc), 1 μg of the blotting. Primary and secondary antibodies used in western blot analysis appropriate expression vectors according to the desirable experimental were: anti-HA antibody F-7 (Santa Cruz Biotechnology; catalog no. sc-7392, conditions and 0.1 μgofaβ-galactosidase expression plasmid to normalize dilution 1:1000), anti-GFP (Thermo Fisher Scientific; catalog no. A-11122; for transfection efficiency. Transfection mixture was replaced 6 h later with dilution 1:5000), anti-rabbit IgG-HRP (Santa Cruz Biotechnology; catalog growth medium and the cells were further cultured for 42 h. Cells were then no. sc-2004; dilution 1:3000) and goat anti-mouse IgG-HRP (Santa Cruz lysed and luciferase activity was determined according to manufacturer’s Biotechnology; catalog no. sc-2005, dilution 1:3000). Signal was visualized specifications (Promega) with a Lumat LB 9507 luminometer (Berthold by Pierce ECL western blotting substrate (Thermo Fisher Scientific; catalog Technologies). To correct for differences in transfection efficiencies, no. 32106). Co-immunopurification experiments for GemC1-p73 interaction luciferase activity was normalized to that of β-galactosidase, which was Journal of Cell Science

8 RESEARCH ARTICLE Journal of Cell Science (2018) 132, jcs228684. doi:10.1242/jcs.228684 measured spectrophotometrically at 420 nm using o-nitrophenyl β-D- tracheae by protease digestion (Pronase, Roche), seeded onto collagen-coated galactopyranoside as substrate. supported, semipermeable membranes (Transwell filters, Corning), and The expression vectors that were used in the NSCs were: cultured at 37°C under 5% CO2 in MTEC complete medium (You and Brody, pCAGGS-GemC1-IRES-GFP, pLV-DEST-p73-GFP, pCMV-E2F5 and 2013) to favor proliferation. In detail, the medium was a 1:1 mixture of their empty vectors as control. DMEM and F-12 (Sigma, D8437), 15 mM HEPES, 4 mM L-glutamine, 3.6 mM NaHCO3, supplemented with penicillin (100 units/ml; Invitrogen) Chromatin immunoprecipitation assay and streptomycin (100 µg/ml; Invitrogen), 10 µg/ml insulin, 5 µg/ml HEK293T cells were co-transfected by using Turbofect (Fermentas) within transferrin, 0.1 µg/ml cholera toxin, 25 ng/ml EGF, 30 µg/ml bovine a 100 mm dish, with a total of 5 µg DNA, containing 1 µg of each of the pituitary extract, retinoic acid (5×10−7 M) and 5% FBS. Two days after following plasmid DNAs: pGL3-p73 promoter, pLV-Dest-p73GFP or confluency, an air-liquid inerface (ALI) culture condition was established by pLVDest-GFP, pLV-Dest-GemC1-IRES-dsRED or pLV-Dest-IRES- adding a serum-free differentiation medium (You and Brody, 2013) only in dsRED, pCMV-E2F5 and pcDNA3.1+/C-(K)DYK-TFDP1. At 48 h after the basal chamber. Ciliogenesis initiates 2–3 days after establishment of transfection, the cells were fixed according to the manufacturer’s protocol ALI. To infect MTECs, 5 days after the initiation of the culture, epithelial (Active Motif ChIP IT High Sensitivity kit), using 16% paraformaldehyde tight junctions were disrupted by treating cells with 12 mM EGTA in 10 mM (PFA) aqueous solution (Electron Microscopy Sciences, catalog no. 15710). HEPES pH 7.4, at 37°C for 20 min. Cells were spin-infected at 1300 g Sonication was performed in a Bioruptor (Diagenode) for 30 cycles (30 s for 80 min with a non-concentrated lentiviral preparation (∼106 infectious on/off, power settings high), using 20 µg chromatin. Subsequently, chromatin units per ml) containing 5 μg/ml hexadimethrine bromide (Sigma-Aldrich). was incubated with 4 µg of either a rabbit anti-GFP antibody (Thermo The infection mix was removed 24 h later, and an ALI condition was set up Fisher Scientific, A11122) or a rabbit control IgG (Sigma-Aldrich, 2 days after the infection. Lentiviral expression vectors were kindly provided catalog no. I5006). The antibody-bound protein–DNA complexes were by Dr M. Goetz. immunoprecipitated through the use of Protein G agarose beads and washed using gravity filtration. Immunoprecipitates were analyzed by quantitative Mouse knockout strain and immunohistochemistry real-time PCR (qPCR; Applied Biosystems, StepOne) using the Kapa SYBR C57/Bl6 and Parkes mice were housed in the animal house of the Fast qPCR kit (KapaBiosystems, KK4604), as previously described University of Patras. Animal care and experiments were approved by the (Karamitros et al., 2015). Primers used to amplify the E2F-containing Veterinary Administrations of the Prefectures of Achaia, Greece, and region were Fw 5′-CAAAAGGCGGCGGGAAGG-3′ and Rv 5′-CCTA- were conducted in strict accordance with EU directives. The mouse line GGCGGGTTATATGGGC-3′. Primers for OR1 (Fw: 5′- TTGGGAGGAA- carrying the knockout (KO) allelle Gmnctm1a(KOMP)Wtsi (http://www. GTGACAACAG-3′;5′-Rv CGTTTCCGTAAGCATCCAAG-3′) were used mousephenotype.org/data/alleles/MGI:2685452/tm1a(KOMP)Wtsi) of as a negative control. The experiment was conducted twice. the GemC1 gene was generated by the trans-NIH Knockout Mouse Project (KOMP), and was obtained from the KOMP repository and In utero electroporation has been previously described by Arbi et al. (2016). In brief, inactivation of In utero electroporations (IUE) were performed in mouse embryos at E14.5 GemC1 was accomplished by the insertion of a lacZ-neo cassette dpc as described previously (Kyrousi et al., 2015). Pregnant females were between exon 2 and exon 3 of the GemC1 gene. A ‘knockout-first’ anesthetized by intraperitoneal (IP) injection of saline solution containing strategy that ablates gene function by the insertion of RNA processing fentanyl (0.05 mg/kg), midazolam (5 mg/kg) and medetomidine (0.5 mg/kg), signals without the deletion of any of the target gene was followed while anesthesia was antagonized by IP injection of atipamezol (antisedan, (Testa et al., 2004). 2.5 mg/kg) and flumazenil (anexate, 0.5 mg/kg). The uterine horns were Brains and tracheas were dissected from neonatal mice (independent of carefully exposed and using a glass microcapillary 1 µl of the plasmids IRES- sex). Brains and tracheas from P0-P3 mice were fixed overnight in 4% PFA. GFP, GemC1-IRES-GFP, together with Fast Green (0.1%, Sigma), into the Brains from P7 mice were isolated after transcardiac perfusion and lateral ventricles of the embryos. For GemC1 overexpression, the total plasmid post-fixed in 4% PFA overnight at 4°C. Subsequently, brains and tracheas concentration was 1 µg. The embryos were subsequently electroporated with 5 were washed with PBS and cryopreserved in 30% sucrose. The brains were pulses applied at 40 V for 50 ms each at intervals of 500 ms using the frozen in 7.5% gelatine plus 15% sucrose and the tracheas in Tissue-Tek. Electroporation system ECM830 (Harvard Apparatus). Following The tissues were sectioned at 10-μm thickness. electroporation, embryos were placed back into the abdominal cavity and For immunohistochemistry, brain coronal and trachea transverse sections left to grow for 2 days before euthanization. Data were obtained from n=12 and derived from newborn mice (Parkes and C57B6) were fixed with 4% PFA n=10 sections overexpressing IRES-GFP and GemC1-IRES-GFP, for 10 min, treated with 0.3% Triton X-100 for 5 min, and immunostained respectively (n=3). as previously described (Spella et al., 2011). In more detail, sections were incubated in blocking solution containing 10% FBS, 3% BSA, 0.1% Tween Lentiviral production 20 in 1× PBS, for 1 h. Samples were incubated with primary antibodies in Lentiviral production was as previously described (Arbi et al., 2016). GFP or blocking solution at 4°C, overnight. Brains from n=4 (P0, n=2; P7, n=2) and GFP-GemC1 was initially cloned into the BamHI/XhoI or KpnI/XbaI tracheas from n=2 mice were examined. restriction site, respectively, of the pENTR1AminusCmR vector. An LR Primary antibodies used were: rabbit anti-p73 (EP436Y) (1:300, Abcam, recombination reaction, using the Gateway LR Clonase II enzyme mix ab40658), chicken anti-GFP (1:1000, Aves Lab, GFP1020), mouse anti- (Invitrogen, 11791), was performed between the attL-containing entry GFP (1:500, Thermo Fisher Scientific, clone 3E6, A1120), mouse anti- clone and the attR-containing destination pLVDest-CAG vector. A second- acetylated α-tubulin (1:1500, Sigma, T6793), rat anti-E-cadherin (1:500, generation packaging system (Kutner et al., 2009) was used. HEK293T cells, Thermo Fisher Scientific, clone ECCD-2, 131900). grown in DMEM (Gibco) with 10% FBS (Gibco), were used as producer cells. The following Alexa Fluor-labeled secondary antibodies (Invitrogen) In short, lentiviral particles were generated by transient co-transfection of were used at 1:1000 in blocking solution, for 1 h: Alexa Fluor 488 goat anti- HEK293T cells with the expression vector carrying either the GFP or the rabbit, Alexa Fluor 488 goat anti-chicken, Alexa Fluor 488 goat anti-mouse, GFP-GEMC1 gene and the two helper plasmids psPAX2 (packaging vector, Alexa Fluor 568 goat anti-rabbit, Alexa Fluor 647 donkey anti-mouse, Addgene #12260) and pMD2.G (envelope vector, Addgene #12259), using Alexa Fluor 568 anti-rat. Turbofect (Fermentas) as transfection reagent. The lentiviral supernatant was DNA was stained either with Hoechst dye 33258 (1:1.500, Sigma) or with harvested 48 h later and filtered with 0.45-µm filter. Draq-5 (1:1.000, Biostatus). Sections were mounted in Mowiol 4-88 (Calbiochem). Mouse tracheal epithelial cells – culture and lentiviral infection Mouse tracheal epithelial cells (MTECs) were isolated from wild-type C57/ Postnatal radial glial cell culture Bl6 mice at age 8 weeks or older, as described previously (You et al., 2002; For culturing of mouse postnatal radial glial cells (pRGCs) in vitro, the walls

Arbi et al., 2016). Briefly, epithelial cells were isolated from n=7 mouse of the lateral ventricles of newborn Parkes mice were dissected, Journal of Cell Science

9 RESEARCH ARTICLE Journal of Cell Science (2018) 132, jcs228684. doi:10.1242/jcs.228684 mechanically dissociated and plated onto poly-D-lysine coated coverslips in gp/index.html) (Tripathi et al., 2015). Raw RNA-sequence data were proliferation medium containing DMEM-High Glucose (Gibco), 10% FBS deposited to the Sequence Read Archive (SRA) under the accession number (Gibco), 1% penicillin/streptomycin (Gibco). pRGCs were kept in PRJNA528250. proliferation medium for 3 days which was subsequently replaced with differentiation medium containing DMEM-High Glucose (Gibco), 2% FBS Acknowledgements (Gibco), 1% penicillin/streptomycin (Gibco). Cells were analyzed 3 days We thank Dr Magdalena Goetz (Ludwig-Maximilians-Universität, München, following differentiation. pRGCs from n=3 mice were examined. Germany) for kindly providing the lentiviral expression vectors. We thank the Advanced Light Microscopy Facility of the Medical School at the University of Patras and the Experimental Animal Facility of the University of Patras for support with Image acquisition experiments, and members of our laboratories for helpful discussions. Images derived from immunohistochemistry were recorded on a confocal fluorescence microscopy Leica TCS SP5 with a Leica DMI6000B Competing interests microscope using 40× and 63× lenses. Digital images were processed The authors declare no competing or financial interests. with Adobe Photoshop, Adobe Illustrator and Fiji software. Author contributions Quantification and statistical analysis Conceptualization: S.T.; Methodology: M.-E.L., M.A., I.L., K.K., A.K., G.L., C.K., A.M., Statistically significant estimates were obtained from N independent T.G., N.J., D.G., V.B., P.K.P., A.P., Z.L., S.T.; Resources: S.T.; Writing - original draft: experiments and n technical replicates, as defined in the figure legends. M.-E.L, M.A., I.L., A.P., Z.L., S.T.; Writing - review & editing: Z.L., S.T.; Supervision: Statistics were performed with two-tailed Mann–Whitney U test or a two- Z.L., S.T.; Project administration: Z.L., S.T.; Funding acquisition: A.P., Z.L., S.T. tailed unpaired Student’s t-test, as indicated. The mean±standard error of the mean (±s.e.m.) was plotted in each graph. Differences in the mean values were Funding considered to be significant at P<0.05 (*P<0.05, **P<0.01, ***P<0.001). This work was supported by the General Secretariat for Research and Technology (GSRT) and Hellenic Foundation for Research and Innovation (HFRI, 1680); and Fondation Sante. Work in A.P.’s lab was supported by UMG core funding and the Native (intrinsic) circular chromosome conformation capture Deutsche Forschungsgemeinschaft (DFG) (grant numbers: PA2456/4-1 and (i4C) analysis PA2456/5-1). i4C was performed on ∼1 million wild-type or GemC1-knockout cells pooled from >4 mice and crosslinked in 1% PFA/PBS at RT for 10 min. Preparation Data availability of the i4C templates, using ApoI as the primary restriction enzyme and the Raw RNA-sequence and i4C data reported in this paper have been deposited to the promoter regions of Foxj1 and p73 as viewpoints, was performed exactly as Short Read Archive (SRA) under accession numbers PRJNA528250 (https://www. described by Brant et al. (2016). PCR-based preparation of i4C libraries for ncbi.nlm.nih.gov/sra/PRJNA528250) and PRJNA528247 (https://www.ncbi.nlm. sequencing on a HiSeq2500 (Illumina) platform was done as previously nih.gov/sra/PRJNA528247), respectively. described by Stadhouders et al. (2013), and using the following viewpoint primers (5′–3′; sequences in lowercase indicate the Illumina adapter): Foxj1- Supplementary information F: aatgatacggcgaccaccgaacactctttccctacacgacgctcttccgatctTAGAGCCTGA- Supplementary information available online at CCAGCCTGCAATGGCTGCAGGAATTC. p73-F: aatgatacggcgaccacc- http://jcs.biologists.org/lookup/doi/10.1242/jcs.228684.supplemental gaacactctttccctacacgacgctcttccgatct-AGCTAGGGAGAGATCATGGCAGT- References AAGGCAGGAATTT. Following sequencing to at least 8 million reads, Arbi, M., Pefani, D.-E., Kyrousi, C., Lalioti, M.-E., Kalogeropoulou, A., analysis of chromosomal interactions was performed by using the fourSig Papanastasiou, A. D., Taraviras, S. and Lygerou, Z. (2016). GemC1 controls method (Williams et al., 2014) to obtain significant interactions for each multiciliogenesis in the airway epithelium. EMBO Rep. 17, 400-413. doi:10.15252/ viewpoint. The i4C data were deposited to the Sequence Read Archive (SRA) embr.201540882 under the accession number PRJNA528247. Arbi, M., Pefani, D.-E., Taraviras, S. and Lygerou, Z. (2018). Controlling centriole numbers: Geminin family members as master regulators of centriole amplification Chromatin immunoprecipitation (ChIP) coupled to qPCR and multiciliogenesis. Chromosoma 127, 151-174. doi:10.1007/s00412-017-0652-7 Balestrini, A., Cosentino, C., Errico, A., Garner, E. and Costanzo, V. (2010). ChIP was performed on <0.5 million cells wild-type or GemC1-knockout GEMC1 is a TopBP1-interacting protein required for chromosomal DNA cells crosslinked in 1% PFA/PBS at RT for 10 min and quenched in replication. Nat. Cell Biol. 12, 484-491. doi:10.1038/ncb2050 0.125 M Glycine. For the preparation of ChIP material the protocol by Ford Boon, M., Wallmeier, J., Ma, L., Loges, N. T., Jaspers, M., Olbrich, H., et al. (2014) was followed using a Bioruptor sonicator and adjusting Dougherty, G. W., Raidt, J., Werner, C., Amirav, I. et al. (2014). MCIDAS fragment size to 300-500 bp; for details see: https://ethanomics.files. mutations result in a mucociliary clearance disorder with reduced generation of wordpress.com/2012/01/chip_covaris4.pdf). multiple motile cilia. Nat. Commun. 5, 4418. doi:10.1038/ncomms5418 For H3K27ac IP a rabbit polyclonal antibody was used (39133, Active Brant, L., Georgomanolis, T., Nikolic, M., Brackley, C. A., Kolovos, P., van Motif). qPCR analysis was performed in biological duplicates (each a pool Ijcken, W., Grosveld, F. G., Marenduzzo, D. and Papantonis, A. (2016). Exploiting native forces to capture chromosome conformation in mammalian cell of cells from more than four mice in technical duplicates) on a LightCycler ′– ′ OlfR nuclei. Mol. Syst. Biol. 12, 891. doi:10.15252/msb.20167311 480 II (ABI) platform using the following primer pairs (5 3 ): -Fb Brooks, E. R. and Wallingford, J. B. (2014). Multiciliated cells. Curr. Biol. 24, (negative control): ATAGTGGCCACATCAGCTCACA; OlfR-Rb (negative R973-R982. doi:10.1016/j.cub.2014.08.047 control): GCAAACGATGACAAATTCTGCCCA; Foxj1-Fc: CCCTCAC- Caillat, C., Pefani, D.-E., Gillespie, P. J., Taraviras, S., Blow, J. J., Lygerou, Z. ATTCATCCCCGCA; Foxj1-Rc: CAAGGGCAGATGGAGAGAGGTG; and Perrakis, A. (2013). The Geminin and Idas coiled coils preferentially form a p73-Fd: CAGCCGCACACCCGAACTC; p73-Rc: CAGGTAGGCAGC- heterodimer that inhibits Geminin function in DNA replication licensing. J. Biol. TCTGGGAC. Chem. 288, 31624-31634. doi:10.1074/jbc.M113.491928 Caillat, C., Fish, A., Pefani, D.-E., Taraviras, S., Lygerou, Z. and Perrakis, A. (2015). The structure of the GemC1 coiled coil and its interaction with the Geminin RNA sequencing and data analysis family of coiled-coil proteins. Acta Crystallogr. D Biol. Crystallogr. 71, 2278-2286. Data from poly(A)+ cDNA libraries generated using the TruSeq RNA doi:10.1107/S1399004715016892 library kit (Illumina) were sequenced read pairs on a GAII platform Engeland, K. (2018). Cell cycle arrest through indirect transcriptional repression by (Illumina); raw reads were mapped to the mouse reference genome (mm19) p53: i have a DREAM. Cell Death Differ. 25, 114-132. doi:10.1038/cdd.2017.172 and analyzed by using the QuickNGS pipeline (Wagle et al., 2015) to obtain Fliegauf, M., Benzing, T. and Omran, H. (2007). When cilia go bad: cilia defects read counts. Then, read counts from the different mouse strains were and ciliopathies. Nat. Rev. Mol. Cell Biol. 8, 880-893. doi:10.1038/nrm2278 ‘ ’ Ford, E., Nikopoulou, C., Kokkalis, A. and Thanos, D. (2014). A method for normalized using the intrinsic top quantile function of the RUVSeq generating highly multiplexed ChIP-seq libraries. BMC Res. Notes 7, 312. doi:10. package (Risso et al., 2014), and differential gene expression analysis was 1186/1756-0500-7-312 performed using DEseq2 (Love et al., 2014). Gene ontology (GO) term Gonzalez-Cano, L., Fuertes-Alvarez, S., Robledinos-Anton, N., Bizy, A., enrichment analyses was performed using Metascape (http://metascape.org/ Villena-Cortes, A., Fariñas, I., Marques, M. M. and Marin, M. C. (2016). p73 Journal of Cell Science

10 RESEARCH ARTICLE Journal of Cell Science (2018) 132, jcs228684. doi:10.1242/jcs.228684

is required for ependymal cell maturation and neurogenic SVZ cytoarchitecture. Spassky, N. and Meunier, A. (2017). The development and functions of Dev. Neurobiol. 76, 730-747. doi:10.1002/dneu.22356 multiciliated epithelia. Nat. Rev. Mol. Cell Biol. 18, 423-436. doi:10.1038/nrm. Julian, L. M. and Blais, A. (2015). Transcriptional control of stem cell fate by E2Fs 2017.21 and pocket proteins. Front. Genet. 6, 161. doi:10.3389/fgene.2015.00161 Spector, R., Robert Snodgrass, S. and Johanson, C. E. (2015). A balanced view Karamitros, D., Patmanidi, A. L., Kotantaki, P., Potocnik, A. J., Bähr-Ivacevic, of the cerebrospinal fluid composition and functions: focus on adult humans. Exp. T., Benes, V., Lygerou, Z., Kioussis, D. and Taraviras, S. (2015). Geminin Neurol. 273, 57-68. doi:10.1016/j.expneurol.2015.07.027 deletion increases the number of fetal hematopoietic stem cells by affecting the Spella, M., Kyrousi, C., Kritikou, E., Stathopoulou, A., Guillemot, F., Kioussis, expression of key transcription factors. Development 142, 70-81. doi:10.1242/ D., Pachnis, V., Lygerou, Z. and Taraviras, S. (2011). Geminin regulates cortical dev.109454 progenitor proliferation and differentiation. Stem Cells Dayt. Ohio 29, 1269-1282. Kutner, R. H., Zhang, X.-Y. and Reiser, J. (2009). Production, concentration and doi:10.1002/stem.678 titration of pseudotyped HIV-1-based lentiviral vectors. Nat. Protoc. 4, 495-505. Stadhouders, R., Kolovos, P., Brouwer, R., Zuin, J., van den Heuvel, A., Kockx, C., Palstra, R.-J., Wendt, K. S., Grosveld, F., van Ijcken, W. et al. (2013). doi:10.1038/nprot.2009.22 Multiplexed chromosome conformation capture sequencing for rapid genome- Kyrousi, C., Arbi, M., Pilz, G.-A., Pefani, D.-E., Lalioti, M.-E., Ninkovic, J., Götz, scale high-resolution detection of long-range chromatin interactions. Nat. Protoc. M., Lygerou, Z. and Taraviras, S. (2015). Mcidas and GemC1 are key regulators 8, 509-524. doi:10.1038/nprot.2013.018 for the generation of multiciliated ependymal cells in the adult neurogenic niche. Stolp, H. B. and Molnár, Z. (2015). Neurogenic niches in the brain: help and Development 142, 3661-3674. doi:10.1242/dev.126342 hindrance of the barrier systems. Front. Neurosci. 9, 20. doi:10.3389/fnins.2015. Kyrousi, C., Lalioti, M.-E., Skavatsou, E., Lygerou, Z. and Taraviras, S. (2016). 00020 Mcidas and GemC1/Lynkeas specify embryonic radial glial cells. Neurogenesis Stubbs, J. L., Vladar, E. K., Axelrod, J. D. and Kintner, C. (2012). Multicilin Austin Tex 3, e1172747. doi:10.1080/23262133.2016.1172747 promotes centriole assembly and ciliogenesis during multiciliate cell Kyrousi, C., Lygerou, Z. and Taraviras, S. (2017). How a radial glial cell decides to differentiation. Nat. Cell Biol. 14, 140-147. doi:10.1038/ncb2406 become a multiciliated ependymal cell. Glia 65, 1032-1042. doi:10.1002/glia. Symeonidou, I.-E., Kotsantis, P., Roukos, V., Rapsomaniki, M.-A., Grecco, 23118 H. E., Bastiaens, P., Taraviras, S. and Lygerou, Z. (2013). Multi-step loading of Love, M. I., Huber, W. and Anders, S. (2014). Moderated estimation of fold change human minichromosome maintenance proteins in live human cells. J. Biol. Chem. and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550. doi:10. 288, 35852-35867. doi:10.1074/jbc.M113.474825 1186/s13059-014-0550-8 Terré, B., Piergiovanni, G., Segura-Bayona, S., Gil-Gómez, G., Youssef, S. A., Ma, L., Quigley, I., Omran, H. and Kintner, C. (2014). Multicilin drives centriole Attolini, C. S.-O., Wilsch-Bräuninger, M., Jung, C., Rojas, A. M., Marjanović, biogenesis via E2f proteins. Genes Dev. 28, 1461-1471. doi:10.1101/gad. M. et al. (2016). GEMC1 is a critical regulator of multiciliated cell differentiation. 243832.114 EMBO J. 35, 942-960. doi:10.15252/embj.201592821 Marshall, C. B., Mays, D. J., Beeler, J. S., Rosenbluth, J. M., Boyd, K. L., Santos Testa, G., Schaft, J., van der Hoeven, F., Glaser, S., Anastassiadis, K., Zhang, Guasch, G. L., Shaver, T. M., Tang, L. J., Liu, Q., Shyr, Y. et al. (2016). p73 is Y., Hermann, T., Stremmel, W. and Stewart, A. F. (2004). A reliable lacZ required for multiciliogenesis and regulates the Foxj1-associated gene network. expression reporter cassette for multipurpose, knockout-first alleles. Genesis 38, Cell Rep. 14, 2289-2300. doi:10.1016/j.celrep.2016.02.035 151-158. doi:10.1002/gene.20012 Medina-Bolıvar,́ C., González-Arnay, E., Talos, F., González-Gómez, M., Moll, Tripathi, S., Pohl, M. O., Zhou, Y., Rodriguez-Frandsen, A., Wang, G., Stein, U. M. and Meyer, G. (2014). Cortical hypoplasia and ventriculomegaly of p73- D. A., Moulton, H. M., DeJesus, P., Che, J., Mulder, L. C. F. et al. (2015). Meta- “ ” deficient mice: developmental and adult analysis. J. Comp. Neurol. 522, and orthogonal integration of influenza OMICs data defines a role for UBR4 in 2663-2679. doi:10.1002/cne.23556 virus budding. Cell Host Microbe 18, 723-735. doi:10.1016/j.chom.2015.11.002 ́ Nemajerova, A., Kramer, D., Siller, S. S., Herr, C., Shomroni, O., Pena, T., Wagle, P., Nikolic, M. and Frommolt, P. (2015). QuickNGS elevates next- Gallinas Suazo, C., Glaser, K., Wildung, M., Steffen, H. et al. (2016). TAp73 is a generation sequencing data analysis to a new level of automation. BMC Genomics 16, 487. doi:10.1186/s12864-015-1695-x central transcriptional regulator of airway multiciliogenesis. Genes Dev. 30, Williams, R. L., Starmer, J., Mugford, J. W., Calabrese, J. M., Mieczkowski, P., 1300-1312. doi:10.1101/gad.279836.116 Yee, D. and Magnuson, T. (2014). fourSig: a method for determining Paez-Gonzalez, P., Abdi, K., Luciano, D., Liu, Y., Soriano-Navarro, M., Rawlins, chromosomal interactions in 4C-Seq data. Nucleic Acids Res. 42, e68-e68. E., Bennett, V., Garcia-Verdugo, J. M. and Kuo, C. T. (2011). Ank3-dependent doi:10.1093/nar/gku156 SVZ niche assembly is required for the continued production of new neurons. You, Y. and Brody, S. L. (2013). Culture and differentiation of mouse tracheal Neuron 71, 61-75. doi:10.1016/j.neuron.2011.05.029 epithelial cells. Methods Mol. Biol. 945, 123-143. doi:10.1007/978-1-62703-125- Pefani, D.-E., Dimaki, M., Spella, M., Karantzelis, N., Mitsiki, E., Kyrousi, C., 7_9 Symeonidou, I.-E., Perrakis, A., Taraviras, S. and Lygerou, Z. (2011). Idas, a You, Y., Richer, E. J., Huang, T. and Brody, S. L. (2002). Growth and differentiation novel phylogenetically conserved geminin-related protein, binds to geminin and is of mouse tracheal epithelial cells: selection of a proliferative population. required for cell cycle progression. J. Biol. Chem. 286, 23234-23246. doi:10.1074/ Am. J. Physiol. Lung Cell. Mol. Physiol. 283, L1315-L1321. doi:10.1152/ jbc.M110.207688 ajplung.00169.2002 Risso, D., Ngai, J., Speed, T. P. and Dudoit, S. (2014). Normalization of RNA-seq Zhou, F., Narasimhan, V., Shboul, M., Chong, Y. L., Reversade, B. and Roy, S. data using factor analysis of control genes or samples. Nat. Biotechnol. 32, (2015). Gmnc is a master regulator of the multiciliated cell differentiation program. 896-902. doi:10.1038/nbt.2931 Curr. Biol. 25, 3267-3273. doi:10.1016/j.cub.2015.10.062 Journal of Cell Science