Journal:

Article Title: pp32 Reduction Induces Differentiation of TSU-Pr1 Cells

doi:

Figure Lengend Snippet: Expression of IL-6 in TSU-Pr1 cells. TSU-Pr1 cells stably transfected with pp32 anti-sense express higher levels of IL-6 message as compared to parental TSU-Pr1 cells and vector-only control by RT-PCR analysis, which validates the cDNA microarray analysis (see Figure 6).

Article Snippet: Microarray Analysis of TSU-Pr1 Cell Lines This procedure was performed at The Johns Hopkins University Oncology Microarray facility by using a human glass 12K cDNA chip.

Techniques: Expressing, Stable Transfection, Transfection, Plasmid Preparation, Control, Reverse Transcription Polymerase Chain Reaction, Microarray

Journal: Bioengineering

Article Title: COVID-19 Diagnostic Strategies Part II: Protein-Based Technologies

doi: 10.3390/bioengineering8050054

Figure Lengend Snippet: The developed protein microarray-based tests for COVID-19 detection.

Article Snippet: PEPperPRINT GmbH [ ] , PEPperCHIP ® Pan-Corona Spike Protein Microarray , Antibodies against S antigen , S proteins derived from seven coronaviruses translated into overlapping peptides , (No info) , (No info) , One array with 4564 peptides in duplicate , RUO , For Serum antibody fingerprint analysis, Immune monitoring and Epitope studies.

Techniques: Microarray, Peptide Microarray, Bioprocessing, Derivative Assay, High Throughput Screening Assay, Enzyme-linked Immunosorbent Assay, Western Blot, Virus, Clinical Proteomics

Journal: bioRxiv

Article Title: Histone H3K27 methyltransferase EZH2 interacts with MEG3-lncRNA to directly regulate integrin signaling and endothelial cell function

doi: 10.1101/2022.05.20.492787

Figure Lengend Snippet: a) Distribution of annotated single hits over MEG3 gene, with statistically filtered EZH2-FLASH reads from two biological replicates in HUVECs. b) The occupancy of EZH2 hits over MEG3 features. Total reads per feature are given with exons being mostly occupies vs introns. c) Proportion of overlapping features over MEG3. The occupancy of EZH2 over each MEG3 exon is shown for two constitutively expressed transcripts. For both given transcripts there is high occupancy of exon 3. d) RNA immunoprecipitation (RIP) for EZH2 and H3K27me3 (repressive chromatin) followed by qPCR analysis. RIP-purified RNA from UV crosslinked HUVECs was used to prepare cDNA for qPCR analysis with primers against MEG3 (exon 3 region). Primers against U1snRNA gene serves as a negative control. Side diagram of EHZ2-MEG3 interacting region is charted as per FLASH hits and sequence. e) Distribution of EZH2 hybrids hits over MEG3 gene. Intermolecular MEG3-RNA interactions found in chimeras are captured by EZH2-FLASH-seq. Hits represent MEG3:MEG3 hybrids (black). IgG hybrids are plotted but are <1. f) Total MEG3:MEG3 hybrid count against predicted free energy of hybridization (dG) for MEG3 interactions ( red lncRNA:MEG3, blue mRNA:MEG3, green MEG3:antisense, purple snoRNA:MEG3) with free hybridization energy cutoff at dG<-10 kcal mol -1 , as captured by EZH2-FLASH-seq ( i ) vs. IgG control ( ii ) .

Article Snippet: Following sonication as described, samples were immunoprecipitated using EZH2 (D2C9) XP(R) Rabbit mAb, (5246S Cell signalling technology), Tri-Methyl-Histone H3 (H3K27me3) (C36B11) Rabbit mAb (9733S, CST) antibodies or IgG control (Normal Rabbit IgG, 2729S, CST) and captured on beads using Protein G Dyneabeads (10003D, Life Technologies).

Techniques: RNA Immunoprecipitation, Purification, Negative Control, Sequencing, Hybridization, Control

Journal: bioRxiv

Article Title: Histone H3K27 methyltransferase EZH2 interacts with MEG3-lncRNA to directly regulate integrin signaling and endothelial cell function

doi: 10.1101/2022.05.20.492787

Figure Lengend Snippet: a. Overview of the critical steps to obtain MEG3-bound genomic loci and intersections with EZH2 and H3K27me3 signals (obtained from GEO databases for HUVECs). In addition, enhancer regions were mapped within the genomic tracks. The intersection between GEO EZH2 ChIP data, GEO H3K27me3 ChIP data and statistically filtered MEG3-ChIRP data from two biological replicates was performed. The number of genes and degree of overlap is obtained between MEG3 and PRC2-dependent genes. The p-values are a result of hypergeometric test. b. Distribution of MEG3 peaks overlapping EZH2-ChIP peaks or H3K27me3-peaks with intersecting reads in relation to (i) gene regions and (ii) gene-type. c. Maximum peak score of ChIP signal for EZH2 and H3K27me3 intersecting the top enriched MEG3 peaks associated with nearest genes. Highest EZH2 peak score is over ITGA4, whereas H3K27me3 was detected in ITGA4, ITGA7, ITGA8 and ITGA9, members of ITGA family. d. Normalized reads from RNA-seq de novo analysis of GEO: GSE71164 dataset on Hg38, and expression of ITGA4 gene between Scr and siEZH2 depleted HUVECs, showing that ITGA4 is targeted by EZH2. Dataset in d and e is compared using Student’s t-test. e. ITGA4 expression from microarray analysis in C2C12 cells depleted of MEG3 (10nM, LNA GapMer) as per GEO dataset: GSE73524. The data shows that ITGA4 is a direct target of MEG3. f. (i) Total number of representable peaks (mRNA, antisense and lncRNA genes) from ChIP-seq analysis of Scr vs. MEG3 KD HUVECs. (ii ) Depletion of MEG3 gene in HUVECs (10nM LNA gapmers) was achieved with relative expression showing ∼70% reduction compared with Scr control. g. (i) Heat map showing distribution of reads and EZH2 densities at all unique RefSeq genes within TSSs ± 3 kb, sorted by EZH2 occupancy, in Control vs. MEG3 deficient (10nM) HUVECs. (ii) Overlap of ChIP-results between MEG3 and EZH2-dependent genes, with overlapped genes belonging to the biological pathway regulating cell adhesion. The common targets had lost or reduced EZH2 ChIP-signal.

Article Snippet: Following sonication as described, samples were immunoprecipitated using EZH2 (D2C9) XP(R) Rabbit mAb, (5246S Cell signalling technology), Tri-Methyl-Histone H3 (H3K27me3) (C36B11) Rabbit mAb (9733S, CST) antibodies or IgG control (Normal Rabbit IgG, 2729S, CST) and captured on beads using Protein G Dyneabeads (10003D, Life Technologies).

Techniques: RNA Sequencing, Expressing, Microarray, ChIP-sequencing, Control

Journal: bioRxiv

Article Title: Histone H3K27 methyltransferase EZH2 interacts with MEG3-lncRNA to directly regulate integrin signaling and endothelial cell function

doi: 10.1101/2022.05.20.492787

Figure Lengend Snippet: a) Computational analysis pipeline used to obtain orthologous peaks in human and intersect regions and genes enriched in repressive chromatin (H3K27me3) from ChIP-seq public dataset GSE114283. Up- and down-regulated genes were obtained associated with the peak region within 2000bp, and relevant function and biological pathway were associated using GREAT and DAVID analysis b) Overlap of the GEO datasets from a (Microarray GSE73524 ) and b (RNA-seq GSE71164 ) and the GSE114283 ChIP-seq reads of H3K27me 3 distribution in mouse MN cells depleted of MEG3 vs. control. ChIP extracted peaks unique to Ctrl vs. MEG3 KD were obtained, and associated mouse gene list composed based on reduction in H3K27me 3 signal. Using gene orthologous analysis in gProfiler we obtained human orthologous targets that was used for data intersection. c) Maximum peak scores of the overlapping signal over ITGA4 promoter, obtained by intersection of EZH2 ChIP signal with MEG3-ChIRP signal at this region. Upon depletion of MEG3 the EZH2 signal is significantly reduced whereby no overlap with MEG3 ChIRP signal is seen. d) Relative expression of ITGA4 in HUVEC measuring the levels of ITGA4 following addition of siRNA (50nM).

Article Snippet: Following sonication as described, samples were immunoprecipitated using EZH2 (D2C9) XP(R) Rabbit mAb, (5246S Cell signalling technology), Tri-Methyl-Histone H3 (H3K27me3) (C36B11) Rabbit mAb (9733S, CST) antibodies or IgG control (Normal Rabbit IgG, 2729S, CST) and captured on beads using Protein G Dyneabeads (10003D, Life Technologies).

Techniques: ChIP-sequencing, Microarray, RNA Sequencing, Control, Expressing

Journal: bioRxiv

Article Title: Histone H3K27 methyltransferase EZH2 interacts with MEG3-lncRNA to directly regulate integrin signaling and endothelial cell function

doi: 10.1101/2022.05.20.492787

Figure Lengend Snippet: a. Venn diagram showing the intersection between statistically filtered FLASH data from two biological replicates of our MEG3-ChIRP-seq-data (green), de novo hg38 analysed GEO RNA-seq data from siEZH2 deficient HUVECs (GSE71164, blue), and EZH2 ChIP-seq following MEG3 KD (yellow) and FLASH-seq transcriptome following EZH2 IP (pink). b. Correlation between gene expression levels and FLASH signal. Gray, expressed RefSeq genes with reproducible FLASH signal consistently detected in RNA-seq. Blue, genes with the highest RNA-seq signals and no reproducible FLASH signal belonging to integrin cell surface interaction pathway. Red , expressed ITGA4 gene, and green, ITGB1 gene, without reproducible FLASH signals. Data are from two biological replicates of each EZH2 FLASH sample and three biological replicates of EZH2 RNA-seq samples (Scr vs. siEZH2, GSE71164). c. Genomic tracks showing ChIRP-seq signal (MEG3 Odd, Even and LacZ) in HUVECs over ITGA4 gene only. The MEG3 binding site is located upstream of the ITGA4 gene in the promoter region, and it overlaps with the H3K27me3 signal and EZH2; as well as downstream within the ITGA4 gene body, where it overlaps with within the EZH2 signal in the intronic region of the gene. d. MEG3-ChIRP followed by qPCR, analysis of MEG3 binding region on ITGA4 in HUVECs. The crosslinked cell lysates were incubated with combined biotinylated probes against MEG3 lncRNA and the binding complexes recovered by magnetic streptavidin-conjugated beads. The qPCR was performed to detect the enrichment of specific region that associated with MEG3, peaks were related to input control and compared vs. the non-biotynilated control. e. ChIP-QPCR enrichment for EZH2 and H3K27me3 over ITGA4 promoter region in HUVECs depleted of MEG3 vs. Control.

Article Snippet: Following sonication as described, samples were immunoprecipitated using EZH2 (D2C9) XP(R) Rabbit mAb, (5246S Cell signalling technology), Tri-Methyl-Histone H3 (H3K27me3) (C36B11) Rabbit mAb (9733S, CST) antibodies or IgG control (Normal Rabbit IgG, 2729S, CST) and captured on beads using Protein G Dyneabeads (10003D, Life Technologies).

Techniques: RNA Sequencing, ChIP-sequencing, Gene Expression, Binding Assay, Incubation, Control, ChIP-qPCR

Journal: bioRxiv

Article Title: Histone H3K27 methyltransferase EZH2 interacts with MEG3-lncRNA to directly regulate integrin signaling and endothelial cell function

doi: 10.1101/2022.05.20.492787

Figure Lengend Snippet: a. ChIP signal enrichment vs . 1% input for EZH2 and H3K27me3 mark over ITGA4 promoter regions in HUVECs treated with A-395 (5µM, 24h) inhibitor of PRC2 vs. Control (DMSO). The expression was measured using two sets of primers against the same promoter region of ITGA4. Representative graphs are average of three qPCR datasets ± SEM. b. ITGA4 expression in the presence of A-395 vs . DMSO control, N=6 independent experiments compared using t -test. c. Measuring the expression levels of ITGA4 upon depletion of MEG3 using LNA GapmeRs (10nM, 48h), data is mean of N=5 independent experiments (biological replicates). d. Representative image of immunofluorescence staining for ITGA4 protein levels in ECs treated with A-395 vs . DMSO, or upon MEG3 depletion like in b . e. Intra-cellular localisation of MEG3 (chromatin associated lncRNA) between different cellular compartments in HUVECs treated with A-395 vs. DMSO, whereby the distribution of MEG3 has shifted upon PRC2 inhibition with A-395; from the nucleus (where it was highly chromatin bound) into the cytoplasm. Representative bars were compared by t-test and on-way Anova. f. MEG3-ChIRP followed by qPCR, N =3, analysis of MEG3 binding over ITGA4 promoter region in HUVECs treated with A-395 (5µM, 24h) vs. DMSO. MEG3-ChIRP HUVEC lysates treated with A-395 resulted in reduced engagement of MEG3 with ITGA4 site compared with either DMSO control or ChIRP with non-biotinylated probes. The non-biotin probes served as a negative control, and we detected the background level <1.

Article Snippet: Following sonication as described, samples were immunoprecipitated using EZH2 (D2C9) XP(R) Rabbit mAb, (5246S Cell signalling technology), Tri-Methyl-Histone H3 (H3K27me3) (C36B11) Rabbit mAb (9733S, CST) antibodies or IgG control (Normal Rabbit IgG, 2729S, CST) and captured on beads using Protein G Dyneabeads (10003D, Life Technologies).

Techniques: Control, Expressing, Immunofluorescence, Staining, Inhibition, Binding Assay, Negative Control

Journal: bioRxiv

Article Title: Histone H3K27 methyltransferase EZH2 interacts with MEG3-lncRNA to directly regulate integrin signaling and endothelial cell function

doi: 10.1101/2022.05.20.492787

Figure Lengend Snippet: a. Measure of cell migratory capacity using ECIS functional analysis in ECs treated with control or A-395 (5µM, 24h) inhibitor. Experiments were performed in duplicates (technical replicates) and four experiments were run for migration assay and six for adhesion (biological replicates). The data showing ECIS trace (left hand side) is mean ±SD as calculated by the ECIS. The graph on the right is mean±SEM with N =6, data was compared using ordinary one-way ANOVA with Dunnett’s multiple comparisons tests. b. Adhesion to Fibronectin, FN (20µg/ml) was used to coat the culture plates and assess adhesion of endothelial cells within 3h of ECIS assay, following cell pre-treatment with A-395, 24h. The difference in resistance change was calculated over 3h. c. Subcutaneous Matrigel plug injection (200µl) into mice ( N =5) treated with DMSO (control, left flange) and A-395 (1mg/ml, right flange) was done for 2 weeks. Matrigel plugs were collected and processed for histology. Staining for H3K27me3 was done, displaying nuclear positivity with strong intensity in control (<0.02% DMSO in water) and the A-395 treatment decreased total H3K27me3 staining, as compared by t-test. d. Staining for arterioles was performed to assess vessel growth as angiogenesis and data was compared using Student’s t-test. The data shows increased area of staining for Isolectin B4 (Iso-B4) dye in A-395 vs. DMSO treated Matrigel plugs with increased neovascularization, P<0.05. e. A-395 has increased the percentage of vessels positive for ITGA4 (red) within the Isolectin B4 positive cells, compared with the DMSO using t -test. f. Graphical abstract. 1 Maternally Expressed Gene–MEG3 is highly expressed with hypoxia and bound to EZH2 in endothelial cells (EC) affected by ischaemic insult. 2 Such MEG3:EZH2 complex assembles onto the target genes to 3 direct the EZH2 activity to “write” H3K27me3 trimethylation repressive mark and block expression of target gene i.e. integrin alpha 4 (ITGA4) and its ability to dimerise with integrin beta 1 (ITGB1), leading to 4 reduced EC function as measured by adhesion and migration. Hence 5 targeted disruptions of MEG3:EZH2 interaction, or inhibition of EZH2 activity could increase EC function under ischaemia.

Article Snippet: Following sonication as described, samples were immunoprecipitated using EZH2 (D2C9) XP(R) Rabbit mAb, (5246S Cell signalling technology), Tri-Methyl-Histone H3 (H3K27me3) (C36B11) Rabbit mAb (9733S, CST) antibodies or IgG control (Normal Rabbit IgG, 2729S, CST) and captured on beads using Protein G Dyneabeads (10003D, Life Technologies).

Techniques: Functional Assay, Control, Migration, Injection, Staining, Activity Assay, Blocking Assay, Expressing, Inhibition

Journal: The EMBO Journal

Article Title: HNF1A recruits KDM6A to activate differentiated acinar cell programs that suppress pancreatic cancer

doi: 10.15252/embj.2019102808

Figure Lengend Snippet: A Representative immunofluorescence images and quantifications showing that 3‐month‐old Hnf1a pKO mice have increased number of KI67 + (red) acinar cell nuclei co‐staining with DAPI (blue) and Amylase (green). Arrows point to KI67 + acinar cells in Hnf1a pKO mouse. Acinar proliferation is represented as the average of the KI67 + /Amylase + cell ratio. Quantifications were performed on 3 random fields from 3 Pdx1 Cre and 3 Hnf1a pKO mice. P ‐values are from two‐tailed Student's t ‐test. Representative H&E stainings of pancreata from Kras G12D and Hnf1a pKO ; Kras G12D mice. B–D Kras G12D and Hnf1a pKO ; Kras G12D mice have normal morphology at 7 days. E–J At 21 days, Hnf1a pKO ; Kras G12D mice show acinar‐to‐ductal metaplasia (dashed encircled areas) and regions with desmoplastic reaction (asterisk), which are not observed in Kras G12D mice (E, H). K–P At 8 weeks, Kras G12D pancreas show occasional abnormal ductal structures (dashed encircled areas in N, which is a magnification of squared dotted box in K) and Hnf1a pKO ; Kras G12D mice (L, M, O, P) present mucinous tubular complexes (black arrows), and more advanced PanINs with luminal budding (open arrows) including foci of spindle cell proliferation (asterisks) and incipient infiltrative growth (black dashed box area in O). Data information: Black dashed boxes in (E, F, K, L and O) indicate magnified areas in (H, G, N, M and P) respectively. Scale bars indicate 200 μm (A), 100 μm (C, E, F, K, L), 50 μm (O), and 20 μm (B, D, G, H–J, M, N, P).

Article Snippet: Endogenous peroxidase and protein blocking was performed with 3% H 2 O 2 diluted in PBS for 10 min and with 1% BSA, 10% normal goat serum (Abcam, Cambridge, UK), and 0.1% Triton X‐100 (Merck KGaA, Darmstadt, Germany) for 60 min. Anti‐HNF1A and anti‐KDM6A stainings were performed at a dilution of 1:250 (Anti‐HNF1A, Abcam ab204306, Cambridge, UK), 1:200 (Anti‐HNF1A, Cell Signaling Technology, 89670, Leiden, The Netherlands), and 1:100 (Anti‐UTX, Cell Signaling Technology 33510S, Denver, USA), respectively.

Techniques: Immunofluorescence, Staining, Two Tailed Test

Journal: The EMBO Journal

Article Title: HNF1A recruits KDM6A to activate differentiated acinar cell programs that suppress pancreatic cancer

doi: 10.15252/embj.2019102808

Figure Lengend Snippet: A Breeding strategy to generate Hnf1a aKO and Ptf1a Cre ;Hnf1a +/+ control mice using Ptf1a Cre and Hnf1a LoxP alleles. B Ptf1a Cre deletes HNF1A efficiently in acinar cells but to a lesser extent in islets of Langerhans. HNF1A IHC and hematoxylin staining in pancreas of control and Hnf1a aKO mice. HNF1A is expressed in acinar and islet cells, but not in ductal cells in normal pancreas (left). HNF1A expression is depleted in acinar cells but largely not in islets in Hnf1a aKO pancreas (right). The squared dotted boxes (top) indicate magnified areas (bottom). Arrows point at ducts, arrow head at HNF1A‐positive acinar cell, and open arrow head at HNF1A positive islet cell. The dotted encircled areas indicate islets of Langerhans. Scale bar (top) 300 μm (bottom) 50 μm. C H&E stainings in pancreas of control (left) and Hnf1 aKO mice (right) showing unaltered pancreatic morphology. Scale bar 300 μm. D Expression of acinar differentiation genes in pancreas from Hnf1a aKO and controls, depicted as box plots with median values and IQR of TPM values. Whiskers extend to highest and lowest data points within 1.5× IQR outside box limits. P ‐values were determined by two‐tailed Student's t ‐test and n = 3 replicates per condition. E GSEA showing increased expression of oncogenic pathways in Hnf1a aKO pancreas. F Western blots (top) and quantifications (bottom) showing increased phospho‐p42 levels in Hnf1a aKO pancreas. Quantification of signal intensities of phospho‐p44/p42 normalized to total‐p44/p42 levels. Data are shown as dots with mean and error bars ± SD. P ‐values were determined by two‐tailed Student's t ‐test. G Distribution of pancreatic HNF1A binding sites in annotated genomic regions. H Venn diagrams illustrating that HNF1A‐bound regions are enriched in regions of active promoters and enhancers. P ‐values and odds ratios were calculated by Fisher's exact test. I Enrichment of known HNF1 motifs in the top 500 most significant HNF1A‐bound ChIP‐seq regions and percentage of regions containing each motif. The “union” is the percentage of regions with at least one motif sequence occurrence. Enrichment P ‐values are calculated using the one‐tailed binomial test. J Genome browser track for Fn1 and Timp1 genes showing upregulated expression in Hnf1a aKO and Kdm6a pKO pancreas, and absence of HNF1A or KDM6A binding in adjacent regions. Plots show TPM values normalized to Hprt with mean and error bars ± SD. P ‐values were determined by two‐tailed Student's t ‐test.

Article Snippet: Endogenous peroxidase and protein blocking was performed with 3% H 2 O 2 diluted in PBS for 10 min and with 1% BSA, 10% normal goat serum (Abcam, Cambridge, UK), and 0.1% Triton X‐100 (Merck KGaA, Darmstadt, Germany) for 60 min. Anti‐HNF1A and anti‐KDM6A stainings were performed at a dilution of 1:250 (Anti‐HNF1A, Abcam ab204306, Cambridge, UK), 1:200 (Anti‐HNF1A, Cell Signaling Technology, 89670, Leiden, The Netherlands), and 1:100 (Anti‐UTX, Cell Signaling Technology 33510S, Denver, USA), respectively.

Techniques: Staining, Expressing, Two Tailed Test, Western Blot, Binding Assay, ChIP-sequencing, Sequencing, One-tailed Test

Journal: The EMBO Journal

Article Title: HNF1A recruits KDM6A to activate differentiated acinar cell programs that suppress pancreatic cancer

doi: 10.15252/embj.2019102808

Figure Lengend Snippet: A Fold change (FC) in transcripts in Hnf1a aKO versus control pancreas, plotted against significance (−Log 10 q; genes significant at q < 0.05 are shown as colored dots above the horizontal line). B GSEA showing that genes specific to differentiated acinar cells were downregulated in Hnf1a aKO pancreas, but not genes specific to islets or duct cells. Upregulated genes were enriched in genes specific to mesenchymal cells. Lineage‐enriched genes were obtained from Muraro et al . C Top functional annotations for differentially expressed genes in Hnf1a aKO pancreas. D GSEA revealed that Hnf1a aKO pancreas showed increased transcripts involved in oncogenic pathways such as EMT, MAPK, KRAS, PI3K‐AKT. E HNF1A promotes transcriptional activation of direct target genes. Left: HNF1A‐bound genes were enriched among genes that showed downregulation in Hnf1a mutants, but not among upregulated genes. P ‐values and odds ratios (O.R.) calculated by Fisher's exact test. Right: Venn diagrams showing overlap of HNF1A‐bound genes with genes that were downregulated and upregulated in Hnf1a mutant pancreas.

Article Snippet: Endogenous peroxidase and protein blocking was performed with 3% H 2 O 2 diluted in PBS for 10 min and with 1% BSA, 10% normal goat serum (Abcam, Cambridge, UK), and 0.1% Triton X‐100 (Merck KGaA, Darmstadt, Germany) for 60 min. Anti‐HNF1A and anti‐KDM6A stainings were performed at a dilution of 1:250 (Anti‐HNF1A, Abcam ab204306, Cambridge, UK), 1:200 (Anti‐HNF1A, Cell Signaling Technology, 89670, Leiden, The Netherlands), and 1:100 (Anti‐UTX, Cell Signaling Technology 33510S, Denver, USA), respectively.

Techniques: Functional Assay, Activation Assay, Mutagenesis

Journal: The EMBO Journal

Article Title: HNF1A recruits KDM6A to activate differentiated acinar cell programs that suppress pancreatic cancer

doi: 10.15252/embj.2019102808

Figure Lengend Snippet: A Human orthologs of genes that were up‐ and downregulated in Hnf1a aKO pancreas were also up‐ and downregulated in human pancreas with low versus high HNF1A expression (lowest versus highest expression deciles, respectively). A random list of 717 genes controlled for similar expression levels was used for comparison. Violin plots include median and interquartile ranges. Dots are average values for each gene. Kruskal–Wallis P < 0.0001. B GSEA demonstrates that down‐ or upregulated genes in Hnf1a aKO mice (downward or upward arrows) showed down‐ or upregulation, respectively, in gene lists ranked by differential expression in non‐classical versus classical PDAC molecular subtypes from the TCGA‐PAAD study (Cancer Genome Atlas Research Network, Electronic Address Aadhe, Cancer Genome Atlas Research N, ). All enrichments had GSEA FDR q ‐values < 0.01. C Analysis of HNF1A function in 121 high‐purity cases of the ICGC‐PACA‐AU cohort identified tumors with most pronounced downregulation of direct HNF1A target genes. We performed GSEA with a gene set of 106 human orthologs of HNF1A direct targets showing downregulation in Hnf1a aKO pancreas. For each tumor sample, we performed differential expression against all other samples and used GSEA to ascertain abnormal expression of the mouse HNF1A‐dependent gene set in the tumor. Samples were ranked by the resulting normalized enrichment score (NES) and classified as either HNF1A LoF samples (purple, NES < 0; P < 0.05), or Control 1 (beige, NES < 0; P > 0.05) and Control 2 (gray, NES > 0). HNF1A LoF samples were predominantly non‐classical tumors (Collisson et al , ; Moffitt et al , ; Bailey et al , ). Putative loss‐of‐function KDM6A mutations ( KDM6A LoF mutants) were found in 19% of HNF1A LoF tumors versus 2% of all others (Fisher's P = 0.005). KDM6A mutations were considered functional if classified as “high” functional impact in ICGC (small ≤ 200‐bp deletions/insertions, single base substitutions), or as likely loss‐of‐function structural variants in Bailey et al , all of which were frame‐shift mutations. Other KDM6A mutations were classified as unknown. Heatmaps show Z ‐score‐normalized expression of deregulated genes in Hnf1a aKO pancreas. We confirmed that 85% of 106 downregulated and 60% of genes of 146 upregulated human orthologs showed differential expression across the 3 HNF1A profiles ( q < 0.05, SAM multiclass analysis). D HNF1A mRNA levels differed in HNF1A LoF and control groups (Kruskal–Wallis, P < 0.01), despite considerable variability and overlap between groups. E KDM6A mRNA levels were downregulated in HNF1A LoF tumors (Kruskal–Wallis, P < 0.001). Data information: Box plots in (D and E) show HNF1A and KDM6A expression in HNF1A LoF tumors ( n = 26) and Control 1 ( n = 39) and Control 2 ( n = 57) tumors. The horizontal central line marks the median. Box limits indicate the first and third quartiles, and whiskers extend to highest and lowest data points within 1.5× interquartile range outside box limits.

Article Snippet: Endogenous peroxidase and protein blocking was performed with 3% H 2 O 2 diluted in PBS for 10 min and with 1% BSA, 10% normal goat serum (Abcam, Cambridge, UK), and 0.1% Triton X‐100 (Merck KGaA, Darmstadt, Germany) for 60 min. Anti‐HNF1A and anti‐KDM6A stainings were performed at a dilution of 1:250 (Anti‐HNF1A, Abcam ab204306, Cambridge, UK), 1:200 (Anti‐HNF1A, Cell Signaling Technology, 89670, Leiden, The Netherlands), and 1:100 (Anti‐UTX, Cell Signaling Technology 33510S, Denver, USA), respectively.

Techniques: Expressing, Functional Assay

Journal: The EMBO Journal

Article Title: HNF1A recruits KDM6A to activate differentiated acinar cell programs that suppress pancreatic cancer

doi: 10.15252/embj.2019102808

Figure Lengend Snippet: A Consensus clustered Z‐score‐normalized gene expression heatmaps of high‐purity TCGA‐PAAD and ICGC‐PACA‐AU human PDAC samples. Clustering was performed with non‐negative matrix factorization based on expression of significantly down‐ and upregulated genes in Hnf1a aKO pancreas. This revealed a cluster (HNF1A cluster 3) with concordant up‐ and downregulation of genes in Hnf1a aKO pancreas, which predominantly matched non‐classical PDAC molecular subtypes (quasimesenchymal, basal, squamous‐like, pink in top tracks), as opposed to classical PDAC subtypes (green in top tracks). Multiclass SAM differentially expressed genes ( q < 0.05) between HNF1A clusters are shown. Genes were hierarchically clustered using complete linkage with one minus Pearson correlation metrics. Along the right side of the heatmaps are green and red indicators of down‐ and upregulated genes in Hnf1a aKO pancreas, respectively. B TP63 expression was increased in HNF1A LoF tumors compared to control tumors. RSEM normalized count data are shown as box plots with interquartile range, median, and whiskers. Box limits indicate the first and third quartiles and whiskers extend to highest and lowest data points within 1.5× IQR outside box limits. HNF1A LoF ( n = 26), Control 1 ( n = 39), and Control 2 ( n = 57) tumors (P, Kruskal–Wallis). C, D Expression of HNF1A and KDM6A , showing downregulation in non‐classical PDAC subtypes (P, Kruskal–Wallis). Dots are RSEM normalized values presented with mean ± SD. Collisson subtypes: Quasimesenchymal (QM, n = 34) and Classical (CL, n = 54). Moffitt subtypes: Basal (BA, n = 65) and Classical (CL, n = 85). Bailey subtypes: Squamous‐like (SQ‐like, n = 31) and Pancreatic Progenitor (PP, n = 53). E, F HNF1A levels are not lower in high histological grade PDAC (E), while KDM6A levels are (F). To determine whether histological grade of human PDAC was associated with expression levels of HNF1A (E) or KDM6A (F) proteins, we evaluated contingency tables of tumor grades versus staining intensities of each case in tissue microarray (TMA) IHC. Tumor grades were scored as either moderately differentiated (G2) or poorly differentiated/high grade (G3), and staining intensities were expressed as an Immuno Reactivity Score (IRS) reflecting either No, Weak, Moderate, or Strong staining intensities (see material and methods for details). Numbers of cases and percentages (in brackets) out of total cases are indicated for each tumor grade and staining intensity. The Chi‐squared test was used to determine the probability of a significant relationship. Chi‐square and P ‐values are shown. N = 217 patients for HNF1A and N = 208 patients for KDM6A.

Article Snippet: Endogenous peroxidase and protein blocking was performed with 3% H 2 O 2 diluted in PBS for 10 min and with 1% BSA, 10% normal goat serum (Abcam, Cambridge, UK), and 0.1% Triton X‐100 (Merck KGaA, Darmstadt, Germany) for 60 min. Anti‐HNF1A and anti‐KDM6A stainings were performed at a dilution of 1:250 (Anti‐HNF1A, Abcam ab204306, Cambridge, UK), 1:200 (Anti‐HNF1A, Cell Signaling Technology, 89670, Leiden, The Netherlands), and 1:100 (Anti‐UTX, Cell Signaling Technology 33510S, Denver, USA), respectively.

Techniques: Expressing, Staining, Microarray

Journal: The EMBO Journal

Article Title: HNF1A recruits KDM6A to activate differentiated acinar cell programs that suppress pancreatic cancer

doi: 10.15252/embj.2019102808

Figure Lengend Snippet: A Efficient deletion of Kdm6a in the pancreatic epithelium at E15.5. KDM6A (red) is ubiquitously expressed in all pancreatic cells. CDH1 (green) marks epithelial cells. Upon deletion, KDM6A staining is lost specifically in CDH1‐expressing epithelial cells but not in mesenchymal cells or in the stomach epithelium (white arrow heads). Scale bar indicates 100 μm. B Kdm6a mutant mice show normal fasting and fed glycemia. The horizontal stroked line indicates blood glucose levels at 250 mg/dl as a reference. C–H The pancreas of Kdm6a pKO mice were histologically normal until 8 weeks of age. At 8 weeks of age, some signs of acinar cell attrition and fat replacement could be observed, as shown in this representative image. Scale bars: 250 μm (10× magnification), 50 μm (40× magnification). I Representative picture (left) showing increased number of KI67 (red) amylase‐expressing acinar cells (green) in Kdm6a pKO pancreas. Scale bar, 250 μm. Quantifications (right) were performed on three pancreatic sections separated by at least 100 μm from 4 control and 4 Kdm6a pKO mice. Acinar cell proliferation was represented as the average of the KI67 + /Amylase + cell ratio ± SD. P ‐values were determined by two‐tailed Student's t‐ test. J GSEA plots showing enrichment of Oncostatin M and “TNFA signaling via NFKB” gene sets among genes upregulated in Kdm6a pKO pancreas. K Western blots (top) and quantifications (bottom) showing increased phospho‐p44/p42 levels in Kdm6a pKO pancreas. Quantification of signal intensities of phospho‐p44/p42 normalized to total‐p44/p42 levels. Data are shown as dots with mean and error bars ± SD. P ‐values were determined by two‐tailed Student's t ‐test. L Most significantly deranged REACTOME pathways in both KDM6A‐ and HNF1A‐deficient pancreas (see also ). M Kdm6a pKO down‐ and upregulated gene sets showed concordant deregulation in KDM6A LoF mutant tumors versus classical PDAC (based on Bailey et al 's signature) (Bailey et al , ). N Tumors with KDM6A‐deficient phenotypes showed decreased KDM6A mRNA. We created a gene set of human orthologs of Kdm6a pKO downregulated genes, and for each high‐purity tumor sample in the ICGC‐PACA‐AU, we used GSEA to test for enrichment of this gene set in gene lists that were rank‐ordered by differential expression in the individual sample versus all other samples. Samples with NES < 0 and P ‐value < 0.05 were considered as having KDM6A LoF phenotypes and were compared against all other samples. Z ‐score‐normalized count data are shown as box plots with IQR, median, and whiskers. Whiskers extend to highest and lowest data points within 1.5× IQR outside box limits. P ‐values were determined by two‐tailed Student's t ‐test. O Gene sets that showed up‐ or downregulation in non‐classical human PDAC showed concordant enrichment in up‐ or downregulated genes in Kdm6a pKO versus control pancreas. GSEA NES and FDR q‐values are shown. P Genomic distribution of KDM6A binding sites in mouse pancreas. Q, R Top: ChIP‐seq and RNA‐seq tracks in control and Kdm6a pKO pancreas, in two loci harboring downregulated genes ( Kif12 , Gprc5c ) in Kdm6a pKO pancreas. Bottom: ChIP‐qPCR validations for regions highlighted in green (R1, R2, R3), showing that Kdm6a mutants have increased H3K27me3 and decreased H3K27ac in most regions. H3K4me1 was also decreased in mutants in distal sites. Error bars show SD, and P ‐values were determined by two‐tailed Student's t ‐test, n = 3. S KDM6A‐bound regions are enriched in active pancreas promoters and enhancers. P ‐values are calculated by Fisher's exact test. T, U Genome Browser examples (top) of HNF1A and KDM6A binding to genes known as negative regulators of EMT: Gstp1 (T) and Deptor in (U) that are downregulated in Hnf1a aKO and Kdm6a pKO pancreas (bottom). Plots show TPM values normalized to Hprt with mean and error bars ± SD. N = 4 per condition and P ‐values were determined by two‐tailed Student's t ‐test.

Article Snippet: Endogenous peroxidase and protein blocking was performed with 3% H 2 O 2 diluted in PBS for 10 min and with 1% BSA, 10% normal goat serum (Abcam, Cambridge, UK), and 0.1% Triton X‐100 (Merck KGaA, Darmstadt, Germany) for 60 min. Anti‐HNF1A and anti‐KDM6A stainings were performed at a dilution of 1:250 (Anti‐HNF1A, Abcam ab204306, Cambridge, UK), 1:200 (Anti‐HNF1A, Cell Signaling Technology, 89670, Leiden, The Netherlands), and 1:100 (Anti‐UTX, Cell Signaling Technology 33510S, Denver, USA), respectively.

Techniques: Staining, Expressing, Mutagenesis, Two Tailed Test, Western Blot, Binding Assay, ChIP-sequencing, RNA Sequencing Assay

Journal: The EMBO Journal

Article Title: HNF1A recruits KDM6A to activate differentiated acinar cell programs that suppress pancreatic cancer

doi: 10.15252/embj.2019102808

Figure Lengend Snippet: A Motif analysis in functional KDM6A‐bound regions, showing top ten de novo motifs ranked by P ‐value determined by HOMER software. B Co‐binding analysis in functional KDM6A‐bound enhancer and promoter regions revealed that HNF1A was the most enriched co‐bound TF among three other acinar cell TFs. Binding regions of TAL1 in a non‐pancreatic cell type and random binding sites were used as negative controls. P ‐values were determined by Fisher's exact test for peak comparisons using all enhancer and promoter regions as background. C The most downregulated genes in Kdm6a pKO pancreas are shown ranked by q‐value and are almost invariably bound by HNF1A and downregulated in Hnf1a aKO pancreas, or known to be direct HNF1A‐dependent target genes from other studies (red and purple, respectively). D, E GSEA analysis on the Hnf1a aKO and Kdm6a pKO ranked‐ordered gene lists versus their reciprocal up‐ or downregulated gene sets, demonstrated that KDM6A and HNF1A regulate similar genes. F Expression changes in Hnf1a aKO and Kdm6a pKO pancreas, showing that genes bound by KDM6A and downregulated in Kdm6a pKO pancreas (red dots) were generally downregulated in Hnf1a aKO pancreas. G HNF1A and KDM6A co‐occupy the same regions in Pah , which is downregulated in Hnf1a and Kdm6a knock‐out pancreas. H Genes that were co‐bound by KDM6A and HNF1A showed greatest downregulation in Kdm6a pKO pancreas, compared with KDM6A‐bound genes that were not bound by HNF1A. Box plots show median and IQR of Log 2 TPM fold‐changes and whiskers extend to highest and lowest data points within 1.5× IQR outside box limits. P ‐values were determined by two‐tailed Student's t ‐tests and n = 4 replicates per condition.

Article Snippet: Endogenous peroxidase and protein blocking was performed with 3% H 2 O 2 diluted in PBS for 10 min and with 1% BSA, 10% normal goat serum (Abcam, Cambridge, UK), and 0.1% Triton X‐100 (Merck KGaA, Darmstadt, Germany) for 60 min. Anti‐HNF1A and anti‐KDM6A stainings were performed at a dilution of 1:250 (Anti‐HNF1A, Abcam ab204306, Cambridge, UK), 1:200 (Anti‐HNF1A, Cell Signaling Technology, 89670, Leiden, The Netherlands), and 1:100 (Anti‐UTX, Cell Signaling Technology 33510S, Denver, USA), respectively.

Techniques: Functional Assay, Software, Binding Assay, Expressing, Knock-Out, Two Tailed Test

Journal: The EMBO Journal

Article Title: HNF1A recruits KDM6A to activate differentiated acinar cell programs that suppress pancreatic cancer

doi: 10.15252/embj.2019102808

Figure Lengend Snippet: A–D Left: Genome Browser examples of loci co‐bound by HNF1A and KDM6A, showing loss of KDM6A binding in HNF1A‐deficient pancreas (region highlighted in green) and decreased RNA levels in HNF1A‐deficient pancreas. Right: ChIP‐qPCRs showing loss of KDM6A and HNF1A binding in highlighted regions in left and qPCRs show downregulation of target genes in Hnf1a ‐KO pancreas. Error bars show SD, and P ‐values were determined by two‐tailed Student's t ‐test, n = 4 for ChIP‐qPCRs and n = 3 for qPCRs.

Article Snippet: Endogenous peroxidase and protein blocking was performed with 3% H 2 O 2 diluted in PBS for 10 min and with 1% BSA, 10% normal goat serum (Abcam, Cambridge, UK), and 0.1% Triton X‐100 (Merck KGaA, Darmstadt, Germany) for 60 min. Anti‐HNF1A and anti‐KDM6A stainings were performed at a dilution of 1:250 (Anti‐HNF1A, Abcam ab204306, Cambridge, UK), 1:200 (Anti‐HNF1A, Cell Signaling Technology, 89670, Leiden, The Netherlands), and 1:100 (Anti‐UTX, Cell Signaling Technology 33510S, Denver, USA), respectively.

Techniques: Binding Assay, Two Tailed Test

Journal: The EMBO Journal

Article Title: HNF1A recruits KDM6A to activate differentiated acinar cell programs that suppress pancreatic cancer

doi: 10.15252/embj.2019102808

Figure Lengend Snippet: A HNF1A binding to chromatin is unaffected in Kdm6a pKO pancreas. Scatterplot showing unchanged HNF1A binding in Kdm6a pKO versus control pancreas (e.g., see also Appendix Fig S6A–D ). B Western blot showing KDM6A depletion in two clones from Kdm6a ‐KO acinar cell lines. C–F qPCR in Kdm6a ‐KO acinar cell lines (KO1 and KO2) shows reduced expression of HNF1A bound genes, while ChIP‐qPCRs for HNF1A show that its binding to those genes is unchanged when KDM6A is depleted. Selected genes and HNF1A binding regions were from Appendix Fig S6A–D . qPCR data are relative mRNA expression of indicated genes versus Hprt . ChIP‐qPCR values indicate fold enrichment relative to control region. Error bars show ± SD, and P ‐values were determined by two‐tailed Student's t ‐test. G GSEA analysis on ranked‐ordered gene list from Kras G12D ;Kdm6a pKO data from Andricovich et al versus gene sets from up‐ or downregulated genes in Hnf1a aKO and Kdm6a pKO pancreas demonstrates that KDM6A and HNF1A regulate similar genes in normal and Kras G12D ‐transformed pancreas. H GSEA, comparing rank‐ordered expression data from Kras G12D ;Kdm6a pKO , Hnf1a aKO , and Kdm6a pKO mice with gene sets from Andricovich et al , shows that most of the pathways that are enriched in Kdm6a ‐deficient pancreatic cancer are dependent on HNF1A and KDM6A function in the non‐tumoral pancreas.

Article Snippet: Endogenous peroxidase and protein blocking was performed with 3% H 2 O 2 diluted in PBS for 10 min and with 1% BSA, 10% normal goat serum (Abcam, Cambridge, UK), and 0.1% Triton X‐100 (Merck KGaA, Darmstadt, Germany) for 60 min. Anti‐HNF1A and anti‐KDM6A stainings were performed at a dilution of 1:250 (Anti‐HNF1A, Abcam ab204306, Cambridge, UK), 1:200 (Anti‐HNF1A, Cell Signaling Technology, 89670, Leiden, The Netherlands), and 1:100 (Anti‐UTX, Cell Signaling Technology 33510S, Denver, USA), respectively.

Techniques: Binding Assay, Western Blot, Clone Assay, Expressing, Two Tailed Test, Transformation Assay

Journal: The EMBO Journal

Article Title: HNF1A recruits KDM6A to activate differentiated acinar cell programs that suppress pancreatic cancer

doi: 10.15252/embj.2019102808

Figure Lengend Snippet: A Co‐immunoprecipitation of endogenous HNF1A and KDM6A followed by Western blot demonstrated that HNF1A is in the same complex as KDM6A. B Western blot showing loss of HNF1A and unchanged KDM6A in Hnf1a −/− pancreas. C Differential binding analysis of KDM6A in Hnf1a −/− versus wild‐type pancreas. Pink dots below zero (1,873 sites) show regions with reduced KDM6A binding, and pink dots above zero (118 sites) are regions with increased binding at FDR < 0.05. D, E Regions that show reduced KDM6A binding in Hnf1a −/− chromatin are strongly bound by HNF1A and are highly enriched in HNF1 motifs. P ‐values in (D) were calculated with two‐tailed Mann–Whitney U‐test and in (E) with Fisher's exact test. F KDM6A binding is markedly reduced in HNF1A‐ and KDM6A‐co‐bound regions in Hnf1a −/− pancreas, but not in other KDM6A‐bound regions. G Genes that loose KDM6A binding in Hnf1a ‐mutant pancreas are predominantly downregulated in Hnf1a aKO pancreas and are direct HNF1A target genes (red dots). H Summary model depicting that HNF1A recruits KDM6A to genomic binding sites, activating an acinar differentiation program that indirectly suppresses core oncogenic pathways. Defective HNF1A or KDM6A function results in failure of this shared program, with increased activity of pathways that, in the presence of KRAS mutations, promote high‐grade non‐classical PDAC with sarcomatoid features.

Article Snippet: Endogenous peroxidase and protein blocking was performed with 3% H 2 O 2 diluted in PBS for 10 min and with 1% BSA, 10% normal goat serum (Abcam, Cambridge, UK), and 0.1% Triton X‐100 (Merck KGaA, Darmstadt, Germany) for 60 min. Anti‐HNF1A and anti‐KDM6A stainings were performed at a dilution of 1:250 (Anti‐HNF1A, Abcam ab204306, Cambridge, UK), 1:200 (Anti‐HNF1A, Cell Signaling Technology, 89670, Leiden, The Netherlands), and 1:100 (Anti‐UTX, Cell Signaling Technology 33510S, Denver, USA), respectively.

Techniques: Immunoprecipitation, Western Blot, Binding Assay, Two Tailed Test, MANN-WHITNEY, Mutagenesis, Activity Assay