412 Diabetes Volume 67, March 2018

The p300 and CBP Transcriptional Coactivators Are Required for b-Cell and a-Cell Proliferation

Chi Kin Wong,1,2 Adam K. Wade-Vallance,2 Dan S. Luciani,2,3 Paul K. Brindle,4 Francis C. Lynn,2,3,5 and William T. Gibson1,2

Diabetes 2018;67:412–422 | https://doi.org/10.2337/db17-0237

p300 (EP300)andCBP(CREBBP)aretranscriptionalco- expression synergistically in b-cells (1). These transcription activators with histone acetyltransferase activity. Various factors also recruit coregulators to fine-tunegeneexpres- b-cell transcription factors can recruit p300/CBP, and sion. p300 and CBP (p300/CBP) are transcription coac- thus the coactivators could be important for b-cell func- tivators that share .60% sequence identity and tion and health in vivo. We hypothesized that p300/CBP exhibit highly similar functions. These coactivators acet- contribute to the development and proper function of pan- ylate lysine residues on histones to modulate chromatin creatic islets. To test this, we bred and studied mice lack- structure or function, and lysine residues on nonhistone ing p300/CBP in their islets. Mice lacking either p300 to modulate their activities (2). Although p300/CBP or CBP in islets developed glucose intolerance attribut- can acetylate most histone proteins, they are absolutely es- able to impaired secretion, together with reduced sential for acetylating histone H3 lysine 27 (3). The H3K27Ac a-andb-cell area and islet insulin content. These pheno- mark tags tissue-specific promoters and enhancers and sig- types were exacerbated in mice with only a single copy nals transcription of the tissue-specifictargetgenes(3,4). of p300 or CBP expressed in islets. Removing p300 in b pancreatic endocrine progenitors impaired proliferation p300/CBP appear to regulate important -cell functions of neonatal a-andb-cells. Mice lacking all four copies in vitro. For instance, p300/CBP coactivate insulin ISLET STUDIES of p300/CBP in pancreatic endocrine progenitors failed expression in vitro by binding synergistically to Pdx1 and to establish a-andb-cell mass postnatally. Transcriptomic NeuroD1/E47 (5). Small interfering RNA knockdown of analyses revealed significant overlaps between p300/ p300/CBP in INS1 cells reduced glucose-stimulated insulin CBP-downregulated and genes downregulated in (6). In contrast, CRISPR-Cas9–mediated Hnf1a-null islets and Nkx2.2-null islets, among others. deletion of p300 in INS1 832/13 cells induced a subtle in- Furthermore, p300/CBP are important for the acetylation crease in glucose-stimulated insulin secretion and reduced of H3K27 at loci downregulated in Hnf1a-null islets. We high glucose-mediated apoptosis (7). Mice with the S436A conclude that p300 and CBP are limiting cofactors for islet variant in both copies of CBP, a mutation that renders CBP development, and hence for postnatal glucose homeosta- unresponsive to insulin-triggered phosphorylation, had in- sis, with some functional redundancy. creased islet mass but relatively normal b-cell function (8). These data left unresolved whether p300/CBP expression in is necessary for establishing glucose ho- The expression of specific transcription factors both deter- meostasis in vivo. We hypothesized that the removal of mines and maintains the identities of pancreatic endocrine p300/CBP from pancreatic endocrine progenitors would cells through activation of endocrine genes. For example, lead to postnatal glucose intolerance due to defects in islet Pdx1, MafA, and NeuroD1 form a transcriptional complex mass and function. In this study, we generated and at the insulin promoters and enhancers to activate insulin phenotyped Neurog3-Cre–driven pancreatic islet-specific

1Department of Medical Genetics, University of British Columbia, Vancouver, Corresponding author: Chi Kin Wong, [email protected]. British Columbia, Canada Received 21 February 2017 and accepted 21 November 2017. 2BC Children’s Hospital Research Institute, Vancouver, British Columbia, Canada This article contains Supplementary Data online at http://diabetes 3Department of Surgery, University of British Columbia, Vancouver, British Co- .diabetesjournals.org/lookup/suppl/doi:10.2337/db17-0237/-/DC1. lumbia, Canada 4St. Jude Children’s Research Hospital, Memphis, TN © 2017 by the American Diabetes Association. Readers may use this article as 5Department of Cellular & Physiological Sciences, University of British Columbia, long as the work is properly cited, the use is educational and not for profit, and the Vancouver, British Columbia, Canada work is not altered. More information is available at http://www.diabetesjournals .org/content/license. diabetes.diabetesjournals.org Wong and Associates 413 p300 and CBP knockout (KO) mice to study the roles of assays were performed on a Biorep Perifusion System per these coactivators in pancreatic islets in vivo. manufacturer instructions using 100 islets per chamber (Biorep Technologies, Miami Lakes, FL). RESEARCH DESIGN AND METHODS Immunofluorescence Staining fi Animals Adult pancreata were xed in 10% formalin for 24 h at 4°C, fi m All procedures were approved by the University of British dehydrated, and embedded in paraf n, and 5- m serial fi Columbia Animal Care Committee. Mice housed at BC sections were made. A total of four to ve sections, each m Children’s Hospital Research Institute were under a 12-h separated by 150 m, were obtained per adult . For light/dark cycle with ad libitum access to standard chow E15.5 and E18.5, sections were obtained from the entire m (Teklad 2918; Envigo, Huntingdon, U.K.) and water. Neurog3- pancreas and sections separated by 30 m were stained. m Cre mice were obtained from Dr. Francis Lynn, BC Children’s For P0 and P7, sections separated by 60 m were obtained. Hospital Research Institute, Vancouver, British Columbia, These sections were stained for insulin, glucagon, somato- fl fl Canada (9). Ep300 / mice were obtained from Dr. Paul statin, ghrelin, chromogranin A, and/or Ki67. Other pro- Brindle, St. Jude Children’s Research Hospital, Memphis, teins were stained using randomly chosen sections. After fl fl TN (10). Crebbp / mice were purchased from The Jackson blocking, the sections were incubated with primary anti- Laboratory (Bar Harbor, ME) (10). All mice were kept on bodies overnight at 4°C followed by incubation with second- C57BL/6J background. Cre-negative littermates from each ary antibodies. Primary antibodies used include rabbit breeding setup were used as wild-type (WT) controls. Timed anti-p300 (N-15 + C-20 1:1, 1/50; Santa Cruz Biotechnology, matings were used to study embryonic day 15.5 (E15.5), E18.5, Dallas, TX), rabbit anti-CBP (1/200; CST America, Framingham, postnatal day 0 (P0), and P7 mice; the morning when a vaginal MA), guinea pig anti-insulin (1/200; Abcam, Cambridge, plug was found on the dams was designated as E0.5. U.K.), mouse anti-glucagon (1/1,000; Abcam), rabbit anti- somatostatin (1/400; Abcam), goat anti-somatostatin (1/200; Metabolic Phenotyping Santa Cruz Biotechnology), goat anti-ghrelin (1/100; Santa For glucose tolerance test, mice were fasted for 5 h and then Cruz Biotechnology), rabbit anti-Ki67 (1/200; CST Amer- injected intraperitoneally with 2 g/kg glucose. For insulin ica), rabbit anti-chromogranin A (1/200; Abcam), goat anti- tolerance test, mice were injected with 0.7 units/kg Humulin chromogranin A (1/200; Santa Cruz Biotechnology), R (Eli Lily, Indianapolis, IN). Blood was sampled from mouse mouse anti-Ngn3 (1/50; Developmental Studies Hybridoma tails, and blood glucose levels were assessed using a OneTouch Bank at the University of Iowa, IowaCity,IA),rabbitanti- Ultra Mini Glucometer (Johnson & Johnson, New Bruns- H3K27Ac (1/200; CST America), and rabbit anti-H3K27me3 wick, NJ). For plasma insulin measurement, mice were (1/200; CST America). The TUNEL assays were performed fasted for 5 h, and blood was sampled from the saphenous with the In Situ Cell Death Detection Kit (Sigma-Aldrich, veins using heparinized capillary tubes before and after St. Louis, MO). Representative images of individual islets glucose injection. Heparinized blood was centrifuged at were taken on a SP5 Confocal Microscope (Leica, Wetzlar, 2,000g for 15 min at 4°C to separate plasma. Body compo- Germany). Images of whole pancreas sections were tiled on sition analysis and metabolic cage experiments were per- a BX61 Fluorescence Microscope (Olympus, Tokyo, Japan) formed as previously described (11). and quantified using Fiji (14). Islet endocrine area were calculated by dividing the corresponding stained area by Analyte Measurement the total pancreas area. For E15.5 studies, the total number fi Insulin was quanti ed using STELLUX Chemiluminescent of cells per population were counted and normalized to Immunoassays (ALPCO, Salem, NH). Glucagon was quanti- total DAPI count per section. fied using Glucagon ELISA (Mercodia, Uppsala, Sweden). Somatostatin was quantified using a Somatostatin EIA Transcriptomic Analyses (enzyme immunosorbent assay) Kit (Phoenix Pharmaceut- Total RNA was extracted from islets using TRIzol Reagent fi icals, Burlingame, CA). Total glucagon-like peptide 1 (GLP-1) (Thermo Fisher Scienti c, Waltham, MA) and the RNeasy was quantified using Multi Species GLP-1 Total ELISA Micro Kit (Qiagen, Hilden, Germany). For RNA sequencing IsletKO IsletKO (Merck Millipore, Burlington, MA). (RNA-seq), six WT, three p300 ,threeCBP ,and three CBPHet;p300KO samples were sequenced in two Ex Vivo Islet Assays batches at the University of British Columbia Biomedical Re- Mouse pancreatic islets were isolated as described pre- search Centre Sequencing Core. RNA samples with an RNA viously (12). For glucose-stimulated insulin secretion assay, integrity number .8.5 as measured on a model 2100 Bio- overnight recovered islets were incubated in Krebs-Ringer analyzer (Agilent Technologies, Santa Clara, CA) were pre- buffer containing 2.8 mmol/L glucose for 1 h at 37°C. After pared into sequencing libraries on NeoPrep using TruSeq the preincubation, 30 islets were incubated in Krebs-Ringer Stranded mRNA Library Prep Kit (Illumina, San Diego, CA). buffer containing either 2.8, 16.7, or 2.8 mmol/L glucose Each library was sequenced to a depth of at least 20 million with 30 mmol/L KCl for 1 h at 37°C. Supernatants were paired end reads on a NextSEq 500 Sequencing System collected for insulin measurement. Fura-2 calcium imaging (Illumina). Quality filtered reads were aligned to reference was performed as described previously (13). Perifusion mouse genome mm10 with TopHat2 (15). Count tables for 414 Roles of p300/CBP in Pancreatic Islets Diabetes Volume 67, March 2018 the aligned reads were generated, and batch effects were They also appeared phenotypically identical to Cre-negative fl fl corrected using SVA (Surrogate Variable Analysis) (16), Ep300 / mice and to mice bearing the Neurog3-Cre trans- followed by calling of differentially expressed genes using gene alone (data not shown). We then made Neurog3-Cre; fl fl DESeq2 with default settings in R (17). Genes were defined Ep300 / mice (p300IsletKO mice). At E15.5, p300 was effec- as differentially expressed based on an adjusted P value tively removed using Neurog3-Cre in up to 95% of chromog- of ,0.05. The Venn diagram for overlapping downregu- ranin A–positive cells, while Ngn3-positive progenitors lated genes was generated using BioVenn (18). Downregu- remained p300 positive (Supplementary Fig. 1). This is likely lated gene lists were uploaded to Webgestalt for gene due to a time lag between the onset of Cre expression and ontology (GO) terms analysis and tar- the onset of recombination. Among the chromogranin get prediction (19), using a reference list of 15,999 islet A–positive cells, recombination occurred in newly formed genes from the WT islet transcriptome for accurate over- b-cells, a-cells, d-cells, and e-cells (Supplementary Fig. 1, representation analyses. The lowest false discovery rate from d-cells; other cell types are not shown). In p300IsletKO mouse 2 Webgestalt was capped at 10 15. Manual gene set enrich- pancreata, p300 was absent in islet nuclei but was expressed ment analyses were performed by applying hypergeomet- normally in the exocrine tissues (Fig. 1A and B). The protein ric test with Bonferroni correction to the overlapped genes levels of CBP were similar between WT and p300-null islets, between different gene sets. All RNA-seq data were deposited indicating the absence of compensatory overexpression of in the Gene Expression Omnibus database (GSE101537). aCBPparalog. For quantitative PCR (qPCR), RNAs were reverse p300IsletKO mice were glucose intolerant at 8 weeks of transcribed using the iScript cDNA Synthesis Kit (Bio-Rad, age but displayed normal insulin tolerance (Fig. 1C and D). Hercules, CA) and quantified by SYBR green–based reaction Plasma insulin levels of p300IsletKO mice were 60% lower using GoTaq qPCR Master Mix (Promega, Madison, WI) and thanthoseofWTmicebothbeforeandafterglucosein- primers (Supplementary Table 1) on an ABI 7500 Real-Time jection (Fig. 1E). Theoretically, defective glucose PCR System (Thermo Fisher Scientific). Data were normal- in p300IsletKO mice could be in part due to the Neurog3-Cre– ized to 18s rRNA, and fold changes were calculated with the mediated recombination outside of islets such as in the ven- ΔΔCT method (20). tromedial hypothalamus and enteroendocrine cells (9,23). We found that WT and p300IsletKO mice had similar body Western Blotting composition, energy expenditure, locomotor activity, and Islet nuclear extracts were prepared using the NE-PER food intake (Supplementary Fig. 2A–D). Also, plasma total Nuclear and Cytoplasmic Extraction Kit (Thermo Fisher IsletKO GLP-1 levels were normal in p300 mice (Supplemen- Scientific). One microgram of nuclear lysate was subject to tary Fig. 2E). In the absence of observable extrapancreatic Western blotting using rabbit anti-p300 (N-15 + C-20 1:1, phenotypes, the glucose intolerance and hypoinsulinemia 1/1,000; Santa Cruz Biotechnology), rabbit anti-CBP (1/1,000; can be attributed to the recombination within pancreatic Santa Cruz Biotechnology), and mouse anti–TATA binding islets rather than other Neurog3-expressing tissues. protein (1/5,000; Abcam), as described previously (21).

Low-Input Native Chromatin Immunoprecipitation p300IsletKO Mice Have Reduced Islet Area and Islet The low-input native chromatin immunoprecipitation (N-ChIP) Insulin Content IsletKO protocol was carried out as previously described (22). One hun- To examine how p300 mice developed glucose intol- dred islets were dispersed in 0.05% trypsin and flash frozen erance and hypoinsulinemia, we first quantified islet cell IsletKO prior to lysis. Buffers were supplemented with 10 mmol/L area in adult mice. Although the weights of p300 sodium butyrate to retain acetylation signals. Cells were lysed mouse pancreata were comparable to WT mice (Supplemen- and digested with MNase for 5 min, as validated using WT tary Fig. 3A), they showed a 25% reduction in a-cell and islet cells. Chromatin equivalent to 10 islets per sample was b-cell area (Fig. 2A). The d-cell area was unaffected. The subjected to ChIP using rabbit anti-H3K27Ac or rabbit anti- reduced a-cell and b-cell areas were attributable to a reduced H3K27me3 (both 2 mL/ChIP;CSTAmerica).ChIP’d DNA was number of islets rather than islet size (Fig. 2B and Supple- quantified by qPCR and expressed as the percentage of input. mentary Fig. 3B). The b-cell-to-a cell ratios were similar between WT and p300IsletKO mouse pancreata (Supplemen- Statistics tary Fig. 3C). Insulin content of p300-null islets was re- 6 fi Data were shown as mean SD. Statistical signi cance was duced by 19%, whereas glucagon content was unchanged tested using the Student t test, one-way ANOVA, or two-way (Fig. 2C). The somatostatin content was elevated by nearly , ANOVA, as appropriate, with P 0.05 considered to be twofold. Fasting plasma glucagon levels of p300IsletKO mice fi statistically signi cant. were unaffected (Supplementary Fig. 3D). To assess the function of p300-null islet explants, we stimulated the islets RESULTS with low glucose, high glucose, or KCl and measured their Mice Lacking p300 in Pancreatic Islets Develop Glucose insulin release. Although both WT and p300-null islets Intolerance Due to Hypoinsulinemia responded similarly to low or high glucose, p300-null islets fl We first characterized Neurog3-Cre; Ep300 /WT mice that had increased insulin secretion and calcium response upon did not develop glucose intolerance up to 24 weeks of age. KCl stimulation (Fig. 2D and E). Glucagon secretion from diabetes.diabetesjournals.org Wong and Associates 415

Figure 1—Mice lacking p300 in pancreatic islets develop glucose intolerance due to hypoinsulinemia. A: Western blotting for p300 and CBP in isolated WT or p300-null islet nuclear extracts. TBP (TATA binding protein) was used as a loading control. The experiment was replicated once. B: Representative immunofluorescence images of p300 and CBP in the pancreatic islets of WT and p300IsletKO mice. n = 3. Scale bar = 50 mm. Insulin, Ins; glucagon, Gcg. C: Glucose tolerance test of 8-week-old WT and p300IsletKO mice. n =8–9. D: Insulin tolerance test of 9-week-old WT and p300IsletKO mice. n =5–7. E: Plasma insulin measurement before and 15 min after glucose injection. n =4–5. Two-way ANOVA for C–E.*P , 0.05; **P , 0.01; ***P , 0.001.

p300-null islets under low-glucose conditions was not sig- Mice With Only a Single Copy of p300 or CBP in Islets nificantly different from that of the WT islets (Supplemen- Develop More Severe Glucose and Islet Phenotypes tary Fig. 3E). Thus, the combined defects in islet mass and Than Mice Lacking p300 or CBP Alone in Islets islet insulin content result in deficient glucose-stimulated Since mice lacking p300 or CBP had similar phenotypes, we IsletKO IsletKO insulin secretion in p300IsletKO mice. hypothesized that p300 or CBP mice lacking an additional copy of p300 or CBP would develop more severe CBPIsletKO Mice Share Similar b-Cell Phenotypes With glucose phenotypes. To test this, we generated Neurog3-Cre; IsletKO fl fl fl fl fl p300 Mice Crebbp /WT; Ep300 / mice and Neurog3-Cre; Crebbp / ; fl To understand whether p300 and CBP function similarly in Ep300 /WT mice (CBPHet;p300KO and CBPKO;p300Het mice, fl/fl pancreatic islets, we generated Neurog3-Cre; Crebbp respectively). These triallelic mice developed severe glucose IsletKO (CBP ) mice. The deletion of CBP in islets was not com- intolerance by 8 weeks of age with no defects in insulin pensated for by the overexpression of p300 (Fig. 3A). tolerance (Fig. 4A and B and Supplementary Fig. 4, IsletKO CBP mice developed glucose phenotypes similar to CBPKO;p300Het mice data). Unlike the biallelic mice, triallelic IsletKO those of p300 mice, including glucose intolerance at p300/CBP mice of either genotype failed to mount an in- 8 weeks of age without insulin resistance and defective in- sulin response to glucose challenge (Fig. 4C). In addition, sulin release upon glucose injection (Fig. 3B–D). Unlike CBPHet;p300KO mice had 58% less a-cell area and 45% less IsletKO IsletKO p300 mice, fasting plasma insulin levels of CBP b-cell area (Fig. 4D). Their islets contained 72% less insulin mice did not differ from those of controls (Fig. 3D). The than WT islets (Fig. 4E). These phenotypes recapitulated IsletKO pancreata of CBP mice had 40% less a-cell area and those seen in p300IsletKO or CBPIsletKO mice despite being 30% less b-cell area, with d-cell area unaffected (Fig. 3E). In more severe. contrast to p300-null islets, CBP-null islets had reduced glu- cagon content but normal somatostatin content compared Expression of p300/CBP Is Necessary for Neonatal with controls (Fig. 3F). Similar to p300-null islets, CBP-null b-Cell and a-Cell Proliferation islets had reduced insulin content, but their b-cell secretory Since reduced a-cell and b-cell areas could be caused by function appeared to be normal (Fig. 3F and G). Overall, defects in differentiation, proliferation, and/or apoptosis, both p300IsletKO and CBPIsletKO mice exhibited reduced we examined these processes throughout pancreas develop- b-cell area and islet insulin contents, features that explain ment in WT and p300IsletKO mice. We first excluded apo- their glucose intolerance. ptosis as a possible cause of islet cell loss by performing 416 Roles of p300/CBP in Pancreatic Islets Diabetes Volume 67, March 2018

Figure 2—p300IsletKO mice display defects in islet mass and islet insulin content. A: Quantification of b-cell, a-cell, and d-cell areas of adult WT and p300IsletKO mice pancreata. n =8forb-cells; n =4–5fora-cells and d-cells. B: Islet density of adult WT and p300IsletKO mice pancreata. n =5. C: Insulin (Ins), glucagon (Gcg), and somatostatin (Sst) content of WT and p300-null islets. n =5–6. D: Glucose-stimulated insulin secretion assays on isolated WT and p300-null islets. n =5–7. E: Fura-2 calcium imaging on isolated WT and p300-null islets. n =3.Studentt test for A–D. Two-way ANOVA for E.*P , 0.05; **P , 0.01.

TUNEL assays on E18.5 p300IsletKO mouse pancreata and the weaning age, in contrast to the triallelic p300/CBP mice, adult p300IsletKO mouse pancreata; apoptotic events in which were observed at the expected Mendelian ratio (Sup- p300-null islets were as rare as in WT islets (data not plementary Table 2). We speculated that these p300/CBP shown). Next, the number of newly differentiated endo- double-KO mice might die shortly after birth because of crine cells and Ngn3-positive endocrine progenitors were a failure to establish sufficient b-cell mass. At P0, some unaffected in E15.5 p300IsletKO mouse pancreata (Supple- double-KO pups survived, but their pancreata lacked mentary Fig. 5A). At E18.5, a-cell, b-cell, and pan-endocrine a-cells and b-cells completely (Fig. 5E and F). Surprisingly, cell areas were normal in p300IsletKO mouse pancreata their d-cell and e-cell populations were unaffected (Fig. 5E). (Supplementary Fig. 5B). At P7, the a-cell and b-cell areas Immunostaining showed that a few e-cells, which are nor- were reduced in p300IsletKO mouse pancreata (Fig. 5A). The mally absent in adult mouse pancreata, persist in the bial- percentages of Ki67+ a-cell and b-cells in these pancreata lelic and triallelic mouse pancreata (Supplementary Fig. 6A). were lower than that of WT; however, the overall percent- Thus, at least one allele of p300 or CBP is necessary for age of Ki67+ cells in the pancreata was unchanged (Fig. 5B normal development of a-cells and b-cells, but not for and C). This indicated that the proliferation of neonatal d-cells or e-cells. a-cells and b-cells was reduced in p300-null islets. Hence, IsletKO the reduced a-cell and b-cell mass in the adult p300 Loss of p300/CBP Impairs Genes Associated With mouse pancreata originated after E18.5 and is attributable Multiple Islet/b-Cell Transcription Factors and Impairs to impaired proliferation. the Coactivation of Hnf1a-Associated Genes In Vivo fl fl We attempted to breed for Neurog3-Cre; Crebbp / ; Because p300/CBP are transcriptional coactivators, the loss fl fl Ep300 / mice (p300/CBP double-KO mice), but we did not of p300/CBP may reduce the expression of genes important observe any of the double-KO mice in a cohort of 59 pups at for islet function or development. We examined gene diabetes.diabetesjournals.org Wong and Associates 417

Figure 3—CBPIsletKO mice share similar phenotypes with p300IsletKO mice. A: Representative immunofluorescence images of p300 and CBP in the pancreatic islets of WT and CBPIsletKO mice. Scale bar = 50 mm. B: Glucose tolerance test of 8-week-old WT and CBPIsletKO mice. n =4–11. C: Insulin tolerance test of 9-week-old WT and CBPIsletKO mice. n =5–8. D: Plasma insulin measurement of WT and CBPIsletKO mice before and 15 min after glucose injection. n =6–7. E: Quantification of b-cell, a-cell, and d-cell areas of adult WT and CBPIsletKO mice pancreata as the percentage of total pancreas area. n =4–5fora-cells and b-cells; n =5–6ford-cells. F: Islet insulin (Ins), glucagon (Gcg), and somatostatin (Sst) content of WT and CBP-null islets as quantified by ELISA. n =4.G: Perifusion assay for insulin secretion of WT and CBP-null islets. n =3.Two- way ANOVA for B–D.Studentt test for E and F.*P , 0.05; **P , 0.01; ***P , 0.001.

expression by performing RNA-seq on islet mRNAs from Hnf1a (Fig. 6C and Supplementary Table 5). We also per- WT, p300IsletKO,CBPIsletKO,andCBPHet;p300KO mice. We formed gene set enrichment analysis by comparing our gene identified 761 (477 down, 284 up), 923 (513 down, sets to published downregulated genes in mouse islets lack- 410 up), and 5,589 (2,411 down, 3,178 up) differentially ing factors important for b-cell development and function, expressed genes in p300-null, CBP-null, and triallelic islets includingPdx1,NeuroD1,Hnf1a,Pax6,MafA,Nkx6.1,and relative to WT islets, respectively (Supplementary Table 3). Nkx2.2 (24–30). Our gene sets overlapped more signifi- We focused our analyses on the downregulated genes. cantly with the gene sets of Hnf1a and Nkx2.2, followed The aggregation of the downregulated gene sets revealed by MafA, Nkx6.1, Pdx1, and NeuroD1 (Fig. 6D and Supple- 230 downregulated genes overlapped between p300-null mentary Tables 6 and 7). islets (48.2%) and CBP-null islets (44.8%) (Fig. 6A). The Because Hnf1a could recruit p300/CBP for coactivation genes downregulated in CBPHet;p300KO islets overlapped (31), we further examined the genes that overlap between with 436 (91.4%) and 437 (85.2%) of the downregulated the Hnf1a gene set and the gene sets we had defined as genes from p300-null and CBP-null islets, respectively. En- downregulated genes in p300-null islets, CBP-null islets, and richment analyses of the Biological Process GO terms on all CBPHet;p300KO islets. Tmem27,aknownHnf1a-mediated three sets suggested three common themes of genes down- regulator of b-cell proliferation (32), was reduced in all three regulated by the loss of p300/CBP: lipid metabolic process, models. Other loci downregulated in Hnf1a-null islets, in- regulation of hormone levels, and ion transport (Fig. 6B and cluding Pklr, Slc2a2,andG6pc2, were also downregulated Supplementary Table 4). Transcription factor target predic- in triallelic islets as validated by qPCR (Fig. 6E) (24). b-Cell tion from Webgestalt showed that all three gene sets were transcription factors were not specifically downregulated significantly enriched for the predicted transcription factor in either p300-null or CBP-null islets, whereas Hnf4a, 418 Roles of p300/CBP in Pancreatic Islets Diabetes Volume 67, March 2018

Figure 4—Triallelic deletion of p300/CBP in islets leads to severe glucose intolerance. A: Intraperitoneal glucose tolerance test of 8-week-old WT and CBPHet;p300KO mice. n =6.B: Insulin tolerance test of adult WT and CBPHet;p300KO mice. n =6–7. C: Plasma insulin measurement of WT and CBPHet;p300KO mice before and 15 min after glucose injection. n =7.D:Quantification of b-cell, a-cell, and d-cell areas of adult WT and CBPHet;p300KO mouse pancreata as percent of total pancreas area. n =4–6fora-cells and b-cells; n =4ford-cells. E: Islet insulin content of WT and CBPHet;p300KO islets as quantified by ELISA. n = 4. Two-way ANOVA for A–C.Studentt test for D and E.**P , 0.01; ***P , 0.001.

Hnf1b,andNeuroD1 were downregulated in triallelic islets the reduced dosages of p300/CBP impaired coactivation of (Supplementary Table 8). Insulin-processing genes were not downregulated genes in Hnf1a-null islets, which we attri- altered in the biallelic mouse islets. Ins1 and Ins2 mRNAs bute to reductions in global and loci-specific H3K27Ac levels. were normally expressed in the biallelic mouse islets, although both were downregulated by .50% in the triallelic islets. DISCUSSION Because p300/CBP coactivate transcription factors in In this study, the loss of either p300 or CBP alone in the part by acetylating H3K27 at target promoters and pancreatic islets was sufficient to perturb whole-body glu- enhancers, we hypothesized that the loss of p300/CBP cose homeostasis. Mice lacking p300 or CBP in islets devel- would reduce H3K27 acetylation at the loci downregulated oped similar b-cell phenotypes, including reduced b-cell in Hnf1a-null islets. We assessed the acetylation and area and insulin content. Mechanistically, p300 and CBP methylation statuses of H3K27 at various loci using low- are known to coactivate Pdx1, NeuroD1, Hnf4a,andHnf1a/ input N-ChIP, and found that there was significantly less b in vitro (33–35). Our RNA-seq data suggested that genes H3K27Ac at the promoters of G6pc2, Hnf4a, Pklr,and downregulated in Hnf1a-null islets became downregulated Tmem27 in the triallelic islets (Fig. 6F and Supplementary once the dosages of either p300 or CBP were reduced in the Fig. 6B, negative loci). Pdx1-associated genes also showed islets. Hnf1a/b are transcription factors that are reduced H3K27Ac at their promoters and enhancers in the critical for pancreas and b-cell development (36,37). In triallelic islets (Fig. 6G). The H3K27Ac levels at these loci particular, impaired Hnf1a coactivation in our mouse mod- were reduced in CBP-null islets, although the reduction els could attenuate b-cell proliferation through genes such did not reach statistical significance. These loci-specific as Tmem27 (32). The role of Hnf1a in a-cells remains un- H3K27Ac levels clearly correlated with the total dosages clear, although high levels of HNF1a were found in FACS- of p300/CBP in the cells. We confirmed an ;60% reduction sorted human a-cells, thereby implying that p300/CBP of H3K27Ac globally in the triallelic islet nuclei (Fig. 6H). might also regulate aspects of a-cell biology, such as prolif- Total and loci-specific H3K27me3 levels were unaffected in eration, through HNF1a (38). p300/CBP bind to Hnf1a/b triallelic islets (Fig. 6H and Supplementary Fig. 6B). Overall, through their transactivation domains, and coactivate their diabetes.diabetesjournals.org Wong and Associates 419

Figure 5—Expression of p300/CBP is necessary for b-cell and a-cell development. A: Quantification of b-cell, a-cell, and d-cell areas of P7 WT and p300IsletKO mouse pancreata as the percentage of total pancreas area. n =4–6. B: Representative immunofluorescence images of insulin, glucagon, and Ki67 in P7 WT and p300IsletKO mouse pancreata. Scale bar = 50 mm. C:Quantification of Ki67+ b-cells, a-cells, and all pancreatic cells in P7 WT and p300IsletKO mouse pancreata as the percentage of total b-cells, a-cells, and total pancreatic cells. n =8.D: Quantification of b-cell, a-cell, d-cell, e-cell, and chromogranin A–positive pan-endocrine cell areas of P0 WT and p300/CBP double-KO (dKO) mouse pancreata as the percentage of total pancreas area. n =3.E: Representative immunofluorescence images of insulin (Ins), glucagon (Gcg), somatostatin (Sst), ghrelin (Ghrl), chromogranin A (ChrA), and DAPI in P0 WT and p300/CBP double-KO mouse pancreata. Scale bar = 50 mm. Student t test for A, C,andD.*P , 0.05; **P , 0.01.

downstream targets by acetylating the histones bound to activator function instead (30,42). In the future, it will be regulatory elements affiliated with these targets (39). The interesting to explore whether p300/CBP interact physically observed loss of H3K27Ac in triallelic islets at loci down- with Nkx2.2 and acetylate the H3K27 residues at Nkx2.2- regulated in Hnf1a-null islets appears to be in line with associated loci, and whether the genomic occupancy of such a mechanism. Nkx2.2 or Hnf1a in islets is affected by p300 deletion. Although Hnf1a might be one of the targets in p300- Both Ins1 and Ins2 mRNAs were reduced in the triallelic null/CBP-null islets, the coactivation of other transcription p300/CBP islets but not in p300-null islets or CBP-null factors could also account for the phenotypes of p300/CBP- islets. Reduced transcriptional activities of MafA and null islets. Our data suggest that p300/CBP do not appear Nkx6.1, which are not known to recruit p300/CBP previ- to have major importance in the development of d-cells ously, might contribute indirectly to the reduced insulin or e-cells. The lack of effect on d-cells and e-cells in the gene expression seen in triallelic mice. Alternatively, p300/ double-KO mice shows striking similarity to the phenotypes CBP might regulate insulin gene expression by binding to of Nkx2.2-null mice (40,41). Although Nkx2.2 is not known Pdx1 and NeuroD1 (6,35). The acetylation of H3K27 at the to interact with p300/CBP, the significant overlapping be- Ins1 promoter correlated with the dosages of p300/CBP tween the gene set of Nkx2.2 and our p300/CBP gene sets present in the islets. The reduced insulin gene expression suggested that Nkx2.2 might mediate some of the pheno- in triallelic islets could be a consequence of less p300/CBP types seen in the p300/CBP mutant mice. Intriguingly, available to b-cell transcription factors, which in turn im- Nkx2.2 is mainly known for its repressor function, so pairs the acetylation of H3K27 at insulin promoters. Taken p300/CBP might be recruited by Nkx2.2 to initiate its together, p300/CBP may coordinate transcriptional networks 420 Roles of p300/CBP in Pancreatic Islets Diabetes Volume 67, March 2018

Figure 6—Loss of p300/CBP impairs coactivation of Hnf1a through reduced H3K27 acetylation. A: Venn diagram of overlapping downregulated genes of p300IsletKO,CBPIsletKO,andCBPHet;p300KO mouse islets compared with WT islets. B: The three Biological Process GO terms commonly overrepresented in the downregulated genes of p300IsletKO,CBPIsletKO,andCBPHet;p300KO mouseislets.Allsignificantly overrepresented GO terms and their associated genes can be found in Supplementary Table 4. C: Transcription factor target analysis by Webgestalt on the downregulated genes of p300IsletKO,CBPIsletKO,andCBPHet;p300KO mouse islets. Hnf1a was commonly overrepresented in all three gene sets. D: Gene set enrichment analysis on downregulated gene sets derived from microarray or RNA-seq data of mice lacking b-cell transcription factors in islets or b-cells. Random 1 and 2 were control gene lists generated randomly from the 15,999 genes in the WT reference list. E:qPCR of islet Hnf1a-associated genes in WT and CBPHet;p300KO mouse islets. n =5–6. F: Low-input N-ChIP for H3K27Ac at Hnf1a-associated genes in WT, CBPIsletKO,andCBPHet;p300KO mouse islets. n =3–5. G: Low-input N-ChIP for H3K27Ac at Pdx1-associated loci in in WT, CBPIsletKO,and CBPHet;p300KO mouse islets. n =3–5. H: Representative immunofluorescence images of insulin, H3K27Ac, and H3K27me3 in WT and CBPHet;p300KO mouse islets. Scale bar = 50 mm. FDR, false discovery rate. Student t test for E.One-wayANOVAforF and G.*P , 0.05; **P , 0.01; ***P , 0.001. diabetes.diabetesjournals.org Wong and Associates 421 in b-cells by coactivating various b-cell transcription fac- 2. Bedford DC, Brindle PK. Is histone acetylation the most important physiological tors, perhaps through Hnf1a/b and/or Nkx2.2. Mutations function for CBP and p300? Aging (Albany NY) 2012;4:247–255 in many of these transcription factors are known to cause 3. Heinz S, Romanoski CE, Benner C, Glass CK. The selection and function of cell fi – monogenic diabetes, including HNF1A, HNF1B, PDX1,and type-speci c enhancers. Nat Rev Mol Cell Biol 2015;16:144 154 4. Bose DA, Donahue G, Reinberg D, Shiekhattar R, Bonasio R, Berger SL. RNA NEUROD1 (43), suggesting that p300/CBP could have - binding to CBP stimulates histone acetylation and transcription. Cell 2017;168:135– evancy to the underlying pathophysiology. 149.e22 Overall, mice lacking p300 or CBP alone in islets 5. Qiu Y, Guo M, Huang S, Stein R. Acetylation of the BETA2 transcription factor by developed glucose intolerance and hypoinsulinemia associ- p300-associated factor is important in insulin gene expression. J Biol Chem 2004; ated with reduced islet area and insulin content. Mice 279:9796–9802 lacking three copies of p300/CBP in islets developed similar 6. Sampley ML, Ozcan S. Regulation of insulin gene transcription by multiple yet exacerbated phenotypes. Mice lacking all copies of histone acetyltransferases. DNA Cell Biol 2012;31:8–14 p300/CBP died postnatally due to their failure to establish 7. Bompada P, Atac D, Luan C, et al. Histone acetylation of glucose-induced b-cell mass. Islet genes mediated by p300/CBP overlapped thioredoxin-interacting protein gene expression in pancreatic islets. Int J Biochem significantly with genes downregulated in islets lacking tran- Cell Biol 2016;81(Pt A):82–91 scription factors such as Hnf1a and Nkx2.2. p300/CBP 8. Hussain MA, Porras DL, Rowe MH, et al. Increased pancreatic beta-cell pro- expression was required to acetylate H3K27 at the loci liferation mediated by CREB binding protein gene activation. Mol Cell Biol 2006;26: 7747–7759 downregulated in Hnf1a-null islets including Slc2a2, Pklr, b 9. Schonhoff SE, Giel-Moloney M, Leiter AB. Neurogenin 3-expressing progenitor Hnf4a, and particularly Tmem27, which could regulate -cell cells in the gastrointestinal tract differentiate into both endocrine and non-endocrine proliferation. Thus, the expression of p300/CBP family of cell types. Dev Biol 2004;270:443–454 coactivators in islets is critical to drive b-cell genesis and to 10. Kasper LH, Fukuyama T, Biesen MA, et al. Conditional knockout mice reveal maintain b-cell proliferation and insulin content. In the distinct functions for the global transcriptional coactivators CBP and p300 in T-cell pancreatic endocrine lineage, p300 and CBP serve as func- development. Mol Cell Biol 2006;26:789–809 tionally similar yet limiting cofactors to coordinate various 11. Wong CK, Botta A, Pither J, Dai C, Gibson WT, Ghosh S. A high-fat diet rich in islet transcription factors and maintain whole-body glucose corn oil reduces spontaneous locomotor activity and induces insulin resistance homeostasis. in mice. J Nutr Biochem 2015;26:319–326 12. Li D-S, Yuan Y-H, Tu H-J, Liang Q-L, Dai L-J. A protocol for islet isolation from mouse pancreas. Nat Protoc 2009;4:1649–1652 13. Luciani DS, White SA, Widenmaier SB, et al. Bcl-2 and Bcl-xL suppress glucose Acknowledgments. The authors thank The Canucks for Kids Childhood signaling in pancreatic b-cells. Diabetes 2013;62:170–182 Diabetes Laboratories at BC Children’s Hospital Research Institute (BCCHR) for 14. Schindelin J, Arganda-Carreras I, Frise E, et al. Fiji: an open-source platform for institutional support and Dr. Jingsong Wang (BCCHR) for technical assistance biological-image analysis. Nat Methods 2012;9:676–682 at the Imaging Core. The authors also thank Ryan Vander Werff and the Uni- 15. Kim D, Pertea G, Trapnell C, Pimentel H, Kelley R, Salzberg SL. TopHat2: versity of British Columbia (UBC) Biomedical Research Centre Sequencing Core for support on RNA-seq experiments, Dr. Julie Brind’Amour and Dr. Matthew accurate alignment of transcriptomes in the presence of insertions, deletions and Lorincz (UBC) for advice on low-input ChIP, and Dr. Lawryn Kasper (St. Jude Child- gene fusions. Genome Biol 2013;14:R36 ren’s Research Hospital) for advice on p300 Western blotting. 16. Leek JT, Johnson WE, Parker HS, Jaffe AE, Storey JD. The sva package for Funding. The salary for C.K.W. is supported by a BCCHR Graduate Student- removing batch effects and other unwanted variation in high-throughput experi- – ship, and the investigator salary for W.T.G. is supported by BCCHR Intramural ments. Bioinformatics 2012;28:882 883 IGAP Award. This study was supported by grants to W.T.G. from the Natural 17. Love MI, Huber W, Anders S. Moderated estimation of fold change and dis- Sciences and Engineering Research Council of Canada (RGPIN 402576-11) and persion for RNA-seq data with DESeq2. Genome Biol 2014;15:550 — the Canadian Institutes of Health Research Institute of Nutrition, Metabolism 18. Hulsen T, de Vlieg J, Alkema W. BioVenn a web application for the com- and Diabetes (MOP-119595 and PJT-148695). parison and visualization of biological lists using area-proportional Venn diagrams. Duality of Interest. No potential conflicts of interest relevant to this article BMC Genomics 2008;9:488 were reported. 19. Wang J, Duncan D, Shi Z, Zhang B. WEB-based GEne SeT AnaLysis Toolkit – Author Contributions. C.K.W. and W.T.G. conceived the study and (WebGestalt): update 2013. Nucleic Acids Res 2013;41:W77 W83. designed the experiments, generated and analyzed the data, wrote the manuscript, 20. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using revised the article’s intellectual content, revised the manuscript, and approved the real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods 2001; – final version of the manuscript. A.K.W.-V. generated and analyzed the data, contrib- 25:402 408 uted to the study design, revised the article’s intellectual content, revised the man- 21. Kasper LH, Boussouar F, Ney PA, et al. A transcription-factor-binding uscript, and approved the final version of the manuscript. D.S.L. helped with the surface of coactivator p300 is required for haematopoiesis. Nature 2002;419: – calcium imaging experiment, contributed to the study design, revised the article’s 738 743 ’ intellectual content, revised the manuscript, and approved the final version of the 22. Brind Amour J, Liu S, Hudson M, Chen C, Karimi MM, Lorincz MC. An ultra- fi manuscript. P.K.B. and F.C.L. contributed to the study design, revised the article’s low-input native ChIP-seq protocol for genome-wide pro ling of rare cell populations. intellectual content, revised the manuscript, and approved the final version of the Nat Commun 2015;6:6033 manuscript. C.K.W. and W.T.G. are the guarantors of this work and, as such, had full 23. Song J, Xu Y, Hu X, Choi B, Tong Q. Brain expression of Cre recombinase driven fi – access to all the data in the study and take responsibility for the integrity of the data by pancreas-speci c promoters. Genesis 2010;48:628 634 and the accuracy of the data analysis. 24. Servitja J-M, Pignatelli M, Maestro MA, et al. Hnf1alpha (MODY3) controls tissue-specific transcriptional programs and exerts opposed effects on cell growth in References pancreatic islets and liver. Mol Cell Biol 2009;29:2945–2959 1. Melloul D, Marshak S, Cerasi E. Regulation of insulin gene transcription. Dia- 25. Gu C, Stein GH, Pan N, et al. Pancreatic beta cells require NeuroD to achieve betologia 2002;45:309–326 and maintain functional maturity. Cell Metab 2010;11:298–310 422 Roles of p300/CBP in Pancreatic Islets Diabetes Volume 67, March 2018

26. Taylor BL, Liu F-F, Sander M. Nkx6.1 is essential for maintaining the functional 35. Qiu Y, Guo M, Huang S, Stein R. Insulin gene transcription is mediated by state of pancreatic beta cells. Cell Reports 2013;4:1262–1275 interactions between the p300 coactivator and PDX-1, BETA2, and E47. Mol Cell Biol 27. Gao T, McKenna B, Li C, et al. Pdx1 maintains b cell identity and function by 2002;22:412–420 repressing an a cell program. Cell Metab 2014;19:259–271 36. Pontoglio M, Sreenan S, Roe M, et al. Defective insulin secretion in hepatocyte 28. Hang Y, Yamamoto T, Benninger RKP, et al. The MafA transcription factor nuclear factor 1alpha-deficient mice. J Clin Invest 1998;101:2215–2222 becomes essential to islet b-cells soon after birth. Diabetes 2014;63:1994–2005 37. Haumaitre C, Barbacci E, Jenny M, Ott MO, Gradwohl G, Cereghini S. Lack of TCF2/ 29. Swisa A, Avrahami D, Eden N, et al. PAX6 maintains b cell identity by re- vHNF1 in mice leads to pancreas agenesis. Proc Natl Acad Sci USA 2005;102:1490–1495 pressing genes of alternative islet cell types. J Clin Invest 2017;127:230–243 38. Bramswig NC, Everett LJ, Schug J, et al. Epigenomic plasticity enables human 30. Gutiérrez GD, Bender AS, Cirulli V, et al. Pancreatic b cell identity requires pancreatic a to b cell reprogramming. J Clin Invest 2013;123:1275–1284 continual repression of non-b cell programs. J Clin Invest 2017;127:244–259 39. Barbacci E, Chalkiadaki A, Masdeu C, et al. HNF1beta/TCF2 mutations impair 31. Soutoglou E, Papafotiou G, Katrakili N, Talianidis I. Transcriptional activation by transactivation potential through altered co-regulator recruitment. Hum Mol Genet hepatocyte nuclear factor-1 requires synergism between multiple coactivator pro- 2004;13:3139–3149 teins. J Biol Chem 2000;275:12515–12520 40. Prado CL, Pugh-Bernard AE, Elghazi L, Sosa-Pineda B, Sussel L. Ghrelin cells 32. Akpinar P, Kuwajima S, Krützfeldt J, Stoffel M. Tmem27: a cleaved and shed replace insulin-producing beta cells in two mouse models of pancreas development. plasma membrane protein that stimulates pancreatic proliferation. Cell Proc Natl Acad Sci USA 2004;101:2924–2929 Metab 2005;2:385–397 41. Mastracci TL, Wilcox CL, Arnes L, et al. Nkx2.2 and Arx genetically interact to 33. Dell H, Hadzopoulou-Cladaras M. CREB-binding protein is a transcriptional regulate pancreatic endocrine cell development and endocrine hormone expression. coactivator for hepatocyte nuclear factor-4 and enhances apolipoprotein gene ex- Dev Biol 2011;359:1–11 pression. J Biol Chem 1999;274:9013–9021 42. Doyle MJ, Sussel L. Nkx2.2 regulates beta-cell function in the mature islet. 34. Ban N, Yamada Y, Someya Y, et al. Hepatocyte nuclear factor-1alpha recruits Diabetes 2007;56:1999–2007 the transcriptional co-activator p300 on the GLUT2 gene promoter. Diabetes 2002; 43. Fajans SS, Bell GI, Polonsky KS. Molecular mechanisms and clinical patho- 51:1409–1418 physiology of maturity-onset diabetes of the young. N Engl J Med 2001;345:971–980