SETBP1 induces transcription of a network of development genes by acting as an epigenetic hub

SETBP1 induces transcription of a network of development genes by acting as an epigenetic hub ARTICLE DOI: 10.1038/s41467-018-04462-8 OPEN SETBP1 induces transcription of a network of development genes by acting as an epigenetic hub 1 1 1 1 1 2 Rocco Piazza , Vera Magistroni , Sara Redaelli , Mario Mauri , Luca Massimino , Alessandro Sessa , 1 3 3 1 1 Marco Peronaci , Maciej Lalowski , Rabah Soliymani , Caterina Mezzatesta , Alessandra Pirola , 2 2 4 5 6 7 Federica Banfi , Alicia Rubio , Delphine Rea , Fabio Stagno , Emilio Usala , Bruno Martino , 8 9 10 11 Leonardo Campiotti , Michele Merli , Francesco Passamonti , Francesco Onida , 12 13 14 2,15 Alessandro Morotti , Francesca Pavesi , Marco Bregni , Vania Broccoli 3 1 Marc Baumann & Carlo Gambacorti-Passerini SETBP1 variants occur as somatic mutations in several hematological malignancies such as atypical chronic myeloid leukemia and as de novo germline mutations in the Schinzel–Giedion syndrome. Here we show that SETBP1 binds to gDNA in AT-rich promoter regions, causing activation of gene expression through recruitment of a HCF1/KMT2A/PHF8 epigenetic complex. Deletion of two AT-hooks abrogates the binding of SETBP1 to gDNA and impairs target gene upregulation. Genes controlled by SETBP1 such as MECOM are significantly upregulated in leukemias containing SETBP1 mutations. Gene ontology analysis of deregu- lated SETBP1 target genes indicates that they are also key controllers of visceral organ development and brain morphogenesis. In line with these findings, in utero brain electro- poration of mutated SETBP1 causes impairment of mouse neurogenesis with a profound delay in neuronal migration. In summary, this work unveils a SETBP1 function that directly affects gene transcription and clarifies the mechanism operating in myeloid malignancies and in the Schinzel–Giedion syndrome caused by SETBP1 mutations. 1 2 Department of Medicine and Surgery, University of Milano-Bicocca and San Gerardo hospital, 20900 Monza, Italy. Stem Cell and Neurogenesis Unit, Division of Neuroscience, San Raffaele Scientific Institute, 20132 Milan, Italy. Department of Biochemistry and Developmental Biology, Faculty of Medicine, Meilahti Clinical Proteomics Core Facility, University of Helsinki, 00290 Helsinki, Finland. Service d’Hématologie Adulte, Hôpital Saint-Louis, 75010 Paris, 5 6 France. Chair and Hematology Section, Ferrarotto Hospital, AOU Policlinico, 95123 Catania, Italy. Azienda Brotzu U.O. Ematologia e CTMO, Ospedale 7 8 Businco, 09121 Cagliari, Italy. UO Ematologia Azienda Ospedaliera “BIANCHI MELACRINO MORELLI”, 89124 Reggio Calabria, Italy. Dipartimento Medicina Clinica e Sperimentale, Università Insubria, 21100 Varese, Italy. Division of Hematology, University Hospital Ospedale di Circolo e Fondazione 10 11 Macchi, 21100 Varese, Italy. Hematology, Dipartimento di Medicina Clinica e Sperimentale, University of Varese, 21100 Varese, Italy. BMT Center - Oncohematology Unit, Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, University of Milan, 20122 Milan, Italy. Department of Clinical and Biological Sciences, University of Torino, 10043 Orbassano (Torino), Italy. Hematology and Bone Marrow Transplantation Unit, IRCCS San Raffaele 14 15 Scientific Institute, 20132 Milan, Italy. Oncology Unit, ASST Valle Olona, Ospedale di Circolo di Busto Arsizio, 21052 Busto Arsizio, Italy. CNR Institute of Neuroscience, 20129 Milan, Italy. These authors contributed equally: Rocco Piazza, Vera Magistroni, Sara Redaelli. Correspondence and requests for materials should be addressed to R.P. (email: rocco.piazza@unimib.it) NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 1 | | | 1234567890():,; ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ecently, we and others demonstrated the involvement of rich regions (Fig. 1a; mean peaks A/T content: 65.8% vs. 59% for 1–9 SETBP1 mutations in several hematological malignancies . the whole human genome; p < 0.0001) in cells expressing both RIn our work, we showed that SETBP1 somatic variants occur SETBP1-G870S and SETBP1-WT. De novo motif discovery in a 4 amino acid mutational hotspot within the so called SKI- identified a SETBP1 consensus binding site (Fig. 1b; AAAATAA/ homology domain. This mutational hotspot is part of a degron T; p = 0.002) largely overlapping the AT-hook consensus motif of motif that specifies substrate recognition by the cognate SCF-β- HMGA1 (AAAATA; http://hocomoco.autosome.ru/motif/ TrCP E3 ubiquitin ligase. SETBP1 mutations cause a functional loss HMGA1_HUMAN.H10MO.D), suggesting that SETBP1 binds of the degron motif targeted by SCF-β-TrCP and responsible for the gDNA through its AT-hook domains. Querying the Catalog of short half-life of the protein. Therefore, these mutations result in an Inferred Sequence Binding Preferences (http://cisbp.ccbr. increased half-life of the mutated SETBP1 protein causing its utoronto.ca/index.php) for murine Setbp1 resulted in a very accumulation and inhibition of the PP2A phosphatase oncosup- similar sequence, with an AAT trinucleotide as the core motif pressor through the SETBP1–SET–PP2A axis . (http://cisbp.ccbr.utoronto.ca/TFreport.php?searchTF = SETBP1 mutations occurring in the same hotspot were pre- T008692_1.02), therefore suggesting that the DNA-binding motif viously found in a germline disease known as the for SETBP1/Setbp1 is conserved across evolution. Independent Schinzel–Giedion syndrome (SGS) . Despite the overlap between ligand-fishing experiments confirmed the ability of SETBP1 to the mutations present in hematological disorders and in SGS, bind to gDNA (Fig. 1c). recent data suggest that somatic SETBP1 mutations found in Peak distribution analysis revealed enrichment around pro- leukemias are more disruptive to the degron than germline var- moters (Fig. 1d; 58% and 65% of the total peak count reside at iants responsible for the onset of SGS . In SGS, germline +/−50 Kb from each transcription start site (TSS) for WT and SETBP1 mutations occur as de novo variants, causing a severe G870S, respectively). However, the binding of SETBP1 was not phenotype characterized by mental retardation associated with restricted to promoters but also present in enhancers, exonic, distorted neuronal layering , multi-organ development intronic, and intergenic regions (Fig. 1e). In line with the abnormalities, and higher than normal risk of tumors . quantitative effect of SETBP1 mutations on SETBP1 protein Although the first studies led to a reliable characterization of stability, SETBP1-G870S and SETBP1-WT ChIP-Seq experiments SETBP1 as an oncogene, several elements of SETBP1 activity revealed a global increase of SETBP1-G870S binding across all the were still unclear, in particular: (1) inhibition of PP2A phos- regions tested (Fig. 1e; Jaccard p = 0.046). phatase alone does not explain the SETBP1-dependent phenotype Peak annotation identified ectopic binding of SETBP1-G870S 10 10 of SGS , and (2) SETBP1 possesses three conserved AT-hooks , to 277 genes (Supplementary Figure 2a-c; Supplementary Data 2) therefore suggesting a role as a DNA-binding protein. resulting in a strong functional enrichment for development- Preliminary evidence that murine Setbp1 is able to bind to related biological processes such as nervous system, heart, and genomic DNA (gDNA) was initially given by Oakley and col- bone development (Fig. 1f). Notably, over one third of the leagues : by transducing murine bone marrow progenitors with binding regions lies within either evolutionary conserved regions, high titer retrovirus expressing Setbp1 followed by chromatin defined as 100 bp gDNA windows characterized by a human- immunoprecipitation (ChIP) experiments, the authors demon- mouse conservation ≥70% , or DNase I hypersensitive clusters, strated binding of Setbp1 to Hoxa9/10 promoters and upregula- thus suggesting that a dysregulation in their activation could lead tion of the two genes. Using a similar murine model, to significant functional consequences (Fig. 1g). Vishwakarma and colleagues showed binding of Setbp1 to the Runx1 promoter; this, however, was associated with down- modulation of RUNX1 expression. SETBP1 upregulates target genes at the transcriptional level. In this study, we analyze the interaction between SETBP1 and To dissect the effect of SETBP1 at the transcriptional level, we gDNA on a global, unbiased scale and demonstrate that SETBP1 generated RNA-Seq profiles for cells overexpressing SETBP1- binds DNA in adenine-thymine (AT)-rich promoter regions, G870S, SETBP1-WT, as well as an Empty control. Comparative causing activation of gene expression through the recruitment of analysis of cells overexpressing G870S vs. WT revealed a very a SET1/KMT2A (MLL1) COMPASS-like complex. similar profile (Fig. 2a, b). This finding supports the expected quantitative mechanism of action of SETBP1 mutations, as pre- 1, 3, 11 viously proposed . Results This model is further supported by the evidence that virtually SETBP1 as a DNA-binding protein. Cristobal et al. showed a all the activating SETBP1 variants identified so far fall exactly clear role for SETBP1 as a natural inhibitor of PP2A . However, within the PEST domain and >95% within 4 amino acids of the SETBP1 possesses three conserved AT-hook domains , respon- extremely short degron linear motif and that the disruption of the sible for binding to the minor groove of AT-rich gDNA regions; PEST domain causes subsequent impairment of the SETBP1−β- 1, 3, 11 thus their presence suggests a role for SETBP1 as a DNA-binding TrCP axis . To confirm this hypothesis in the context of protein. transcriptional regulation, we analyzed the intersection between To test this hypothesis, we generated isogenic 293 FLP-In cell SETBP1 peaks generated in our ChIP experiments and ENCODE lines harboring wild-type (WT) and mutated (G870S) SETBP1 in transcription factor binding sites (TFBSs). In the presence of fusion with the V5 tag. Other isogenic models, such as CRISPR- mutated SETBP1, we detected an increase in the total number of Cas9, could not be used given the absence of shared TFBS; however, the putative binding partners did not immunoprecipitation-grade anti-SETBP1 antibodies. change between SETBP1-WT and mutated lines (Supplementary The G870S variant was chosen as it is the most frequent Figure 3). These data highlight the presence of a quantitative mutation found in myelodysplastic/myeloproliferative disorders, rather than a qualitative effect for SETBP1 mutations, thus together with the D868N . The G870S line showed similar supporting the reduced degradation model. SETBP1 transcript levels but increased SETBP1 protein compared Subsequent comparative RNA-Seq analysis of SETBP1-G870S to the WT one (Supplementary Figure 1a, b), as expected . Anti- vs. Empty showed the presence of 2687 differentially expressed V5 ChIP-Seq experiments revealed a total of 3065 broad genomic genes (DEGs), 57% of which were downregulated and 43% regions as being bound by SETBP1-G870S (Supplementary upregulated (Fig. 2a, b; Supplementary Data 3). The intersection Data 1). The presence of broad peaks occurred mainly in AT- between genes bound by SETBP1 in promoter regions and DEGs 2 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE ab c d 6e-07 WT A/T content 68% *** V5-SETBP1-G870S – – + + + Empty + + – – – 3e-07 64% Oligonucleotide probe T U T U Beads WT enriched motif G870S enriched motif only 2 2 0e+00 V5 60% 1 1 V5-INPUT G870S 0 0 56% Lamin B1-INPUT 2.0e-07 52% 5.0e-08 –50 –30 0 30 50 Distance from TSS ef g 10,000 40% 100% SETBP1-G870S: GO biological processes Skeletal system development Mesenchyme development 30% 75% Stem cell development Palate development Cartilage morphogenesis 20% 50% Nervous system development Cardiac ventr. morphogenesis Organ morphogenesis 10% 25% Neurogenesis BMP signalling pathway 0 20 40 60 80 100 120 140 160 180 0% 0% –Log10 binom. Bonferroni prob. WT G870S WT G870S CGI No CGI Fig. 1 Interaction of SETBP1 with genomic DNA. a A/T content comparison between SETBP1 WT/G870S-binding regions and the reference genome. The frequency of A/T in SETBP1 target regions was compared with A/T frequency in the entire genome. Pearson's chi-squared test was used to test the significance of the difference between the two proportions. Actual numbers can be found in Supplementary Table 1. b SETBP1 consensus binding site revealed by de novo motif discovery. c Nuclear cell lysate oligonucleotide pulldown experiment. Target (T) and non-target (U) biotinylated probe oligonucleotides were designed according to ChIP-Seq data. Empty beads were used as control for non-specific binding. Pulldown was performed on nuclear extract from FLP-In SETBP1-G870S transfectants or Empty lines. Lamin B1 was used as a loading control. d Peak distribution density according to the distance from gene transcription start sites. e Peak quantitation in the different genomic regions, reflecting the position of binding sites relative to the next known gene. f Gene Ontology (GO) biological process functional enrichment of SETBP1 target genes. g Left, percentage of WT and G870S regions covered by CpG islands (CGIs), evolutionary conserved regions (ECRs), and DNase I hypersensitivity (DHS) at single-nucleotide resolution. Right, percentage of SETBP1 target genes having CpG islands within their promoter. ***p < 0.001 (false discovery rate (FDR) < 0.001) revealed 105 co-occurring being part of a nucleosome-remodeling complex involved in genes (Fig. 2c, Supplementary Data 4). Of them, 99 (94.3%; p < transcriptional activation (Fig. 3d; Supplementary Data 5). In line −6 1×10 ) were upregulated, suggesting a primary role for SETBP1 with the previous findings, correlation of SETBP1-binding as a positive inducer of gene expression (Fig. 2c). Relative regions as well as related epigenetic marks between SETBP1- quantification on a subset of target genes (SKIDA1, NFE2L2, G870S and SETBP1-WT was nearly perfect (r = 0.985 and p < PDE4D, FBXO8, CEP44, COBLL1, BMP5, ERBB4, CDKN1B)on 0.00001 for anti-V5; r = 0.873 and p < 0.00001 for anti-H3K9Ac; SETBP1-WT, SETBP1-G870S, and Empty cells by mean of Supplementary Figure 4). quantitative polymerase chain reaction (Q-PCR) confirmed the Co-immunoprecipitation (Co-IP)/proteomics experiments differential expression detected by RNA-Seq (Fig. 2d). directed against SETBP1 (Supplementary Data 6) revealed direct Functional annotation of DEGs and Gene Set Enrichment interaction of both WT and mutated SETBP1 with HCF1, a core Analyses (Fig. 2e, f) showed significant enrichment for ontologies protein of the SET1/KMT2A complex, responsible for H3K4 related to cell differentiation and tissue development, thus mono- and di-methylation . These results were confirmed by suggesting that SETBP1-mediated transcriptional deregulation independent immunoprecipitation/western blot experiments may play an important role in the onset of SGS . (Fig. 4a) and by acceptor photobleaching fluorescence resonance energy transfer (FRET) assays (Fig. 4b). In silico linear domain analysis revealed the presence of a putative HCF1-binding motif SETBP1 is part of a multiprotein epigenetic complex. AT- (HBM) occurring at position 991–994 of SETBP1 (Supplementary hook-containing proteins are often part of large chromatin 19–21 Figure 5). Deletion of this motif in WT and mutated cells remodeling complexes . ChIP-Seq profiles of a set of histone (SETBP1ΔHBM and SETBP1-G870S-ΔHBM; Supplementary marks associated with gene expression (H3K4me2, H3K4me3, Figure 6) caused a complete and specific abrogation of H3K9ac, H3K27ac, and H3K36me3) revealed peak distribution SETBP1/HCF1 interaction in both lines (Fig. 4b); conversely, enrichment around the promoter regions for all the tested marks the known SETBP1–β-TrCP interaction present only in the WT with the exception of H3K36me3, which was enriched in gene protein was not affected by the deletion (Fig. 4c). To assess bodies, as expected (Fig. 3a). Differential enrichment analysis in whether the loss of the SETBP1-HCF1 interaction could result in SETBP1-G870S vs. Empty showed a correlation between SETBP1 the impairment of the SETBP1 transcriptional machinery, we promoter occupancy and increase of H3K4me2 and H3K9ac −4 −16 analyzed by Q-PCR the expression levels of a set of upregulated (Fig. 3b, c; p = 1×10 and p = 1.2 × 10 , respectively). A genes: the deletion of HBM resulted in a complete normalization significant transcriptional upregulation was detected for the genes of the expression level for all the tested genes (Fig. 4d; MECOM p bound by SETBP1-G870S and displaying an increase in = 0.01; BMP5 p < 0.001, PDE4D p < 0.001, ERBB4 p = 0.036). H3K4me2 and H3K9ac, corroborating the hypothesis of SETBP1 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 3 | | | Genome WT G870S TSS region Enhancers Exons Introns Intergenic CGIs ECRs DHS WT G870S Number of peaks 58% 65% 66% Bits Bits Region coverage Genomic regions Genomic regions 2% 1% 34% 28% 31% 21% Number of genes with CGI 91% 9% 86% 14% ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ab c MA-plot: G870S/WT MA-plot: G870S/Empty RNA UP 165 genes 1157 genes RNA DOWN –5 –5 Empty 0 1524 117 genes 1530 genes WT G870S 1e-01 1e+01 1e+03 1e+05 1e-01 1e+01 1e+03 1e+05 PC1: 95% variance Mean expression Mean expression SETBP1 G870S Anatomical structure arrangement p <1E-5 –0.2 *** –0.4 *** –0.6 *** *** 'G870S' (positively correlated) 5.0 2.5 *** *** ** Zero cross at 12176 –2.5 *** 'Empty' (negatively correlated) *** –5.0 0 5000 10,000 15,000 20,000 25,000 30,000 Rank in ordered dataset Brain morphogenesis p <1E-5 –0.2 –0.6 EW M E WM E W M E WM E W M E WM E W M E W M E W M –0.8 SKIDA1 NFE2L2 PDE4D FBXO8 CEP44 COBLL1 BMP5 ERBB4 CDKN1B 'G870S' (positively correlated) 5.0 2.5 Zero cross at 12176 e –2.5 SETBP1-G870S: GO biological processes 'Empty' (negatively correlated) –5.0 Neg. regulation of leukocyte differentiation 0 5000 10,000 15,000 20,000 25,000 30,000 Neg. regulation of stem cell differentiation Rank in ordered dataset Granulocyte migration Myeloid leukocyte mediated immunity Enodochondral ossification Endodermal cell differentiation Mesodermal cell differentiation p <1E-5 Regulation of cardiac muscle cell differentiation Positive regulation of chondrocyte differentiation Cell morphogenesis involved in neuron differentiation –0.2 Positive regulation of osteoclast differentiation Endochondral ossification Regulation of granulocyte chemotaxis –0.4 Forebrain cell migration CNS neuron differentiation Negative regulation of osteoblast differentiation Cell differentiation involved in kidney development Cell proliferation in forebrain 'G870S' (positively correlated) 5.0 Dendrite morphogenesis CNS neuron axonogenesis 2.5 Brain morphogenesis Zero cross at 12176 Forebrain regionalization –2.5 Cranial nerve morphogenesis 'Empty' (negatively correlated) –5.0 CNS projection neuron axongenesis 0 5000 10,000 15,000 20,000 25,000 30,000 Anatomical structure arrangement Rank in ordered dataset 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Enrichment profile Hits Ranking metric scores –Log10 FDR Fig. 2 Effect of SETBP1 G870S expression at the transcriptional level. a MA plot showing the expression data of SETBP1-G870S vs. SETBP1-WT (left) and SETBP1-G870S vs. Empty cells (right) as a function of log ratios (M) and mean average gene counts. b Variance distribution of 3 Empty, 3 SETBP1-WT, and 3 SETBP1-G870S clones emphasized by Principal Component Analysis 1 (PC1) showing 95% of total variance. c Venn diagram showing the number of differentially expressed genes being directly bound by SETBP1 within their promoter region (red circle). d Q-PCR analysis of a subset of SETBP1 DEGs identified by RNA-Seq: E (Empty), W (SETBP1_wt), M (SETBP1_G870S). The housekeeping gene GUSB was used as an internal reference. Experiments were performed in triplicate; statistical analysis was performed using t-test. Error bars represent the standard error. **p < 0.01; ***p < 0.001. e Dysregulated GO biological process revealed by functional enrichment analysis of the differentially expressed genes resulting from G870S mutation. f Gene set enrichment analysis displaying three of the most enriched categories. Genes are shown as a function of the enrichment score (y axis in the upper part) and relative gene expression (x axis) Finally, a direct interaction between SETBP1 and KMT2A was germline mutations in two PHD family members, PHF8 and confirmed by Co-IP experiments (Fig. 4e). PHF6, have been identified as the cause of two X-linked mental Plant homeodomain (PHD) fingers interact with methylated retardation syndromes, namely Siderius X-linked mental retarda- 25 26 histone tails and are typically found in proteins responsible for tion and Borjeson–Forssman–Lehmann syndrome , and the epigenetic modulation of gene expression . Interestingly, somatic PHF6 mutations have been found in T cell acute 4 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | Relative mRNA expression level Log2 fold change PC2: 2% variance Ranked list Enrichment Ranked list Enrichment Ranked list Enrichment metric score metric score metric score NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE a d Relative 12 H3K4me2 12 H3K4me3 12 H3K36me3 Row min Row max 10 10 10 8 8 8 6 6 6 4 4 4 2 2 2 Approved symbol 0 0 0 SENP6 –15 Kb –9 Kb –3 Kb 3 Kb 9 Kb 15 Kb –15 Kb –9 Kb –3 Kb 3 Kb 9 Kb 15 Kb –15 Kb –9 Kb –3 Kb 3 Kb 9 Kb 15 Kb ARRDC3 NCOA2 CDKN2A 12 12 H3K9Ac H3K27Ac BMP5 10 10 8 8 KLHL1 6 6 MGAT2 4 4 CDKN1B 2 2 CDKN2C 0 0 COMMD3 –15 Kb –9 Kb –3 Kb 3 Kb 9 Kb 15 Kb –15 Kb –9 Kb –3 Kb 3 Kb 9 Kb 15 Kb LAMA2 BMI1 BMPR1B DNAJC1 30 6 *** NS NS TFPI 25 5 PCDH7 RNF217 20 4 COBLL1 TRDMT1 MEF2C 6 EDIL3 10 2 SYDE2 5 1 PDE4D SLC38A11 0 0 0 NEK7 ANK3 01 2 34 0 1 2 3 4 012 345 RBMS3 SETBP1 G870S/Empty (Log2 FC) SETBP1 G870S/Empty (Log2 FC) SETBP1 G870S/Empty (Log2 FC) NFIB 50 5 TBX18 NS *** PRKACB RNF145 40 4 DST ALX1 30 3 HGF SPAG6 20 2 FBXO8 ZSWIM2 10 1 LRRIQ1 MECOM 0 0 TRDN PTEN 012 345 0 1 2 345 NR2F1 SETBP1 G870S/Empty (Log2 FC) SETBP1 G870S/Empty (Log2 FC) PTPRC FAM172A c BAZ2B Empty RFESD SETBP1 WT ATP6V1G3 SETBP1 G870S NFE2L2 SETBP1 WT peaks RHOBTB3 SETBP1 G870S peaks FAM188A BMP5 Refseq genes G3BP1 H3K4me2 ZFYVE16 H3K4me2 G870S METAP1D H3K4me3 VMP1 H3K4me3 G870S RNF43 H3K36me3 PPP3CC H3K36me3 G870S RBM20 H3K9Ac H3K9Ac G870S SENP6 H3K27Ac H3K27Ac G870S Fig. 3 SETBP1-mediated epigenetic modulation. a Histone modification ChIP-Seq peak distribution densities plotted according to their distance from gene transcription start sites. b Epigenetic changes resulting from the presence of SETBP1-G870S expressed as function of SETBP1 differential DNA binding (ChIP-Seq G870S/Empty fold change in x axis) vs. histone modification differential enrichment (G870S/Empty fold change in y axis). ***p < 0.001 c SETBP1 ChIP-Seq coverage track and peak alignment to the hg19 reference genome are superimposed to the different histone methylation ChIP-Seq coverage tracks within the BMP5 locus. d Gene expression heatmap of the subset of SETBP1 targets harboring increased H3K4me2 and H3K9ac activation marks generated on three Empty and three SETBP1-G870S clones 27 28 lymphoblastic and acute myeloid leukemias (AMLs) . PHF8 is Taken globally, these data suggest a link between altered PHD also responsible for brain and craniofacial development in activity and SGS phenotype. zebrafish . While the role of PHF6 as an epigenetic modulator is less clear, PHF8 acts as a lysine demethylase. It can bind di- and SETBP1 epigenetic machinery requires functional AT-hooks. tri-methylated H3K4 in the context of KMT2A complexes The identification of an AAAATAA/T consensus binding motif through its N-terminal PHD finger , exerting its demethylating closely resembling that of HMGA1 suggests that SETBP1 may activity on H4 Lysine 20. ChIP against H4K20me1 revealed a bind to gDNA thanks to the presence of its AT-hook domains. To significant decrease in H4K20 mono-methylation in cells test this hypothesis, we generated new isogenic lines carrying expressing mutated SETBP1 for all the tested SETBP1 target ΔAT1 deletions of the first (SETBP1-G870S ), second (SETBP1- genes (Fig. 4f), suggesting that the SETBP1 complex possesses ΔAT2 ΔAT1,2 G870S ), and both (SETBP1-G870S ) AT-hook domains H4K20 demethylase activity. Co-IP experiments confirmed the (Supplementary Figure 7). The third AT-hook was not deleted as interaction of SETBP1 with PHF8 and PHF6, indicating that both it resides within the SET-binding domain. Combined disruption PHD members are part of the SETBP1 complex (Fig. 4g, h). of AT-hooks 1 and 2 resulted in a marked reduction of mRNA NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 5 | | | H3K9Ac G870S/Empty H3K4me2 G870S/Empty Number of sites Number of sites (Log2 FC) (Log2 FC) [log2] [log2] H3K4me3 G870S/Empty H3K27Ac G870S/Empty (Log2 FC) (Log2 FC) H3K36me3 G870S/Empty (Log2 FC) Empty1 Empty2 Empty3 G870S1 G870S2 G870S3 SETBP1 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ab c FRET: SETBP1-HCF1 FRET: SETBP1-bTrCP 130 130 Lysate IP:HCF1 IP:lgG Lysate IP:V5 120 120 WB 110 110 V5 HCF1 100 100 HCF1 V5 90 90 12 3 4 5 67 8 9 10 12 3 4 5 67 8 9 10 Frame number Frame number SETBP1-WT SETBP1 G870S SETBP1-WT SETBP1 G870S SETBP1 ΔHBM SETBP1 G870S ΔHBM SETBP1 ΔHBM SETBP1 G870S ΔHBM ef 0.10 H4K20me1 Lysate IP:V5 0.08 WB 0.06 KMT2A 0.04 V5 ** 0.02 ** *** *** * ** ** *** BMP5 NFE2L2 SKIDA1 PDE4D COBLL1 MECOM BMP5 PDE4D ERBB4 Empty SETBP1 G870S SETBP1 G870S ΔHBM Empty SETBP1 G870S gh Promoter Lysate IP:V5 Lysate IP:V5 HCF1 WB PHF8/6 PHF8 PHF6 MLL1 V5 V5 CDS AT-rich POLR2A H3K4me H4K20me ** * *** *** *** *** *** *** *** ** *** *** *** *** *** *** *** * ** MECOM BMP5 PDE4D ERBB4 MECOM BMP5 PDE4D ERBB4 Empty SETBP1 G870S ΔATH1 SETBP1 G870S SETBP1 G870S ΔATH2 SETBP1 G870S ΔATH1,2 Fig. 4 SETBP1 interacts with the SET1/KMT2A COMPASS-like complex. a Co-immunoprecipitation was performed against the HCF1 protein (left) or the V5 flag (right) and blotted with an anti-V5 or HCF1 antibody. b, c FRET analysis showing physical interaction between SETBP1 and HCF1. Positive FRET signal was recorded in both couples of HCF1 and SETBP1 WT or G870S (b, green and blue lines), conversely no FRET signal was recorded for HCF1 and SETBP1ΔHBM or G870SΔHBM (b, red and orange lines). FRET between β-TrCP and SETBP1 variants was assayed to demonstrate that ΔHBM did not modify the known SETBP1–β-TrCP interaction (c). Acceptor photobleaching was performed after the third acquired frame and indicated with gray bars in both the graphs. Bars represent the standard error of three experiments. d Relative expression of SETBP1 target genes as assessed by Q-PCR in empty (black), SETBP1-G870S (orange), and SETBP1-G870S-ΔHBM (light orange) lines. e Co-immunoprecipitation was performed against the V5 flag and blotted with an anti-KMT2A antibody. f ChIP against H4K20me1 followed by Q-PCR on a set of SETBP1 target genes performed on Empty (black bars) and SETBP1- G870S cells (orange bars). g Co-immunoprecipitation was performed against the V5 flag and blotted with anti-PHF8 and anti-PHF6 antibodies. h Proposed model for SETBP1 epigenetic network. i Relative expression of SETBP1 target genes in cells transduced with empty vector, SETBP1-G870S, or SETBP1- G870S carrying deletion of the first (ΔATH1), second (ΔATH2), and both (ΔATH1,2) AT-hooks. j ChIP against SETBP1-G870S in cells transduced with empty vector, SETBP1-G870S, or SETBP1-G870S carrying deletion of the first (ΔATH1), second (ΔATH2), or both (ΔATH1,2) AT-hooks, followed by Q- PCR on a set of SETBP1 target genes. In panels d, f, i and j, statistical analysis was performed using t-test. Bars represent the standard error of three experiments. *p < 0.05; **p < 0.01; ***p < 0.001 6 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | Empty G870S Empty G870S G870S Empty G870S Empty G870S Empty G870S Empty G870S Empty Empty G870S G870S Empty Empty G870S G870S Relative mRNA expression level Relative mRNA expression level % Donor fluorescence % Input % Input % Donor fluorescence NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE expression for all the target genes under analysis (Fig. 4i; ERBB4 p = 0.003), indicating that SETBP1 binding to gDNA MECOM p = 0.014; BMP5 p < 0.001; PDE4D p < 0.001; ERBB4- depends on the presence of functional AT-hooks (Fig. 4j). ΔAT1 ΔAT1,2 G870S p < 0.001; ERBB4-G870S p = 0.1). ChIP per- The fact that SETBP1 is commonly found in DNase I formed on target genes followed by Q-PCR confirmed the same hypersensitive cluster regions together with its ability to interact effect (MECOM p < 0.001; BMP5 p < 0.001; PDE4D p < 0.001; with an epigenetic activator complex containing H3K4 ab 293FLP 293FLP ** 0.4 ** 0.3 2000 0.2 0.1 0 0.0 Empty WT G870S Empty WT G870S 0.4 *** 0.3 0.2 0.1 MECOM Empty SETBP1 G870S Empty SETBP1 WT SETBP1 G870S SETBP1 WT peaks SETBP1 G870S peaks H3K4me2 H3K4me2 G870S H3K4me3 H3K4me3 G870S H3K36me3 H3K36me3 G870S H3K9Ac H3K9Ac G870S H3K27Ac H3K27Ac G870S MLL (GSM1897369) de 293FLP aCML patients Group CTRL MUT Group Gene *** f CDC25C 300 r = 0.921 –0.2E+00 CIT FKBP15 TYMS ATPIF1 GINS4 –0.6E+00 RSU1 CD63 TPI1 ME2 CDKN3 –1.0E+01 KIF20A DHRS1 STIL TSPAN13 –1.4E+01 HAUS1 HMOX1 KDM1B BUB1 NAGK 0 –1.8E+01 GMNN 05 10 15 MFF WT MUT WIPI1 Log2 RNA-Seq aCML patients SETBP1 MUT WT SETBP1 Gene HMOX1 DHRS1 CDC25C BUB1 CDKN3 CIT KIF20A ATPIF1 GINS4 TYMS GMNN STIL HAUS1 TPI1 TSPAN13 CD63 FKBP15 NAGK MFF ME2 RSU1 KDM1B WIPI1 Relative Row min Row max NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 7 | | | MECOM Rel. Expr. (RPKM) % Input Empty_1 Empty_2 Empty_3 G870S_1 G870S_2 G870S_3 MECOM Rel. Expr. (RPM) CMLPh-001 CMLPh-002 CMLPh-004 CMLPh-006 CMLPh-007 CMLPh-008 CMLPh-009 CMLPh-010 CMLPh-011 CMLPh-012 CMLPh-014 CMLPh-016 CMLPh-017 CMLPh-018 CMLPh-020 CMLPh-021 MECOM Rel. Expr. (AU) CMLPh-024 CMLPh-025 Log2 Q-PCR CMLPh-028 CMLPh-034 CMLPh-035 CMLPh-003 CMLPh-005 CMLPh-013 CMLPh-015 CMLPh-019 CMLPh-023 CMLPh-026 CMLPh-029 CMLPh-030 CMLPh-036 CMLPh-037 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 methyltransferase activity suggests that SETBP1 should be able to and myeloid differentiation (Supplementary Data 10). In control gene expression through modulation of chromatin accordance with its previously observed transcriptional activity , accessibility. To test this hypothesis, we generated ATAC-Seq we show here that MECOM target genes are also differentially (assay for transposase-accessible chromatin using sequencing) expressed (p < 0.0001, Fig. 5d) in cells expressing SETBP1-G870S. experiments on the isogenic 293 FLP-In Empty, SETBP1-WT, To confirm these results in fresh leukemic cells, RNA-Seq and Q- and SETBP1-G870S cell lines. We assessed the ATAC-Seq signal PCR analyses were performed in 32 atypical chronic myeloid in a region comprised between −2000 bp and +10,000 bp from leukemia (aCML) cases (11 positive and 21 negative for SETBP1 the TSS of all the human genes. In line with the expected high somatic mutations). The results show that SETBP1-positive accessibility of TSS regions, ATAC-Seq analysis revealed a peak at patients express higher levels of MECOM (Fig. 5e, f; p = TSS in all the lines under test (Supplementary Figure 8a, b; 0.0002) and MECOM target genes (Fig. 5g; p < 0.0001), with Supplementary Data 7–9). Subsequently, we investigated whether excellent correlation between RNA-Seq and Q-PCR data (Fig. 5f). a correlation could be found between the intensity of the relative ATAC-Seq signal in SETBP1-G870S vs. Empty or SETBP1-WT Mutated SETBP1 delays neuronal migration to the cortex. vs. Empty at TSS and gene bodies and the relative gene Upon their general developmental delay, Schinzel–Giedion expression, as assessed by RNA-Seq in the same lines. Analysis patients present heavy neurological impairment characterized by at global gene level performed throughout a set of iterations (50) microcephaly, altered neuronal layering, underdeveloped corpus characterized by progressively more stringent differential expres- callosum, ventriculomegaly, cortical atrophy, or dysplasia that sion fold-change filtering criteria revealed a significant linear 11, 12 cause seizures and severe mental retardation . Additionally, relationship between differential ATAC-Seq signal at TSS and the mRNA of SETBP1 gene is expressed in both germinal (high RNA expression for both SETBP1-G870S and SETBP1-WT level of expression) and differentiated area (low level) of different (Supplementary Figure 8c). Taken globally, these data indicate brain regions including cerebral cortex (Supplementary Fig- that genes that are significantly upregulated in the presence of ure 10). To gain insight into the function of the mutated form of either WT or mutated SETBP1 exhibit an increased chromosomal SETBP1 in nervous system development, we transduced radial accessibility in their TSS regions as well as in gene bodies. glial progenitors of cerebral cortices of E13.5 mouse embryos with expression plasmids encoding SETBP1-WT, SETBP1-G870S, ΔAT1,2 MECOM is a direct transcriptional target of SETBP1. In 2013, SETBP1-G870S , and SETBP1-G870S-ΔHBM using an in Makishima and colleagues reported that the presence of mutated utero electroporation system (Fig. 6a) . The transduced cells and SETBP1 was associated with upregulation of MECOM expres- their progeny can be easily traced along their development thanks sion ; however, a clear mechanistic explanation of this finding to the green fluorescent protein (GFP) expression. Two days after was missing. MECOM, located on the long arm of chromosome 3, the procedure, almost the totality of the SETBP1-G870S electro- is a transcription factor able to recruit both coactivators and porated cells remained stacked in the deep part of the developing corepressors . It is expressed in hematopoietic stem cells, playing cortical wall while many control cells (transduced with GFP only) an important role in hematopoiesis and in hematopoietic stem were already migrating in the outward cortical region where post- cell self-renewal . MECOM is often overexpressed, as a result of mitotic neurons reside (Fig. 6b; Supplementary Movie 1 and 2). chromosomal translocations, in myelodysplastic syndromes and In addition, in SETBP1-G870S-overexpressing tissue, the apical in approximately 10% of AML cases, being a strong negative domain of the cortical layer where the neural progenitors are 33, 34 prognostic marker for therapy response and survival . located was severely disorganized as shown by aberrant locali- In line with Makishima’s report, differential expression RNA- zation of TBR2+ intermediate progenitors that lost their typical Seq (Fig. 5a; p = 0.04 and p = 0.08 for SETBP1-WT and SETBP1- strip arrangement in the basal part of the ventricular zone G870S vs. Empty, respectively) and Q-PCR analyses (Fig. 5b; p = (Fig. 6c). Five days after surgery, many control GFP+ neurons 0.01 and p = 0.001 for SETBP1-WT and SETBP1-G870S vs. were detected in the mature cerebral mantel zone, contributing to Empty, respectively) showed that MECOM is upregulated in 293 the fiber tract of the corpus callosum (Fig. 6d, arrow), while the FLP-In cells expressing WT or mutated SETBP1. A similar vast majority of cells electroporated with SETBP1-G870S were upregulation was detected in the human myeloid TF-1 cell line incorrectly located at the deepest cortical tissue (65 vs. 32% of the transduced with SETBP1-G870S (Supplementary Figure 9; p = control condition; t-test p < 0.001 in bins 1–3) with only few 0.0001). ChIP-Seq data showed that both SETBP1-WT and neurons projecting to the contralateral hemisphere through the SETBP1-G870S bind to the MECOM promoter, highlighting corpus callosum (Fig. 6d). In line with the proposed quantitative MECOM as a direct target of SETBP1 transcriptional activity model for SETBP1 mutations, overexpression of SETBP1-WT (Fig. 5c). MECOM modulates the expression of a significant showed similar, albeit reduced, effects (Fig. 6d). Overexpression number of genes involved in hematopoietic stem cell proliferation of mutated SETBP1 carrying either AT-hooks 1 and 2 double Fig. 5 Analysis of MECOM expression and downstream targets. a Box plot showing the RNA-Seq differential expression analysis of MECOM in the 293 FLP- In Empty, SETBP1-WT, and SETBP1-G870S cell models. The top and bottom of each box represent the first and third quartile, respectively; the internal line represents the median; the dot represents the mean. Experiments were performed in triplicate. b Q-PCR analysis of MECOM expression in the 293 FLP-In Empty, SETBP1-WT, and SETBP1-G870S cell models. The top and bottom of each box represent the first and third quartile, respectively; the internal line represents the median; the small square represents the mean. Experiments were performed in triplicate; statistical analysis was performed using t-test. c SETBP1 ChIP-Seq coverage track and peak alignment to the hg19 reference genome (blue tracks) are superimposed to the different histone methylation tracks and KMT2A (MLL) ChIP-Seq coverage track within the MECOM locus. The boxed histogram represents an independent ChIP experiment performed against the V5 flag in the FLP-In cells followed by a Q-PCR directed against the predicted SETBP1-G870S-binding locus on the MECOM promoter. ChIP was performed in triplicate; statistical analysis was performed using t-test. d Gene expression heatmap of MECOM target genes in three Empty/SETBP1-G870S FLP-In clones. e Differential MECOM expression as read counts per million of mapped reads (RPM) in 32 aCML patients carrying WT (21) or mutated (11) SETBP1. The top and bottom of the box represent the first and third quartile, respectively; the internal line represents the median. Statistical analysis was performed using t-test. f Linear correlation of MECOM expression as assessed by RNA-Seq (x axis) and Q-PCR (y axis). r represents the Pearson linear correlation coefficient. g Gene expression heatmap of MECOM target genes in 32 aCML patients carrying WT (21) or mutated SETBP1 (11). Error bars represent the standard error. *p < 0.05; **p < 0.01; ***p < 0.001 8 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** ** *** * ** **** **** *** **** **** * **** **** **** **** **** **** * **** **** **** **** *** NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE ab c IEU E13.5 - analysis E15.5 SETBP1 SETBP1 GFP GFP G870S G870S GFP or SETBP1 (WT or mutants) + GFP v v v d e DAPI/GFP IEU E13.5 - analysis E18.5 Bin: SETBP1 GFP p G870S ns cp ns ns ns ns iz ns vz/svz ns ns 0.0 0.1 0.2 0.3 0.4 0.5 Percentage of GFP cells Fig. 6 In utero electroporation of GFP and SETBP1-G870S. a Schematic representation of the electroporation procedure. b Snapshots from 3D reconstruction resulting from 2-photon microscopy on either GFP- or SETBP1-electroporated cortices (2 days) after tissue clarification (X-Clarity system) showed defects in radial migration of the SETBP1-G870S misexpressing cells. p pial side, v ventricular side. The GFP signal in the pial membrane (white arrows) and along the thickness of the SETBP1-G870S tissue (red arrow) is due to the basal processes of the GFP+ radial glia cells located in the deepest part of the organ. c Immunohistochemistry for GFP and TBR2 on coronal section of 2 days electroporated tissues. d Five-day electroporated cortices and quantification of the migration of the GFP+ cells from apical (bin #1) to pial part of the organ (bin #5); arrow indicates GFP+ corpus callosum. Statistical analysis was performed using two-way ANOVA; error bars represent standard error. *p < 0.05; **p < 0.01; ***p < 0.001; ***p<0.0001. e Immunohistochemistry for GFP and SATB2 marker on coronal section of 5 days electroporated tissues. In the insets on the right, the images of SETBP1- G870S DAPI (up), GFP, SATB2, and merge of GFP/SATB2 (bottom) are shown. cp cortical plate, iz intermediate zone, svz subventricular zone, vz ventricular zone. Bars: c, e: 100 μm, d: 250 μm ΔAT1,2 knockout (SETBP1-G870S ) or deletion of the HCF1- modulator, showing that SETBP1 interacts with gDNA through binding domain (SETBP1ΔHBM) largely restored the normal its AT-hook domains, forming a multiprotein complex including migration pattern (Fig. 6d), indicating that the presence of HCF1, KMT2A, PHF8, and PHF6, which results in increased functional domains responsible for DNA interaction or multi- chromatin accessibility, as assessed by ATAC-Seq (Supplemen- protein complex recruitment is required to modulate neuronal tary Figure 8), and transcriptional activation. The altered tran- migration. Despite the impaired migration, the cells expressing scription caused by the increased levels of SETBP1-G870S was SETBP1-G870S retained the capability of differentiating into also shown to be involved in the pathogenesis of SGS, aCML, and neurons as confirmed by the expression of the mature neuronal related myeloid malignancies, as revealed by in utero electro- marker SATB2 (Fig. 6e). These findings demonstrate that the poration of SETBP1-G870S in ventricular central nervous system increase in SETBP1 levels during brain morphogenesis has a high cells and by the analysis of aCML samples. impact on the dynamics of both neuronal proliferation and Oakley et al. previously showed that, in murine bone marrow, migration that can be responsible for the neuroanatomical defects progenitors transduced with high-titer retrovirus-expressing described in SGS. Setbp1 are able to bind to Hoxa9/10 and Myb promoters and to 13, 37 14 upregulate these genes . Vishwakarma and colleagues , using a similar murine model, demonstrated binding of Setbp1 to the Runx1 promoter , which, however, was associated with Runx1 Discussion downmodulation. Interestingly, in our human 293 FLP-In model This work describes the results of a next-generation sequencing- we found a similar downregulation (Supplementary Figure 11a). based, unbiased approach performed to investigate the function ChIP experiments, however, demonstrated only a very weak of SETBP1 and its pathological variant SETBP1-G870S. We binding of SETBP1 to the human RUNX1 promoter (Supple- demonstrate the ability of human SETBP1 to directly bind gDNA mentary Figure 11b), likely due to the expression of SETBP1 and that the binding occurs preferentially but not exclusively in being much lower in our model than in traditional high-titer gene promoter regions, as SETBP1 has also been detected in retroviral transduction systems or to differences in species spe- enhancers, exonic, intronic, and intergenic regions. In this work, cificity. To test this hypothesis, we repeated ChIP experiments we focused on the potential role of SETBP1 as a transcriptional NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 9 | | | IEU E13.5 - analysis E18.5 SETBP1-G870S SETBP1-WT GFP SETBP1-G870S SETBP1-G870S IEU E13.5 - analysis E15.5 ΔHBM ΔATH1,2 GFP/TBR2 GFP/SATB2 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 using a transient transfection model, achieving a 4.2- and 6.9-fold constitute the only genetic alteration in SGS, somatic SETBP1 increase in SETBP1-WT and G870S expression, respectively, mutations are typically present in conjunction with several other when compared with FLP-In (Supplementary Figure 11c). somatic events in myeloid neoplasms, in which SETBP1 muta- Indeed, ChIP experiments performed using transient SETBP1 tions usually represent late events . transfectants showed a significant increase in the binding to In summary, we showed here that SETBP1 is able to recruit a RUNX1 promoter (Supplementary Figure 11d). This, however, HCF1/KMT2A/PHF8/PHF6 transcriptional activator complex, was accompanied by an increase in RUNX1 expression (Supple- thus identifying SETBP1 as a transcriptional activator, beyond its mentary Figure 11e), confirming that SETBP1 promotes upre- original SET-binding activity. Further studies will be required to gulation of gene expression and suggesting that RUNX1 gain further insights into this complex network and to dissect the downmodulation is not a direct effect of SETBP1. Further studies functional role of SETBP1–gDNA interaction in intergenic will be required to assess whether the differences between our regions. model and the murine model developed by Vishwakarma and colleagues are species specific. DEG analysis between SETBP1- Methods G870S and Empty cells in our FLP-In model and in aCML Patients. Diagnosis of aCML and related diseases was performed according to the patients (11 positive and 21 negative for SETBP1 somatic muta- World Health Organization 2008 classification. All patients provided written informed consent, which was approved by the institutional ethics committee. This tions) failed to reveal significant changes in the expression of study was conducted in accordance with the Declaration of Helsinki. RUNX1, HOXA9/10,or MYB (Supplementary Figure 12), corro- borating the hypothesis that the effect of SETBP1 as a tran- Cell-lines. 293T, TF-1, and the 293 FLP-In™ cell-lines were purchased from ATCC scription factor is species/tissue specific. (Manassas, VA, USA), DSMZ (Braunschweig, Germany), and Thermo Fisher In line with evidence indicating that KMT2A has strong H3 Scientific (Waltham, MA, USA), respectively, and maintained following the mono- and di-methylation but very weak tri-methylation activ- manufacturers' instructions. ity , no significant H3K4me3 enrichment could be found in promoters occupied by SETBP1, although the intersection Plasmids and transfections. Stable 293 FLP-In Empty, 293 FLP-In SETBP1wt, between SETBP1 promoter occupancy and transcriptome analysis and 293 FLP-In SETBP1-G870S cell lines were prepared cotransfecting pOG44 together with the increase in H3K9Ac and the direct interaction (Thermo Fisher Scientific) and pFRT-SETBP1 vectors with Fugene6 reagent and −1 were maintained in standard medium with 100 μgml Hygromycin. Stable TF-1 of SETBP1 with PHF8 demethylase reveal that SETBP1 is part of cell lines were retrovirally infected using phoenix packaging cells transfected with a transcriptional activator complex. The exact explanation of this 10 μg of MIGR1-SETBP1 Gly870Ser or WT or with empty MIGR1 vector using complex histone pattern is still unclear, given also our limited FuGENE6 (Promega); retroviruses were collected after 3 days of culture. Transient knowledge about the functions of the H3K4me2 mark. Never- 293T transfectants were prepared using the pcDNA6.2-SETBP1wt, pcDNA6.2- G870S, or empty vectors with Fugene6 reagent (Promega, Madison, WI, USA) theless, several lines of evidence suggest that H3K4me2 is strongly 38, 39 following standard protocols. The deletion of the HBM site (aa991-994; enriched in lineage-specific gene promoters . The abundance NM_015559.2) within the SETBP1 coding sequence was performed with the fol- of genes associated with multi-organ development among those lowing primers SETBP1_HCF1del_for (CAGCATTTTTCGGATTAATTTTC regulated by SETBP1 is particularly interesting in the context of CGGTGCCATATATCCAGTATG) and SETBP1_HCF1del_rev (CATACTGGAT ATATGGCACCGGAAAATTAATCCGAAAAATGCTG); the deletion of the AT- SGS, whose hallmark is the presence of multi-organ development hooks 1 and 2 were performed using the following primers: SETBP1_ATH1del_for abnormalities. Indeed, analysis of the Gene Ontologies associated (CAGTCTTACTGTGATCACTCCACTCACAGTCGAGACGATTCATG), with DEGs in SETBP1-G870S cells highlights the presence of a SETBP1 ATH1del_rev (CATGAATCGTCTCGACTGTGAGTGGAGTGATCACA much stronger association with SGS than with myeloid malig- GTAAGACTG), SETBP1_ATH2del_for (GTAGGACTTCAGACTTGAA- GACCATGACAAAGGTGCC), and SETBP1 ATH2del_rev (GGCACCTTTGTC nancies, despite the fact that SETBP1 somatic mutations are 1–9 ATGGTCTTCAAGTCTGAAGTCCTAC) as previously described . pCGN- detected in a large number of clonal myeloid disorders .In HCF1-fl was a gift from Winship Herr (Addgene plasmid #53309). aCML and related malignancies, the hyperactivation of the SETBP1–SET axis caused by the stabilization of mutated SETBP1 ChIP sequencing (ChIP-Seq). ChIP was performed as previously reported (GEO protein and leading to the inhibition of the PP2A oncosuppressor accession number: GSE86335). Briefly, proteins were crosslinked with 0.4% for- probably plays a major functional role in driving the leukemic maldehyde and cells were lysed. Chromatin was fragmented with a Bioruptor phenotype, while the interaction with gDNA and the subsequent sonicator system (Diagenode, SA, USA) and subsequently immunoprecipitated with H3K4me3 (ab8580, Abcam, UK), H3K4me2 (C15410035C, Diagenode), modulation of gene expression likely plays a critical role in the H3K36me3 (Ab9050, Abcam), H3K27Ac (Ab4729, Abcam), H3K9Ac (39137, onset of SGS, in which the SETBP1-mediated inhibition of PP2A Active Motif, Carlsbad, CA, USA), and H4K20me1 (Mab147-010, Diagenode) is probably insufficient to recapitulate the complex phenotype of antibodies or anti-V5 agarose beads (Sigma-Aldrich). After immunoprecipitation, the SGS syndrome . However, the transcriptional activity of DNA was purified and libraries were prepared for sequencing following the Illu- SETBP1 likely contributes to the leukemic phenotype too. In mina ChIP-Seq protocol (TrueSeq ChIP library prep kit IP-202-1012) with an Illumina HiSeq2500 in single read mode (Galseq, Monza, Italy). Validation of 2013, Makishima and colleagues reported that cases with SETBP1 ChIP-Seq data was performed amplifying the immunoprecipitated DNA with Sybr- mutations were characterized by high expression of the MECOM Green Q-PCR. Input was used as a loading control. Primer sequences were: 3, 40 oncogene . Here we show that MECOM is a direct target of BMP5_Fw (CAACCCTGCTGGGAAAGAAGAG), BMP5_Rw (TCATCAAGCT SETBP1 transcriptional activity, which explains the original AACTTAGGCACAAC), NFE2L2_Fw (AACCAGAAGAATACAATCCCAATG), NFE2L2_Rw (AAGAAGTTTCTGCTCATCCTTTGTAG), PDE4D_Fw (CCTT observation. Interestingly, Goyama and colleagues previously GAGCCAACCTTCTCCTTC), PDE4D_Rw (CACCCAAAGACATGACCAA showed that SETBP1 is one of the genes whose expression is most CCTC), SKIDA1_Fw (TTCAAGTATCACGTTACTGTTTGC), SKIDA1_Rw strongly reduced in MECOM−/− hematopoietic cells , sug- (GTCACTTATTCAGCCACGCAGAC), COBLL1_Fw (TCTAATTGGTGGCAG gesting that SETBP1 is also one of the MECOM downstream GTTTAAGC), COBLL1_Rw (TGTCTGTCAGGTGTAAAGAATCATC), ERBB4_Fw (ACAAACTCCTCCAAACTGCTACTG), ERBB4_Rw (GTGATCCA targets. Taken together, these findings suggest the presence of a TTGGAAACTGTAAATGC), RUNX1P2_Fw (CCTATGCAAACGAGCTGAGG), positive feedback loop occurring between MECOM and SETBP1, RUNX1P2_Rw(GCTCTATGAATGAGAGTGCCTG), MECOM_Fw whose biological importance and effects will require further stu- (CTCCCAAATGTCTTAATCGTGTCG), and MECOM_Rw dies. These findings highlight a potentially critical role for this (TTCGGACCCTTTGGCTAGATTGTG). ChIP-Seq analyses were performed using MACS v. 1.4.2 using the –no model transcriptional complex and its downstream effectors in the parameter to skip the model building step. The ratio of the intersection and the oncogenesis of SETBP1-positive myeloid disorders, therefore similarity of the two genomic interval sets was calculated via bedtools Jaccard suggesting that a strict dichotomous model is too simplistic to v2.26 statistics . fully recapitulate and explain both phenotypes. It is also impor- Fold enrichment of either SETBP1 or histone methylation ChIP-Seq tant to note that, while de novo germline SETBP1 mutations experiments in G870S or WT background were calculated with MACS using 10 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE narrow or broad (SETBP1) peak calling parameters . Transcription factor Co-immunoprecipitation. One hundred million HEK 293T cells were transiently association strength and relative fold change were calculated as previously transfected with pcDNA 6.2 V5-SETBP1-WT, SETBP1 G870S, or with the empty reported . Downstream statistics, namely, multivariate analysis with linear vector as control. Forty eight hours after transfection, cells were collected, washed correlation assumptions, were performed with the IBM SPSS statistical package. twice with cold PBS, lysed in Buffer 1 (Pipes, pH 8.0, 5 mM, KCl 85 mM, NP-40 0.5%), supplemented with protease inhibitors (Halt™ Protease Inhibitor Cocktail, Thermo Fisher Scientific), kept for 10 min on ice, homogenized with douncer ATAC-Seq. Cells (100,000/sample) were washed once in cold phosphate-buffered homogenizer (10 hits), and centrifuged at 700 × g for 10 min. The pellet was then saline (PBS) 1×, spun at 800 × g for 5’ at 4 °C and resuspended in cold PBS in the resuspended in Buffer 2 (Tris HCl, pH 8.0, 50 mM, sodium dodecyl sulfate 0.1%, presence of proteasome inhibitors and incubated on ice for 10’. Cells were cen- deoxicholate 0.5%) supplemented with protease inhibitors, sonicated with Bior- trifuged at 800 × g at 4 °C for 10’ and resuspended in: 2× TD buffer (25 µl), Tn5 uptor Next Gen (Diagenode) (5 cycles 30 s ON, +30 s OFF) to promote gDNA Transposase (2.5 µl; Illumina), and water (up to 50 µl). The sample was incubated disruption and centrifuged at 18,000 × g for 10 min. The supernatant representing at 37 °C for 30’ and purified using the SPRI AMPure XP beads. Post-tagmentation the nuclear fraction was quantified with Bradford assay and a total of 1 µg of amplification was performed using Nextera primers (Illumina) and Herculase II protein was loaded on 100 µl V5-agarose beads (Sigma-Aldrich) and incubated polymerase (Agilent) using a standard protocol. Quality of the ATAC-Seq libraries under rotation overnight at 4 °C. Beads were then washed three times with PBS was assessed using a Tape Station (Agilent) and by agar gel. Quantification was +protease inhibitors, and elution was performed with 7 M Urea, 2 M Thiourea, performed using a QuBit Fluorometer (ThermoFisher). and 4% CHAPS for subsequent mass spectrometric analysis or with Laemmli buffer for western blot analysis. All chemicals were purchased from Sigma Aldrich. ATAC-Seq analysis. ATAC-Seq fastq files were initially aligned to the hg38 48 Immunoblot analysis. Primary antibodies were V5 (ab27671 330 Abcam, Cam- human reference genome using BWA with standard parameters. BAM files were bridge, UK; dilution 1:2000), HCF1 (A301-399A Bethyl Laboratories, Inc., Mon- sorted and indexed using Samtools . tgomery, TX, USA; dilution 1:1000), MLL1 (14689 Cell Signaling Technology, BAM files from individual replicates were initially merged together and Danvers, MA, USA; dilution 1:1000), PHF8 (A301-772A Bethyl Laboratories, Inc., subsequently processed using our ATAC-Seq tool. Briefly, all the gene start Montgomery, TX, USA; dilution 1:500), and PHF6 (A301-451A Bethyl Labora- positions, gene end positions, chromosome, TSS, and gene strands were annotated tories, Inc., Montgomery, TX, USA; dilution 1:500). Secondary antibody was anti- (Gencode24). Coverage counts at single-nucleotide resolution were generated for mouse anti-rabbit horseradish peroxidase conjugated (Biorad, Hercules, CA, USA). each Gene/TSS from basesBeforeTSS (2000) to basesAfterTSS (10,000). Coverage at The uncropped scan of western blots related to PHF6 immunoprecipitation are each position was then normalized using the total coverage of each input BAM. To shown in Supplementary Figure 13 as an example. plot ATAC-Seq heatmaps, normalized coverage data were binned (binSize = 200 bp) and sorted in decreasing order according to the intensity of the ATAC signal throughout the entire region (sum of the binned signals from basesBeforeTSS to Proteomics data analysis. The protein samples were digested in Amicon Ultra-0.5 basesAfterTSS). To plot line graphs, normalized counts were summed at individual centrifugal filters using modified FASP method . The peptides were separated with bins across the entire gene set from basesBeforeTSS to basesAfterTSS. Final binned the nanoAcquity UPLC system (Waters) equipped with a 5-µm Symmetry C18 counts were then normalized by the gene set size. trapping column, 180 µm×20 mm, reverse-phase (Waters), followed by an analy- To test for the presence of a correlation between chromatin accessibility and tical 1.7-µm, 75 µm×250 mm BEH-130 C18 reversed-phase column (Waters), in a RNA expression, ATAC-Seq and RNA-Seq data generated from the same lines single-pump trapping mode. The parameters of the HD-MSE runs were described were compared using the following approach: normalized ATAC-Seq signal data previously . Protein identifications were performed with ProteinLynx Global generated for a region comprised between basesBeforeTSS (1500) to basesAfterTSS Server (PLGS v3.0) as described . Database searches were carried out against (5000) for each gene in G870S and Empty lines were initially calculated. A G870S/ UniProt human protein database (release_07072015, 71907 entries) with Ion Empty coverage ratio was then calculated throughout the entire gene set. In Accounting algorithm and using the following parameters: peptide and fragment parallel, normalized G870S/Empty RNA-Seq Log2 fold-change expression ratio tolerance: automatic, maximum protein mass: 500 kDa, minimum fragment ions was computed. Log2 fold-change data were filtered in Log2_Threshold_Cycles (50) matches per protein: 7, minimum fragment ions matches per peptide: 3, minimum iterations with progressively more stringent criteria, by increasing the Log2 fold- peptide matches per protein: 1, primary digest reagent: trypsin, missed cleavages change filter from Log2_Threshold_Start (0) at Log2_Threshold_Step (+0.1 increase allowed: 2, fixed modification: carbamidomethylation C, variable modifications: for each Log2_Threshold_Cycle) steps. For each iteration, a Pearson correlation was deamidation (N, Q), oxidation of Methionine (M) and FDR < 4%. generated between normalized ATAC-Seq ratios and filtered RNA-Seq fold changes. Immunofluorescence microscopy. Transfected HEK 293 cells, seeded on glass −1 coverslips coated with poly-D-lysine (0.1 mg ml ), were washed twice with PBS and fixed for 10 min at 25 °C with 4% (w/v) p-formaldehyde in 0.12 M sodium Code availability. Code of the ATAC-Seq tool and of the Epigenetic Marks phosphate buffer, pH 7.4, incubated for 1 h with primary antibodies in GDB buffer Correlation tool will be made available upon request. [0.02 M sodium phosphate buffer, pH 7.4, containing 0.45 M NaCl, 0.2% (w/v) bovine gelatin] followed by staining with conjugated secondary anti IgG antibodies for 1 h. After two washes with PBS, coverslips were mounted on glass slides with a Epigenetic marks correlation tool. Anti-V5 and anti-H3K9Ac 293T-SETBP1-WT 90% (v/v) glycerol/PBS solution. and G870S coverage files were used as input. Coverage was binned (BigWig) in 1000 bp regions for all the ChIP-Seq experiments. Individual bins for all the regions across the entire genome were paired and used for Pearson correlation analysis as FRET. FRET measurements were performed with the laser-induced acceptor 52, 53 well as for XY plots. photobleaching method . In our experimental set-up, FRET couples analyzed were made up of transiently transfected GFP-fused SETBP1 WT or mutated forms in combination with transiently transfected Flag-tagged β-TrCP or endogenous Oligonucleotide pulldown assay. Biotinylated oligonucleotides were synthesized HCF1. Forty eight hours after transfection, HEK 293 cells were labeled with proper by Metabion International AG (Steinkirchen, Germany). Target and non-target primary and secondary antibodies and imaged for FRET analysis. Used FRET (unrelated) oligonucleotides were designed according to ChIP-Seq from SETBP1 couples were GFP signal as donor fluorochrome and Alexa Fluor 555-conjugated bound (target-T, chr6: 55,742,037-55,742,136; hg19) and unbound (unrelated-U, secondary antibodies as acceptor fluorochrome . Briefly, three images were cap- chr6:55,775,477-55,775,576; hg19) regions. In the pulldown experiment, 25 µl of tured before bleaching in the 488 nm and the 561 nm channels using the line-by- Pierce-streptavidin magnetic beads (Thermo Fisher Scientific) were washed using a line sequential mode without any averaging steps to reduce basal bleaching. magnetic stand and resuspended in 200 µl of B/W buffer (10 mM Tris HCl, pH 7.5, Bleaching of the acceptor was performed within region of interest (ROI) identified 1 mM EDTA, 2 M NaCl). Each double-stranded biotinylated DNA probe (200 nM) in the nuclear areas with a positive colocalization between the FRET couples using was bound to the beads for 20 min at room temperature (RT) on a rotating device. 30 pulses of a full power 20 mW 561-nm laser line (each pulse 1.28 µs/pixel). After One sample without probe was used as a control for unspecific binding. Beads were bleaching, seven images were acquired in the same channels without any time delay washed with TE buffer and resuspended in 100 µl of BS/THES buffer (22 mM Tris to obtain a full curve. The number of bleaching steps, laser intensity, and acqui- HCl, pH 7.5, 4.4 mM EDTA, 8.9% Sucrose, 62 mM NaCl, 10 mM Hepes, 5 mM sition parameters were held constant throughout each experiment. FRET signal CaCl , 50 mM KCl, 12% Glycerol) supplemented with HALT protease inhibitor was quantified by measuring the average intensities of ROIs in the donor and cocktail (Thermo Fisher Scientific). In all, 15 µg of nuclear extract from 293 FLP-In acceptor fluorochrome channels before and after bleaching using the ImageJ transfectants was diluted in 400 µl of BS/THES buffer and, after collecting 30 µl as software (http://rsbweb.nih.gov/ij/). To determine any change of fluorescence INPUT fraction, was added to the beads together with 1 µg of non-specific DNA intensities not due to FRET occurring during the measurements, a distinct competitor LightShift Poly(dI-dC) (Thermo Scientific). Samples were incubated at unbleached sentinel ROI of approximately the same size of the bleached ROI was 4 °C for 30 min on a rotating device. Beads were then washed four times with BS/ measured in parallel, and all the results were normalized according to the back- THES buffer supplemented with 1 µg of Poly(dI-dC). Final wash was done with ground bleaching recorded in the sentinel. Proper controls were performed to 500 µl of buffer (100 mM NaCl, 25 mM Tris HCl) after incubation for 1 min on a verify that no artefacts were generated in the emission spectra throughout the rotating device at RT. Elution of bound proteins was performed with 50 µl of experimental set-up due to sample overheating. Twenty measurements from three Laemmli Buffer. Lamin-B1 was used as a loading control. different experiments were performed for each experimental condition. NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 11 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 RNA-Seq. RNA libraries were generated starting from 1 µg of total RNA extracted References from 5 × 10 cells using TRIzol (Invitrogen, Life Technologies). RNA quality was 1. Piazza, R. et al. Recurrent SETBP1 mutations in atypical chronic myeloid assessed by using a Tape Station instrument (Agilent). To avoid over- leukemia. Nat. Genet. 45,18–24 (2013). representation of 3′-ends, only high-quality RNA with a RNA Integrity Number ≥8 2. Sakaguchi, H. et al. Exome sequencing identifies secondary mutations of was used. RNA was processed according to the TruSeq Stranded mRNA Library SETBP1 and JAK3 in juvenile myelomonocytic leukemia. Nat. Genet. 45, Prep Kit protocol. The libraries were sequenced on an Illumina HiSeq 3000 with 76 937–941 (2013). bp paired-end reads using Illumina TruSeq technology (GEO accession number: 3. Makishima, H. et al. Somatic SETBP1 mutations in myeloid malignancies. GSE86335). Nat. Genet. 45, 942–946 (2013). Image processing and basecall was performed using the Illumina Real Time 4. Meggendorfer, M. et al. SETBP1 mutations occur in 9% of MDS/MPN and in Analysis Software. Fastq files were aligned to the human genome (GRCh38/hg39) 4% of MPN cases and are strongly associated with atypical CML, monosomy by using STAR , a splice junction mapper for RNA-Seq data, together with the 7, isochromosome i(17)(q10), ASXL1 and CBL mutations. Leukemia 27, corresponding splice junctions Ensembl GTF annotation, using the following 1852–1860 (2013). parameters: 5. Inoue, D. et al. SETBP1 mutations drive leukemic transformation in ASXL1- --runThreadN 8 --outReadsUnmapped Fastx --outFilterType BySJout mutated MDS. Leukemia 29, 847–857 (2015). --outSAMattributes NH HI AS nM MD --outFilterMultimapNmax 20 6. Fabiani, E. et al. SETBP1 mutations in 106 patients with therapy-related --alignSJoverhangMin 8 --alignSJDBoverhangMin 1 --outFilterMismatchNmax myeloid neoplasms. Haematologica 99, e152–153 (2014). 999 --outFilterMismatchNoverLmax 0.04 --alignIntronMin 20 --alignIntronMax 7. Patnaik, M. M. et al. ASXL1 and SETBP1 mutations and their prognostic 1000000 --alignMatesGapMax 1000000 --alignTranscriptsPerReadNmax 100000 contribution in chronic myelomonocytic leukemia: a two-center study of 466 --quantMode TranscriptomeSAM GeneCounts --limitBAMsortRAM 16620578182 patients. Leukemia 28, 2206–2212 (2014). --outSAMtype BAM SortedByCoordinate --chimSegmentMin 20 8. Thol, F. et al. SETBP1 mutation analysis in 944 patients with MDS and AML. --chimJunctionOverhangMin 10. Leukemia 27, 2072–2075 (2013). 9. Wang, X. et al. GATA2 and secondary mutations in familial myelodysplastic Quantitative real-time PCR. Total RNA was extracted with Trizol reagent fol- syndromes and pediatric myeloid malignancies. Haematologica 100, e398–401 lowing standard protocol (Thermo Fisher Scientific). One microgram of RNA was (2015). used to synthesize cDNA using Reverse Transcription Reagents (Thermo Fisher 10. Hoischen, A. et al. De novo mutations of SETBP1 cause Schinzel-Giedion Scientific) after pretreatment with DNAseI to avoid contamination from gDNA. syndrome. Nat. Genet. 42, 483–485 (2010). Real-time Q-PCR was performed using TaqMan® Brilliant II QPCR Master Mix 11. Acuna-Hidalgo, R. et al. Overlapping SETBP1 gain-of-function mutations in (Agilent Technologies, Santa Clara, CA, USA) on a Stratagene-MX3005P (Agilent Schinzel-Giedion syndrome and hematologic malignancies. PLoS Genet. 13, Technologies, Santa Clara, USA) under standard conditions. The housekeeping e1006683 (2017). gene GUSB was used as an internal reference . TaqMan® Gene Expression Assays 12. Lach, B. & Arredondo, J. Cobblestone lissencephaly in Schinzel-Giedion (Thermo Fisher Scientific) were used: SKIDA1 (Hs01096520), NFE2L2 syndrome. J. Child Neurol. 28, 259–263 (2013). (Hs00975961), PDE4D (Hs03988495), FBXO8 (Hs00942619), CEP44 13. Oakley, K. et al. Setbp1 promotes the self-renewal of murine myeloid (Hs00604612), COBLL1 (Hs01117513), BMP5 (Hs00234930), ERBB4 progenitors via activation of Hoxa9 and Hoxa10. Blood 119, 6099–6108 (Hs00955525), CDKN1B (Hs01597588), SETBP1 (Hs00210203), RUNX1 (2012). (Hs01021971), MECOM (Hs00602795). 14. Vishwakarma, B. A. et al. Runx1 repression by histone deacetylation is critical for Setbp1-induced mouse myeloid leukemia development. Leukemia 30, Differential expression analysis. The gene counts generated by the --quantMode 200–208 (2016). GeneCounts parameter were used to calculate gene expression. All the subsequent 15. Cristobal, I. et al. SETBP1 overexpression is a novel leukemogenic mechanism steps were performed using R scripts. These scripts were automatically generated that predicts adverse outcome in elderly patients with acute myeloid leukemia. through metaprogramming, using a dedicated tool (StarCounts2DESeq2) devel- Blood 115, 615–625 (2010). oped by RP using C# as metalanguage. Differential expression analyses were per- 16. Kulakovskiy, I. V. et al. HOCOMOCO: expansion and enhancement of the formed using DESeq2 . Analysis of DEGs characterized by the co-occurrence of collection of transcription factor binding sites models. Nucleic Acids Res. 44, both H3K4me2 and H3K9Ac marks was performed by multivariate regression D116–125 (2016). analysis using the following filtering criteria: transcription factor and histone mark 17. Weirauch, M. T. et al. Determination and inference of eukaryotic relative coverage enrichment >0.5 (log2 FC G870S/Empty); differential gene transcription factor sequence specificity. Cell 158, 1431–1443 (2014). expression p-value < 0.05 (G870S vs. Empty, DESeq2). 18. Ovcharenko, I., Nobrega, M. A., Loots, G. G. & Stubbs, L. ECR Browser: a tool for visualizing and accessing data from comparisons of multiple vertebrate In utero electroporation. SETBP1-G870S was cloned in the pCAG expression genomes. Nucleic Acids Res. 32, W280–286 (2004). vector . In utero electroporation was carried out at E13.5 to target the expression 19. Bourachot, B., Yaniv, M. & Muchardt, C. The activity of mammalian brm/ vectors into the ventricular radial glial cells of the mouse embryos, as previously SNF2alpha is dependent on a high-mobility-group protein I/Y-like DNA 58, 59 described . Briefly, uterine horns of E13.5 pregnant dams were exposed by binding domain. Mol. Cell. Biol. 19, 3931–3939 (1999). −1 midline laparotomy after anesthetization with 312 mg kg Avertin. DNA plasmid 20. Cairns, B. R. et al. Two functionally distinct forms of the RSC nucleosome- (1 µl, corresponding to 3 µg) mixed with 0.03% fast-green dye in PBS was injected remodeling complex, containing essential AT hook, BAH, and into the telencephalic vesicle using a micropipette pulled through the uterine wall bromodomains. Mol. Cell 4, 715–723 (1999). and amniotic sac. Platinum tweezer-style electrodes (7 mm) were placed outside 21. Xiao, H. et al. Dual functions of largest NURF subunit NURF301 in the uterus over the telencephalon, and four pulses of 40 mV for 50 ms were applied nucleosome sliding and transcription factor interactions. Mol. Cell 8, 531–543 at 950 ms intervals using a BTX square wave electroporator. The uterus was then (2001). placed back into the abdomen, the cavity was filled with warm sterile PBS, and the 22. Rao, R. C. & Dou, Y. Hijacked in cancer: the KMT2 (MLL) family of abdominal muscle and skin incisions were closed with silk sutures. methyltransferases. Nat. Rev. Cancer 15, 334–346 (2015). All experiments were conducted after approval and following the guidelines of 23. Dinkel, H. et al. ELM 2016--data update and new functionality of the animal care and use committee from San Raffaele Scientific Institute. eukaryotic linear motif resource. Nucleic Acids Res. 44, D294–300 (2016). 24. Sanchez, R. & Zhou, M. M. The PHD finger: a versatile epigenome reader. MECOM analysis. To identify MECOM direct targets among the SETBP1 DEGs, Trends Biochem. Sci. 36, 364–372 (2011). we took advantage of previously published RNA-Seq and ChIP-Seq data in the 25. Siderius, L. E. et al. X-linked mental retardation associated with cleft lip/palate context of MECOM overexpression . The resulting genes were functionally maps to Xp11.3-q21.3. Am. J. Med. Genet. 85, 216–220 (1999). annotated with FunRich and filtered in accordance with their expression within 26. Lower, K. M. et al. Mutations in PHF6 are associated with Borjeson- the bone marrow (Hypergeometric enrichment test p = 0.0013), thus resulting in a Forssman-Lehmann syndrome. Nat. Genet. 32, 661–665 (2002). total of 23 MECOM direct targets. 27. Van Vlierberghe, P. et al. PHF6 mutations in T-cell acute lymphoblastic leukemia. Nat. Genet. 42, 338–342 (2010). 28. Van Vlierberghe, P. et al. PHF6 mutations in adult acute myeloid leukemia. Data availability. All relevant sequencing data are available at the NCBI GEO Leukemia 25, 130–134 (2011). Server, accession code: GSE86335. 29. Qi, H. H. et al. Histone H4K20/H3K9 demethylase PHF8 regulates zebrafish brain and craniofacial development. Nature 466, 503–507 (2010). Received: 14 December 2016 Accepted: 27 April 2018 30. Buenrostro, J. D., Giresi, P. G., Zaba, L. C., Chang, H. Y. & Greenleaf, W. J. Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position. Nat. Methods 10, 1213–1218 (2013). 12 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE 31. Wieser, R. The oncogene and developmental regulator EVI1: expression, 56. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and biochemical properties, and biological functions. Gene 396, 346–357 (2007). dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014). 32. Kataoka, K. & Kurokawa, M. Ecotropic viral integration site 1, stem cell self- 57. Niwa, H., Yamamura, K. & Miyazaki, J. Efficient selection for high-expression renewal and leukemogenesis. Cancer Sci. 103, 1371–1377 (2012). transfectants with a novel eukaryotic vector. Gene 108, 193–199 (1991). 33. Barjesteh van Waalwijk van Doorn-Khosrovani, S. et al. High EVI1 expression 58. Saito, T. & Nakatsuji, N. Efficient gene transfer into the embryonic mouse predicts poor survival in acute myeloid leukemia: a study of 319 de novo AML brain using in vivo electroporation. Dev. Biol. 240, 237–246 (2001). patients. Blood 101, 837–845 (2003). 59. Sessa, A., Mao, C. A., Hadjantonakis, A. K., Klein, W. H. & Broccoli, V. Tbr2 34. Groschel, S. et al. High EVI1 expression predicts outcome in younger adult directs conversion of radial glia into basal precursors and guides neuronal patients with acute myeloid leukemia and is associated with distinct amplification by indirect neurogenesis in the developing neocortex. Neuron cytogenetic abnormalities. J. Clin. Oncol. 28, 2101–2107 (2010). 60,56–69 (2008). 35. Bard-Chapeau, E. A. et al. Ecotopic viral integration site 1 (EVI1) regulates 60. Pathan, M. et al. FunRich: an open access standalone functional enrichment multiple cellular processes important for cancer and is a synergistic partner for and interaction network analysis tool. Proteomics 15, 2597–2601 (2015). FOS protein in invasive tumors. Proc. Natl. Acad. Sci. USA 109, 2168–2173 61. Zhu, L. et al. ASH1L links histone H3 lysine 36 dimethylation to MLL (2012). leukemia. Cancer Discov. 6, 770–783 (2016). 36. Sessa, A. et al. Tbr2-positive intermediate (basal) neuronal progenitors safeguard cerebral cortex expansion by controlling amplification of pallial Acknowledgements glutamatergic neurons and attraction of subpallial GABAergic interneurons. This work was supported by Associazione Italiana Ricerca sul Cancro 2013 (IG-14249) to Genes Dev. 24, 1816–1826 (2010). C.G.-P., Associazione Italiana; Ricerca sul Cancro 2015 (IG-17727) to R.P., Telethon 37. Nguyen, N. et al. Myb expression is critical for myeloid leukemia development grant (GGP15096), and the Italian Ministry of Health Young investigator grant (# GR- induced by Setbp1 activation. Oncotarget 7, 86300-86312 (2016). 2013-02355540) to A.S. 38. Pekowska, A., Benoukraf, T., Ferrier, P. & Spicuglia, S. A unique H3K4me2 profile marks tissue-specific gene regulation. Genome Res. 20, 1493–1502 (2010). Author contributions 39. Zhang, J., Parvin, J. & Huang, K. Redistribution of H3K4me2 on neural tissue R.P.: Conceived and designed the experiments, performed the experiments, analyzed the specific genes during mouse brain development. BMC Genomics 13(Suppl 8), data, wrote the paper. V.M. and S.R.: performed experiments, analyzed the data, critically S5 (2012). revised the manuscript. M.Mauri, M.P., and C.M.: performed experiments, critically 40. Makishima, H. Somatic SETBP1 mutations in myeloid neoplasms. Int. J. revised the manuscript. L.M.: performed bioinformatics analyses, critically revised the Hematol. 105, 732–742 (2017). manuscript. A.S., F.B., and A.R.: performed in utero electroporation experiments, criti- 41. Goyama, S. et al. Evi-1 is a critical regulator for hematopoietic stem cells and cally revised the manuscript. M.L., R.S., and M.Baumann: performed mass-spectrometry transformed leukemic cells. Cell Stem Cell 3, 207–220 (2008). experiments, critically revised the manuscript. A.P.: performed library preparation for 42. Redaelli, S. et al. Activity of bosutinib, dasatinib, and nilotinib against 18 NGS analyses, critically revised the manuscript. D.R., F.S., E.U., B.M., L.C., M.Merli, F. imatinib-resistant BCR/ABL mutants. J. Clin. Oncol. 27, 469–471 (2009). Passamonti, F.O., A.M., F.Pavesi, and M.Bregni: contributed reagents, critically revised 43. Piazza, R. et al. Epigenetic silencing of the proapoptotic gene BIM in the manuscript. V.B.: conceived in utero electroporation experiments, critically revised anaplastic large cell lymphoma through an MeCP2/SIN3a deacetylating the manuscript. C.G.-P.: conceived and designed the experiments, wrote the paper. complex. Neoplasia 15, 511–522 (2013). 44. Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 (2008). Additional information 45. Quinlan, A. R. & Hall, I. M. BEDTools: a flexible suite of utilities for Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- comparing genomic features. Bioinformatics 26, 841–842 (2010). 018-04462-8. 46. Feng, J., Liu, T., Qin, B., Zhang, Y. & Liu, X. S. Identifying ChIP-seq enrichment using MACS. Nat. Protoc. 7, 1728–1740 (2012). Competing interests: The authors declare no competing interests. 47. Ouyang, Z., Zhou, Q. & Wong, W. H. ChIP-Seq of transcription factors Reprints and permission information is available online at http://npg.nature.com/ predicts absolute and differential gene expression in embryonic stem cells. Proc. Natl. Acad. Sci. USA 106, 21521–21526 (2009). reprintsandpermissions/ 48. Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows- Wheeler transform. Bioinformatics 25, 1754–1760 (2009). Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in 49. Li, H. et al. The Sequence Alignment/Map format and SAMtools. published maps and institutional affiliations. Bioinformatics 25, 2078–2079 (2009). 50. Scifo, E. et al. Proteomic analysis of the palmitoyl protein thioesterase 1 interactome in SH-SY5Y human neuroblastoma cells. J. Proteomics 123,42–53 (2015). Open Access This article is licensed under a Creative Commons 51. Makela, J. et al. Peroxisome proliferator-activated receptor-gamma Attribution 4.0 International License, which permits use, sharing, coactivator-1alpha mediates neuroprotection against excitotoxic brain injury adaptation, distribution and reproduction in any medium or format, as long as you give in transgenic mice: role of mitochondria and X-linked inhibitor of apoptosis appropriate credit to the original author(s) and the source, provide a link to the Creative protein. Eur. J. Neurosci. 43, 626–639 (2016). Commons license, and indicate if changes were made. The images or other third party 52. Busnelli, M., Mauri, M., Parenti, M. & Chini, B. Analysis of GPCR material in this article are included in the article’s Creative Commons license, unless dimerization using acceptor photobleaching resonance energy transfer indicated otherwise in a credit line to the material. If material is not included in the techniques. Methods Enzymol. 521, 311–327 (2013). article’s Creative Commons license and your intended use is not permitted by statutory 53. Konig, P. et al. FRET-CLSM and double-labeling indirect regulation or exceeds the permitted use, you will need to obtain permission directly from immunofluorescence to detect close association of proteins in tissue sections. the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Lab. Invest. 86, 853–864 (2006). licenses/by/4.0/. 54. Ishikawa-Ankerhold, H. C., Ankerhold, R. & Drummen, G. P. Advanced fluorescence microscopy techniques--FRAP, FLIP, FLAP, FRET and FLIM. Molecules 17, 4047–4132 (2012). © The Author(s) 2018 55. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013). NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 13 | | | http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nature Communications Springer Journals

Loading next page...
 
/lp/springer_journal/setbp1-induces-transcription-of-a-network-of-development-genes-by-lJyPnJP0Ks
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Science, Humanities and Social Sciences, multidisciplinary; Science, Humanities and Social Sciences, multidisciplinary; Science, multidisciplinary
eISSN
2041-1723
D.O.I.
10.1038/s41467-018-04462-8
Publisher site
See Article on Publisher Site

Abstract

ARTICLE DOI: 10.1038/s41467-018-04462-8 OPEN SETBP1 induces transcription of a network of development genes by acting as an epigenetic hub 1 1 1 1 1 2 Rocco Piazza , Vera Magistroni , Sara Redaelli , Mario Mauri , Luca Massimino , Alessandro Sessa , 1 3 3 1 1 Marco Peronaci , Maciej Lalowski , Rabah Soliymani , Caterina Mezzatesta , Alessandra Pirola , 2 2 4 5 6 7 Federica Banfi , Alicia Rubio , Delphine Rea , Fabio Stagno , Emilio Usala , Bruno Martino , 8 9 10 11 Leonardo Campiotti , Michele Merli , Francesco Passamonti , Francesco Onida , 12 13 14 2,15 Alessandro Morotti , Francesca Pavesi , Marco Bregni , Vania Broccoli 3 1 Marc Baumann & Carlo Gambacorti-Passerini SETBP1 variants occur as somatic mutations in several hematological malignancies such as atypical chronic myeloid leukemia and as de novo germline mutations in the Schinzel–Giedion syndrome. Here we show that SETBP1 binds to gDNA in AT-rich promoter regions, causing activation of gene expression through recruitment of a HCF1/KMT2A/PHF8 epigenetic complex. Deletion of two AT-hooks abrogates the binding of SETBP1 to gDNA and impairs target gene upregulation. Genes controlled by SETBP1 such as MECOM are significantly upregulated in leukemias containing SETBP1 mutations. Gene ontology analysis of deregu- lated SETBP1 target genes indicates that they are also key controllers of visceral organ development and brain morphogenesis. In line with these findings, in utero brain electro- poration of mutated SETBP1 causes impairment of mouse neurogenesis with a profound delay in neuronal migration. In summary, this work unveils a SETBP1 function that directly affects gene transcription and clarifies the mechanism operating in myeloid malignancies and in the Schinzel–Giedion syndrome caused by SETBP1 mutations. 1 2 Department of Medicine and Surgery, University of Milano-Bicocca and San Gerardo hospital, 20900 Monza, Italy. Stem Cell and Neurogenesis Unit, Division of Neuroscience, San Raffaele Scientific Institute, 20132 Milan, Italy. Department of Biochemistry and Developmental Biology, Faculty of Medicine, Meilahti Clinical Proteomics Core Facility, University of Helsinki, 00290 Helsinki, Finland. Service d’Hématologie Adulte, Hôpital Saint-Louis, 75010 Paris, 5 6 France. Chair and Hematology Section, Ferrarotto Hospital, AOU Policlinico, 95123 Catania, Italy. Azienda Brotzu U.O. Ematologia e CTMO, Ospedale 7 8 Businco, 09121 Cagliari, Italy. UO Ematologia Azienda Ospedaliera “BIANCHI MELACRINO MORELLI”, 89124 Reggio Calabria, Italy. Dipartimento Medicina Clinica e Sperimentale, Università Insubria, 21100 Varese, Italy. Division of Hematology, University Hospital Ospedale di Circolo e Fondazione 10 11 Macchi, 21100 Varese, Italy. Hematology, Dipartimento di Medicina Clinica e Sperimentale, University of Varese, 21100 Varese, Italy. BMT Center - Oncohematology Unit, Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, University of Milan, 20122 Milan, Italy. Department of Clinical and Biological Sciences, University of Torino, 10043 Orbassano (Torino), Italy. Hematology and Bone Marrow Transplantation Unit, IRCCS San Raffaele 14 15 Scientific Institute, 20132 Milan, Italy. Oncology Unit, ASST Valle Olona, Ospedale di Circolo di Busto Arsizio, 21052 Busto Arsizio, Italy. CNR Institute of Neuroscience, 20129 Milan, Italy. These authors contributed equally: Rocco Piazza, Vera Magistroni, Sara Redaelli. Correspondence and requests for materials should be addressed to R.P. (email: rocco.piazza@unimib.it) NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 1 | | | 1234567890():,; ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ecently, we and others demonstrated the involvement of rich regions (Fig. 1a; mean peaks A/T content: 65.8% vs. 59% for 1–9 SETBP1 mutations in several hematological malignancies . the whole human genome; p < 0.0001) in cells expressing both RIn our work, we showed that SETBP1 somatic variants occur SETBP1-G870S and SETBP1-WT. De novo motif discovery in a 4 amino acid mutational hotspot within the so called SKI- identified a SETBP1 consensus binding site (Fig. 1b; AAAATAA/ homology domain. This mutational hotspot is part of a degron T; p = 0.002) largely overlapping the AT-hook consensus motif of motif that specifies substrate recognition by the cognate SCF-β- HMGA1 (AAAATA; http://hocomoco.autosome.ru/motif/ TrCP E3 ubiquitin ligase. SETBP1 mutations cause a functional loss HMGA1_HUMAN.H10MO.D), suggesting that SETBP1 binds of the degron motif targeted by SCF-β-TrCP and responsible for the gDNA through its AT-hook domains. Querying the Catalog of short half-life of the protein. Therefore, these mutations result in an Inferred Sequence Binding Preferences (http://cisbp.ccbr. increased half-life of the mutated SETBP1 protein causing its utoronto.ca/index.php) for murine Setbp1 resulted in a very accumulation and inhibition of the PP2A phosphatase oncosup- similar sequence, with an AAT trinucleotide as the core motif pressor through the SETBP1–SET–PP2A axis . (http://cisbp.ccbr.utoronto.ca/TFreport.php?searchTF = SETBP1 mutations occurring in the same hotspot were pre- T008692_1.02), therefore suggesting that the DNA-binding motif viously found in a germline disease known as the for SETBP1/Setbp1 is conserved across evolution. Independent Schinzel–Giedion syndrome (SGS) . Despite the overlap between ligand-fishing experiments confirmed the ability of SETBP1 to the mutations present in hematological disorders and in SGS, bind to gDNA (Fig. 1c). recent data suggest that somatic SETBP1 mutations found in Peak distribution analysis revealed enrichment around pro- leukemias are more disruptive to the degron than germline var- moters (Fig. 1d; 58% and 65% of the total peak count reside at iants responsible for the onset of SGS . In SGS, germline +/−50 Kb from each transcription start site (TSS) for WT and SETBP1 mutations occur as de novo variants, causing a severe G870S, respectively). However, the binding of SETBP1 was not phenotype characterized by mental retardation associated with restricted to promoters but also present in enhancers, exonic, distorted neuronal layering , multi-organ development intronic, and intergenic regions (Fig. 1e). In line with the abnormalities, and higher than normal risk of tumors . quantitative effect of SETBP1 mutations on SETBP1 protein Although the first studies led to a reliable characterization of stability, SETBP1-G870S and SETBP1-WT ChIP-Seq experiments SETBP1 as an oncogene, several elements of SETBP1 activity revealed a global increase of SETBP1-G870S binding across all the were still unclear, in particular: (1) inhibition of PP2A phos- regions tested (Fig. 1e; Jaccard p = 0.046). phatase alone does not explain the SETBP1-dependent phenotype Peak annotation identified ectopic binding of SETBP1-G870S 10 10 of SGS , and (2) SETBP1 possesses three conserved AT-hooks , to 277 genes (Supplementary Figure 2a-c; Supplementary Data 2) therefore suggesting a role as a DNA-binding protein. resulting in a strong functional enrichment for development- Preliminary evidence that murine Setbp1 is able to bind to related biological processes such as nervous system, heart, and genomic DNA (gDNA) was initially given by Oakley and col- bone development (Fig. 1f). Notably, over one third of the leagues : by transducing murine bone marrow progenitors with binding regions lies within either evolutionary conserved regions, high titer retrovirus expressing Setbp1 followed by chromatin defined as 100 bp gDNA windows characterized by a human- immunoprecipitation (ChIP) experiments, the authors demon- mouse conservation ≥70% , or DNase I hypersensitive clusters, strated binding of Setbp1 to Hoxa9/10 promoters and upregula- thus suggesting that a dysregulation in their activation could lead tion of the two genes. Using a similar murine model, to significant functional consequences (Fig. 1g). Vishwakarma and colleagues showed binding of Setbp1 to the Runx1 promoter; this, however, was associated with down- modulation of RUNX1 expression. SETBP1 upregulates target genes at the transcriptional level. In this study, we analyze the interaction between SETBP1 and To dissect the effect of SETBP1 at the transcriptional level, we gDNA on a global, unbiased scale and demonstrate that SETBP1 generated RNA-Seq profiles for cells overexpressing SETBP1- binds DNA in adenine-thymine (AT)-rich promoter regions, G870S, SETBP1-WT, as well as an Empty control. Comparative causing activation of gene expression through the recruitment of analysis of cells overexpressing G870S vs. WT revealed a very a SET1/KMT2A (MLL1) COMPASS-like complex. similar profile (Fig. 2a, b). This finding supports the expected quantitative mechanism of action of SETBP1 mutations, as pre- 1, 3, 11 viously proposed . Results This model is further supported by the evidence that virtually SETBP1 as a DNA-binding protein. Cristobal et al. showed a all the activating SETBP1 variants identified so far fall exactly clear role for SETBP1 as a natural inhibitor of PP2A . However, within the PEST domain and >95% within 4 amino acids of the SETBP1 possesses three conserved AT-hook domains , respon- extremely short degron linear motif and that the disruption of the sible for binding to the minor groove of AT-rich gDNA regions; PEST domain causes subsequent impairment of the SETBP1−β- 1, 3, 11 thus their presence suggests a role for SETBP1 as a DNA-binding TrCP axis . To confirm this hypothesis in the context of protein. transcriptional regulation, we analyzed the intersection between To test this hypothesis, we generated isogenic 293 FLP-In cell SETBP1 peaks generated in our ChIP experiments and ENCODE lines harboring wild-type (WT) and mutated (G870S) SETBP1 in transcription factor binding sites (TFBSs). In the presence of fusion with the V5 tag. Other isogenic models, such as CRISPR- mutated SETBP1, we detected an increase in the total number of Cas9, could not be used given the absence of shared TFBS; however, the putative binding partners did not immunoprecipitation-grade anti-SETBP1 antibodies. change between SETBP1-WT and mutated lines (Supplementary The G870S variant was chosen as it is the most frequent Figure 3). These data highlight the presence of a quantitative mutation found in myelodysplastic/myeloproliferative disorders, rather than a qualitative effect for SETBP1 mutations, thus together with the D868N . The G870S line showed similar supporting the reduced degradation model. SETBP1 transcript levels but increased SETBP1 protein compared Subsequent comparative RNA-Seq analysis of SETBP1-G870S to the WT one (Supplementary Figure 1a, b), as expected . Anti- vs. Empty showed the presence of 2687 differentially expressed V5 ChIP-Seq experiments revealed a total of 3065 broad genomic genes (DEGs), 57% of which were downregulated and 43% regions as being bound by SETBP1-G870S (Supplementary upregulated (Fig. 2a, b; Supplementary Data 3). The intersection Data 1). The presence of broad peaks occurred mainly in AT- between genes bound by SETBP1 in promoter regions and DEGs 2 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE ab c d 6e-07 WT A/T content 68% *** V5-SETBP1-G870S – – + + + Empty + + – – – 3e-07 64% Oligonucleotide probe T U T U Beads WT enriched motif G870S enriched motif only 2 2 0e+00 V5 60% 1 1 V5-INPUT G870S 0 0 56% Lamin B1-INPUT 2.0e-07 52% 5.0e-08 –50 –30 0 30 50 Distance from TSS ef g 10,000 40% 100% SETBP1-G870S: GO biological processes Skeletal system development Mesenchyme development 30% 75% Stem cell development Palate development Cartilage morphogenesis 20% 50% Nervous system development Cardiac ventr. morphogenesis Organ morphogenesis 10% 25% Neurogenesis BMP signalling pathway 0 20 40 60 80 100 120 140 160 180 0% 0% –Log10 binom. Bonferroni prob. WT G870S WT G870S CGI No CGI Fig. 1 Interaction of SETBP1 with genomic DNA. a A/T content comparison between SETBP1 WT/G870S-binding regions and the reference genome. The frequency of A/T in SETBP1 target regions was compared with A/T frequency in the entire genome. Pearson's chi-squared test was used to test the significance of the difference between the two proportions. Actual numbers can be found in Supplementary Table 1. b SETBP1 consensus binding site revealed by de novo motif discovery. c Nuclear cell lysate oligonucleotide pulldown experiment. Target (T) and non-target (U) biotinylated probe oligonucleotides were designed according to ChIP-Seq data. Empty beads were used as control for non-specific binding. Pulldown was performed on nuclear extract from FLP-In SETBP1-G870S transfectants or Empty lines. Lamin B1 was used as a loading control. d Peak distribution density according to the distance from gene transcription start sites. e Peak quantitation in the different genomic regions, reflecting the position of binding sites relative to the next known gene. f Gene Ontology (GO) biological process functional enrichment of SETBP1 target genes. g Left, percentage of WT and G870S regions covered by CpG islands (CGIs), evolutionary conserved regions (ECRs), and DNase I hypersensitivity (DHS) at single-nucleotide resolution. Right, percentage of SETBP1 target genes having CpG islands within their promoter. ***p < 0.001 (false discovery rate (FDR) < 0.001) revealed 105 co-occurring being part of a nucleosome-remodeling complex involved in genes (Fig. 2c, Supplementary Data 4). Of them, 99 (94.3%; p < transcriptional activation (Fig. 3d; Supplementary Data 5). In line −6 1×10 ) were upregulated, suggesting a primary role for SETBP1 with the previous findings, correlation of SETBP1-binding as a positive inducer of gene expression (Fig. 2c). Relative regions as well as related epigenetic marks between SETBP1- quantification on a subset of target genes (SKIDA1, NFE2L2, G870S and SETBP1-WT was nearly perfect (r = 0.985 and p < PDE4D, FBXO8, CEP44, COBLL1, BMP5, ERBB4, CDKN1B)on 0.00001 for anti-V5; r = 0.873 and p < 0.00001 for anti-H3K9Ac; SETBP1-WT, SETBP1-G870S, and Empty cells by mean of Supplementary Figure 4). quantitative polymerase chain reaction (Q-PCR) confirmed the Co-immunoprecipitation (Co-IP)/proteomics experiments differential expression detected by RNA-Seq (Fig. 2d). directed against SETBP1 (Supplementary Data 6) revealed direct Functional annotation of DEGs and Gene Set Enrichment interaction of both WT and mutated SETBP1 with HCF1, a core Analyses (Fig. 2e, f) showed significant enrichment for ontologies protein of the SET1/KMT2A complex, responsible for H3K4 related to cell differentiation and tissue development, thus mono- and di-methylation . These results were confirmed by suggesting that SETBP1-mediated transcriptional deregulation independent immunoprecipitation/western blot experiments may play an important role in the onset of SGS . (Fig. 4a) and by acceptor photobleaching fluorescence resonance energy transfer (FRET) assays (Fig. 4b). In silico linear domain analysis revealed the presence of a putative HCF1-binding motif SETBP1 is part of a multiprotein epigenetic complex. AT- (HBM) occurring at position 991–994 of SETBP1 (Supplementary hook-containing proteins are often part of large chromatin 19–21 Figure 5). Deletion of this motif in WT and mutated cells remodeling complexes . ChIP-Seq profiles of a set of histone (SETBP1ΔHBM and SETBP1-G870S-ΔHBM; Supplementary marks associated with gene expression (H3K4me2, H3K4me3, Figure 6) caused a complete and specific abrogation of H3K9ac, H3K27ac, and H3K36me3) revealed peak distribution SETBP1/HCF1 interaction in both lines (Fig. 4b); conversely, enrichment around the promoter regions for all the tested marks the known SETBP1–β-TrCP interaction present only in the WT with the exception of H3K36me3, which was enriched in gene protein was not affected by the deletion (Fig. 4c). To assess bodies, as expected (Fig. 3a). Differential enrichment analysis in whether the loss of the SETBP1-HCF1 interaction could result in SETBP1-G870S vs. Empty showed a correlation between SETBP1 the impairment of the SETBP1 transcriptional machinery, we promoter occupancy and increase of H3K4me2 and H3K9ac −4 −16 analyzed by Q-PCR the expression levels of a set of upregulated (Fig. 3b, c; p = 1×10 and p = 1.2 × 10 , respectively). A genes: the deletion of HBM resulted in a complete normalization significant transcriptional upregulation was detected for the genes of the expression level for all the tested genes (Fig. 4d; MECOM p bound by SETBP1-G870S and displaying an increase in = 0.01; BMP5 p < 0.001, PDE4D p < 0.001, ERBB4 p = 0.036). H3K4me2 and H3K9ac, corroborating the hypothesis of SETBP1 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 3 | | | Genome WT G870S TSS region Enhancers Exons Introns Intergenic CGIs ECRs DHS WT G870S Number of peaks 58% 65% 66% Bits Bits Region coverage Genomic regions Genomic regions 2% 1% 34% 28% 31% 21% Number of genes with CGI 91% 9% 86% 14% ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ab c MA-plot: G870S/WT MA-plot: G870S/Empty RNA UP 165 genes 1157 genes RNA DOWN –5 –5 Empty 0 1524 117 genes 1530 genes WT G870S 1e-01 1e+01 1e+03 1e+05 1e-01 1e+01 1e+03 1e+05 PC1: 95% variance Mean expression Mean expression SETBP1 G870S Anatomical structure arrangement p <1E-5 –0.2 *** –0.4 *** –0.6 *** *** 'G870S' (positively correlated) 5.0 2.5 *** *** ** Zero cross at 12176 –2.5 *** 'Empty' (negatively correlated) *** –5.0 0 5000 10,000 15,000 20,000 25,000 30,000 Rank in ordered dataset Brain morphogenesis p <1E-5 –0.2 –0.6 EW M E WM E W M E WM E W M E WM E W M E W M E W M –0.8 SKIDA1 NFE2L2 PDE4D FBXO8 CEP44 COBLL1 BMP5 ERBB4 CDKN1B 'G870S' (positively correlated) 5.0 2.5 Zero cross at 12176 e –2.5 SETBP1-G870S: GO biological processes 'Empty' (negatively correlated) –5.0 Neg. regulation of leukocyte differentiation 0 5000 10,000 15,000 20,000 25,000 30,000 Neg. regulation of stem cell differentiation Rank in ordered dataset Granulocyte migration Myeloid leukocyte mediated immunity Enodochondral ossification Endodermal cell differentiation Mesodermal cell differentiation p <1E-5 Regulation of cardiac muscle cell differentiation Positive regulation of chondrocyte differentiation Cell morphogenesis involved in neuron differentiation –0.2 Positive regulation of osteoclast differentiation Endochondral ossification Regulation of granulocyte chemotaxis –0.4 Forebrain cell migration CNS neuron differentiation Negative regulation of osteoblast differentiation Cell differentiation involved in kidney development Cell proliferation in forebrain 'G870S' (positively correlated) 5.0 Dendrite morphogenesis CNS neuron axonogenesis 2.5 Brain morphogenesis Zero cross at 12176 Forebrain regionalization –2.5 Cranial nerve morphogenesis 'Empty' (negatively correlated) –5.0 CNS projection neuron axongenesis 0 5000 10,000 15,000 20,000 25,000 30,000 Anatomical structure arrangement Rank in ordered dataset 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Enrichment profile Hits Ranking metric scores –Log10 FDR Fig. 2 Effect of SETBP1 G870S expression at the transcriptional level. a MA plot showing the expression data of SETBP1-G870S vs. SETBP1-WT (left) and SETBP1-G870S vs. Empty cells (right) as a function of log ratios (M) and mean average gene counts. b Variance distribution of 3 Empty, 3 SETBP1-WT, and 3 SETBP1-G870S clones emphasized by Principal Component Analysis 1 (PC1) showing 95% of total variance. c Venn diagram showing the number of differentially expressed genes being directly bound by SETBP1 within their promoter region (red circle). d Q-PCR analysis of a subset of SETBP1 DEGs identified by RNA-Seq: E (Empty), W (SETBP1_wt), M (SETBP1_G870S). The housekeeping gene GUSB was used as an internal reference. Experiments were performed in triplicate; statistical analysis was performed using t-test. Error bars represent the standard error. **p < 0.01; ***p < 0.001. e Dysregulated GO biological process revealed by functional enrichment analysis of the differentially expressed genes resulting from G870S mutation. f Gene set enrichment analysis displaying three of the most enriched categories. Genes are shown as a function of the enrichment score (y axis in the upper part) and relative gene expression (x axis) Finally, a direct interaction between SETBP1 and KMT2A was germline mutations in two PHD family members, PHF8 and confirmed by Co-IP experiments (Fig. 4e). PHF6, have been identified as the cause of two X-linked mental Plant homeodomain (PHD) fingers interact with methylated retardation syndromes, namely Siderius X-linked mental retarda- 25 26 histone tails and are typically found in proteins responsible for tion and Borjeson–Forssman–Lehmann syndrome , and the epigenetic modulation of gene expression . Interestingly, somatic PHF6 mutations have been found in T cell acute 4 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | Relative mRNA expression level Log2 fold change PC2: 2% variance Ranked list Enrichment Ranked list Enrichment Ranked list Enrichment metric score metric score metric score NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE a d Relative 12 H3K4me2 12 H3K4me3 12 H3K36me3 Row min Row max 10 10 10 8 8 8 6 6 6 4 4 4 2 2 2 Approved symbol 0 0 0 SENP6 –15 Kb –9 Kb –3 Kb 3 Kb 9 Kb 15 Kb –15 Kb –9 Kb –3 Kb 3 Kb 9 Kb 15 Kb –15 Kb –9 Kb –3 Kb 3 Kb 9 Kb 15 Kb ARRDC3 NCOA2 CDKN2A 12 12 H3K9Ac H3K27Ac BMP5 10 10 8 8 KLHL1 6 6 MGAT2 4 4 CDKN1B 2 2 CDKN2C 0 0 COMMD3 –15 Kb –9 Kb –3 Kb 3 Kb 9 Kb 15 Kb –15 Kb –9 Kb –3 Kb 3 Kb 9 Kb 15 Kb LAMA2 BMI1 BMPR1B DNAJC1 30 6 *** NS NS TFPI 25 5 PCDH7 RNF217 20 4 COBLL1 TRDMT1 MEF2C 6 EDIL3 10 2 SYDE2 5 1 PDE4D SLC38A11 0 0 0 NEK7 ANK3 01 2 34 0 1 2 3 4 012 345 RBMS3 SETBP1 G870S/Empty (Log2 FC) SETBP1 G870S/Empty (Log2 FC) SETBP1 G870S/Empty (Log2 FC) NFIB 50 5 TBX18 NS *** PRKACB RNF145 40 4 DST ALX1 30 3 HGF SPAG6 20 2 FBXO8 ZSWIM2 10 1 LRRIQ1 MECOM 0 0 TRDN PTEN 012 345 0 1 2 345 NR2F1 SETBP1 G870S/Empty (Log2 FC) SETBP1 G870S/Empty (Log2 FC) PTPRC FAM172A c BAZ2B Empty RFESD SETBP1 WT ATP6V1G3 SETBP1 G870S NFE2L2 SETBP1 WT peaks RHOBTB3 SETBP1 G870S peaks FAM188A BMP5 Refseq genes G3BP1 H3K4me2 ZFYVE16 H3K4me2 G870S METAP1D H3K4me3 VMP1 H3K4me3 G870S RNF43 H3K36me3 PPP3CC H3K36me3 G870S RBM20 H3K9Ac H3K9Ac G870S SENP6 H3K27Ac H3K27Ac G870S Fig. 3 SETBP1-mediated epigenetic modulation. a Histone modification ChIP-Seq peak distribution densities plotted according to their distance from gene transcription start sites. b Epigenetic changes resulting from the presence of SETBP1-G870S expressed as function of SETBP1 differential DNA binding (ChIP-Seq G870S/Empty fold change in x axis) vs. histone modification differential enrichment (G870S/Empty fold change in y axis). ***p < 0.001 c SETBP1 ChIP-Seq coverage track and peak alignment to the hg19 reference genome are superimposed to the different histone methylation ChIP-Seq coverage tracks within the BMP5 locus. d Gene expression heatmap of the subset of SETBP1 targets harboring increased H3K4me2 and H3K9ac activation marks generated on three Empty and three SETBP1-G870S clones 27 28 lymphoblastic and acute myeloid leukemias (AMLs) . PHF8 is Taken globally, these data suggest a link between altered PHD also responsible for brain and craniofacial development in activity and SGS phenotype. zebrafish . While the role of PHF6 as an epigenetic modulator is less clear, PHF8 acts as a lysine demethylase. It can bind di- and SETBP1 epigenetic machinery requires functional AT-hooks. tri-methylated H3K4 in the context of KMT2A complexes The identification of an AAAATAA/T consensus binding motif through its N-terminal PHD finger , exerting its demethylating closely resembling that of HMGA1 suggests that SETBP1 may activity on H4 Lysine 20. ChIP against H4K20me1 revealed a bind to gDNA thanks to the presence of its AT-hook domains. To significant decrease in H4K20 mono-methylation in cells test this hypothesis, we generated new isogenic lines carrying expressing mutated SETBP1 for all the tested SETBP1 target ΔAT1 deletions of the first (SETBP1-G870S ), second (SETBP1- genes (Fig. 4f), suggesting that the SETBP1 complex possesses ΔAT2 ΔAT1,2 G870S ), and both (SETBP1-G870S ) AT-hook domains H4K20 demethylase activity. Co-IP experiments confirmed the (Supplementary Figure 7). The third AT-hook was not deleted as interaction of SETBP1 with PHF8 and PHF6, indicating that both it resides within the SET-binding domain. Combined disruption PHD members are part of the SETBP1 complex (Fig. 4g, h). of AT-hooks 1 and 2 resulted in a marked reduction of mRNA NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 5 | | | H3K9Ac G870S/Empty H3K4me2 G870S/Empty Number of sites Number of sites (Log2 FC) (Log2 FC) [log2] [log2] H3K4me3 G870S/Empty H3K27Ac G870S/Empty (Log2 FC) (Log2 FC) H3K36me3 G870S/Empty (Log2 FC) Empty1 Empty2 Empty3 G870S1 G870S2 G870S3 SETBP1 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ab c FRET: SETBP1-HCF1 FRET: SETBP1-bTrCP 130 130 Lysate IP:HCF1 IP:lgG Lysate IP:V5 120 120 WB 110 110 V5 HCF1 100 100 HCF1 V5 90 90 12 3 4 5 67 8 9 10 12 3 4 5 67 8 9 10 Frame number Frame number SETBP1-WT SETBP1 G870S SETBP1-WT SETBP1 G870S SETBP1 ΔHBM SETBP1 G870S ΔHBM SETBP1 ΔHBM SETBP1 G870S ΔHBM ef 0.10 H4K20me1 Lysate IP:V5 0.08 WB 0.06 KMT2A 0.04 V5 ** 0.02 ** *** *** * ** ** *** BMP5 NFE2L2 SKIDA1 PDE4D COBLL1 MECOM BMP5 PDE4D ERBB4 Empty SETBP1 G870S SETBP1 G870S ΔHBM Empty SETBP1 G870S gh Promoter Lysate IP:V5 Lysate IP:V5 HCF1 WB PHF8/6 PHF8 PHF6 MLL1 V5 V5 CDS AT-rich POLR2A H3K4me H4K20me ** * *** *** *** *** *** *** *** ** *** *** *** *** *** *** *** * ** MECOM BMP5 PDE4D ERBB4 MECOM BMP5 PDE4D ERBB4 Empty SETBP1 G870S ΔATH1 SETBP1 G870S SETBP1 G870S ΔATH2 SETBP1 G870S ΔATH1,2 Fig. 4 SETBP1 interacts with the SET1/KMT2A COMPASS-like complex. a Co-immunoprecipitation was performed against the HCF1 protein (left) or the V5 flag (right) and blotted with an anti-V5 or HCF1 antibody. b, c FRET analysis showing physical interaction between SETBP1 and HCF1. Positive FRET signal was recorded in both couples of HCF1 and SETBP1 WT or G870S (b, green and blue lines), conversely no FRET signal was recorded for HCF1 and SETBP1ΔHBM or G870SΔHBM (b, red and orange lines). FRET between β-TrCP and SETBP1 variants was assayed to demonstrate that ΔHBM did not modify the known SETBP1–β-TrCP interaction (c). Acceptor photobleaching was performed after the third acquired frame and indicated with gray bars in both the graphs. Bars represent the standard error of three experiments. d Relative expression of SETBP1 target genes as assessed by Q-PCR in empty (black), SETBP1-G870S (orange), and SETBP1-G870S-ΔHBM (light orange) lines. e Co-immunoprecipitation was performed against the V5 flag and blotted with an anti-KMT2A antibody. f ChIP against H4K20me1 followed by Q-PCR on a set of SETBP1 target genes performed on Empty (black bars) and SETBP1- G870S cells (orange bars). g Co-immunoprecipitation was performed against the V5 flag and blotted with anti-PHF8 and anti-PHF6 antibodies. h Proposed model for SETBP1 epigenetic network. i Relative expression of SETBP1 target genes in cells transduced with empty vector, SETBP1-G870S, or SETBP1- G870S carrying deletion of the first (ΔATH1), second (ΔATH2), and both (ΔATH1,2) AT-hooks. j ChIP against SETBP1-G870S in cells transduced with empty vector, SETBP1-G870S, or SETBP1-G870S carrying deletion of the first (ΔATH1), second (ΔATH2), or both (ΔATH1,2) AT-hooks, followed by Q- PCR on a set of SETBP1 target genes. In panels d, f, i and j, statistical analysis was performed using t-test. Bars represent the standard error of three experiments. *p < 0.05; **p < 0.01; ***p < 0.001 6 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | Empty G870S Empty G870S G870S Empty G870S Empty G870S Empty G870S Empty G870S Empty Empty G870S G870S Empty Empty G870S G870S Relative mRNA expression level Relative mRNA expression level % Donor fluorescence % Input % Input % Donor fluorescence NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE expression for all the target genes under analysis (Fig. 4i; ERBB4 p = 0.003), indicating that SETBP1 binding to gDNA MECOM p = 0.014; BMP5 p < 0.001; PDE4D p < 0.001; ERBB4- depends on the presence of functional AT-hooks (Fig. 4j). ΔAT1 ΔAT1,2 G870S p < 0.001; ERBB4-G870S p = 0.1). ChIP per- The fact that SETBP1 is commonly found in DNase I formed on target genes followed by Q-PCR confirmed the same hypersensitive cluster regions together with its ability to interact effect (MECOM p < 0.001; BMP5 p < 0.001; PDE4D p < 0.001; with an epigenetic activator complex containing H3K4 ab 293FLP 293FLP ** 0.4 ** 0.3 2000 0.2 0.1 0 0.0 Empty WT G870S Empty WT G870S 0.4 *** 0.3 0.2 0.1 MECOM Empty SETBP1 G870S Empty SETBP1 WT SETBP1 G870S SETBP1 WT peaks SETBP1 G870S peaks H3K4me2 H3K4me2 G870S H3K4me3 H3K4me3 G870S H3K36me3 H3K36me3 G870S H3K9Ac H3K9Ac G870S H3K27Ac H3K27Ac G870S MLL (GSM1897369) de 293FLP aCML patients Group CTRL MUT Group Gene *** f CDC25C 300 r = 0.921 –0.2E+00 CIT FKBP15 TYMS ATPIF1 GINS4 –0.6E+00 RSU1 CD63 TPI1 ME2 CDKN3 –1.0E+01 KIF20A DHRS1 STIL TSPAN13 –1.4E+01 HAUS1 HMOX1 KDM1B BUB1 NAGK 0 –1.8E+01 GMNN 05 10 15 MFF WT MUT WIPI1 Log2 RNA-Seq aCML patients SETBP1 MUT WT SETBP1 Gene HMOX1 DHRS1 CDC25C BUB1 CDKN3 CIT KIF20A ATPIF1 GINS4 TYMS GMNN STIL HAUS1 TPI1 TSPAN13 CD63 FKBP15 NAGK MFF ME2 RSU1 KDM1B WIPI1 Relative Row min Row max NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 7 | | | MECOM Rel. Expr. (RPKM) % Input Empty_1 Empty_2 Empty_3 G870S_1 G870S_2 G870S_3 MECOM Rel. Expr. (RPM) CMLPh-001 CMLPh-002 CMLPh-004 CMLPh-006 CMLPh-007 CMLPh-008 CMLPh-009 CMLPh-010 CMLPh-011 CMLPh-012 CMLPh-014 CMLPh-016 CMLPh-017 CMLPh-018 CMLPh-020 CMLPh-021 MECOM Rel. Expr. (AU) CMLPh-024 CMLPh-025 Log2 Q-PCR CMLPh-028 CMLPh-034 CMLPh-035 CMLPh-003 CMLPh-005 CMLPh-013 CMLPh-015 CMLPh-019 CMLPh-023 CMLPh-026 CMLPh-029 CMLPh-030 CMLPh-036 CMLPh-037 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 methyltransferase activity suggests that SETBP1 should be able to and myeloid differentiation (Supplementary Data 10). In control gene expression through modulation of chromatin accordance with its previously observed transcriptional activity , accessibility. To test this hypothesis, we generated ATAC-Seq we show here that MECOM target genes are also differentially (assay for transposase-accessible chromatin using sequencing) expressed (p < 0.0001, Fig. 5d) in cells expressing SETBP1-G870S. experiments on the isogenic 293 FLP-In Empty, SETBP1-WT, To confirm these results in fresh leukemic cells, RNA-Seq and Q- and SETBP1-G870S cell lines. We assessed the ATAC-Seq signal PCR analyses were performed in 32 atypical chronic myeloid in a region comprised between −2000 bp and +10,000 bp from leukemia (aCML) cases (11 positive and 21 negative for SETBP1 the TSS of all the human genes. In line with the expected high somatic mutations). The results show that SETBP1-positive accessibility of TSS regions, ATAC-Seq analysis revealed a peak at patients express higher levels of MECOM (Fig. 5e, f; p = TSS in all the lines under test (Supplementary Figure 8a, b; 0.0002) and MECOM target genes (Fig. 5g; p < 0.0001), with Supplementary Data 7–9). Subsequently, we investigated whether excellent correlation between RNA-Seq and Q-PCR data (Fig. 5f). a correlation could be found between the intensity of the relative ATAC-Seq signal in SETBP1-G870S vs. Empty or SETBP1-WT Mutated SETBP1 delays neuronal migration to the cortex. vs. Empty at TSS and gene bodies and the relative gene Upon their general developmental delay, Schinzel–Giedion expression, as assessed by RNA-Seq in the same lines. Analysis patients present heavy neurological impairment characterized by at global gene level performed throughout a set of iterations (50) microcephaly, altered neuronal layering, underdeveloped corpus characterized by progressively more stringent differential expres- callosum, ventriculomegaly, cortical atrophy, or dysplasia that sion fold-change filtering criteria revealed a significant linear 11, 12 cause seizures and severe mental retardation . Additionally, relationship between differential ATAC-Seq signal at TSS and the mRNA of SETBP1 gene is expressed in both germinal (high RNA expression for both SETBP1-G870S and SETBP1-WT level of expression) and differentiated area (low level) of different (Supplementary Figure 8c). Taken globally, these data indicate brain regions including cerebral cortex (Supplementary Fig- that genes that are significantly upregulated in the presence of ure 10). To gain insight into the function of the mutated form of either WT or mutated SETBP1 exhibit an increased chromosomal SETBP1 in nervous system development, we transduced radial accessibility in their TSS regions as well as in gene bodies. glial progenitors of cerebral cortices of E13.5 mouse embryos with expression plasmids encoding SETBP1-WT, SETBP1-G870S, ΔAT1,2 MECOM is a direct transcriptional target of SETBP1. In 2013, SETBP1-G870S , and SETBP1-G870S-ΔHBM using an in Makishima and colleagues reported that the presence of mutated utero electroporation system (Fig. 6a) . The transduced cells and SETBP1 was associated with upregulation of MECOM expres- their progeny can be easily traced along their development thanks sion ; however, a clear mechanistic explanation of this finding to the green fluorescent protein (GFP) expression. Two days after was missing. MECOM, located on the long arm of chromosome 3, the procedure, almost the totality of the SETBP1-G870S electro- is a transcription factor able to recruit both coactivators and porated cells remained stacked in the deep part of the developing corepressors . It is expressed in hematopoietic stem cells, playing cortical wall while many control cells (transduced with GFP only) an important role in hematopoiesis and in hematopoietic stem were already migrating in the outward cortical region where post- cell self-renewal . MECOM is often overexpressed, as a result of mitotic neurons reside (Fig. 6b; Supplementary Movie 1 and 2). chromosomal translocations, in myelodysplastic syndromes and In addition, in SETBP1-G870S-overexpressing tissue, the apical in approximately 10% of AML cases, being a strong negative domain of the cortical layer where the neural progenitors are 33, 34 prognostic marker for therapy response and survival . located was severely disorganized as shown by aberrant locali- In line with Makishima’s report, differential expression RNA- zation of TBR2+ intermediate progenitors that lost their typical Seq (Fig. 5a; p = 0.04 and p = 0.08 for SETBP1-WT and SETBP1- strip arrangement in the basal part of the ventricular zone G870S vs. Empty, respectively) and Q-PCR analyses (Fig. 5b; p = (Fig. 6c). Five days after surgery, many control GFP+ neurons 0.01 and p = 0.001 for SETBP1-WT and SETBP1-G870S vs. were detected in the mature cerebral mantel zone, contributing to Empty, respectively) showed that MECOM is upregulated in 293 the fiber tract of the corpus callosum (Fig. 6d, arrow), while the FLP-In cells expressing WT or mutated SETBP1. A similar vast majority of cells electroporated with SETBP1-G870S were upregulation was detected in the human myeloid TF-1 cell line incorrectly located at the deepest cortical tissue (65 vs. 32% of the transduced with SETBP1-G870S (Supplementary Figure 9; p = control condition; t-test p < 0.001 in bins 1–3) with only few 0.0001). ChIP-Seq data showed that both SETBP1-WT and neurons projecting to the contralateral hemisphere through the SETBP1-G870S bind to the MECOM promoter, highlighting corpus callosum (Fig. 6d). In line with the proposed quantitative MECOM as a direct target of SETBP1 transcriptional activity model for SETBP1 mutations, overexpression of SETBP1-WT (Fig. 5c). MECOM modulates the expression of a significant showed similar, albeit reduced, effects (Fig. 6d). Overexpression number of genes involved in hematopoietic stem cell proliferation of mutated SETBP1 carrying either AT-hooks 1 and 2 double Fig. 5 Analysis of MECOM expression and downstream targets. a Box plot showing the RNA-Seq differential expression analysis of MECOM in the 293 FLP- In Empty, SETBP1-WT, and SETBP1-G870S cell models. The top and bottom of each box represent the first and third quartile, respectively; the internal line represents the median; the dot represents the mean. Experiments were performed in triplicate. b Q-PCR analysis of MECOM expression in the 293 FLP-In Empty, SETBP1-WT, and SETBP1-G870S cell models. The top and bottom of each box represent the first and third quartile, respectively; the internal line represents the median; the small square represents the mean. Experiments were performed in triplicate; statistical analysis was performed using t-test. c SETBP1 ChIP-Seq coverage track and peak alignment to the hg19 reference genome (blue tracks) are superimposed to the different histone methylation tracks and KMT2A (MLL) ChIP-Seq coverage track within the MECOM locus. The boxed histogram represents an independent ChIP experiment performed against the V5 flag in the FLP-In cells followed by a Q-PCR directed against the predicted SETBP1-G870S-binding locus on the MECOM promoter. ChIP was performed in triplicate; statistical analysis was performed using t-test. d Gene expression heatmap of MECOM target genes in three Empty/SETBP1-G870S FLP-In clones. e Differential MECOM expression as read counts per million of mapped reads (RPM) in 32 aCML patients carrying WT (21) or mutated (11) SETBP1. The top and bottom of the box represent the first and third quartile, respectively; the internal line represents the median. Statistical analysis was performed using t-test. f Linear correlation of MECOM expression as assessed by RNA-Seq (x axis) and Q-PCR (y axis). r represents the Pearson linear correlation coefficient. g Gene expression heatmap of MECOM target genes in 32 aCML patients carrying WT (21) or mutated SETBP1 (11). Error bars represent the standard error. *p < 0.05; **p < 0.01; ***p < 0.001 8 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** **** ** *** * ** **** **** *** **** **** * **** **** **** **** **** **** * **** **** **** **** *** NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE ab c IEU E13.5 - analysis E15.5 SETBP1 SETBP1 GFP GFP G870S G870S GFP or SETBP1 (WT or mutants) + GFP v v v d e DAPI/GFP IEU E13.5 - analysis E18.5 Bin: SETBP1 GFP p G870S ns cp ns ns ns ns iz ns vz/svz ns ns 0.0 0.1 0.2 0.3 0.4 0.5 Percentage of GFP cells Fig. 6 In utero electroporation of GFP and SETBP1-G870S. a Schematic representation of the electroporation procedure. b Snapshots from 3D reconstruction resulting from 2-photon microscopy on either GFP- or SETBP1-electroporated cortices (2 days) after tissue clarification (X-Clarity system) showed defects in radial migration of the SETBP1-G870S misexpressing cells. p pial side, v ventricular side. The GFP signal in the pial membrane (white arrows) and along the thickness of the SETBP1-G870S tissue (red arrow) is due to the basal processes of the GFP+ radial glia cells located in the deepest part of the organ. c Immunohistochemistry for GFP and TBR2 on coronal section of 2 days electroporated tissues. d Five-day electroporated cortices and quantification of the migration of the GFP+ cells from apical (bin #1) to pial part of the organ (bin #5); arrow indicates GFP+ corpus callosum. Statistical analysis was performed using two-way ANOVA; error bars represent standard error. *p < 0.05; **p < 0.01; ***p < 0.001; ***p<0.0001. e Immunohistochemistry for GFP and SATB2 marker on coronal section of 5 days electroporated tissues. In the insets on the right, the images of SETBP1- G870S DAPI (up), GFP, SATB2, and merge of GFP/SATB2 (bottom) are shown. cp cortical plate, iz intermediate zone, svz subventricular zone, vz ventricular zone. Bars: c, e: 100 μm, d: 250 μm ΔAT1,2 knockout (SETBP1-G870S ) or deletion of the HCF1- modulator, showing that SETBP1 interacts with gDNA through binding domain (SETBP1ΔHBM) largely restored the normal its AT-hook domains, forming a multiprotein complex including migration pattern (Fig. 6d), indicating that the presence of HCF1, KMT2A, PHF8, and PHF6, which results in increased functional domains responsible for DNA interaction or multi- chromatin accessibility, as assessed by ATAC-Seq (Supplemen- protein complex recruitment is required to modulate neuronal tary Figure 8), and transcriptional activation. The altered tran- migration. Despite the impaired migration, the cells expressing scription caused by the increased levels of SETBP1-G870S was SETBP1-G870S retained the capability of differentiating into also shown to be involved in the pathogenesis of SGS, aCML, and neurons as confirmed by the expression of the mature neuronal related myeloid malignancies, as revealed by in utero electro- marker SATB2 (Fig. 6e). These findings demonstrate that the poration of SETBP1-G870S in ventricular central nervous system increase in SETBP1 levels during brain morphogenesis has a high cells and by the analysis of aCML samples. impact on the dynamics of both neuronal proliferation and Oakley et al. previously showed that, in murine bone marrow, migration that can be responsible for the neuroanatomical defects progenitors transduced with high-titer retrovirus-expressing described in SGS. Setbp1 are able to bind to Hoxa9/10 and Myb promoters and to 13, 37 14 upregulate these genes . Vishwakarma and colleagues , using a similar murine model, demonstrated binding of Setbp1 to the Runx1 promoter , which, however, was associated with Runx1 Discussion downmodulation. Interestingly, in our human 293 FLP-In model This work describes the results of a next-generation sequencing- we found a similar downregulation (Supplementary Figure 11a). based, unbiased approach performed to investigate the function ChIP experiments, however, demonstrated only a very weak of SETBP1 and its pathological variant SETBP1-G870S. We binding of SETBP1 to the human RUNX1 promoter (Supple- demonstrate the ability of human SETBP1 to directly bind gDNA mentary Figure 11b), likely due to the expression of SETBP1 and that the binding occurs preferentially but not exclusively in being much lower in our model than in traditional high-titer gene promoter regions, as SETBP1 has also been detected in retroviral transduction systems or to differences in species spe- enhancers, exonic, intronic, and intergenic regions. In this work, cificity. To test this hypothesis, we repeated ChIP experiments we focused on the potential role of SETBP1 as a transcriptional NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 9 | | | IEU E13.5 - analysis E18.5 SETBP1-G870S SETBP1-WT GFP SETBP1-G870S SETBP1-G870S IEU E13.5 - analysis E15.5 ΔHBM ΔATH1,2 GFP/TBR2 GFP/SATB2 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 using a transient transfection model, achieving a 4.2- and 6.9-fold constitute the only genetic alteration in SGS, somatic SETBP1 increase in SETBP1-WT and G870S expression, respectively, mutations are typically present in conjunction with several other when compared with FLP-In (Supplementary Figure 11c). somatic events in myeloid neoplasms, in which SETBP1 muta- Indeed, ChIP experiments performed using transient SETBP1 tions usually represent late events . transfectants showed a significant increase in the binding to In summary, we showed here that SETBP1 is able to recruit a RUNX1 promoter (Supplementary Figure 11d). This, however, HCF1/KMT2A/PHF8/PHF6 transcriptional activator complex, was accompanied by an increase in RUNX1 expression (Supple- thus identifying SETBP1 as a transcriptional activator, beyond its mentary Figure 11e), confirming that SETBP1 promotes upre- original SET-binding activity. Further studies will be required to gulation of gene expression and suggesting that RUNX1 gain further insights into this complex network and to dissect the downmodulation is not a direct effect of SETBP1. Further studies functional role of SETBP1–gDNA interaction in intergenic will be required to assess whether the differences between our regions. model and the murine model developed by Vishwakarma and colleagues are species specific. DEG analysis between SETBP1- Methods G870S and Empty cells in our FLP-In model and in aCML Patients. Diagnosis of aCML and related diseases was performed according to the patients (11 positive and 21 negative for SETBP1 somatic muta- World Health Organization 2008 classification. All patients provided written informed consent, which was approved by the institutional ethics committee. This tions) failed to reveal significant changes in the expression of study was conducted in accordance with the Declaration of Helsinki. RUNX1, HOXA9/10,or MYB (Supplementary Figure 12), corro- borating the hypothesis that the effect of SETBP1 as a tran- Cell-lines. 293T, TF-1, and the 293 FLP-In™ cell-lines were purchased from ATCC scription factor is species/tissue specific. (Manassas, VA, USA), DSMZ (Braunschweig, Germany), and Thermo Fisher In line with evidence indicating that KMT2A has strong H3 Scientific (Waltham, MA, USA), respectively, and maintained following the mono- and di-methylation but very weak tri-methylation activ- manufacturers' instructions. ity , no significant H3K4me3 enrichment could be found in promoters occupied by SETBP1, although the intersection Plasmids and transfections. Stable 293 FLP-In Empty, 293 FLP-In SETBP1wt, between SETBP1 promoter occupancy and transcriptome analysis and 293 FLP-In SETBP1-G870S cell lines were prepared cotransfecting pOG44 together with the increase in H3K9Ac and the direct interaction (Thermo Fisher Scientific) and pFRT-SETBP1 vectors with Fugene6 reagent and −1 were maintained in standard medium with 100 μgml Hygromycin. Stable TF-1 of SETBP1 with PHF8 demethylase reveal that SETBP1 is part of cell lines were retrovirally infected using phoenix packaging cells transfected with a transcriptional activator complex. The exact explanation of this 10 μg of MIGR1-SETBP1 Gly870Ser or WT or with empty MIGR1 vector using complex histone pattern is still unclear, given also our limited FuGENE6 (Promega); retroviruses were collected after 3 days of culture. Transient knowledge about the functions of the H3K4me2 mark. Never- 293T transfectants were prepared using the pcDNA6.2-SETBP1wt, pcDNA6.2- G870S, or empty vectors with Fugene6 reagent (Promega, Madison, WI, USA) theless, several lines of evidence suggest that H3K4me2 is strongly 38, 39 following standard protocols. The deletion of the HBM site (aa991-994; enriched in lineage-specific gene promoters . The abundance NM_015559.2) within the SETBP1 coding sequence was performed with the fol- of genes associated with multi-organ development among those lowing primers SETBP1_HCF1del_for (CAGCATTTTTCGGATTAATTTTC regulated by SETBP1 is particularly interesting in the context of CGGTGCCATATATCCAGTATG) and SETBP1_HCF1del_rev (CATACTGGAT ATATGGCACCGGAAAATTAATCCGAAAAATGCTG); the deletion of the AT- SGS, whose hallmark is the presence of multi-organ development hooks 1 and 2 were performed using the following primers: SETBP1_ATH1del_for abnormalities. Indeed, analysis of the Gene Ontologies associated (CAGTCTTACTGTGATCACTCCACTCACAGTCGAGACGATTCATG), with DEGs in SETBP1-G870S cells highlights the presence of a SETBP1 ATH1del_rev (CATGAATCGTCTCGACTGTGAGTGGAGTGATCACA much stronger association with SGS than with myeloid malig- GTAAGACTG), SETBP1_ATH2del_for (GTAGGACTTCAGACTTGAA- GACCATGACAAAGGTGCC), and SETBP1 ATH2del_rev (GGCACCTTTGTC nancies, despite the fact that SETBP1 somatic mutations are 1–9 ATGGTCTTCAAGTCTGAAGTCCTAC) as previously described . pCGN- detected in a large number of clonal myeloid disorders .In HCF1-fl was a gift from Winship Herr (Addgene plasmid #53309). aCML and related malignancies, the hyperactivation of the SETBP1–SET axis caused by the stabilization of mutated SETBP1 ChIP sequencing (ChIP-Seq). ChIP was performed as previously reported (GEO protein and leading to the inhibition of the PP2A oncosuppressor accession number: GSE86335). Briefly, proteins were crosslinked with 0.4% for- probably plays a major functional role in driving the leukemic maldehyde and cells were lysed. Chromatin was fragmented with a Bioruptor phenotype, while the interaction with gDNA and the subsequent sonicator system (Diagenode, SA, USA) and subsequently immunoprecipitated with H3K4me3 (ab8580, Abcam, UK), H3K4me2 (C15410035C, Diagenode), modulation of gene expression likely plays a critical role in the H3K36me3 (Ab9050, Abcam), H3K27Ac (Ab4729, Abcam), H3K9Ac (39137, onset of SGS, in which the SETBP1-mediated inhibition of PP2A Active Motif, Carlsbad, CA, USA), and H4K20me1 (Mab147-010, Diagenode) is probably insufficient to recapitulate the complex phenotype of antibodies or anti-V5 agarose beads (Sigma-Aldrich). After immunoprecipitation, the SGS syndrome . However, the transcriptional activity of DNA was purified and libraries were prepared for sequencing following the Illu- SETBP1 likely contributes to the leukemic phenotype too. In mina ChIP-Seq protocol (TrueSeq ChIP library prep kit IP-202-1012) with an Illumina HiSeq2500 in single read mode (Galseq, Monza, Italy). Validation of 2013, Makishima and colleagues reported that cases with SETBP1 ChIP-Seq data was performed amplifying the immunoprecipitated DNA with Sybr- mutations were characterized by high expression of the MECOM Green Q-PCR. Input was used as a loading control. Primer sequences were: 3, 40 oncogene . Here we show that MECOM is a direct target of BMP5_Fw (CAACCCTGCTGGGAAAGAAGAG), BMP5_Rw (TCATCAAGCT SETBP1 transcriptional activity, which explains the original AACTTAGGCACAAC), NFE2L2_Fw (AACCAGAAGAATACAATCCCAATG), NFE2L2_Rw (AAGAAGTTTCTGCTCATCCTTTGTAG), PDE4D_Fw (CCTT observation. Interestingly, Goyama and colleagues previously GAGCCAACCTTCTCCTTC), PDE4D_Rw (CACCCAAAGACATGACCAA showed that SETBP1 is one of the genes whose expression is most CCTC), SKIDA1_Fw (TTCAAGTATCACGTTACTGTTTGC), SKIDA1_Rw strongly reduced in MECOM−/− hematopoietic cells , sug- (GTCACTTATTCAGCCACGCAGAC), COBLL1_Fw (TCTAATTGGTGGCAG gesting that SETBP1 is also one of the MECOM downstream GTTTAAGC), COBLL1_Rw (TGTCTGTCAGGTGTAAAGAATCATC), ERBB4_Fw (ACAAACTCCTCCAAACTGCTACTG), ERBB4_Rw (GTGATCCA targets. Taken together, these findings suggest the presence of a TTGGAAACTGTAAATGC), RUNX1P2_Fw (CCTATGCAAACGAGCTGAGG), positive feedback loop occurring between MECOM and SETBP1, RUNX1P2_Rw(GCTCTATGAATGAGAGTGCCTG), MECOM_Fw whose biological importance and effects will require further stu- (CTCCCAAATGTCTTAATCGTGTCG), and MECOM_Rw dies. These findings highlight a potentially critical role for this (TTCGGACCCTTTGGCTAGATTGTG). ChIP-Seq analyses were performed using MACS v. 1.4.2 using the –no model transcriptional complex and its downstream effectors in the parameter to skip the model building step. The ratio of the intersection and the oncogenesis of SETBP1-positive myeloid disorders, therefore similarity of the two genomic interval sets was calculated via bedtools Jaccard suggesting that a strict dichotomous model is too simplistic to v2.26 statistics . fully recapitulate and explain both phenotypes. It is also impor- Fold enrichment of either SETBP1 or histone methylation ChIP-Seq tant to note that, while de novo germline SETBP1 mutations experiments in G870S or WT background were calculated with MACS using 10 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE narrow or broad (SETBP1) peak calling parameters . Transcription factor Co-immunoprecipitation. One hundred million HEK 293T cells were transiently association strength and relative fold change were calculated as previously transfected with pcDNA 6.2 V5-SETBP1-WT, SETBP1 G870S, or with the empty reported . Downstream statistics, namely, multivariate analysis with linear vector as control. Forty eight hours after transfection, cells were collected, washed correlation assumptions, were performed with the IBM SPSS statistical package. twice with cold PBS, lysed in Buffer 1 (Pipes, pH 8.0, 5 mM, KCl 85 mM, NP-40 0.5%), supplemented with protease inhibitors (Halt™ Protease Inhibitor Cocktail, Thermo Fisher Scientific), kept for 10 min on ice, homogenized with douncer ATAC-Seq. Cells (100,000/sample) were washed once in cold phosphate-buffered homogenizer (10 hits), and centrifuged at 700 × g for 10 min. The pellet was then saline (PBS) 1×, spun at 800 × g for 5’ at 4 °C and resuspended in cold PBS in the resuspended in Buffer 2 (Tris HCl, pH 8.0, 50 mM, sodium dodecyl sulfate 0.1%, presence of proteasome inhibitors and incubated on ice for 10’. Cells were cen- deoxicholate 0.5%) supplemented with protease inhibitors, sonicated with Bior- trifuged at 800 × g at 4 °C for 10’ and resuspended in: 2× TD buffer (25 µl), Tn5 uptor Next Gen (Diagenode) (5 cycles 30 s ON, +30 s OFF) to promote gDNA Transposase (2.5 µl; Illumina), and water (up to 50 µl). The sample was incubated disruption and centrifuged at 18,000 × g for 10 min. The supernatant representing at 37 °C for 30’ and purified using the SPRI AMPure XP beads. Post-tagmentation the nuclear fraction was quantified with Bradford assay and a total of 1 µg of amplification was performed using Nextera primers (Illumina) and Herculase II protein was loaded on 100 µl V5-agarose beads (Sigma-Aldrich) and incubated polymerase (Agilent) using a standard protocol. Quality of the ATAC-Seq libraries under rotation overnight at 4 °C. Beads were then washed three times with PBS was assessed using a Tape Station (Agilent) and by agar gel. Quantification was +protease inhibitors, and elution was performed with 7 M Urea, 2 M Thiourea, performed using a QuBit Fluorometer (ThermoFisher). and 4% CHAPS for subsequent mass spectrometric analysis or with Laemmli buffer for western blot analysis. All chemicals were purchased from Sigma Aldrich. ATAC-Seq analysis. ATAC-Seq fastq files were initially aligned to the hg38 48 Immunoblot analysis. Primary antibodies were V5 (ab27671 330 Abcam, Cam- human reference genome using BWA with standard parameters. BAM files were bridge, UK; dilution 1:2000), HCF1 (A301-399A Bethyl Laboratories, Inc., Mon- sorted and indexed using Samtools . tgomery, TX, USA; dilution 1:1000), MLL1 (14689 Cell Signaling Technology, BAM files from individual replicates were initially merged together and Danvers, MA, USA; dilution 1:1000), PHF8 (A301-772A Bethyl Laboratories, Inc., subsequently processed using our ATAC-Seq tool. Briefly, all the gene start Montgomery, TX, USA; dilution 1:500), and PHF6 (A301-451A Bethyl Labora- positions, gene end positions, chromosome, TSS, and gene strands were annotated tories, Inc., Montgomery, TX, USA; dilution 1:500). Secondary antibody was anti- (Gencode24). Coverage counts at single-nucleotide resolution were generated for mouse anti-rabbit horseradish peroxidase conjugated (Biorad, Hercules, CA, USA). each Gene/TSS from basesBeforeTSS (2000) to basesAfterTSS (10,000). Coverage at The uncropped scan of western blots related to PHF6 immunoprecipitation are each position was then normalized using the total coverage of each input BAM. To shown in Supplementary Figure 13 as an example. plot ATAC-Seq heatmaps, normalized coverage data were binned (binSize = 200 bp) and sorted in decreasing order according to the intensity of the ATAC signal throughout the entire region (sum of the binned signals from basesBeforeTSS to Proteomics data analysis. The protein samples were digested in Amicon Ultra-0.5 basesAfterTSS). To plot line graphs, normalized counts were summed at individual centrifugal filters using modified FASP method . The peptides were separated with bins across the entire gene set from basesBeforeTSS to basesAfterTSS. Final binned the nanoAcquity UPLC system (Waters) equipped with a 5-µm Symmetry C18 counts were then normalized by the gene set size. trapping column, 180 µm×20 mm, reverse-phase (Waters), followed by an analy- To test for the presence of a correlation between chromatin accessibility and tical 1.7-µm, 75 µm×250 mm BEH-130 C18 reversed-phase column (Waters), in a RNA expression, ATAC-Seq and RNA-Seq data generated from the same lines single-pump trapping mode. The parameters of the HD-MSE runs were described were compared using the following approach: normalized ATAC-Seq signal data previously . Protein identifications were performed with ProteinLynx Global generated for a region comprised between basesBeforeTSS (1500) to basesAfterTSS Server (PLGS v3.0) as described . Database searches were carried out against (5000) for each gene in G870S and Empty lines were initially calculated. A G870S/ UniProt human protein database (release_07072015, 71907 entries) with Ion Empty coverage ratio was then calculated throughout the entire gene set. In Accounting algorithm and using the following parameters: peptide and fragment parallel, normalized G870S/Empty RNA-Seq Log2 fold-change expression ratio tolerance: automatic, maximum protein mass: 500 kDa, minimum fragment ions was computed. Log2 fold-change data were filtered in Log2_Threshold_Cycles (50) matches per protein: 7, minimum fragment ions matches per peptide: 3, minimum iterations with progressively more stringent criteria, by increasing the Log2 fold- peptide matches per protein: 1, primary digest reagent: trypsin, missed cleavages change filter from Log2_Threshold_Start (0) at Log2_Threshold_Step (+0.1 increase allowed: 2, fixed modification: carbamidomethylation C, variable modifications: for each Log2_Threshold_Cycle) steps. For each iteration, a Pearson correlation was deamidation (N, Q), oxidation of Methionine (M) and FDR < 4%. generated between normalized ATAC-Seq ratios and filtered RNA-Seq fold changes. Immunofluorescence microscopy. Transfected HEK 293 cells, seeded on glass −1 coverslips coated with poly-D-lysine (0.1 mg ml ), were washed twice with PBS and fixed for 10 min at 25 °C with 4% (w/v) p-formaldehyde in 0.12 M sodium Code availability. Code of the ATAC-Seq tool and of the Epigenetic Marks phosphate buffer, pH 7.4, incubated for 1 h with primary antibodies in GDB buffer Correlation tool will be made available upon request. [0.02 M sodium phosphate buffer, pH 7.4, containing 0.45 M NaCl, 0.2% (w/v) bovine gelatin] followed by staining with conjugated secondary anti IgG antibodies for 1 h. After two washes with PBS, coverslips were mounted on glass slides with a Epigenetic marks correlation tool. Anti-V5 and anti-H3K9Ac 293T-SETBP1-WT 90% (v/v) glycerol/PBS solution. and G870S coverage files were used as input. Coverage was binned (BigWig) in 1000 bp regions for all the ChIP-Seq experiments. Individual bins for all the regions across the entire genome were paired and used for Pearson correlation analysis as FRET. FRET measurements were performed with the laser-induced acceptor 52, 53 well as for XY plots. photobleaching method . In our experimental set-up, FRET couples analyzed were made up of transiently transfected GFP-fused SETBP1 WT or mutated forms in combination with transiently transfected Flag-tagged β-TrCP or endogenous Oligonucleotide pulldown assay. Biotinylated oligonucleotides were synthesized HCF1. Forty eight hours after transfection, HEK 293 cells were labeled with proper by Metabion International AG (Steinkirchen, Germany). Target and non-target primary and secondary antibodies and imaged for FRET analysis. Used FRET (unrelated) oligonucleotides were designed according to ChIP-Seq from SETBP1 couples were GFP signal as donor fluorochrome and Alexa Fluor 555-conjugated bound (target-T, chr6: 55,742,037-55,742,136; hg19) and unbound (unrelated-U, secondary antibodies as acceptor fluorochrome . Briefly, three images were cap- chr6:55,775,477-55,775,576; hg19) regions. In the pulldown experiment, 25 µl of tured before bleaching in the 488 nm and the 561 nm channels using the line-by- Pierce-streptavidin magnetic beads (Thermo Fisher Scientific) were washed using a line sequential mode without any averaging steps to reduce basal bleaching. magnetic stand and resuspended in 200 µl of B/W buffer (10 mM Tris HCl, pH 7.5, Bleaching of the acceptor was performed within region of interest (ROI) identified 1 mM EDTA, 2 M NaCl). Each double-stranded biotinylated DNA probe (200 nM) in the nuclear areas with a positive colocalization between the FRET couples using was bound to the beads for 20 min at room temperature (RT) on a rotating device. 30 pulses of a full power 20 mW 561-nm laser line (each pulse 1.28 µs/pixel). After One sample without probe was used as a control for unspecific binding. Beads were bleaching, seven images were acquired in the same channels without any time delay washed with TE buffer and resuspended in 100 µl of BS/THES buffer (22 mM Tris to obtain a full curve. The number of bleaching steps, laser intensity, and acqui- HCl, pH 7.5, 4.4 mM EDTA, 8.9% Sucrose, 62 mM NaCl, 10 mM Hepes, 5 mM sition parameters were held constant throughout each experiment. FRET signal CaCl , 50 mM KCl, 12% Glycerol) supplemented with HALT protease inhibitor was quantified by measuring the average intensities of ROIs in the donor and cocktail (Thermo Fisher Scientific). In all, 15 µg of nuclear extract from 293 FLP-In acceptor fluorochrome channels before and after bleaching using the ImageJ transfectants was diluted in 400 µl of BS/THES buffer and, after collecting 30 µl as software (http://rsbweb.nih.gov/ij/). To determine any change of fluorescence INPUT fraction, was added to the beads together with 1 µg of non-specific DNA intensities not due to FRET occurring during the measurements, a distinct competitor LightShift Poly(dI-dC) (Thermo Scientific). Samples were incubated at unbleached sentinel ROI of approximately the same size of the bleached ROI was 4 °C for 30 min on a rotating device. Beads were then washed four times with BS/ measured in parallel, and all the results were normalized according to the back- THES buffer supplemented with 1 µg of Poly(dI-dC). Final wash was done with ground bleaching recorded in the sentinel. Proper controls were performed to 500 µl of buffer (100 mM NaCl, 25 mM Tris HCl) after incubation for 1 min on a verify that no artefacts were generated in the emission spectra throughout the rotating device at RT. Elution of bound proteins was performed with 50 µl of experimental set-up due to sample overheating. Twenty measurements from three Laemmli Buffer. Lamin-B1 was used as a loading control. different experiments were performed for each experimental condition. NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 11 | | | ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 RNA-Seq. RNA libraries were generated starting from 1 µg of total RNA extracted References from 5 × 10 cells using TRIzol (Invitrogen, Life Technologies). RNA quality was 1. Piazza, R. et al. Recurrent SETBP1 mutations in atypical chronic myeloid assessed by using a Tape Station instrument (Agilent). To avoid over- leukemia. Nat. Genet. 45,18–24 (2013). representation of 3′-ends, only high-quality RNA with a RNA Integrity Number ≥8 2. Sakaguchi, H. et al. Exome sequencing identifies secondary mutations of was used. RNA was processed according to the TruSeq Stranded mRNA Library SETBP1 and JAK3 in juvenile myelomonocytic leukemia. Nat. Genet. 45, Prep Kit protocol. The libraries were sequenced on an Illumina HiSeq 3000 with 76 937–941 (2013). bp paired-end reads using Illumina TruSeq technology (GEO accession number: 3. Makishima, H. et al. Somatic SETBP1 mutations in myeloid malignancies. GSE86335). Nat. Genet. 45, 942–946 (2013). Image processing and basecall was performed using the Illumina Real Time 4. Meggendorfer, M. et al. SETBP1 mutations occur in 9% of MDS/MPN and in Analysis Software. Fastq files were aligned to the human genome (GRCh38/hg39) 4% of MPN cases and are strongly associated with atypical CML, monosomy by using STAR , a splice junction mapper for RNA-Seq data, together with the 7, isochromosome i(17)(q10), ASXL1 and CBL mutations. Leukemia 27, corresponding splice junctions Ensembl GTF annotation, using the following 1852–1860 (2013). parameters: 5. Inoue, D. et al. SETBP1 mutations drive leukemic transformation in ASXL1- --runThreadN 8 --outReadsUnmapped Fastx --outFilterType BySJout mutated MDS. Leukemia 29, 847–857 (2015). --outSAMattributes NH HI AS nM MD --outFilterMultimapNmax 20 6. Fabiani, E. et al. SETBP1 mutations in 106 patients with therapy-related --alignSJoverhangMin 8 --alignSJDBoverhangMin 1 --outFilterMismatchNmax myeloid neoplasms. Haematologica 99, e152–153 (2014). 999 --outFilterMismatchNoverLmax 0.04 --alignIntronMin 20 --alignIntronMax 7. Patnaik, M. M. et al. ASXL1 and SETBP1 mutations and their prognostic 1000000 --alignMatesGapMax 1000000 --alignTranscriptsPerReadNmax 100000 contribution in chronic myelomonocytic leukemia: a two-center study of 466 --quantMode TranscriptomeSAM GeneCounts --limitBAMsortRAM 16620578182 patients. Leukemia 28, 2206–2212 (2014). --outSAMtype BAM SortedByCoordinate --chimSegmentMin 20 8. Thol, F. et al. SETBP1 mutation analysis in 944 patients with MDS and AML. --chimJunctionOverhangMin 10. Leukemia 27, 2072–2075 (2013). 9. Wang, X. et al. GATA2 and secondary mutations in familial myelodysplastic Quantitative real-time PCR. Total RNA was extracted with Trizol reagent fol- syndromes and pediatric myeloid malignancies. Haematologica 100, e398–401 lowing standard protocol (Thermo Fisher Scientific). One microgram of RNA was (2015). used to synthesize cDNA using Reverse Transcription Reagents (Thermo Fisher 10. Hoischen, A. et al. De novo mutations of SETBP1 cause Schinzel-Giedion Scientific) after pretreatment with DNAseI to avoid contamination from gDNA. syndrome. Nat. Genet. 42, 483–485 (2010). Real-time Q-PCR was performed using TaqMan® Brilliant II QPCR Master Mix 11. Acuna-Hidalgo, R. et al. Overlapping SETBP1 gain-of-function mutations in (Agilent Technologies, Santa Clara, CA, USA) on a Stratagene-MX3005P (Agilent Schinzel-Giedion syndrome and hematologic malignancies. PLoS Genet. 13, Technologies, Santa Clara, USA) under standard conditions. The housekeeping e1006683 (2017). gene GUSB was used as an internal reference . TaqMan® Gene Expression Assays 12. Lach, B. & Arredondo, J. Cobblestone lissencephaly in Schinzel-Giedion (Thermo Fisher Scientific) were used: SKIDA1 (Hs01096520), NFE2L2 syndrome. J. Child Neurol. 28, 259–263 (2013). (Hs00975961), PDE4D (Hs03988495), FBXO8 (Hs00942619), CEP44 13. Oakley, K. et al. Setbp1 promotes the self-renewal of murine myeloid (Hs00604612), COBLL1 (Hs01117513), BMP5 (Hs00234930), ERBB4 progenitors via activation of Hoxa9 and Hoxa10. Blood 119, 6099–6108 (Hs00955525), CDKN1B (Hs01597588), SETBP1 (Hs00210203), RUNX1 (2012). (Hs01021971), MECOM (Hs00602795). 14. Vishwakarma, B. A. et al. Runx1 repression by histone deacetylation is critical for Setbp1-induced mouse myeloid leukemia development. Leukemia 30, Differential expression analysis. The gene counts generated by the --quantMode 200–208 (2016). GeneCounts parameter were used to calculate gene expression. All the subsequent 15. Cristobal, I. et al. SETBP1 overexpression is a novel leukemogenic mechanism steps were performed using R scripts. These scripts were automatically generated that predicts adverse outcome in elderly patients with acute myeloid leukemia. through metaprogramming, using a dedicated tool (StarCounts2DESeq2) devel- Blood 115, 615–625 (2010). oped by RP using C# as metalanguage. Differential expression analyses were per- 16. Kulakovskiy, I. V. et al. HOCOMOCO: expansion and enhancement of the formed using DESeq2 . Analysis of DEGs characterized by the co-occurrence of collection of transcription factor binding sites models. Nucleic Acids Res. 44, both H3K4me2 and H3K9Ac marks was performed by multivariate regression D116–125 (2016). analysis using the following filtering criteria: transcription factor and histone mark 17. Weirauch, M. T. et al. Determination and inference of eukaryotic relative coverage enrichment >0.5 (log2 FC G870S/Empty); differential gene transcription factor sequence specificity. Cell 158, 1431–1443 (2014). expression p-value < 0.05 (G870S vs. Empty, DESeq2). 18. Ovcharenko, I., Nobrega, M. A., Loots, G. G. & Stubbs, L. ECR Browser: a tool for visualizing and accessing data from comparisons of multiple vertebrate In utero electroporation. SETBP1-G870S was cloned in the pCAG expression genomes. Nucleic Acids Res. 32, W280–286 (2004). vector . In utero electroporation was carried out at E13.5 to target the expression 19. Bourachot, B., Yaniv, M. & Muchardt, C. The activity of mammalian brm/ vectors into the ventricular radial glial cells of the mouse embryos, as previously SNF2alpha is dependent on a high-mobility-group protein I/Y-like DNA 58, 59 described . Briefly, uterine horns of E13.5 pregnant dams were exposed by binding domain. Mol. Cell. Biol. 19, 3931–3939 (1999). −1 midline laparotomy after anesthetization with 312 mg kg Avertin. DNA plasmid 20. Cairns, B. R. et al. Two functionally distinct forms of the RSC nucleosome- (1 µl, corresponding to 3 µg) mixed with 0.03% fast-green dye in PBS was injected remodeling complex, containing essential AT hook, BAH, and into the telencephalic vesicle using a micropipette pulled through the uterine wall bromodomains. Mol. Cell 4, 715–723 (1999). and amniotic sac. Platinum tweezer-style electrodes (7 mm) were placed outside 21. Xiao, H. et al. Dual functions of largest NURF subunit NURF301 in the uterus over the telencephalon, and four pulses of 40 mV for 50 ms were applied nucleosome sliding and transcription factor interactions. Mol. Cell 8, 531–543 at 950 ms intervals using a BTX square wave electroporator. The uterus was then (2001). placed back into the abdomen, the cavity was filled with warm sterile PBS, and the 22. Rao, R. C. & Dou, Y. Hijacked in cancer: the KMT2 (MLL) family of abdominal muscle and skin incisions were closed with silk sutures. methyltransferases. Nat. Rev. Cancer 15, 334–346 (2015). All experiments were conducted after approval and following the guidelines of 23. Dinkel, H. et al. ELM 2016--data update and new functionality of the animal care and use committee from San Raffaele Scientific Institute. eukaryotic linear motif resource. Nucleic Acids Res. 44, D294–300 (2016). 24. Sanchez, R. & Zhou, M. M. The PHD finger: a versatile epigenome reader. MECOM analysis. To identify MECOM direct targets among the SETBP1 DEGs, Trends Biochem. Sci. 36, 364–372 (2011). we took advantage of previously published RNA-Seq and ChIP-Seq data in the 25. Siderius, L. E. et al. X-linked mental retardation associated with cleft lip/palate context of MECOM overexpression . The resulting genes were functionally maps to Xp11.3-q21.3. Am. J. Med. Genet. 85, 216–220 (1999). annotated with FunRich and filtered in accordance with their expression within 26. Lower, K. M. et al. Mutations in PHF6 are associated with Borjeson- the bone marrow (Hypergeometric enrichment test p = 0.0013), thus resulting in a Forssman-Lehmann syndrome. Nat. Genet. 32, 661–665 (2002). total of 23 MECOM direct targets. 27. Van Vlierberghe, P. et al. PHF6 mutations in T-cell acute lymphoblastic leukemia. Nat. Genet. 42, 338–342 (2010). 28. Van Vlierberghe, P. et al. PHF6 mutations in adult acute myeloid leukemia. Data availability. All relevant sequencing data are available at the NCBI GEO Leukemia 25, 130–134 (2011). Server, accession code: GSE86335. 29. Qi, H. H. et al. Histone H4K20/H3K9 demethylase PHF8 regulates zebrafish brain and craniofacial development. Nature 466, 503–507 (2010). Received: 14 December 2016 Accepted: 27 April 2018 30. Buenrostro, J. D., Giresi, P. G., Zaba, L. C., Chang, H. Y. & Greenleaf, W. J. Transposition of native chromatin for fast and sensitive epigenomic profiling of open chromatin, DNA-binding proteins and nucleosome position. Nat. Methods 10, 1213–1218 (2013). 12 NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications | | | NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-04462-8 ARTICLE 31. Wieser, R. The oncogene and developmental regulator EVI1: expression, 56. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and biochemical properties, and biological functions. Gene 396, 346–357 (2007). dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014). 32. Kataoka, K. & Kurokawa, M. Ecotropic viral integration site 1, stem cell self- 57. Niwa, H., Yamamura, K. & Miyazaki, J. Efficient selection for high-expression renewal and leukemogenesis. Cancer Sci. 103, 1371–1377 (2012). transfectants with a novel eukaryotic vector. Gene 108, 193–199 (1991). 33. Barjesteh van Waalwijk van Doorn-Khosrovani, S. et al. High EVI1 expression 58. Saito, T. & Nakatsuji, N. Efficient gene transfer into the embryonic mouse predicts poor survival in acute myeloid leukemia: a study of 319 de novo AML brain using in vivo electroporation. Dev. Biol. 240, 237–246 (2001). patients. Blood 101, 837–845 (2003). 59. Sessa, A., Mao, C. A., Hadjantonakis, A. K., Klein, W. H. & Broccoli, V. Tbr2 34. Groschel, S. et al. High EVI1 expression predicts outcome in younger adult directs conversion of radial glia into basal precursors and guides neuronal patients with acute myeloid leukemia and is associated with distinct amplification by indirect neurogenesis in the developing neocortex. Neuron cytogenetic abnormalities. J. Clin. Oncol. 28, 2101–2107 (2010). 60,56–69 (2008). 35. Bard-Chapeau, E. A. et al. Ecotopic viral integration site 1 (EVI1) regulates 60. Pathan, M. et al. FunRich: an open access standalone functional enrichment multiple cellular processes important for cancer and is a synergistic partner for and interaction network analysis tool. Proteomics 15, 2597–2601 (2015). FOS protein in invasive tumors. Proc. Natl. Acad. Sci. USA 109, 2168–2173 61. Zhu, L. et al. ASH1L links histone H3 lysine 36 dimethylation to MLL (2012). leukemia. Cancer Discov. 6, 770–783 (2016). 36. Sessa, A. et al. Tbr2-positive intermediate (basal) neuronal progenitors safeguard cerebral cortex expansion by controlling amplification of pallial Acknowledgements glutamatergic neurons and attraction of subpallial GABAergic interneurons. This work was supported by Associazione Italiana Ricerca sul Cancro 2013 (IG-14249) to Genes Dev. 24, 1816–1826 (2010). C.G.-P., Associazione Italiana; Ricerca sul Cancro 2015 (IG-17727) to R.P., Telethon 37. Nguyen, N. et al. Myb expression is critical for myeloid leukemia development grant (GGP15096), and the Italian Ministry of Health Young investigator grant (# GR- induced by Setbp1 activation. Oncotarget 7, 86300-86312 (2016). 2013-02355540) to A.S. 38. Pekowska, A., Benoukraf, T., Ferrier, P. & Spicuglia, S. A unique H3K4me2 profile marks tissue-specific gene regulation. Genome Res. 20, 1493–1502 (2010). Author contributions 39. Zhang, J., Parvin, J. & Huang, K. Redistribution of H3K4me2 on neural tissue R.P.: Conceived and designed the experiments, performed the experiments, analyzed the specific genes during mouse brain development. BMC Genomics 13(Suppl 8), data, wrote the paper. V.M. and S.R.: performed experiments, analyzed the data, critically S5 (2012). revised the manuscript. M.Mauri, M.P., and C.M.: performed experiments, critically 40. Makishima, H. Somatic SETBP1 mutations in myeloid neoplasms. Int. J. revised the manuscript. L.M.: performed bioinformatics analyses, critically revised the Hematol. 105, 732–742 (2017). manuscript. A.S., F.B., and A.R.: performed in utero electroporation experiments, criti- 41. Goyama, S. et al. Evi-1 is a critical regulator for hematopoietic stem cells and cally revised the manuscript. M.L., R.S., and M.Baumann: performed mass-spectrometry transformed leukemic cells. Cell Stem Cell 3, 207–220 (2008). experiments, critically revised the manuscript. A.P.: performed library preparation for 42. Redaelli, S. et al. Activity of bosutinib, dasatinib, and nilotinib against 18 NGS analyses, critically revised the manuscript. D.R., F.S., E.U., B.M., L.C., M.Merli, F. imatinib-resistant BCR/ABL mutants. J. Clin. Oncol. 27, 469–471 (2009). Passamonti, F.O., A.M., F.Pavesi, and M.Bregni: contributed reagents, critically revised 43. Piazza, R. et al. Epigenetic silencing of the proapoptotic gene BIM in the manuscript. V.B.: conceived in utero electroporation experiments, critically revised anaplastic large cell lymphoma through an MeCP2/SIN3a deacetylating the manuscript. C.G.-P.: conceived and designed the experiments, wrote the paper. complex. Neoplasia 15, 511–522 (2013). 44. Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 (2008). Additional information 45. Quinlan, A. R. & Hall, I. M. BEDTools: a flexible suite of utilities for Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- comparing genomic features. Bioinformatics 26, 841–842 (2010). 018-04462-8. 46. Feng, J., Liu, T., Qin, B., Zhang, Y. & Liu, X. S. Identifying ChIP-seq enrichment using MACS. Nat. Protoc. 7, 1728–1740 (2012). Competing interests: The authors declare no competing interests. 47. Ouyang, Z., Zhou, Q. & Wong, W. H. ChIP-Seq of transcription factors Reprints and permission information is available online at http://npg.nature.com/ predicts absolute and differential gene expression in embryonic stem cells. Proc. Natl. Acad. Sci. USA 106, 21521–21526 (2009). reprintsandpermissions/ 48. Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows- Wheeler transform. Bioinformatics 25, 1754–1760 (2009). Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in 49. Li, H. et al. The Sequence Alignment/Map format and SAMtools. published maps and institutional affiliations. Bioinformatics 25, 2078–2079 (2009). 50. Scifo, E. et al. Proteomic analysis of the palmitoyl protein thioesterase 1 interactome in SH-SY5Y human neuroblastoma cells. J. Proteomics 123,42–53 (2015). Open Access This article is licensed under a Creative Commons 51. Makela, J. et al. Peroxisome proliferator-activated receptor-gamma Attribution 4.0 International License, which permits use, sharing, coactivator-1alpha mediates neuroprotection against excitotoxic brain injury adaptation, distribution and reproduction in any medium or format, as long as you give in transgenic mice: role of mitochondria and X-linked inhibitor of apoptosis appropriate credit to the original author(s) and the source, provide a link to the Creative protein. Eur. J. Neurosci. 43, 626–639 (2016). Commons license, and indicate if changes were made. The images or other third party 52. Busnelli, M., Mauri, M., Parenti, M. & Chini, B. Analysis of GPCR material in this article are included in the article’s Creative Commons license, unless dimerization using acceptor photobleaching resonance energy transfer indicated otherwise in a credit line to the material. If material is not included in the techniques. Methods Enzymol. 521, 311–327 (2013). article’s Creative Commons license and your intended use is not permitted by statutory 53. Konig, P. et al. FRET-CLSM and double-labeling indirect regulation or exceeds the permitted use, you will need to obtain permission directly from immunofluorescence to detect close association of proteins in tissue sections. the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Lab. Invest. 86, 853–864 (2006). licenses/by/4.0/. 54. Ishikawa-Ankerhold, H. C., Ankerhold, R. & Drummen, G. P. Advanced fluorescence microscopy techniques--FRAP, FLIP, FLAP, FRET and FLIM. Molecules 17, 4047–4132 (2012). © The Author(s) 2018 55. Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013). NATURE COMMUNICATIONS (2018) 9:2192 DOI: 10.1038/s41467-018-04462-8 www.nature.com/naturecommunications 13 | | |

Journal

Nature CommunicationsSpringer Journals

Published: Jun 6, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off