ARTICLE https://doi.org/10.1038/s41467-019-10799-5 OPEN Time series modeling of cell cycle exit identifies Brd4 dependent regulation of cerebellar neurogenesis Clara Penas1,8,9, Marie E. Maloof 1,9, Vasileios Stathias1, Jun Long2, Sze Kiat Tan 1, Jose Mier3, Yin Fang4, Camilo Valdes5, Jezabel Rodriguez-Blanco2, Cheng-Ming Chiang6, David J. Robbins2, Daniel J. Liebl3, Jae K. Lee3, Mary E. Hatten4, Jennifer Clarke7 & Nagi G. Ayad1 1234567890():,; Cerebellar neuronal progenitors undergo a series of divisions before irreversibly exiting the cell cycle and differentiating into neurons. Dysfunction of this process underlies many neu- rological diseases including ataxia and the most common pediatric brain tumor, medullo- blastoma. To better define the pathways controlling the most abundant neuronal cells in the mammalian cerebellum, cerebellar granule cell progenitors (GCPs), we performed RNA- sequencing of GCPs exiting the cell cycle. Time-series modeling of GCP cell cycle exit identified downregulation of activity of the epigenetic reader protein Brd4. Brd4 binding to the Gli1 locus is controlled by Casein Kinase 1δ (CK1 δ)-dependent phosphorylation during GCP proliferation, and decreases during GCP cell cycle exit. Importantly, conditional deletion of Brd4 in vivo in the developing cerebellum induces cerebellar morphological deficits and ataxia. These studies define an essential role for Brd4 in cerebellar granule cell neurogenesis and are critical for designing clinical trials utilizing Brd4 inhibitors in neurological indications. 1 Department of Psychiatry and Behavioral Sciences, Center for Therapeutic Innovation, Sylvester Comprehensive Cancer Center, Miami Project to Cure Paralysis, University of Miami Miller School of Medicine, Miami, FL 33136, USA. 2 Department of Surgery, University of Miami Miller School of Medicine, Miami, FL 33136, USA. 3 Department of Neurosurgery, Miami Project to Cure Paralysis, University of Miami Miller School of Medicine, Miami, FL 33136, USA. 4 Laboratory of Development Neurobiology, The Rockefeller University, New York, NY 10065, USA. 5 Computing and Information Sciences, Florida International University, Miami, FL 33199, USA. 6 Simmons Comprehensive Cancer Center, Department of Biochemistry and Department of Pharmacology, University of Texas Southwestern Medical Center, Dallas, TX 75390, USA. 7 Department of Statistics, University of Nebraska, Lincoln, NE 68588, USA. 8 Present address: Institut de Neurociències, Departament de Biologia Cellular, Fisiologia i Immunologia, Universitat Autònoma de Barcelona, CIBERNED, Bellaterra 08193, Spain. 9These authors contributed equally: Clara Penas, Marie E. Maloof. Correspondence and requests for materials should be addressed to N.G.A. (email: nayad@miami.edu) NATURE COMMUNICATIONS | (2019)10:3028 | https://doi.org/10.1038/s41467-019-10799-5 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10799-5 D uring postnatal mammalian development, granule cell cycle regulator, Cyclin b1, and the bHLH transcription factor progenitors (GCPs) undergo symmetric divisions in the important for maintaining GCPs in a proliferative state, Atoh1, external germinal layer (EGL) of the brain and exit the decreased by 24 h post plating. By contrast, neuron-specific class cell cycle within a narrow time frame, resulting in rapid cellular III beta-tubulin (Tuj1) and the axonal growth marker, Gap43, expansion and differentiation1. However, the fundamental increased at 24 h. To determine the exact timing of changes in mechanisms controlling irreversible GCP cell-cycle exit have not cellular pathways during cell-cycle exit, we performed short-time- been elucidated. It is essential we discover these mechanisms to series modeling16 with gene ontology clustering analysis of all understand cerebellar development, as defects in GCP expansion mRNAs expressed at 0, 2, 4, 6, 12, 24, and 48 h after plating have been linked to cerebellar ataxia and the most common (Fig. 1d). Interestingly, downregulation of cell proliferation pediatric brain tumor, medulloblastoma2. Epigenetic modifiers pathways (Cluster #47, Fig. 1d; Supplementary Fig. 1, Supple- control gene expression without changing DNA sequence. These mentary Data 1) occurred at the same time as upregulation of modifiers include histone acetyltransferases (HATs) and histone neuronal development pathways (Cluster #75, Fig. 1d; Supple- deacetylases (HDACs). Histone acetyltransferases attach acetyl mentary Fig. 1, Supplementary Data 1), suggesting that the two groups to lysine residues on histone proteins, while HDACs processes may be temporally and mechanistically linked (Fig. 1d, remove those modifications3. Histones normally bind DNA Supplementary Fig. 1). Importantly, the SHH pathway, which is a molecules via their positively charged lysine and arginine tails3. major regulator of GCP expansion17, is downregulated beginning Histone–DNA binding initiates DNA compaction and tran- at 2 h of GCP cell-cycle exit, with the SHH effectors Gli1 and Gli2 scriptional silencing. Histone tail acetylation reduces this positive nearing basal levels by 24 h, which is recapitulated in vivo charge, attenuates DNA binding and compaction and allows (Supplementary Figs. 1, 2). transcription. Histone deacetylation has the opposite effect, thus Our prior studies demonstrated that the epigenetic reader allowing histone–DNA binding and reducing transcription. protein Brd4 regulates Gli1 levels in mouse embryonic fibroblasts Working in combination with HATs (writers) and HDACs (MEFs) by directly binding the Gli1 locus11. Brd4 is part of a (erasers) are histone-reader proteins, which bind to acetylated family of bromodomain and extraterminal domain proteins lysines on histones, recruit transcriptional complexes, and med- (BETs) that bind to acetylated lysines on histones and recruit iate gene transcription. Among the “reader” proteins are bro- transcriptional complexes to induce transcription of various genes modomain and extra-terminal domain (BET) proteins. BET involved in cell proliferation, signaling, and inflammation18. To proteins include Brd2, Brd3, Brd4, and BrdT. Tissue expression test whether Brd4-dependent regulation of Gli1 changes during studies suggest that Brd4 is heavily expressed in the brain, making GCP cell-cycle exit, we performed Brd4 chromatin immunopre- it a likely candidate for modulating neurogenesis in the nervous cipitation (ChIP) analysis in GCPs exiting the cell cycle. Brd4 is system4. Recent sequencing studies have implicated many epi- expressed in the developing cerebellum during P6-9 when GCPs genetic regulators in medulloblastoma5. The epigenetic reader are proliferating and exiting the cell cycle (Supplementary Fig. 3). protein Brd4 has been implicated in various cancers, including As observed in Fig. 2a, Brd4 binding to the Gli1 locus decreased medulloblastoma6–13. Brd4 controls expression of the medullo- dramatically within the first 2 h of GCP cell-cycle exit. Brd4 blastoma essential gene MYC in G3 medulloblastomas, which activity has been shown to be regulated by phosphorylation19 and have poor prognosis as well as GLI1 and GLI2 levels in Sonic therefore, we tested whether Brd4 phosphorylation decreases hedgehog (SHH)-driven medulloblastomas, which have inter- during cell-cycle exit. Brd4 phosphorylation decreases with the mediate prognosis. Highly selective Brd4 inhibitors have been same kinetics as Brd4 binding to the Gli1 locus during cell-cycle developed that reduce MYC, GLI1, and GLI2 levels. These inhi- exit (Fig. 2b, c; Supplementary Fig. 1). bitors have gone into clinical trials for multiple cancer indica- tions, and one Brd4 inhibitor has received fast-track designation from the FDA for myelofibrosis14,15. However, it is unclear Casein Kinase 1δ inhibition or deletion reduces Brd4 binding whether these inhibitors can be given to children suffering from to the Gli1 locus. We have previously demonstrated that Casein medulloblastoma, as we do not fully understand the role of Brd4 Kinase 1 delta (CK1δ) is required for GCP proliferation, and its during normal development. To address this issue, we deleted protein levels decrease with the same kinetics as Brd4 phos- Brd4 in the developing cerebellum in mice and find that it is phorylation during cell-cycle exit20 (Fig. 2b). Therefore, we essential for cerebellar growth. Brd4 knockout leads to cerebellar hypothesized that CK1δ may control Brd4 phosphorylation. ataxia that is linked to defects in cerebellar development starting Indeed, in vitro phosphorylation assays using purified Brd4 and at postnatal day 3. These studies suggest that Brd4 inhibitors CK1δ demonstrated that Brd4 is a CK1δ substrate in vitro may need to be given during a short developmental window in (Fig. 2d, e). Earlier studies have shown that Brd4 phosphorylation children to reduce potential negative effects on cerebellar on serines 492/494 relieves Brd4 autoinhibition, thereby allowing development. it to bind to chromatin19. We found that serines 492 and 494 are necessary for CK1δ-mediated phosphorylation of Brd4 as mutating both to alanine abrogated the ability of CK1δ to Results phosphorylate Brd4 in vitro (Fig. 2d–f). To determine whether Brd4 phosphorylation decreases during cell-cycle exit of CK1δ controls Brd4 activity in GCPs, we incubated proliferating granule cell progenitors. GCPs undergo cell-cycle exit and dif- GCPs with the selective CK1δ inhibitor, SR-127721, or vehicle in ferentiation when plated on poly-D-lysine/laminin coated plates. the presence of SHH and measured Brd4 phosphorylation on We utilized an in vitro system to isolate purified GCPs at various serines 492/494 at 24 h (Fig. 3a, b). SR-1277 treatment decreased times during differentiation and performed RNA sequencing on phospho-S492/494 Brd4 levels relative to the total Brd4 (Fig. 3a, purified cell populations from postnatal day (P) 6 mice to b), which correlated with decreased Brd4 binding to the Gli1 determine their characteristics during the exit process (Fig. 1). locus as measured by Brd4 ChIP analysis (Fig. 3c). Importantly, We found that GCPs exited the cell cycle within 24 h of plating as CK1δ inhibition reduced Brd4 association with the Gli1 locus to a judged by PI-FACS analysis (Fig. 1a) and EdU incorporation similar extent as treatment with the selective Brd4 inhibitor, (Fig. 1b). The mRNA expression of several proliferation markers I-BET15122, suggesting that CK1δ activity is required for main- decreased during this time period while those of differentiation taining Brd4 in an active state (Fig. 3c). In agreement with this, markers increased (Fig. 1c). For example, the levels of the cell- conditional deletion of CK1δ in GCPs in vivo reduced Brd4 2 NATURE COMMUNICATIONS | (2019)10:3028 | https://doi.org/10.1038/s41467-019-10799-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10799-5 ARTICLE a 100 G1 b S 30 % EdU-positive cells 75 G2 % cells 20 50 25 10 0 0 0 2 6 24 0 2 6 24 Hours post-plating Hours post-plating c Cyclin b1 Atoh1 Tuj1 Gap43 1.5 2.0 Relative mRNA levels Relative mRNA levels Relative mRNA levels Relative mRNA levels 5 3 1.5 4 1.0 2 3 1.0 0.5 2 0.5 1 1 0.0 0.0 0 0 0 2 6 24 0 2 6 24 0 2 6 24 0 2 6 24 Hours post-plating Hours post-plating Hours post-plating Hours post-plating d #47 #75 0.0 5.0 –2.5 Log2FC Log2FC 2.5 –5.0 –7.5 0.0 0 10 20 30 40 48 0 10 20 30 40 48 Time (h) Time (h) Mitotic cell cycle process Nervous system development DNA metabolic process Neurogenesis DNA conformation change Neuron projection development Nuclear division Cell morphogenesis involved in differentiation Mitotic nuclear division Neuron development DNA replication Cell morphogenesis Nucleosome assembly Neuron differentiation Chromatin assembly Cell morphogenesis involved in neuron differentiation Protein-DNA complex assembly Establishment of synaptic vesicle localization Nucleosome organization Synaptic vesicle transport 0 5 10 15 20 25 30 35 40 0 1 2 3 4 5 6 –log(adj.p) –log(adj.p) Fig. 1 Timing of cycle exit in GCPs. a GCPs exit the cell cycle within 24 h of plating as judged by PI-FACS analysis. FACS analysis was performed on GCPs purified from P6 mice and plated for the time intervals indicated. FlowJo software was then used to assess the percentage of cells in the G1, S, or G2/M phase. b Purified GCPs exit the cell cycle within 24 h of plating as judged by EdU incorporation studies. GCPs purified from P6 mice were processed for EdU incorporation, which was normalized to the total number of cells labeled with Hoechst staining. c Purified GCPs from P6 mice were plated and processed for the RNA expression at the time intervals indicated for the proliferative markers, Ccnb1 and Atoh1, and the differentiation markers, Tuj1 and Gap43. qRT- PCR was performed and normalized to Gapdh. d Short-time series modeling of mRNAs during cell-cycle exit. Plots represent mRNA expression profiles during GCP cell-cycle exit. Representative clusters (#47, #75 and their related cellular processes) are shown. The red line represents the consensus for each cluster. The gray lines represent individual mRNA expression profiles. The associated biological processes as defined by DAVID41 in each cluster are shown. The identity of each gene in the cluster can be found in Supplementary Data 1. The results are shown as the average values of three independent experiments and are represented as the mean ± SEM. A one-way ANOVA followed by Bonferroni multiple comparison testing was performed (*p < 0.05, **p < 0.01). Source data can be found in source data graphs under tabs for a–c phosphorylation on serines 492/494 (Fig. 3d), and binding to the receptor Smoothened also reduced Gli1 levels and Brd4 binding Gli1 locus (Fig. 3e). Consistent with a direct effect on Brd4, CK1δ to the Gli1 locus (Supplementary Fig. 4). Collectively, these stu- inhibition in Supressor of Fused17 deleted (Sufu−/−) cells that dies suggest that CK1δ-mediated phosphorylation of Brd4 on contain active SHH signaling independent of the membrane serines 492/494 potentiates Brd4 localization to the Gli1 locus NATURE COMMUNICATIONS | (2019)10:3028 | https://doi.org/10.1038/s41467-019-10799-5 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10799-5 a Brd4 ChIP, Gli1 locus b c Phospho-Brd4 Hours S492/494/total Brd4 –3250/–2501 post-plating +1250/+2000 4 0 2 6 24 1.0 Fold respect to input and lgG protein levels 3 Phospho-Brd4 S492/494 Relative 2 Brd4 0.5 1 Ck1δ 0 Gapdh 0 2 6 24 0 2 6 24 0.0 Hours post-plating 0 2 6 24 Hours post-plating d Ck1δ + + + + e Radioactive assay WT-PDID BRD4 + + S484A/S488A-PDID-BRD4 + + S492A/S494A-PDID-BRD4 + + 37 kDa 1.5 Relative protein 225 kDa levels 150 kDa 1.0 90 kDa 0.5 75 kDa Silver stain 50 kDa 0.0 T 88 94 37 kDa W /4 /4 84 92 S4 S4 PDID f Ck1δ + + + + WT-PDID BRD4 + + S484A/S488A-PDID-BRD4 + + S492A/S494A-PDID-BRD4 + + Anti-Phospho-S492/S494 37 kDa 37 kDa Anti-BRD4 30 kDa Anti-CK1δ Fig. 2 Brd4 phosphorylation on S492/494 and binding to the Gli1 locus is regulated by CK1δ and is reduced during GCP cell-cycle exit. a Brd4 association to the Gli1 locus decreases during GCP cell-cycle exit. White bars represent the association to a region far away from the transcription start site of Gli1, and black bars represent the association to a proximal region to the transcription start site (TSS). Brd4 occupancy was analyzed by ChIP-qPCR and then normalized to control ChIP performed using rabbit IgG. b, c Brd4 phosphorylation at serines 492/494 decreases during GCP cell-cycle exit. d CK1δ phosphorylates Brd4 in the PDID domain. The PDID domain of WT-Brd4, S484A/S488A-Brd4, or S492A/S494A-Brd4 were mixed with recombinant CK1δ and 32P-ATP, and the extent of radioactivity bound to WT or mutant Brd4 was determined after SDS-PAGE and autoradiography. Silver staining showed that recombinant WT-Brd4 and S484A/S488A-Brd4 had reduced electrophoretic mobility in the presence of CK1δ, while the S492 A/S494A- Brd4 mutant migrated slower, indicating lower phosphorylation. Both S492A/S492A are required for efficient phosphorylation of Brd4 by CK1δ. e Plot represents the quantification of the phosphorylation levels from the radioactive assay with respect to the total amount of protein quantified from the silver staining. f Confirmation of phosphorylation at S492/494 with an anti-phoshpo-Brd4 antibody. Recombinant WT-Brd4, S484A/S488A-Brd4, or S492A/ S494A-Brd4 were mixed with recombinant CK1δ and ATP, and the extent of phosphorylation determined by anti-phospho-S492/494 western analysis. Note that CK1δ phosphorylation of WT-Brd4, S484A/S488A-Brd4 leads to signal in the anti-phospho-S492/494 western analysis while no signal is observed with the S492A/S494A-Brd4 mutant, as anticipated. The results are shown as the average values of three independent experiments and are represented as the mean ± SEM. A one-way ANOVA followed by Bonferroni multiple comparison testing was performed (*p < 0.05, ***p < 0.001). Source data can be found in source data graphs under tabs for a, b, e and in source data Fig. 1 and Fig. 2 during GCP proliferation. By contrast, during GCP cell-cycle exit, Supplementary Fig. 5). Further, I-BET151 treatment of cerebellar rapid downregulation of CK1δ activity is associated with reduced slices ex vivo reduced EdU incorporation in the EGL (Fig. 4c, d). Brd4 phosphorylation on serines 492/494, decreased Brd4 bind- Finally, treatment of P8 pups with a brain penetrant BET inhi- ing to the Gli1 locus, and attenuated Gli1 expression. bitor, JQ123, also reduced cerebellar GCP proliferation in vivo (Fig. 4e, f). Collectively, these studies suggest that pharmacolo- gical inhibition of Brd4 reduces GCP expansion in the developing Brd4 Inhibition reduces granule cell progenitor proliferation cerebellum. in vitro, ex vivo, and in vivo. Downregulation of Brd4 phos- phorylation and binding during GCP cell-cycle exit suggests that it is an essential regulator of SHH signaling in GCPs, and that Brd4 deletion reduces GCP proliferation and induces ataxia. disrupting Brd4 activity may limit GCP proliferation. To test this, To determine whether genetic disruption of Brd4 affects cerebellar we measured the effect of BET inhibition on GCP proliferation development, we conditionally deleted Brd4 in GCPs by breeding via the Brd4 inhibitor I-BET151. I-BET151 treatment reduced Brd4fl/fl mice to Tg (Atoh1-Cre) mice, which express Cre under GCP proliferation and Gli1 expression in vitro (Fig. 4a, b; the Atoh1 promoter after embryonic day 13.524 (Fig. 5a). Tg 4 NATURE COMMUNICATIONS | (2019)10:3028 | https://doi.org/10.1038/s41467-019-10799-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10799-5 ARTICLE a b Phospho-Brd4 S492/494 Fig. 3 CK1δ inhibition or deletion reduces Brd4 phosphorylation on S492/ 494 and binding to the Gli1 locus. a CK1δ is required for Brd4 phosphorylation in vitro. CK1δ inhibition via the highly selective CK1δ SR-1277 1.0 Relative protein levels SHH + inhibitor (SR1277) reduces Brd4 phosphorylation at S492/494. GCPs were SHH isolated from P7 mice and subsequently treated with SHH + DMSO or SHH + SR-1277 for 24 h. Lysates were resolved and subsequently probed Phospho-Brd4 S492/494 for phospho-Brd4 S492/494, total Brd4, or Gapdh control. b The ratio of 0.5 Brd4 phospho-Brd4 S492/494 to total Brd4 from a) was quantified. Note that SR-1277 significantly reduced phosphorylation of Brd4 at S492/494. Gapdh c CK1δ or Brd4 inhibition reduces Brd4 binding to Gli1 locus, as measured 0.0 by Brd4 ChIP analysis. GCPs were incubated with either I-BET151 (Brd4 DMSO SR-1277 inhibitor) or SR-1277 (CK1δ inhibitor), and the extent of Brd4 bound to the c Brd4 ChIP, Gli1 locus Gli1 locus was measured after anti-Brd4 ChIP analysis. The amount of Brd4 Gli1 locus was normalized to the input DNA and the amount bound using IgG control. 3 DMSO was used as a vehicle control. Note that Brd4 bound close to TSS of Fold respect to input –3250/–2501 Gli1 and that I-BET151 or SR-1277 was able to displace it from the Gli1 locus. +1250/+2000 2 d Conditional deletion of CK1δ in GCPs in vivo reduces phosphorylation of and IgG Brd4 at S492/494. CK1δ was deleted in GCPs by breeding CK1δfl/fl mice to Tg (Atoh1-Cre) mice and the extent of phospho-Brd4 levels determined via 1 western analysis. The total Brd4 and Gapdh were included as loading controls. Note that CK1δ is efficiently deleted in Tg (Atoh1-Cre +);CK1δfl/fl 0 mice and that phosphorylation at serines S494/494 was lower in these animals relative to Tg (Atoh1-Cre-);CK1δfl/fl mice. e CK1δ knockout reduces SO SR 151 7 SO -1 1 7 27 SR 15 27 M M Brd4 binding to the Gli1 locus. GCPs were purified from Tg (Atoh1-Cre +); -1 ET ET D D I-B I-B CK1δfl/fl mice or Tg (Atoh1-Cre-);CK1δfl/fl mice, and the amount of Brd4 d e bound to the Gli1 locus was measured. Note that CK1δ deletion reduced Brd4 ChIP, Brd4 binding to the Gli1 locus. The results are shown as the average values Csnk1d fl/fl Gli1 locus of three independent experiments and are represented as the mean ± SEM. Tg (Atoh1-Cre): – + A paired t test was performed (*p < 0.05). Source data can be found in Phospho-Brd4 S492/494 1.0 source data graphs under tabs for b, c, e and in source data Fig. 3 and Fig. 4 Fold respect to input and IgG Brd4 Ck1δ 0.5 Gapdh in Tg (Atoh1-Cre−);Brd4fl/fl mice (Fig. 6c; Supplementary Movies 1–3). Taken together, these studies demonstrate that Brd4 0.0 is an essential regulator of GCP proliferation and cerebellar devel- Tg (Atoh1-Cre): – + opment in vivo. Csnk1d fl/fl Discussion We report a novel in vivo function for the epigenetic-reader protein (Atoh1-Cre +);Brd4fl/fl or Tg (Atoh1-Cre-);Brd4fl/fl mice were Brd4. Brd4 controls granule cell progenitor expansion in the attained, and Brd4 expression in isolated GCPs was analyzed via developing cerebellum. Brd4 deletion leads to defects in cerebellar qRT-PCR and western blot analysis. As seen in Fig. 5b, c, Brd4 morphology, which leads to ataxia. Brd4 activity is temporally mRNA and protein levels were lower in Tg (Atoh1-Cre +);Brd4fl/fl regulated during cerebellar granule cell development, as both CK1δ- mice relative to Tg (Atoh1-Cre−);Brd4fl/fl mice, suggesting that dependent phosphorylation of Brd4 on serines 492/494 and Brd4 efficient deletion of Brd4 occurs upon Cre expression. Reduced binding to the Gli1 locus decrease during GCP cell-cycle exit. Brd4 was correlated with lower levels of the positive effectors of Multiple transcriptional and posttranslational mechanisms have the SHH pathway Gli1 and Gli2, but not the negative regulator been shown to be required for cell-cycle exit in GCPs in the Gli3 (Fig. 5b). Furthermore, the level of the cell-cycle protein developing cerebellum25–28. To our knowledge, this is the first and Gli1 target, cyclin D1, was decreased in GCPs from Tg report demonstrating that the activity of an epigenetic reader pro- (Atoh1-Cre +);Brd4fl/fl mice relative to Tg (Atoh1-Cre−);Brd4fl/fl tein is modulated during cell-cycle exit in the developing nervous mice, suggesting that Brd4 is an essential regulator of cell pro- system. Interestingly, we find that Brd4 loss from the Gli1 locus liferation in GCPs (Fig. 5c; Supplementary Fig. 6). Although occurs early during cell-cycle exit (within 2 h), suggesting that it other BET proteins are expressed during GCP proliferation may be an initiating event in the differentiation of cerebellar granule (Supplementary Fig. 3), they do not compensate for Brd4 loss cells. Future studies are needed to determine how temporally con- (Supplementary Fig. 6). Indeed, GCPs from Tg (Atoh1-Cre +); trolling Brd4 activity is linked with phosphorylation, epigenetic, and Brd4fl/fl mice proliferated less than those from Tg (Atoh1-Cre−); ubiquitin pathways that induce cell-cycle exit in the developing Brd4fl/fl mice (Fig. 5d, Supplementary Fig. 6). Consistent with cerebellum25–28. Interestingly, Brd4 phosphorylation down- decreased GCP proliferation, cerebella from Tg (Atoh1-Cre +); regulation during cell-cycle exit correlates with decreases in Gli2 Brd4fl/fl mice were smaller than those from Tg (Atoh1-Cre-); levels in a cluster of cell-cycle genes that decreases after postnatal Brd4fl/fl mice (Fig. 6a, b) as well as had abherant cerebellar layer day 7 in vivo (Supplementary Fig. 2)29. formation that persisted throughout early postnatal development We have defined an essential role for Brd4 in the developing (Fig. 6b; Supplementary Figs. 7, 8). Importantly, reduced cerebellar cerebellum. Although other studies have showed that Brd4 is size correlated with behavioral deficits in Tg (Atoh1-Cre +);Brd4fl/fl involved in learning and memory later in development and other mice, which exhibited symptoms of cerebellar ataxia not evident developmental or cellular processes4,23,30–37, ours is the first to NATURE COMMUNICATIONS | (2019)10:3028 | https://doi.org/10.1038/s41467-019-10799-5 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10799-5 a b Gli1 In vitro (Cpm) Relative mRNA levels 30,000 1.0 3H-Thymidine 20,000 0.5 10,000 0 0.0 10 O 50 M nM μM SO 50 M nM μM S n n M M 1 0 0 1 0 0 D 10 D I-BET 151 I-BET 151 c Ex vivo d DMSO I-BET151 % EdU-positive cells 25 20 EdU 15 10 EGL 5 DAPI 0 DMSO I-BET 151 IGL e In vivo f JQ1 (mg/Kg) DMSO 10 25 50 30 % EdU positive cells EdU 20 EGL 10 DAPI IGL 0 DMSO 10 25 50 JQ1 (mg/Kg) Fig. 4 BET bromodomain protein inhibition reduces Gli1 mRNA levels in purified GCPs and GCP proliferation in vitro, ex vivo, and in vivo. a BET protein inhibition reduces GCP proliferation in vitro. GCPs were purified from P6 mice and incubated with the indicated concentrations of I-BET151 or DMSO control for 24 h. Proliferation was then measured using a 3H-thymidine incorporation assay. b BET protein inhibition reduces GCP Gli1 mRNA levels in vitro. Purified GCPs were incubated as described in a. qRT-PCR was performed and normalized to Gapdh. c Cerebellar organotypic slices were incubated in the presence of either DMSO or 1 μM I-BET151 for 24 h, after which time EdU was added for 2 h. Images were then acquired with a confocal microscope and quantified using ImageJ. d Quantification of c. e Increasing doses of JQ1 yielded a gradual decrease in EdU incorporation in vivo in P8 pups. JQ1 was given to pups for 1 day at 10, 25, or 50 mg/kg every 12 h. Then EdU was injected subcutaneously 1 h prior to perfusion. f Quantification of e. The results are shown as the average values of three independent experiments and are represented as the mean ± SEM. A one-way ANOVA followed by Bonferroni multiple comparison testing (a, b, f) or a paired t test (d) was performed (*p < 0.05, **p < 0.01, ***p < 0.001). Source data can be found in source data graphs under tabs for a, b, d, f demonstrate that Brd4 is required for cerebellar growth. Brd4 is tumors such as medulloblastoma are associated with deficits in required for GCP proliferation by controlling the SHH pathway procedural learning38–40, which could be exacerbated with pro- effectors, Gli1 and Gli2. Brd4 inhibition or deletion reduces GCP longed Brd4 inhibitor usage. Our findings suggest that ther- proliferation and responsiveness to SHH signaling. Brd4 deletion apeutic use of Brd4 inhibitors may need to be given during a decreases cerebellum size and induces symptoms of cerebellar temporal window, as Brd4 may be required for proper cerebellar ataxia either as a direct effect of neuron loss or indirectly through development in humans. the aberrant cerebellum morphology (Fig. 6). Brd4 deletion reduces levels of Gli2, an essential gene required for vertebrate Methods development32,33. Therefore, Brd4’s essential role in cerebellar Animal husbandry. All mice were housed in an American Association of granule cell development is likely to be related to Gli2 activity Laboratory Animal Care–accredited facility at the University of Miami and were maintained in accordance with NIH guidelines. Animal use was approved by the required for proper GCP expansion. Brd4, like Gli2, is a target in Institutional Animal Care and Use Committee of the University of Miami. developmental diseases and cancer, and Brd4 inhibitors may be useful in clinical settings in children10–13. However, cerebellum- GCP isolation and compound treatment. GCPs were purified from cerebellar associated developmental disorders and pediatric cerebellar cortex of P6-8 CD1 and Tg (Atoh1-Cre +);CK1δfl/fl mice by using Percoll gradient 6 NATURE COMMUNICATIONS | (2019)10:3028 | https://doi.org/10.1038/s41467-019-10799-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10799-5 ARTICLE a b N.S. FRT IacZ FRT 6 N.S. 4 IoxP neo IoxP 5 IoxP 2.0 FLP Atoh1-Cre–; Brd4fl/fl Relative mRNA levels FRT 1.5 4 IoxP 5 IoxP 6 Atoh1-Cre+; Brd4fl/fl CRE 1.0 FRT 4 IoxP 6 0.5 0.0 Brd4 Brd4 Gli1 Gli2 Gli3 exon4-5 exon5-6 c d N.S. 550 Tg (Atoh1-Cre–); Brd4fl/fl 500 Brd4fl/fl Tg (Atoh1-cre): – + 450 % EDU-positive cells MWM 400 Tg (Atoh1-Cre+); vs. Cre- No SHH 250 kDA Brd4fl/fl Brd4 350 250 kDA 300 Phospho-Brd4 S492/494 37 kDA 250 CyclinD1 200 37 kDA 150 Gapdh 100 50 0 –SHH +SHH Fig. 5 Conditional deletion of Brd4 in GCPs inhibits the sonic hedgehog pathway and SHH mediated proliferation. a Strategy utilized for deleting Brd4 conditionally in the cerebellum. Brd4tm1a(EUCOMM)Wtsi heterozygous mice were obtained through the Knockout Mouse Project Repository at Baylor University. These mice were bred to Tg (ACTFLPe) mice to create Brd4fl/+ mice. Brd4fl/+ mice were bred to Tg (Atoh1-cre) mice to obtain conditional cre expression in the granule cell lineage. Tg (Atoh1-cre) + /-;Brd4fl/ + mice were then backcrossed to generate Tg (Atoh1-cre + );Brd4fl/fl and Tg (Atoh1-cre-); Brd4fl/fl littermates. b GCPs from Tg (Atoh1-Cre + );Brd4fl/fl mice express less Brd4, Gli1 and Gli2 than Tg (Atoh1-Cre-);Brd4fl/fl mice. RNA was extracted from isolated GCPs from P8 mice, and qRT-PCR was performed and normalized to Gapdh. Primers flanking both flox sites were used to verify loss of Brd4 exon 5. c GCPs from Tg (Atoh1-Cre + );Brd4fl/fl mice have less total Brd4, phospho-Brd4 S492/494, and Cyclin D1 than Tg (Atoh1-Cre-);Brd4fl/fl mice. GCPs from P8 mice were isolated, and nuclear extracts or whole cell lysates were resolved by SDS-PAGE and western blot analysis. Gapdh was used as a loading control. d GCPs from Tg (Atoh1-Cre +);Brd4fl/fl mice proliferate less than Tg (Atoh1-Cre-);Brd4fl/fl GCPs. Following purification, GCPs were plated for 48 h with SHH, incubated with EdU and with or without SHH for 2 h, then plated with fresh media on poly-D-lysine/laminin coated coverslips for 3 h. Cells were fixed for EdU detection, reaggregates were imaged with a confocal laser-scanning microscope, and EdU positive cells were quantified with ImageJ. The results are the average of three independent experiments and are represented as the mean ± SEM. A one-way ANOVA followed by Tukey’s multiple comparison testing (5B) or a two-way ANOVA followed by Bonferonni’s multiple comparison testing (5D) was performed (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, N.S. no significance). Source data can be found in source data graphs under tabs for Fig. 5b, d and in source data Fig. 5 sedimentation to yield an enriched GCP fraction20. The cells were then pre-plated NanoDrop or Agilent Bioanalyzer and confirmed to have RIN numbers > 8.5. on a Petri dish to remove contaminating glia. Purified GCPs were resuspended in Sequencing was performed on Illumina HiSeq2000 with three samples per lane, medium (BME, 1.5% glucose, 20 mM glutamine, 10% horse serum, 5% fetal bovine generating an average of 95 M 2 × 100 bp reads per sample. After quality filtering serum, 1% penicillin/streptomycin) and then plated. For proliferation assays, and trimming with FastQC remaining RNA-seq reads were aligned to the Ensembl GCPs were cultured in suspension in the presence of mouse recombinant SHH mouse genome (v.87) using TopHat (v.2.1). For all samples, 80–90% of reads (0.25 ng/mL, 464-SH, R&D Systems); for cell-cycle exit and differentiation assay, aligned successfully. Differential expression analysis was performed by CuffDiff 2.2 GCPs were plated on poly-D-lysine (100 μg/ml, P6407, Sigma)-laminin (20 μg/ml, using the “classic-fpkm” parameter for the normalization method, and the “pooled” L2020, Sigma)–coated plates. For treatment with compounds, 100 nM of SR- 1277, parameter for the dispersion method. Sequencing results from this study have been the indicated amounts of I-BET151, or DMSO were added to the culture medium deposited in under accession number SRP146255. for 24 h. Cerebellar organotypic slice culture and treatment. Cerebella were isolated Functional annotation analysis. For each STEM profile, we performed functional from P8 CD1 mice. Sagittal slices (250 μm) of cerebellar cortex were generated annotation analysis using DAVID41 against the Level 5 Gene Ontology Biological using a Leica VT1000S vibratome, and slices were plated on Millipore culture Processes terms (GOTERM_BP_5). To adjust for the false discovery rate, we only inserts in six-well culture dishes containing 1.5 ml of serum-free medium (Basal considered terms with a Benjamini–Hochberg adjusted p-value of 0.05. Medium Eagle (Gibco), 0.45% D-( + )-glucose solution (Sigma), 1× ITS supple- ment (Sigma), 2 mM L-glutamine (Gibco), 100 U/ml penicillin/streptomycin (Gibco). The slices were then submerged in 2.5 ml of medium containing DMSO or I-BET151 (1 μM) for 24 h, after which 1 ml was removed so that the slices were no Flow cytometry. For flow-cytometric analysis, isolated GCPs were removed from longer submerged, and the medium was below the insert. dishes, washed with PBS and 1% BSA, and fixed with 10% ethanol in PBS overnight at 4 °C. GCPs were then stained with 69 μM propidium iodide in 38 μM sodium citrate buffer and 1 μM RNAse A at 37 °C for 30 min. The number of cells in G1, S, RNA sequencing. Extracted RNA was sent to the John P. Hussman Institute for or G2 phases was determined using a fluorescence-activated cell-sorting device Human Genomics for sequencing. RNA quality was tested by ThermoScientific (LSRII, Becton Dickinson) and analyzed by FlowJo software. NATURE COMMUNICATIONS | (2019)10:3028 | https://doi.org/10.1038/s41467-019-10799-5 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10799-5 a Tg (Atoh1-Cre–);Brd4fl/fl V Tg (Atoh1-Cre+);Brd4fl/fl VI VI IV VII V VII III IV II VIII III II VIII IX IX I X X b Lobe IX Tag1 Calbindin DRAQ5 Merge EGL ML PL IGL Tg (Atoh1-Cre–);Brd4fl/fl Lobe IX EGL ML PL IGL Tg (Atoh1-Cre+);Brd4fl/fl Lobe II EGL ML PL IGL Tg (Atoh1-Cre–);Brd4fl/fl Lobe II EGL PL IGL Tg (Atoh1-Cre+);Brd4fl/fl c 8 80 20 Ataxia rank score 6 60 Time to fall (s) 15 Speed (RPM) 4 40 10 2 20 5 0 0 0 Tg (Atoh1-Cre–); Tg (Atoh1-Cre+); Tg (Atoh1-Cre–); Tg (Atoh1-Cre+); Tg (Atoh1-Cre–); Tg (Atoh1-Cre+); Brd4fl/fl Brd4fl/fl Brd4fl/fl Brd4fl/fl Brd4fl/fl Brd4fl/fl Fig. 6 Conditional deletion of Brd4 in the developing cerebellum induces cerebellar deficits and ataxia. Tg (Atoh1-Cre + );Brd4fl/fl mice have disrupted cerebellar morphology relative to Tg (Atoh1-Cre-);Brd4fl/fl mice. Representative images are from P8 cerebellum mid-sagittal sections of approximately the same location in vermis. a Cresyl violet stain shows reduced cerebellum size in Tg (Atoh1-Cre + );Brd4fl/fl mice versus Tg (Atoh1-Cre);Brd4fl/fl mice. Cerebellar lobes are indicated I–X. Dotted line in Tg (Atoh1-Cre + );Brd4fl/fl divides the anterior and poster cerebellum to emphasize changes to proliferation and structure and proliferation, respectively, in each region. Scale = 500 µm. b Brd4 loss in GCPs disrupts Purkinje cell development. Confocal images are from the posterior lobe IX and anterior lobe II. Tag1 is a marker for migrating GCPs, Calbindin is a marker for Purkinje cells, and DRAQ5 is a DNA marker. Scale = 50 µM. EGL = external granule layer, ML = molecular layer, PL = Purkinje cell layer, IGL = internal granule layer. c Tg (Atoh1-Cre + );Brd4fl/fl mice exhibit symptoms of cerebellar ataxia. Adult (≥ P28) Tg (Atoh1-Cre + );Brd4fl/fl mice or Tg (Atoh1-Cre + );Brd4fl/fl mice were assessed for ataxia symptoms using multiple rank composite tests. Scores from each test were pooled with higher scores indicating symptom severity. Rotarod testing was performed, and latency to fall and speed at fall were recorded for each group after 5 days of training. The results are the average of three independent experiments, and are represented as the mean ± SEM. For ataxia, rank scores a Mann–Whitney U test and for rotarod testing an unpaired t test were performed (*p < 0.05, **p < 0.01). Source data can be found in source data graphs under tab for Fig. 6c 8 NATURE COMMUNICATIONS | (2019)10:3028 | https://doi.org/10.1038/s41467-019-10799-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10799-5 ARTICLE 3H-thymidine assay. GCPs were plated (3 × 105 cells/well) in 96-well dishes. Table 1 Accession numbers for genes used in quantitative DMSO or I-BET151 was added to the medium, and cells were maintained in culture for 24 h, in the presence of SHH. Then, 1 μCi [methyl-3H]-thymidine PCR (Amersham) was added to each well, and cells were harvested 22 h later and analyzed using TopCount (Perkin–Elmer). Gene TaqMan® Gene Expression Master Mix (Applied Biosystems) accession number EdU incorporation assay. For the in vivo study, P8 CD1 mice were injected Cyclin B1 Mm01322149_mH intraperitoneally twice with 10, 25, or 50 mg/Kg of JQ1 during 24 h. Two hours Atoh1 Mm00476035_s1 after the last injection, 50 mg/Kg of EdU was administered subcutaneously over the top of the neck. One hour after EdU administration, pups were perfused with 4% Tuj1 Mm00727586_s1 paraformaldehyde (PFA). Brains were processed for immunohistochemistry, and Gap43 Mm00500404_m1 cerebella were cut in 20 μM sections with a cryostat. For the ex vivo proliferation Gapdh Mm99999915_g1 assays, 1 ml of 25 μM EdU (Invitrogen) was added on top of the slices after 24 h in Brd4, exon 4-5 Mm01348074_m1 culture. Thus, the final concentration was 20 μM EdU per 2.5 -ml medium. Then, Brd4, exon 5-6 Mm00480392_m1 slices were fixed for 2 h with 4% PFA. For in vitro proliferation assays, GCPs were Brd4, exon 7-8 Mm00480394_m1 treated with EdU (20 μM) for 2 h. The cells were then washed with PBS, plated in Brd2 Mm01271171_g1 poly-D-lysine/laminin-coated dishes for 2–3 h, and further fixed with 4% PFA/30% Brd3 Mm00469733_m1 sucrose. Then, sections, slices, or cells were permeabilized and stained using the Click- Gli1 Mm00494654_m1 iT® EdU Alexa Fluor® 594 Imaging Kit (Invitrogen). Samples were imaged using Gli2 Mm01293117_m1 the z-stack of a confocal laser-scanning microscope (Olympus, FV1000), and the Gli3 Mm00492337_m1 images were analyzed using Fiji software (ImageJ). Cyclin A1 Mm00432337_m1 Cyclin D1 Mm00432359_m1 Immunohistochemistry. Mice were perfused with 4% PFA, and cerebella were extracted. Cerebella were fixed in 4% PFA for 2 h, embedded in 30% sucrose in PBS, and cut into 20-μm sections with a cryostat (Leica). Sections were then permeabilized and blocked in 0.5% Triton X-100, 5% fetal bovine serum for 1 h at In the case of the radioactivity assay, 5 μCi [γ-32P]ATP (Perkin–Elmer, room temperature and incubated overnight at 4 °C with the following primary BLU002H250UC) was added to each reaction. The resolved SDS-PAGE gel was antibodies: rabbit anti-Brd4 (1/1000, Bethyl Laboratories Inc.), mouse or rabbit exposed, and radioactive signal was quantified using a Cyclone phosphor imaging anti-calbindin (1/1000, Swant300, CB38), and mouse anti-Tag1 (1/1). The slices system (Perkin–Elmer). were then washed with PBS, and incubated for 4 h at room temperature with the following secondary antibodies as appropriate: Alexa Fluor® 488 goat anti–rabbit IgG, Alexa Fluor® 594 goat anti-mouse IgG or Alexa Fluor® 594 goat anti-mouse Chromatin immunoprecipitation (ChIP). GCPs were collected, cross-linked with IgM (all 1/500, Invitrogen). Sections were then washed with PBS and incubated formaldehyde, and further lysed and sonicated. The chromatin was immunopre- with Hoechst stain (Invitrogen) or DRAQ5 (AbCam) before mounting using cipitated with antibodies for Brd4 (Bethyl Laboratories Inc., A301-985A50) and ProLong Gold Antifade mounting medium (Invitrogen). Confocal images were negative control antibody IgG (Abcam, ab37415). DNA–protein cross-links were acquired with a confocal laser-scanning microscope and were further analyzed with reversed, and DNA was purified to be used in the quantitative amplification of Gli1 Fiji software (ImageJ). locus with SYBR Green (−3250 ~−2501, forward primer, TGGCTCACAACCATC CTGTA, reverse primer, GAGATGCCCTTGCTTCTGTC; + 1251 ~+ 2000, for- ward primer, ACCCAGGAATCCAAGGTGTC, reverse primer, Protein extract preparation, antibodies, and western blot analysis. Cells were TCCTGAAAGCAGGCAGTAGC) (Table 2). homogenized, and extracts were prepared using lysis buffer (NER buffer from the NER-PER nuclear and cytoplasmic extraction reagents (Thermo Fisher Scientific), 1× protease inhibitor cocktail, 1 μM microcystin-LR). The soluble fraction was Generation of Brd4fl/fl mice. Brd4tm1a(EUCOMM)Wtsi (MGI ID: 4441798) hetero- recovered by centrifugation at 16,000 × g for 10 min at 4 °C. Protein concentration zygous mice were obtained through the Knockout Mouse Project Repository at was measured with the BCA Protein Assay Kit (Pierce Biotechnology), and 30 μg of Baylor University from Dr. John Seavitt. These mice were bred to B6.Cg-Tg protein from each sample was resolved by SDS-PAGE. The resolved bands were (ACTFLPe)9205Dym/J mice (The Jackson Laboratory Stock #005703) to remove transferred onto a nitrocellulose membrane and subjected to western blotting with the neomycin resistance gene and LacZ reporter gene (forward primer, the appropriate antibodies. CTTGGGTGGAGAGGCTATTC, reverse primer, AGGTGAGATGACAGGAGA The following primary antibodies were used: mouse anti-CK1δ (C-8) (1/1000, TC) to create Brd4tm1c heterozygous mice. In parallel, Brd4tm1c heterozygous mice Santa Cruz Biotechnology, sc-55553), goat anti-CK1δ (N-19) (1/500, sc-6475, Santa were bred to homozygosity (hereafter referred to as Brd4fl/fl) or to B6.Cg-Tg (Atoh1- Cruz Biotechnology), rabbit-anti C-terminal Brd4 (1/2000)19, rabbit anti-Brd4 (1/ cre)1Bfri/J mice (The Jackson Laboratory Stock #011104, hereafter referred to as 1000, A301-985A50, Bethyl Laboratories Inc.), rabbit anti-phospho-Brd4 492/494 Tg (Atoh1-cre)) to obtain conditional cre expression in the granule cell lineage. (1/2000, ABE1453, Millipore), mouse anti-Gapdh (1/5000, NB300-221, Novus Brd4fl/fl and Tg (Atoh1-cre + /−);Brd4fl/+ mice were crossed to generate Tg (Atoh1- Biolechne), goat anti-CK2α (1/500, sc-6479, Santa Cruz Biotechnology), mouse cre + /−);Brd4fl/fl or Tg (Atoh1-cre−/−);Brd4fl/fl littermates (Table 3). Tg (Atoh1- anti-Flag (1/5000, A8592, Sigma), and rabbit anti-cyclin D1 (1/1000, ab134175, cre + /−);Brd4fl/fl do not breed successfully, therefore, Tg (Atoh1-cre) + /-;Brd4fl/+ Abcam). The following secondary antibodies were used: anti-goat IgG–HRP (1/ were maintained for experiments. 1000, 7074, Cell Signaling), anti-mouse IgG–HRP (1/1000, NXA931, GE Genotype primers: Cre forward: AGAACCTGAAGATGTTCGCG; Cre reverse: Healthcare) and anti-rabbit IgG–HRP (1/1000, NA9340V, GE Healthcare). GGCTATACGTAACAGGGTGT; Brd4 forward 1: TTTGACCTCTGCTCGTGTA GTG; Brd4 forward 2: ACCGCGTCGAGAAGTTCCTATT; Brd4 reverse: CATTG TACCCAGGCTCCTTTCA. RNA isolation and qRT-PCR. Cells were lysed in 1 mL of TriZol Reagent (Invi- trogen), and the RNA was purified with the RNeasy Mini Kit (Qiagen). RNA was then reverse-transcribed with a High Capacity cDNA Reverse Transcripiton Kit Behavioral testing. For ataxia rank composite scoring, we followed the composite (Applied Biosystems). TaqMan probes were designed with the TaqMan Gene scoring method for ataxia outlined in Guyenet et al.42 with modifications. Briefly, Expression Assay tool (Applied Biosystems). The qRT-PCR was performed using a blinded researchers scored Tg (Atoh1-cre + /-);Brd4fl/fl or Tg (Atoh1-cre−/−); TaqMan® Gene Expression Master Mix (Applied Biosystems) in a CFX384 Brd4fl/fl mice on the ledge test, on the hindlimb clasping test, and on the gait test. TouchTM Real-Time PCR Detection System (Bio-Rad). Fold change in gene These three composite tests are used most frequently to establish ataxia severity. expression was estimated using the computed tomography comparative method Composite scores were averaged and compared in Tg (Atoh1-cre + /−);Brd4fl/fl and normalizing to the Gapdh computed tomography values and relative to control or Tg (Atoh1-cre−/−);Brd4fl/fl mice. We used a rotarod paradigm to examine samples (Table 1). mouse balance, coordination and muscle function simultaneously. Mice were trained with rotarod twice daily for 5 days in the accelerating mode (5–40 rotations per minute over 2.5 min), and the latency to fall and speed were recorded after the In vitro phosphorylation of Brd4 with CK1δ. Purified bacterial Brd4 training period. Average latency times and speed for the testing interval were phosphorylation-dependent interaction domain (PDID) proteins were used for this compared in Tg (Atoh1-cre + /−);Brd4fl/fl or Tg (Atoh1-cre−/−);Brd4fl/fl mice. study (PDID wild-type, PDID S484/488 A, PDID S492/494 A). In each reaction, 100 ng of PDID protein was incubated with 200 μM of ATP, 1000U of CSNK1D (NEB, P6030S), and 1X NEBbuffer for protein kinases for 30 min at 30 °C. Laemmli Differentially expressed gene analysis. The gene-collapsed Affymetrix gene sample buffer was added to terminate the reactions, and the samples were heated to expression data were downloaded from GEO (GSE74400) and Differentially 95°C and resolved by SDS-PAGE. The resolved bands were transferred onto a Expressed Genes (DEGs) compared with P0 were identified using limma43. The nitrocellulose membrane and subjected to western blotting with the appropriate expression levels of subset of cell-cycle and neuronal-related genes belonging to the antibodies. STEM-generated Profile 47 were then plotted (Supplementary Fig. 2) showing a NATURE COMMUNICATIONS | (2019)10:3028 | https://doi.org/10.1038/s41467-019-10799-5 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10799-5 Table 2 Primers for amplification of Gli1 locus in the Chromatin immunoprecipitation assay Locus amplified Forward primer Reverse primer −3250 ~ −2501 TGGCTCACAACCATCCTGTA GAGATGCCCTTGCTTCTGTC + 1251 ~+ 2000 ACCCAGGAATCCAAGGTGTC TCCTGAAAGCAGGCAGTAGC Table 3 Primers for genotyping transgenic mice for Cre and flox site Type Forward primer Reverse primer To remove the neomycin resistance gene and LacZ reporter CTTGGGTGGAGAGGCTATTC AGGTGAGATGACAGGAGATC To create Brd4tm1c heterozygous mice CTTGGGTGGAGAGGCTATTC AGGTGAGATGACAGGAGATC To genotype Cre mice AGAACCTGAAGATGTTCGCG GGCTATACGTAACAGGGTGT To genotype Brd4 mice Forward 1: TTTGACCTCTGCTCGTGTAGTG CATTGTACCCAGGCTCCTTTCA Forward 2: ACCGCGTCGAGAAGTTCCTATT similar expression pattern to that from Supplementary Fig. 1. Furthermore, to 2. Hatten, M. E. & Roussel, M. F. Development and cancer of the cerebellum. quantify the overlap between the DEGs of the two data sets, we calculated the Trends Neurosci. 34, 134–142 (2011). pairwise percentage of overlap among all time points, as shown in Supplementary 3. Kouzarides, T. Chromatin modifications and their function. Cell 128, 693–705 Fig. 2. The percentage of overlap is defined as the number of DEGs that were (2007). common between the two data sets divided by the total number of DEGs in the 4. Murray, K. D., Rubin, C. M., Jones, E. G. & Chalupa, L. M. Molecular corresponding RNA-seq Timeseries timepoint. correlates of laminar differences in the macaque dorsal lateral geniculate nucleus. J. Neurosci. 28, 12010–12022 (2008). Mouse embryonic fibroblast (MEF) cultures. MEF Sufu−/− cells11 were main- 5. Northcott, P. A. et al. The whole-genome landscape of medulloblastoma tained in DMEM media with 10% neonatal calf serum and 1% penicillin/strepto- subtypes. Nature 547, 311–317 (2017). mycin. Overall, 1 × 105 cells per well were plated in a 12-well plate. In all, 100 nM 6. Bandopadhayay, P. et al. BET bromodomain inhibition of MYC-amplified of SR-653234 or SR-1277 CK1δ inhibitors were added into the media. DMSO was medulloblastoma. Clin. Cancer Res 20, 912–925 (2014). used as a vehicle control, 100 nM of GDC0449 as negative control, and 10 μM 7. Fielitz, K. et al. Characterization of pancreatic glucagon-producing tumors and GANT-61 as positive control. Twenty-four hours later, RNA was extracted and pituitary gland tumors in transgenic mice overexpressing MYCN in hGFAP- SHH target gene expression was examined by Taqman probe-based qRT-PCR. positive cells. Oncotarget 7, 74415–74426 (2016). Brd4 ChIP analysis was performed as described as above. 8. Han, Y. et al. Synergistic activity of BET inhibitor MK-8628 and PLK inhibitor Volasertib in preclinical models of medulloblastoma. Cancer Lett. 445, 24–33 siRNA transfections in GCP cultures. GCPs were isolated from P6 CD1 pups, (2019). and 7.5 × 105 cells per well were plated in suspension in a 48-well plate. Cells were 9. Henssen, A. et al. BET bromodomain protein inhibition is a therapeutic transfected with 1 µM SMARTpool Accell Mouse Brd4 siRNA (57261, E-041493- option for medulloblastoma. Oncotarget 4, 2080–2095 (2013). 00, Dharmacon GE Helathcare) or 1 µM SMARTpool Accell Mouse GFP Scramble 10. Lin, C. Y. et al. Active medulloblastoma enhancers reveal subgroup-specific siRNA (D-001950-01-05, Dharmacon GE Helathcare) in low-serum media (Accell cellular origins. Nature 530, 57–62 (2016). siRNA delivery media (Dharmacon GE Helathcare), 1.5% glucose, 20 mM gluta- 11. Long, J. et al. The BET bromodomain inhibitor I-BET151 acts downstream of mine, 2% horse serum, 1% fetal bovine serum, 1% penicillin/streptomycin) with smoothened protein to abrogate the growth of hedgehog protein-driven mouse recombinant SHH (0.25 ng/mL). Cells were collected 72 h post transfection, cancers. J. Biol. Chem. 289, 35494–35502 (2014). and RNA isolation and qRT-PCR were performed as described. 12. Tang, Y. et al. Epigenetic targeting of Hedgehog pathway transcriptional output through BET bromodomain inhibition. Nat. Med. 20, 732–740 (2014). 13. Venkataraman, S. et al. Inhibition of BRD4 attenuates tumor cell self-renewal Statistical analysis. All experiments were conducted independently at least three and suppresses stem cell signaling in MYC driven medulloblastoma. times. Statistical analysis was performed with Prism software (Graphpad). Fig- Oncotarget 5, 2355–2371 (2014). ures 1b, c, 2c, e, 4a, b, f: one-way ANOVA followed by Bonferroni multiple 14. A Phase 2 Study of CPI-0610 With and Without Ruxolitinib in Patients With comparison testing (p < 0.05). Figures 3b, e, 4d: paired t test (p < 0.05). Figure 5b: Myelofibrosis - Full Text View - ClinicalTrials.gov, https://clinicaltrials.gov/ct2/ one-way ANOVA followed by Tukey’s multiple comparison testing (p < 0.05). Figure 5d: two-way ANOVA followed by Bonferonni’s multiple comparison testing show/NCT02158858. (p < 0.05). Figure 6c: for ataxia rank scores Mann–Whitney test (p < 0.05) and for 15. Blum, K. A. et al. 41OA phase I study of CPI-0610, a bromodomain and extra rotarod testing unpaired t test (p < 0.05). terminal protein (BET) inhibitor in patients with relapsed or refractory lymphoma. Ann. Oncol. 29, mdy048 (2018). 16. Ernst, J., Nau, G. J. & Bar-Joseph, Z. Clustering short time series gene Reporting summary. Further information on research design is available in expression data. Bioinformatics 21(Suppl 1), i159–i168 (2005). the Nature Research Reporting Summary linked to this article. 17. Pak, E. & Segal, R. A. Hedgehog signal transduction: key players, oncogenic drivers, and cancer therapy. Dev. Cell 38, 333–344 (2016). Data availability 18. Belkina, A. C. & Denis, G. V. BET domain co-regulators in obesity, The authors declare that all transcriptional data supporting the findings of this study are inflammation and cancer. Nat. Rev. Cancer 12, 465–477 (2012). freely available from the databases provided within this paper. Sequencing results from 19. Wu, S. Y., Lee, A. Y., Lai, H. T., Zhang, H. & Chiang, C. M. Phospho switch this study have been deposited in under accession number SRP146255. All other data triggers Brd4 chromatin binding and activator recruitment for gene-specific supporting the findings of this study are available from the corresponding author, Dr. targeting. Mol. Cell 49, 843–857 (2013). Nagi G. Ayad, upon request. 20. Penas, C. et al. Casein kinase 1delta is an APC/C(Cdh1) substrate that regulates cerebellar granule cell neurogenesis. Cell Rep. 11, 249–260 (2015). Received: 14 September 2018 Accepted: 9 May 2019 21. Penas, C. et al. Casein kinase 1delta-dependent Wee1 protein degradation. J. Biol. Chem. 289, 18893–18903 (2014). 22. Dawson, M. A. et al. Inhibition of BET recruitment to chromatin as an effective treatment for MLL-fusion leukaemia. Nature 478, 529–533 (2011). 23. Rudman, M. D. et al. Bromodomain and extraterminal domain-containing protein inhibition attenuates acute inflammation after spinal cord injury. Exp. References Neurol. 309, 181–192 (2018). 1. Espinosa, J. S. & Luo, L. Timing neurogenesis and differentiation: insights 24. Yang, H., Xie, X., Deng, M., Chen, X. & Gan, L. Generation and from quantitative clonal analyses of cerebellar granule cells. J. Neurosci. 28, characterization of Atoh1-Cre knock-in mouse line. Genesis 48, 407–413 2301 (2008). (2010). 10 NATURE COMMUNICATIONS | (2019)10:3028 | https://doi.org/10.1038/s41467-019-10799-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-10799-5 ARTICLE 25. Roussel, M. F. & Hatten, M. E. Cerebellum development and Acknowledgements medulloblastoma. Curr. Top. Dev. Biol. 94, 235–282 (2011). We acknowledge funding from the NIH via grants NS067289, R56102590 to N.G.A. We 26. Lasorella, A. et al. Degradation of Id2 by the anaphase-promoting complex also acknowledge support from the Epigenetics Program of the Sylvester Comprehensive couples cell cycle exit and axonal growth. Nature 442, 471–474 (2006). Cancer Center at the University of Miami. NIH grant CA103867, CPRIT grants 27. Puram, S. V. & Bonni, A. Novel functions for the anaphase-promoting RP190077 and RP180349, and Welch Foundation grant I-1805 to C-M.C. We thank Dr. complex in neurobiology. Semin Cell Dev. Biol. 22, 586–594 (2011). David Rowich for providing the Atoh1-Cre mice. We thank the Moraes laboratory for 28. Huang, J. & Bonni, A. A decade of the anaphase-promoting complex in the providing assistance with the phosphorylation assays. We thank Dr. Bradnerʼs laboratory nervous system. Genes Dev. 30, 622–638 (2016). for providing the JQ1 compound used in these studies. We thank all members of the 29. Zhu, X. et al. Role of Tet1/3 genes and chromatin remodeling genes in Center for Therapeutic Innovation and the Lemmon-Bixby laboratories at the University cerebellar circuit formation. Neuron 89, 100–112 (2016). of Miami for helpful discussions. 30. Fernandez-Alonso, R. et al. Brd4-Brd2 isoform switching coordinates pluripotent exit and Smad2-dependent lineage specification. EMBO Rep. 18, 1108–1122 (2017). Author contributions C.P. designed research, performed experiments, developed methods, analyzed data, and 31. Hsieh, M. C. et al. Bromodomain-containing protein 4 activates voltage-gated wrote the paper. M.E.M. designed research, performed experiments, developed methods, sodium channel 1.7 transcription in dorsal root ganglia neurons to mediate analyzed data, and wrote the paper. V.S. designed research, performed experiments, thermal hyperalgesia in rats. Anesthesiology 127, 862–877 (2017). developed methods, analyzed data, and wrote the paper. Y.F., J.L., J.M., J.R-B. and S.K.T. 32. Korb, E., Herre, M., Zucker-Scharff, I., Darnell, R. B. & Allis, C. D. BET performed experiments. C.V. analyzed data. C.-M.C. provided reagents. J.K.L, D.L., protein Brd4 activates transcription in neurons and BET inhibitor Jq1 blocks D.J.R. and M.E.H. analyzed data and wrote the paper. J.C. designed research, analyzed memory in mice. Nat. Neurosci. 18, 1464–1473 (2015). data, and wrote the paper. N.G.A. designed research, analyzed data, and wrote the paper. 33. Korb, E. et al. Excess translation of epigenetic regulators contributes to fragile X syndrome and is alleviated by Brd4 inhibition. Cell 170, 1209–1223 e1220 (2017). Additional information 34. Li, J. et al. BET bromodomain inhibition promotes neurogenesis while inhibiting Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- gliogenesis in neural progenitor cells. Stem Cell Res. 17, 212–221 (2016). 019-10799-5. 35. Takahashi, K. et al. Spinal bromodomain-containing protein 4 contributes to neuropathic pain induced by HIV glycoprotein 120 with morphine in rats. Competing interests: The authors declare no competing interests. Neuroreport 29, 441–446 (2018). 36. Houzelstein, D. et al. Growth and early postimplantation defects in mice Reprints and permission information is available online at http://npg.nature.com/ deficient for the bromodomain-containing protein Brd4. Mol. Cell Biol. 22, reprintsandpermissions/ 3794–3802 (2002). 37. Lee, J. E. et al. Brd4 binds to active enhancers to control cell identity gene Peer review information: Nature Communications thanks the anonymous reviewer(s) induction in adipogenesis and myogenesis. Nat. Commun. 8, 2217 (2017). for their contribution to the peer review of this work. Peer reviewer reports are available 38. Mostofsky, Stewart H. Goldberg, Melissa C. Landa, Rebecca J. Denckla, Martha B. Evidence for a deficit in procedural learning in children and Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in adolescents with autism: Implications for cerebellar contribution, 752–759 published maps and institutional affiliations. https://www.cambridge.org/core/article/evidence-for-a-deficit-in-procedural- learning-in-children-and-adolescents-with-autism-implications-for- cerebellar-contribution/21353573F61E19D225F9E525BA721C1B Open Access This article is licensed under a Creative Commons 39. Quintero-Gallego, E. A., Gómez, C. M., Casares, E., Márquez, J. & Pérez- Attribution 4.0 International License, which permits use, sharing, Santamaría, F. Declarative and procedural learning in children and adaptation, distribution and reproduction in any medium or format, as long as you give adolescents with posterior fossa tumours. Behav. Brain Funct. 2, 9–9 (2006). appropriate credit to the original author(s) and the source, provide a link to the Creative 40. Baillieux, H., Smet, H. J. D., Paquier, P. F., De Deyn, P. P. & Mariën, P. Cerebellar neurocognition: Insights into the bottom of the brain. Clin. Neurol. Commons license, and indicate if changes were made. The images or other third party Neurosurg. 110, 763–773 (2008). material in this article are included in the article’s Creative Commons license, unless 41. Huang da, W., Sherman, B. T. & Lempicki, R. A. Systematic and integrative indicated otherwise in a credit line to the material. If material is not included in the analysis of large gene lists using DAVID bioinformatics resources. Nat. Protoc. article’s Creative Commons license and your intended use is not permitted by statutory 4, 44–57 (2009). regulation or exceeds the permitted use, you will need to obtain permission directly from 42. Guyenet, S. J. et al. A simple composite phenotype scoring system for the copyright holder. To view a copy of this license, visit http://creativecommons.org/ evaluating mouse models of cerebellar ataxia. J. Vis. Exp. https://doi.org/ licenses/by/4.0/. 10.3791/1787 (2010). 43. Ritchie, M. E. et al. limma powers differential expression analyses for RNA- sequencing and microarray studies. Nucleic Acids Res. 43, e47–e47 (2015). © The Author(s) 2019 NATURE COMMUNICATIONS | (2019)10:3028 | https://doi.org/10.1038/s41467-019-10799-5 | www.nature.com/naturecommunications 11