Glucose and Insulin Treatment of Insulinoma Cells Results In

Glucose and Insulin Treatment of Insulinoma Cells Results in Transcriptional Regulation of a Common Set of Genes Mitsuru Ohsugi,1 Corentin Cras-Me´neur,1 Yiyong Zhou,1 Wesley Warren,2 Ernesto Bernal-Mizrachi,1 and M. Alan Permutt1

Glucose and insulin are important regulators of islet ␤-cell growth and function by activating signaling pathways resulting in transcriptional changes that lead to ancreatic islet ␤-cells can be regulated by multi- adaptive responses. Several immediate early genes have ple stimuli, including nutrients and growth fac- been shown to be rapidly induced by glucose-activated tors. ␤-Cell proliferation and function are depolarization in islet ␤-cells. The current studies ad- controlled by plasma glucose concentration and dress aspects of glucose-regulated transcription: 1) the P by growth factors acting via multiple intracellular signal- number and characteristics of these genes, 2) if depolarization is the major mechanism, and 3) if glucose- ing pathways (1). Changes in gene expression that result stimulated insulin secretion is responsible, because from the activation of these signaling pathways are likely insulin per se can activate transcription. Here, the responsible for the adaptation of ␤-cells to physiological expression profiles of glucose-responsive insulinoma and pathological states. However, large gaps in our knowl- cells 45 min after the addition of glucose, KCl to induce edge currently exist regarding the changes in gene expres- depolarization, or insulin were assessed by endocrine sion and the molecular mechanisms mediating these ␤-cell pancreas cDNA microarrays. Glucose activated more responses to nutrients and growth factors. than 90 genes, representing diverse gene ontology func- Some genes likely to be involved in chronic glucose tions, and most were not previously known to be glucose ␤ responsive. KCl activated 80% of these same glucose- regulation of islet -cell mass or function have been regulated genes and, along with the effects of pretreat- identified (2–6). We have focused on early signaling events ment with diazoxide, suggested that glucose signaling is initiated by glucose treatment of insulinoma cells that mediated primarily via depolarization. There were >150 result in rapid transient activation of a number of imme- genes activated by insulin, and remarkably 71% were diate early genes (IEGs). These include Egr1, Egr2, c-fos, also regulated by glucose. Preincubation with a phos- and c-jun, known to respond to growth factor stimulation phatidylinositol (PI) 3-kinase inhibitor resulted in al- in a number of other tissues (7). The signaling pathways most total inhibition of depolarization and insulin- for induction of IEGs exhibit considerable stimulus and activated transcriptional responses. Thus, through gene expression profiling, these data demonstrate that glu- tissue specificity and in general involve activation of cose and insulin rapidly activate a PI 3-kinase pathway, kinase/phosphatase cascades (8). Initial glucose-mediated resulting in transcription of a common set of genes. This signaling can represent the first step in elucidating long- is consistent with glucose activation of gene transcrip- term changes in gene expression and islet physiology. tion either directly or indirectly through a paracrine/ These signaling pathways, limited to the initial kinase/ autocrine effect via insulin release. These results phosphatase cascades, are critical for understanding how illustrate that expression gene profiling can contribute the ␤-cell responds to its environment. The events occur- ␤ to the elucidation of important -cell biological func- ring from the time the stimulus reaches the ␤-cell until the tions. Diabetes 53:1496–1508, 2004 signal is transmitted to the nucleus to activate or repress transcription of a particular set of genes may be crucial in understanding the defects in islet growth in diabetic From the 1Division of Endocrinology, Metabolism, and Lipid Research, subjects or the adverse consequences of glucose toxicity Department of Internal Medicine, Washington University School of Medicine, St. Louis, Missouri; and the 2Genome Sequencing Center, Washington Univer- on ␤-cell function. sity School of Medicine, St. Louis, Missouri. IEGs are often transcription factors that in turn activate Address correspondence and reprint requests to M. Alan Permutt, MD, Division of Endocrinology, Metabolism, and Lipid Research, Washington expression of downstream target genes, thus generating University School of Medicine, 660 S. Euclid Ave., Campus Box 8127, St. Louis, distinct biological responses by inducing specific long- MO 63110. E-mail: [email protected]. term programs of gene expression. In pancreatic ␤-cells, Received for publication 5 March 2004 and accepted 19 March 2004. M.O. and C.C.-M. contributed equally to this study. activation of expression of these IEGs was shown to ϩ Additional information for this article can be found in an online appendix at depend on depolarization activation of voltage-gated Ca2 http://diabetes.diabetesjournals.org. channels and subsequent influx of extracellular Ca2ϩ. This DMEM, Dulbecco’s modified Eagle’s medium; EPCon, Endocrine Pancreas 2ϩ Consortium; EST, expressed sequence tag; FBS, fetal bovine serum; GO, gene resulted in activation of Ca -regulated kinases, including ontology; IEG, immediate early gene; KATP channel, ATP-sensitive potassium calmodulin-dependent kinase IV and protein kinase A, channel; PI, phosphatidylinositol; qRT-PCR, quantitative RT-PCR; SSC, sodium chloride–sodium citrate. leading to phosphorylation and activation of several tran- © 2004 by the American Diabetes Association. scription factors (cAMP-responsive element binding pro-

1496 DIABETES, VOL. 53, JUNE 2004 M. OHSUGI AND ASSOCIATES

FIG. 1. A: Experimental scheme. MIN6 insulinoma cells were incubated in 5 mmol/l glucose, 5% FBS DMEM, for 18 h before stimulation. The “unstimulated” sample was harvested with- out any addition. Stimulated samples were harvested, and total RNA was extracted 45 min after addition of 25 mmol/l glucose, 50 mmol/l KCl, or 100 nmol/l insulin. B: Pairing scheme. RNA from four samples was labeled with either Cy3 or Cy5 fluorescent dye, and then a pair of samples was hybridized to a cDNA microarray. A graphic representation of the pairing scheme is shown here. An arrow indicates hybridization to a cDNA microarray and RNA labeling with Cy3 or Cy5 fluorescent dye as indicated. Two double-sided arrows with different colors of arrowheads indicate dye-flip hybridization. C: Distribution of the standard deviations for a “self-vs.-self” experiment derived from 9,700 cDNA probe as described in RESEARCH DESIGN AND (12 ؍ microarrays (n METHODS tein, serum response factor, and Elk-1) (9,10). The results the present experiments extend our knowledge of IEGs of these experiments defined the rapid glucose-signaling regulated by glucose, by KCl-induced depolarization, and pathways for a small number of IEGs whose transcription by insulin through use of high-resolution custom cDNA is rapidly activated by glucose, but the results now pose microarrays that contain clones from the Endocrine Pan- additional questions addressed by the current study. creas Consortium (EPCon: http://www.cbil.upenn.edu/ Animal models perfused with glucose for 4–5 days, or EPConDB). The arrays used for these experiments contain transgenic animals overexpressing a particular gene (11), up to 9,700 cDNAs with Ͼ3,000 novel clones not currently result in more readily measured physiological changes, yet available on commercial arrays (12–14). The results of this the sequence of molecular events leading to these physi- work suggest that glucose activation of IEGs is mediated ological changes are difficult to discern. This result high- primarily via depolarization and that glucose and insulin lights the desirability of beginning to dissect these activate an overlapping set of genes. Further, both of these mechanisms using other models. Thus, we designed ex- growth stimuli appear to activate transcription through a periments using insulinoma cells to elucidate early tran- phosphatidylinositol (PI) 3-kinase–dependent pathway. scriptional responses to islet growth factors. The results of These results further illustrate how monitoring expression

DIABETES, VOL. 53, JUNE 2004 1497 1498

TABLE 1 PANCREATIC IN REGULATION GENE Gene regulation by glucose, ranked according to fold change, is compared with that by KCl-induced depolarization and with insulin treatment

GenBank accession LocusLink G Ͼ U G Ͼ U K Ͼ U K Ͼ U I Ͼ U I Ͼ U number ID Name Symbol ratio 95% CI ratio 95% CI ratio 95% CI

Top 50 glucose upregulated genes (54 clones including duplicates) AA537033 13653 Early growth response 1 Egr1 3.91 (2.55–5.99) 7.42 (4.86–11.33) 0.93 (0.77–1.12) AA958974 15936 Immediate early response 2 Ier2 2.92 (2.57–3.33) 7.89 (6.82–9.13) 0.99 (0.86–1.14) BI790969 15901 Inhibitor of DNA binding 1 Idb1 2.55 (2.36–2.76) 2.35 (2.07–2.68) 0.98 (0.88–1.09) AA869400 15936 Immediate early response 2 Ier2 2.54 (2.33–2.77) 6.68 (5.83–7.66) 0.94 (0.84–1.05) AI646026 15937 Immediate early response 3 Ier3 2.25 (2.13–2.39) 1.55 (1.41–1.7) 0.90 (0.84–0.96) AA119154 233895 cDNA sequence BC006909 BC006909 2.07 (1.83–2.35) 5.81 (4.5–7.5) 0.95 (0.88–1.02)

W10821 1.84 (1.52–2.23) 3.77 (3.07–4.62) 0.93 (0.85–1.01) ␤ -CELLS BI319352 74155 RIKEN cDNA 1300002F13 gene 1300002F13Rik 1.69 (1.5–1.9) 1.42 (1.27–1.59) 1.09 (0.96–1.24) AA123373 15902 Inhibitor of DNA binding 2 Idb2 1.61 (1.52–1.71) 1.47 (1.4–1.55) 1.28 (1.23–1.32) AA517679 76266 RIKEN cDNA 0610043A03 gene 0610043A03Rik 1.59 (1.38–1.83) 1.74 (1.67–1.8) 1.79 (1.54–2.08) AI574282 76983 RIKEN cDNA 3110021P21 gene 3110021P21Rik 1.58 (1.39–1.79) 1.59 (1.42–1.77) 1.66 (1.46–1.88) AI785720 20384 Splicing factor, arginine/serine-rich 5 (SRp40, Sfrs5 1.57 (1.45–1.7) 1.48 (1.41–1.54) 1.44 (1.39–1.5) FIRS) BI788712 12338 Calpain 6 Capn6 1.57 (1.39–1.78) 1.58 (1.45–1.73) 1.73 (1.59–1.87) AA882421 11702 S-adenosylmethionine decarboxylase 1 Amd1 1.55 (1.45–1.66) 1.48 (1.36–1.61) 1.21 (1.09–1.35) BI319623 67004 Mus musculus adult male colon cDNA, RIKEN 1.50 (1.37–1.65) 1.62 (1.16–2.26) 1.64 (1.2–2.23) full-length enriched library, clone:90 BI966990 57394 Kidney-specific membrane protein Nx17-pending 1.50 (1.41–1.6) 1.50 (1.4–1.6) 1.61 (1.52–1.7) AA763015 59048 RIKEN cDNA 1500002I11 gene 1500002I11Rik 1.48 (1.3–1.69) 1.49 (1.31–1.69) 1.54 (1.38–1.73) BI900192 71701 RIKEN cDNA 1200003F12 gene 1200003F12Rik 1.48 (1.41–1.55) 1.78 (1.64–1.93) 1.63 (1.43–1.86) AI255428 15902 Inhibitor of DNA binding 2 Idb2 1.47 (1.31–1.66) 1.11 (0.87–1.4) 1.20 (1.01–1.42) AA522172 66882 Basic leucine zipper and W2 domains 1 Bzw1 1.47 (1.39–1.56) 1.44 (1.35–1.54) 1.46 (1.36–1.57) AI646970 66142 Cytochrome c oxidase subunit VIIb Cox7b 1.45 (1.38–1.53) 1.45 (1.36–1.54) 1.58 (1.53–1.64) BG141716 11723 Amylase 2, pancreatic Amy2 1.44 (1.26–1.65) 1.45 (1.24–1.71) 1.63 (1.42–1.88) AA274974 67180 RIKEN cDNA 2610311I19 gene 2610311I19Rik 1.43 (1.34–1.51) 1.45 (1.37–1.52) 1.35 (1.27–1.43) BG141813 15936 Immediate early response 2 Ier2 1.42 (1.17–1.73) 3.42 (2.71–4.32) 0.94 (0.78–1.14) BI792554 1.42 (1.21–1.68) 1.26 (1.18–1.33) 1.22 (1.01–1.48) BI794624 19177 Proteasome (prosome, macropain) subunit, beta Psmb7 1.41 (1.28–1.56) 1.40 (1.19–1.66) 1.16 (1.07–1.27) type 7 AI327322 13006 Chondroitin sulfate proteoglycan 6 Cspg6 1.41 (1.18–1.68) 1.62 (1.36–1.93) 1.70 (1.53–1.89) BI441452 66142 Cytochrome c oxidase subunit VIIb Cox7b 1.41 (1.35–1.46) 1.41 (1.37–1.44) 1.56 (1.49–1.63) BG797876 53861 Zinc finger protein 265 Zfp265 1.40 (1.26–1.57) 1.23 (1.15–1.31) 1.63 (1.51–1.77) BG797897 21402 Transcription elongation factor B (SIII), Tceb1l 1.40 (1.32–1.49) 1.36 (1.29–1.43) 1.51 (1.44–1.58) polypeptide 1-like AA867229 79264 Cerebral cavernous malformations 1 Ccm1 1.40 (1.29–1.52) 1.57 (1.38–1.78) 1.21 (0.84–1.75) BI319581 17184 Matrin 3 Matr3 1.39 (1.27–1.53) 1.60 (1.38–1.85) 1.76 (1.55–1.99) BI963248 73274 RIKEN cDNA 1700034P14 gene 1700034P14Rik 1.39 (1.27–1.53) 1.51 (1.46–1.56) 1.43 (1.23–1.65) IBTS O.5,JN 2004 JUNE 53, VOL. DIABETES, AA275042 53315 N-sulfotransferase Sultn 1.39 (1.34–1.45) 1.35 (1.21–1.51) 1.37 (1.27–1.48) BI437713 69144 RIKEN cDNA 1810035L17 gene 1810035L17Rik 1.38 (1.27–1.51) 1.37 (1.28–1.46) 1.49 (1.37–1.63) AI099189 234358 Hypothetical protein LOC234358 LOC234358 1.37 (1.28–1.47) 1.37 (1.24–1.51) 1.49 (1.37–1.62) AI644134 1.37 (1.26–1.49) 1.55 (1.38–1.73) 1.52 (1.34–1.72) W08076 66384 Signal recognition particle 19 Srp19 1.37 (1.3–1.44) 1.36 (1.27–1.47) 1.42 (1.3–1.54) BI466176 211378 RIKEN cDNA 6720489N17 gene 6720489N17Rik 1.37 (1.2–1.56) 1.32 (1.25–1.4) 1.44 (1.3–1.6) BI794667 22057 Transducer of ErbB-2.1 Tob1 1.36 (1.27–1.47) 1.69 (1.46–1.97) 1.09 (1.03–1.15) BI790762 53861 Zinc finger protein 265 Zfp265 1.36 (1.22–1.52) 1.37 (1.2–1.56) 1.74 (1.57–1.93) BM022660 12417 Chromobox homolog 3 (Drosophila HP1 Cbx3 1.36 (1.25–1.48) 1.45 (1.3–1.61) 1.54 (1.38–1.72) gamma) AA267992 70273 RIKEN cDNA 2310051E17 gene 2310051E17Rik 1.36 (1.2–1.54) 1.66 (1.44–1.9) 1.29 (1.12–1.49) IBTS O.5,JN 2004 JUNE 53, VOL. DIABETES, BI437804 1.36 (1.23–1.5) 0.95 (0.85–1.07) 1.02 (0.87–1.19) AA203882 68514 RIKEN cDNA 1110008L20 gene 1110008L20Rik 1.35 (1.2–1.53) 2.56 (2.19–3) 1.08 (0.99–1.17) BI900177 72111 RIKEN cDNA 2010305F15 gene 2010305F15Rik 1.35 (1.23–1.48) 1.35 (1.24–1.47) 1.43 (1.33–1.55) BI440010 27362 DnaJ (Hsp40) homolog, subfamily B, member 9 Dnajb9 1.35 (1.24–1.47) 1.43 (1.22–1.69) 1.30 (1.13–1.5) BG655381 109082 RIKEN cDNA 1110064L07 gene 1110064L07Rik 1.35 (1.11–1.63) 1.37 (1.09–1.73) 1.33 (1.02–1.74) AA154133 1.34 (1.29–1.4) 0.94 (0.88–1.01) 1.07 (1.01–1.14) AA616918 213895 Hypothetical protein LOC213895 LOC213895 1.34 (1.15–1.57) 1.32 (1.17–1.5) 1.17 (0.97–1.42) BI964951 66118 RIKEN cDNA 1110005A23 gene 1110005A23Rik 1.34 (1.2–1.49) 1.44 (1.31–1.58) 1.66 (1.47–1.87) AA186117 101685 RIKEN cDNA 5830435K17 gene 5830435K17Rik 1.34 (1.13–1.59) 1.30 (1.09–1.54) 1.10 (0.95–1.28) BI901606 22438 Inactive X specific transcripts Xist 1.34 (1.29–1.39) 1.06 (0.96–1.17) 1.16 (1.09–1.23) BI465567 14910 Gene trap ROSA 26, Philippe Soriano Gt(ROSA)26Sor 1.33 (1.26–1.41) 1.21 (1.11–1.32) 1.35 (1.26–1.45) Twenty downregulated genes (21 clones) AI891573 20226 Seryl-aminoacyl-tRNA synthetase 1 Sars1 1.30 (1.12–1.51) 1.47 (0.96–2.23) 1.34 (1.01–1.77) BI793660 107271 Tyrosyl-tRNA synthetase Yars 1.30 (1.14–1.49) 1.43 (1.11–1.84) 1.24 (1.01–1.51) BG321905 21881 Transketolase Tkt 1.30 (1.1–1.54) 1.53 (1.28–1.84) 1.33 (1.14–1.54) AA816062 16362 Interferon regulatory factor 1 Irf1 1.31 (1.2–1.15) 1.08 (1.02–1.15) 1.01 (0.95–1.06) Nuclear factor of kappa light chain gene BI899955 18035 enhancer in B-cells inhibitor, alpha Nfkbia 1.31 (1.22–1.41) 0.74 (0.69–0.8) 1.04 (0.95–1.15) BI793672 19708 Requiem Req 1.31 (1.1–1.57) 1.50 (1.02–2.19) 1.19 (0.87–1.62) W98440 12444 Cyclin D2 Ccnd2 1.32 (1.17–1.48) 1.30 (1.23–1.38) 1.18 (1.05–1.33) BI963259 14312 Bromodomain containing 2 Brd2 1.32 (1.11–1.56) 1.31 (1–1.71) 1.19 (0.94–1.5) BI319630 66867 High mobility group 20A Hmg20a 1.33 (1.07–1.64) 1.08 (0.82–1.43) 1.13 (0.87–1.47) AI505476 110196 Famesyl diphosphate synthetase Fdps 1.33 (1.19–1.49) 1.51 (1.22–1.88) 1.20 (1.08–1.33) AA986494 13629 Eukaryotic translation elongation factor 2 Eef2 1.34 (1.2–1.49) 1.46 (1.17–1.83) 1.40 (1.21–1.61) BI465646 56700 RIKEN cDNA 0610031J06 gene 0610031J06Rik 1.34 (1.05–1.71) 1.56 (1.27–1.93) 1.43 (1.07–1.92) BG311049 56224 Transmembrane 4 superfamily member 9 Tm4sf9-pending 1.34 (1.19–1.52) 1.40 (1–1.96) 1.35 (1.02–1.79) BI788746 20016 RNA polymerase 1-1 Rpo1-1 1.35 (1.15–1.58) 1.16 (0.93–1.47) 1.13 (0.9–1.42) BI900196 20226 Seryl-aminoacyl-tRNA synthetase 1 Sars1 1.35 (1.2–1.51) 1.38 (1.11–1.72) 1.26 (1.08–1.47) ARP1 actin-related protein 1 homolog A AI156875 54130 (yeast) Actr1a 1.36 (1.07–1.72) 1.64 (1.11–2.41) 1.37 (1.09–1.72) AA118297 56370 Transgelin 3 Tagin3 1.37 (1.21–1.55) 1.10 (0.97–1.26) 1.01 (0.88–1.16) AI528846 20742 Spectrin beta 2 Spnb2 1.37 (1.01–1.86) 1.27 (0.87–1.87) 1.24 (0.92–1.67) BI790926 14779 Glutathione peroxidase 4 Gpx4 1.37 (1.01–1.87) 1.00 (0.66–1.51) 1.19 (0.85–1.68) BG797208 18952 Septin 4 Sept4 1.37 (1.07–1.77) 1.49 (0.99–2.25) 1.30 (0.92–1.85) AI323295 13198 DNA-damage inducible transcript 3 Ddit3 2.11 (2.06–2.16) 1.21 (1.1–1.32) 1.08 (1.06–1.09) Fold changes by glucose (G), KCl (K), or insulin (I) relative to unstimulated (U) and 95% CIs are shown. The genes activated both by glucose and insulin are represented in bold. Of note, because of the conversion process from log fold change to numerical fold change, 95% CI is asymmetrical. LocusLink ID, gene name, and gene symbol are shown. .OSG N ASSOCIATES AND OHSUGI M. 1499 GENE REGULATION IN PANCREATIC ␤-CELLS

FIG. 2. Hierarchical clustering (see RESEARCH DESIGN AND METHODS). A: Transcriptional changes by glucose (G) were compared with those by KCl. B: Transcriptional changes by glucose were compared with those by insulin (I). Each horizontal row corresponds to a single gene. Red indicates fold-2.72 ؍) that the gene is upregulated, while green corresponds to downregulation. The color ranges were blunted at log fold change of 1 change). Transcriptional changes by glucose (G), KCl (K), or insulin (I) relative to unstimulated (U) were shown. gene profiles can serve to elucidate important ␤-cell and then purified and concentrated using Microcon microconcentrators biological functions. (Millipore) according to the manufacturer’s protocol. Microarray construction. After the sequencing of cDNA libraries derived from various mouse pancreas tissues gathered by the EPCon (12,13), ex- RESEARCH DESIGN AND METHODS pressed sequence tags (ESTs) were evaluated for similarity to the existing Cell culture and RNA extraction. The MIN6 insulinoma cell line was a gift nonredundant GeneBank entries (http://www.ncbi.nlm.nih.gov/Genbank/ from Dr. Jun-ichi Miyazaki (Osaka University, Japan). MIN6 cells were index.html) and for redundancy within the EST clone set using standard maintained in Dulbecco’s modified Eagle’s medium (DMEM) containing 25 BLAST analyses (17). mmol/l glucose, supplemented with 15% heat-inactivated fetal bovine serum A nonredundant set of clones was selected and PCR amplified in prepara- (FBS), 100 units/ml penicillin, 100 ␮g/ml streptomycin, 100 ␮g/ml L-glutamine, tion for the microarray slide production. PCR products were purified using ␮ ␤ and 5 l/l -mercaptoethanol in humidified 5% CO2, 95% air at 37°C (15). MIN6 Millipore 96-well plates, DNA quantity was determined using Picogreen cells were used between passages 27 and 33. Insulin secretion of the cells was (Molecular Probes) reagents, and, finally, each of the PCR products was checked by a standard static incubation method, and those maintaining normalized with Millipore purified water to a standard concentration of 400 glucose responsiveness with normal glucose threshold for islet ␤-cells (be- ng/␮l. Before microarray spotting, all PCR products were adjusted to a final tween 5 and 10 mmol/l glucose) as well as maximum secretion capacity DNA concentration of 200 ng/␮l in a spotting buffer of 3ϫ sodium chloride– (maximum secretion at least five times higher than the basal secretion) were sodium citrate (SSC) and 0.75 mol/l betaine. PCR products were spotted onto used for experiments. Cells were grown in 150-mm dishes, and once they epoxy-coated slides (MWG Biotech) using an ArrayMaker2 arrayer (designed reached 60% confluence, medium was switched to DMEM containing 5 mmol/l by P. Brown’s laboratory; Stanford University). Spotted slides were incubated glucose and 5% FBS for 18 h before stimulation (Fig. 1A). The stimulating 12–14 h at 40% humidity in a 42°C oven. After incubation, the slides were agent (25 mmol/l glucose, 50 mmol/l KCl, or 100 nmol/l insulin) was added cross-linked using the UV Stratalinker 2400 (Stratagene) at 700 ␮J ϫ100 and directly to the medium, and cells were incubated for 45 min. When indicated, processed by the following steps: 1) gently shaking the slides in a 0.2% SDS LY294002 (50 ␮mol/l; Cell Signaling, Beverly, MA) was added to the medium bath for 2 min; 2) three room temperature water bath washes, each at 1 min; 15 min before treatment with stimulating agents. After stimulation, cells were 3)a50°C water bath for 20 min; 4)a95°C water bath for 2 min; and 5) harvested and total RNA was extracted from cells using TRIzol reagent or spinning the slides dry. The slides were stored in a desiccant cabinet for future RNeasy columns according to the manufacturer’s protocol (Invitrogen and hybridization experiments. Qiagen, respectively). The different clones on the array were identified through BLASTn similarity Labeling of RNA transcripts. First-strand cDNA was generated by oligo-dT matches using the NCBI BLAST (http://www.ncbi.nlm.nih.gov/blast) and WU- primed reverse transcription (Superscript II; Invitrogen) using the 3DNA Array BLAST version 2.0 (17) against the nonredundant subsets of the public Mouse 50 kit (Genisphere) (16). Modified oligo-dT primers were used in which a databases and RefSeq (http://www.ncbi.nlm.nih.gov/RefSeq), and only fluorophore/dendrimer specific oligo sequence was attached to the 5Ј end of matches with an E value below at least 1 ϫ 10Ϫ50 and a score superior to 200 the dT primer. For RNA expression level comparison, samples were paired were considered as a match. Annotation was further confirmed comparing the

1500 DIABETES, VOL. 53, JUNE 2004 M. OHSUGI AND ASSOCIATES N rmtoidpneteprmns( n )wsaaye yqTPRa niae for indicated as qRT-PCR by analyzed was 2) and (1 experiments independent two from RNA K a mmol/l), (0.6 diazoxide depolarization, of inhibitor an of presence and absence the in KCl and glucose by activated genes of Comparison 2 TABLE l412 (0.99 1.28 Klf4 (1.27 1.36 (2.36 2.55 Tob1 (2.57 2.92 Id1 (2.55 Ier2 3.91 Egr1 results to the annotation provided on the EPCon website (http://www.cbil. upenn.edu/EPConDB) for the PancChip 5.0 and manually curating discrepan-

eprmn 1) (experiment cies. Further annotation (ofﬁcial names and symbols, gene ontology [GO]

Microarray functions) was gathered from Source (http://source.stanford.edu) and in some cases imported directly from LocusLink (http://www.ncbi.nlm.nih.gov/ LocusLink). – – – – –

.5 .8( 2.58 (0.89 1.65 1.65) (1.49 2.84 1.47) (3.9 7.40 2.76) (3.03 5.20 3.33) 5.99) Hybridization. Two hybridizations were carried out in a sequential manner. The primary hybridization was performed by adding 38 ␮l of sample to the microarray under a supported glass coverslip (Erie Scientiﬁc) at 50°C for

5mo/ lcs 5mo/ glucose mmol/l 25 glucose mmol/l 25 16–20 h at high humidity in the dark. Before the secondary hybridization, eprmn 1) (experiment slides were gently submerged into 2ϫ SSC, 0.2% SDS for 12 min; transferred

qRT-PCR ϫ ϫ

Ϫ to 2 SSC for 12 min; transferred to 0.2 SSC for 12 min; transferred to 95%

0.25 ethanol for 2 min; and then spun dry by centrifugation. Secondary hybridiza- – 09 .5(2.32 8.25 – – 10.9) –

.2 .9(0.76 1.59 (1.23 2.44 2.42) 4.19) (1.9 5.72 7.37) tions were carried out using the complimentary capture reagents provided in –

.1 .1(0.51 2.21 5.41) the 3DNA Array 50 kit (Genisphere). Data acquisition. Slides were scanned on a Perkin Elmer ScanArray Express HT scanner to detect Cy3 and Cy5 ﬂuorescence. Laser power was kept

eprmn 2) (experiment constant for Cy3/Cy5 scans, and photo-multiplier tube was varied for each

qRT-PCR experiment based on background ﬂuorescence. Gridding and analysis of images was performed using QuantArray (Perkin Elmer). Each spot was

– deﬁned on a pixel-by-pixel basis using a modiﬁed Mann-Whitney statistical – – (0.31 – 0.63 – 9.54)

.2 .0(0.06 1.20 (0.51 1.20 3.92) (0.54 1.07 2.42) (0.72 3.65) 1.91 14.17) test. Pairing scheme. To compare each of the conditions with every other condition, samples were paired as indicated in Fig. 1B. For each pair,

eprmn 1) (experiment transcripts from one condition were labeled with either Cy3 or Cy5 and hybridized with the Cy5- or Cy3-labeled transcripts from the partner condi- qRT-PCR tion. Another pair of RNAs was also labeled inverting dyes (“dye-ﬂip” hybridization). This sample pairing and dye ﬂip required a total of 12 – – – – –

.4 .0(0.19 0.70 (0.7 1.34 2.34) (0.59 1.9) 1.14 (0.75 1.69 1.61) (0.89 3.1) 1.74 0.94) microarrays per experiment (each arrow in Fig. 1B represents one slide of microarray with a pair of RNAs). As a result, each sample for an experiment was hybridized six times: against three other conditions twice, once with Cy3,

eprmn 2) (experiment and the other with Cy5. RNA from the ﬁrst experiment was hybridized to two ϩ

qRT-PCR sets of microarray, a 5,700 clone set and then to a 9,700 clone microarray

izxd 0mo/ C 0mo/ KCl mmol/l 50 KCl mmol/l 50 diazoxide consisting of a superset of the 5,700 clone set; therefore, in each case, 12 slides

– were used. Therefore, a total of 24 microarray slides was used for these .8 .9(1.46 1.69 – 1.98) – – –

.1 .1(1.47 2.01 1.21) (2.07 2.35 (6.82 1.68) 7.89 (4.86 2.64) 7.42 2.59) analyses. Normalization. In the microarray studies performed herein, the probes that gave a signal intensity Ն2.0-fold above the corresponding background intensity in both Cy5 and Cy3 channels were chosen for calculation of the fold ﬁ eprmn 1) (experiment egns hw ob ciae ymcora nlss(N rmeprmn 1). experiment from (RNA analysis microarray by activated be to shown genes, ve

Microarray change. The intensity of each spot was ﬁrst adjusted by subtracting background intensity. Then, the log ratio of Cy5 and Cy3 channel intensity of each spot was calculated, and the median of the log ratio from all probes was – – – – –

.6 .4(0.37 4.84 (1.7 2.98 (1.12 2.76) 3.21 1.97) (7 15.09 2.68) (9.3 9.13) 16.61 11.33) subtracted from each log ratio (the global normalization procedure [18]). This is based on the assumption that the expression levels of most genes are unchanged, and thus log fold change should center at zero. Indeed, data from all hybridizations indicated that the expression of the majority of genes

eprmn 1) (experiment (ϳ95%, data not shown) were unchanged.

qRT-PCR The log ratios from six pairs of dye-ﬂip hybridizations were used to estimate probe-speciﬁc measurement variation, usually derived from self-vs.- – 31)1.0(6.77 18.20 23.18)

– – self experiments (19). Probe-speciﬁc measurement variation was subtracted .6 .6(1.68 3.26 – 4.26) – (7.8 14.36 23.93)

.1 .3(1.58 5.33 9.31) (1.43 3.08 5.31) from log fold change to calculate the ﬁnal normalized log fold change (probe-speciﬁc normalization). Statistical analysis. From our pairing scheme, two direct and two indirect ratios were obtained when two conditions were compared. When glucose- eprmn 2) (experiment stimulated and -unstimulated samples were compared, for example, two ratios

qRT-PCR were acquired from dye-ﬂip hybridizations directly comparing those two samples. Two additional ratios were calculated indirectly from eight hybrid- –

– – – (0.21 1.78 – 20.92) izations involving all glucose, KCl, insulin, and unstimulated samples. These .8 .0(0.11 0.80 (0.84 1.50 9.08) (0.6 1.21 4.84) (1.26 4.74) 2.60 29.62) four ratios were used to calculate average fold change as well as standard error. The 95% CI was calculated with attention to the change of distribution

ATP when converting log fold change to fold change (20). Criteria were set to

eprmn 1) (experiment assess changes in gene expression that included a fold change of 1.3 or more

hne ciao yqRT-PCR by activator channel compared with “unstimulated.” This criterion was selected after observing qRT-PCR that the mean variance of all probes derived from a single self-to-self hybridization from the ﬁrst 5,700 and 9,700 probe sets was 0.12 (Fig. 1C). – – (0.01 0.93 – 1.81) – –

.9 .7( 1.07 (0.36 1.32 1.49) 2.15) (0.09 2.26 (0.04 2.46 3.95) 3.36) Because we had four measurements, that mean variance was further reduced to 0.06 (ϭ 0.12/͌4). With this small variance, the false-positive rate is 0.0032%. With similar multiple hybridization and cDNA microarray, a fold change of 1.3 was used to identify signiﬁcant gene expression change in other studies ϩ eprmn 2) (experiment (21–23). Additionally, signiﬁcant fold changes required that the 95% CI of the diazoxide qRT-PCR fold change, two standard errors away from the mean fold change, excluded Ϫ

0.25 1. Thus, any fold changes meeting these criteria could be considered signiﬁ-

– – – – cant with P Ͻ 0.05. 2.28) 1.85) 4.43) 4.88) –

2.39) Hierarchical clustering. Hierarchical clusters were performed using the Genesis software version 1.3 (Institute for Biomedical Engineering, Graz University of Technology, Graz, Austria) (24). Heat maps generated by

DIABETES, VOL. 53, JUNE 2004 1501 GENE REGULATION IN PANCREATIC ␤-CELLS

TABLE 3 Validation of microarray data by qRT-PCR and biological replication

Glucose Microarray qRT-PCR qRT-PCR qRT-PCR (experiment 1) (experiment 1) (experiment 2) (experiment 3) Ier2 2.82 (2.57–3.33) 6.60 (5.12–8.08) 6.61 (2.89–10.33) 2.83 (1.09–4.57) Egr1 3.91 (2.55–5.99) 7.60 (5.64–9.56) 4.14 (2.66–5.62) 3.77 (2.69–4.85) Kir/Gem 1.27 (1.18–1.49) 2.60 (1.9–3.3) 2.14 (1.24–3.04) 1.27 (0.79–1.75) Id1 2.55 (2.36–2.76) 3.22 (2.54–3.9) 2.37 (1.25–3.49) 2.17 (1.29–3.05) Id2 1.61 (1.5–1.71) 1.74 (1.28–2.2) 2.30 (1.22–3.38) 1.74 (0.76–2.72) Capn6 1.57 (1.39–1.78) 1.64 (0.98–2.29) 1.69 (1.03–2.35) Matr3 1.39 (1.27–1.53) 1.80 (1.2–2.4) 1.75 (1.03–2.46) Tob1 1.36 (1.27–1.47) 1.65 (0.96–2.33) 1.79 (1.03–2.56) RIKEN cDNA I110008L20 gene 1.35 (1.7–1.53) 1.85 (1.15–2.55) 1.48 (0.83–2.13) Dnajb9 1.35 (1.74–1.47) 1.87 (1.3–2.45) 1.33 (0.94–1.73) Klf4 1.28 (0.99–1.65) 2.13 (1.24–3.02) 1.65 (0.85–2.44) Susp1 1.23 (1.15–1.37) 1.59 (0.98–2.2) 1.27 (0.67–1.87) DNA segment, Chr 4, Wayne 1.28 (1.16–1.4) 1.21 (0.83–1.58) 0.97 (0.6–1.35) State University 53 The same RNA samples analyzed by microarray shown in Tables 1 and 2 were also tested by qRT-PCR (experiment 1). Genes chosen for analysis had significant fold changes on microarray for glucose, KCl, or insulin. RNA samples from two additional experiments were also analyzed by qRT-PCR (experiments 2 and 3). Fold changes along with their 95% CIs are represented. hierarchical clustering were created using the average linkage clustering hybridized to two different sets of microarrays (5,700 or (Euclidian distances) directly from the logarithmic ratios (base e), and all 9,700 cDNA probe sets) providing four to eight observa- color ranges for the “heat maps” were adjusted to a maximum ratio of 1. Quantitative real-time RT-PCR. Total RNA was isolated, and 1 ␮g was used tions for each pair. Criteria for significant alteration of to prepare cDNA, primed with random hexamers, and reverse transcribed expression were defined as 1) fold stimulation or inhibi- with Superscript II (Invitrogen) according to manufacture’s protocol. Quanti- tion of Ն1.3 and 2) 95% CI of the fold change excluding 1.0. tative RT-PCR (qRT-PCR) was performed by monitoring in real time the Based on the small variances we observed (Fig. 1C), a increase in fluorescence of the SYBR Green dye (ABI) as described (25,26) Ͻ using the ABI 7000 sequence detection system (Applied Biosystems). For 1.3-fold change would yield false-positive findings 0.01% comparison of transcript levels between samples, a standard curve of cycle (see RESEARCH DESIGN AND METHODS). Independent confirma- thresholds for serial dilutions of a cDNA sample was established and then tion of these fold changes was obtained on a subset of the used to calculate the relative abundance of each gene. Values were then genes with altered expression as described later. normalized to the relative amounts of 18S ribosomal RNA, which were obtained from a similar standard curve. All PCRs were performed at least in Gene expression in response to glucose. Stimulation replicates of four. Standard error of the quantity of transcript normalized to by glucose resulted in a significant increase in mRNA the amount of 18S ribosomal RNA was calculated from a formula with levels for 72 of the ϳ9,700 cDNAs on the arrays. The 50 consideration of error propagation. When gene expression levels of two genes most increased by glucose treatment are shown in conditions were compared, the ratio was expressed with standard error calculated from the same formula. Specificity of each primer pair was Table 1. The complete list of regulated genes is shown in confirmed by melting curve analysis and agarose gel electrophoresis of PCR Table 1S in the online appendix. Glucose-stimulated genes products. Sequences of primers used in this study are included in an online of diverse GO functions, described more fully below, appendix at http://diabetes.diabetesjournals.org. included two immediate early genes previously identified as glucose responsive in islet ␤-cells, early growth re- RESULTS sponse 1 (Egr1) (10,28), and an inhibitor of DNA binding 1 Experimental protocol and cDNA microarray analy- (Idb1) (29). Several other immediate early genes found to sis. The MIN6 insulinoma cell line provides an excellent respond to mitogenic stimuli in other tissues (immediate experimental model, because these cells retain gene ex- early response 2 [Ier2] and 3 [Ier3] and inhibitor of DNA pression and insulin secretory responses to physiological binding 2 [Idb2]) were also shown to respond to glucose changes in glucose concentrations (3,4,6,15,27). Cells were treatment. Interestingly, among the glucose-regulated starved for glucose and serum overnight and then either genes, the overwhelming majority was not previously harvested at the 0 time point (unstimulated) or further known to be glucose responsive. Glucose treatment also incubated with 25 mmol/l glucose for 45 min. Another set resulted in significant downregulation of 21 genes (Table of cells was treated with 50 mmol/l KCl to ascertain the 1), of which only one, Chop (Ddit3), was previously effects of depolarization on gene expression. A third set of described as glucose regulated (3). Chop has been incrim- cells was treated with insulin (100 nmol/l) to examine inated in cell cycle arrest and apoptosis (30). At least 13 of potential autocrine/paracrine effects on gene expression the glucose-responsive genes do not yet have a match in (Fig. 1A). Concentrations of those stimuli were chosen to LocusLink or UniGene. elicit maximal short-term responses, after observing that Glucose effect on gene transcription is mostly medi- the media of insulinoma cells contains 17.4–39.8 nmol/l ated through depolarization. Because islet ␤-cell glu- insulin (M.O., M.A.P., unpublished data). At the end of the cose metabolism is accompanied by closure of ATP- 45-min incubation, RNA was extracted and pairs of RNA sensitive potassium (KATP) channels followed by samples (e.g., unstimulated vs. stimulated) labeled with depolarization, we next sought to study the genes regu- either Cy3 or Cy5 were hybridized together (in duplicate lated by KCl-induced depolarization and the degree of with dye flip) onto cDNA microarrays, as illustrated in Fig. overlap of these genes with those activated by glucose. A 1B. For this experiment, the same RNA samples were total of 202 genes were activated by KCl treatment (139

1502 DIABETES, VOL. 53, JUNE 2004 M. OHSUGI AND ASSOCIATES

TABLE 3 Continued

KCl Insulin Microarray qRT-PCR QRT-PCR qRT-PCR Microarray qRT-PCR qRT-P CR qRT-PCR (experiment 1) (experiment 1) (experiment 2) (experiment 3) (experiment 1) (experiment 1) (experiment 2) (experiment 3) 7.89 (6.82–9.13) 21.0 (15.58–26.34) 8.29 (2.41–14.17) 6.14 (2.08–10.2) 0.99 (0.86–1.14) 1.12 (0.82–1.42) 1.30 (0.24–2.36) 1.38 (0–2.76) 7.42 (4.86–11.33) 11.0 (8.32–13.72) 6.86 (4.18–9.54) 7.23 (4.15–10.31) 0.93 (0.77–1.12) 1.10 (0.8–1.4) 0.98 (0.4–1.56) 0.98 (0.1–1.86) 2.97 (2.45–3.61) 10.9 (2.82–14.06) 6.58 (3.4–9.76) 6.90 (3.08–10.72) 0.86 (0.6–0.95) 1.37 (1.01–1.73) 1.60 (0.4–2.8) 0.70 (0.32–1.08) 2.35 (2.07–2.68) 3.27 (2.49–4.05) 2.40 (1.14–3.66) 2.43 (1.17–3.69) 0.98 (0.88–1.09) 1.25 (0.95–1.55) 1.21 (0.55–1.87) 1.27 (0.61–1.93) 1.45 (1.40–1.55) 1.62 (1.24–2) 1.50 (0.58–2.42) 1.46 (0.72–2.2) 1.28 (1.23–1.32) 1.49 (1.13–1.85) 1.02 (0.52–1.82) 1.22 (0–2.44) 1.58 (1.45–1.73) 3.63 (1.93–5.32) 3.26 (1.77–4.74) 1.73 (1.59–1.87) 1.65 (0.86–2.44) 1.98 (1.03–2.92) 1.60 (1.38–1.85) 1.89 (1.09–2.69) 1.46 (0.56–2.36) 1.76 (1.55–1.99) 1.85 (1.01–2.68) 1.66 (1.02–2.3) 1.69 (1.46–1.97) 2.38 (1.33–3.42) 2.05 (1.27–2.84) 1.08 (1.03–1.15) 0.73 (0.32–1.13) 1.46 (0.79–2.13) 2.56 (2.19–3) 1.95 (1.22–2.69) 2.26 (1.36–3.17) 1.08 (0.99–1.17) 1.37 (0.67–2.08) 1.40 (0.71–2.1) 1.43 (1.22–1.69) 1.42 (0.25–2.09) 1.51 (1.01–2.02) 1.30 (1.13–1.5) 1.15 (0.76–1.53) 2.14 (1.38–2.91) 2.01 (1.47–2.76) 4.14 (2.31–5.97) 4.48 (2.74–6.22) 0.86 (0.58–1.76) 0.79 (0.2–1.39) 1.06 (0.56–1.56) 1.57 (1.46–1.60) 1.62 (1.15–2.09) 1.03 (1.28–2.38) 1.55 (1.45–1.67) 1.90 (1.25–2.55) 1.72 (1.16–2.28) 1.53 (1.31–1.79) 2.03 (1.25–2.81) 1.52 (0.99–2.06) 1.77 (1.65–1.89) 1.57 (1.05–2.08) 1.69 (1.11–2.28)

upregulated and 63 downregulated [Table 1S in the online glucose-regulated genes was expressed to the same extent appendix]). Notably, examination of Table 1 revealed that by insulin treatment. 47 out of the 50 glucose regulated genes listed here were Validation of gene expression profiles by quantita- also activated by KCl. KCl treatment also resulted in a tive RT-PCR (qRT-PCR) and by biological replicates. decrease in mRNA levels for at least half of the glucose- The observation by microarray analysis that genes ap- repressed genes (Table 1). Comparison of all the glucose- peared to be activated by both glucose/depolarization and or KCl-regulated genes revealed that Ͼ80% of genes acti- insulin required validation by independent means. This vated by glucose were also activated by KCl (Table 1S). was performed by qRT-PCR on RNA samples from a This is graphically represented in a hierarchical cluster subset of genes that were shown to be activated on the analysis, as shown in Fig. 2A. The results of these exper- microarrays (experiment 1). As can be seen in Table 3, in iments suggested that depolarization is a major mecha- general, the fold changes noted with qRT-PCR were larger nism of transcriptional regulation induced by glucose. than those observed with microarray, yet there was a To further examine the role of depolarization and glu- highly significant correlation (R2 ϭ 0.90, data not shown) cose-mediated gene expression, diazoxide, a KATP channel between the levels of gene expression measured by the activator, was used to assess the glucose-activated gene two methods for experiment 1. expression in the absence of depolarization. A subset of Additionally, biological replication was sought by qRT- genes expressed within all ranges of fold induction was PCR analysis of RNA obtained from two additional inde- analyzed with qRT-PCR, as shown in Table 2. For all five pendent experiments (experiments 2 and 3). As shown in genes tested here, glucose- and KCl-induced gene activa- Table 3, a total of 13 genes was chosen for this analysis tion was markedly reduced by diazoxide. These data from genes with various expression profiles that appeared therefore further substantiate the role of depolarization in to be activated on the microarrays by at least one stimulus. gene regulation by glucose. These results clearly demonstrated that small fold changes Overlap between glucose- and insulin-induced gene (e.g., Ͻ2) can be observed by microarray and validated expression. There were 129 genes whose expression was with qRT-PCR and that the changes in gene expression significantly increased by insulin treatment after a 45-min after stimulation by all three agents for this representative incubation and 25 genes whose expression was signifi- set of genes were reproducible in three independent cantly suppressed. Comparisons to the top 50 genes acti- experiments. Importantly, these results confirmed that vated by glucose stimulation are shown in Table 1 some genes could rapidly be activated by both glucose and (complete list in Table 1S). Whereas some of these genes insulin under these conditions. had been previously reported to be regulated by insulin Assessment of the PI 3-kinase pathway in depolariza- treatment in other tissues (31), none had been noted to be tion and insulin-mediated gene expression. The obser- altered by insulin treatment in insulinoma cells. vation of a common set of genes activated by glucose and Examination of the genes whose expression was signif- insulin was consistent with glucose/depolarization and icantly affected by insulin treatment revealed a surprising insulin inducing gene transcription via completely sepa- result, i.e., there appeared to be considerable overlap rate pathways that target the same genes or that the two between the genes regulated by insulin and by glucose, stimuli activate pathways that converge on a common as can be readily seen in Table 1, where glucose- and signaling pathway that activates the same genes. To dif- insulin-activated genes are compared (bold). The expres- ferentiate between the two possibilities, gene expression sion profiles after the two treatments are also graphically was evaluated in another set of experiments using a assessed in Fig. 2B by a hierarchical cluster analysis. pharmacological inhibitor (LY294002) for PI 3-kinase, Immediately apparent is that 8 of the top 10 genes acti- known to be involved in insulin signaling (32). KCl and vated by glucose were not at all stimulated by insulin insulin treatment were compared in this experiment, be- treatment. On the other hand, most of the remainder of the cause KCl induced depolarization and glucose stimulated

DIABETES, VOL. 53, JUNE 2004 1503 GENE REGULATION IN PANCREATIC ␤-CELLS

FIG. 3. Comparison of genes activated by KCl and insulin in the absence and presence of an inhibitor of PI 3-kinase, LY294002 (LY) (50 ␮mol/l). The inhibitor was added to the medium 15 min before KCl or insulin treatment. Representation of gene expression by fold change ranked according to KCl (A) and insulin (B) is shown. The dashed lines represent the 1.3-fold change cutoff.

1504 DIABETES, VOL. 53, JUNE 2004 M. OHSUGI AND ASSOCIATES

FIG. 4. Representation of the distribution of the GO functions (see RESEARCH DESIGN AND METHODS) of the genes concomitantly regulated by glucose and insulin. The top-level molecular functions are represented in the pie chart. Some genes bear multiple functions, which have all been accounted for in their different categories. considerably common sets of genes, but the fold changes sion resulting from each stimulus. Through technical rep- by KCl were more robust (Table 1). After 45 min of KCl or licates and dye-flip hybridizations, we ensured that the insulin treatment, the cells were harvested and analysis technical variations on our experiments could be minimal was performed in identical fashion to the previous exper- (33), allowing us to reliably detect fold changes of Ն1.3 for iments. As shown in Fig. 3, the genes were represented in Ͼ99.9% of the probes. The significance of these somewhat log fold changes to allow a representation of the tran- modest changes is further supported by the consistency of scripts on a same scale for both up- and downregulation. results from duplicate probes, the low false-positive rate The addition of LY294002 resulted in almost complete with self-vs.-self analysis, the qRT-PCR data, and the elimination of gene regulation by each stimulus. These biological replicates. results suggest that glucose, acting through depolarization, A number of biologically relevant observations were and insulin share common signaling pathways leading to made by the results of these experiments. 1) Over 90 gene expression and that this PI 3-kinase–dependent potential candidates either up- or downregulated by glu- pathway seems to play a major role. cose were uncovered in this study. Many of these genes Pleiotropic functions of transcripts induced by glu- are currently unknown ESTs and are not yet listed in cose/depolarization and insulin. From earlier studies, UniGene or AllGenes (DoTS assemblies). 2) By comparing genes acutely regulated by glucose and depolarization gene expression profiles after glucose- and KCl-induced were mostly transcription factors or transcriptional regu- depolarization, there was considerable overlap of genes lators, such as Egr1, Egr2, c-myc, c-fos, Idb1, and Idb2. In activated by both agents, consistent with the hypothesis the current experiments, the expression of a common set that glucose activation of rapidly responding genes is of genes was activated or inhibited by glucose and insulin. mediated predominantly via depolarization. These results More than half of these genes can be characterized accord- are in accord with those of earlier studies showing that ing to GO functions (Fig. 4). Two major categories com- expression of several early response genes depended on prised over 70% of the functions, with catalytic (i.e., depolarization (10,29). This premise was further confirmed enzymatic) activity comprising the largest. Binding activ- by the results of the study of the measurements of gene ity, the second largest category observed, includes ligands, expression in the presence of the depolarization inhibitor receptors, and DNA binding proteins. The next most diazoxide in a subset of these genes. 3) The current results abundant categories included genes classified as involved showed that exogenous insulin treatment at 100 nmol/l in translational or transcriptional regulator activity. Nota- activated transcription of a number of newly described bly, genes suppressed by glucose/depolarization and by early response genes in insulinoma cells. The most notable insulin included several genes classified as pro-apoptotic, finding, however, was that glucose and insulin treatments e.g., Chop and Scotin. The major conclusion from this each activated a common subset of genes among several analysis is that glucose/depolarization and insulin treat- thousands on the arrays. ment resulted in alteration of genes not limited to tran- Validation of microarray gene expression results. The scriptional regulators but rather genes with pleiotropic validity of the observed microarray results ideally requires effects in terms of GO functions. both technical validation and biological replication. Tech- nical validation can be seen in that several clones corre- DISCUSSION sponding to the same gene were present on our microarray New findings. In this study, the glucose-induced early set, for example, splicing factor arginine/serine rich transcriptional responses of mouse insulinoma cells were (Sfrs5) and inhibitor of DNA binding 2 (Idb2) (Table 1). We observed as a model for defining initial signaling events also used qRT-PCR for validation of the microarray results that ultimately link the ␤-cell’s environment with long- and observed excellent correlation between microarray term effects such as proliferation and function. The results and qRT-PCR results testing the same samples (Table 3). of the current experiments have potentially important Further, biological replication was achieved by two addi- biological implications, although the conclusions are to tional independent experiments in which expression of a some extent based on small fold changes in gene expres- subset of genes was measured by qRT-PCR (Table 3).

DIABETES, VOL. 53, JUNE 2004 1505 GENE REGULATION IN PANCREATIC ␤-CELLS

Limitations of the study. Although a short 45-min treatment is suitable for evaluating initial transcriptional responses, the long-term changes that may result from these early changes remain to be determined. When MIN6 cells were incubated in high glucose for 4 h, preliminary results indicated that the majority of genes regulated at 45 min returned to baseline. This finding was consistent with characteristics of early response genes (8). On the other hand, new sets of genes not previously regulated at 45 min were now found to be regulated at 4 h (M.O., C.C.-M., Y.Z., E.B.-M., M.A.P., unpublished data). This may suggest cascades of gene expression in which genes responding early to stimuli may initiate or suppress expression of other genes. The maximum concentrations of stimuli for glucose, KCl, and insulin were used. One may raise an issue regarding potential effect of osmotic stress as a glucose stimulus. In previous publications (10,29) and unpublished observations (E.B.-M. and M.A.P.), we have shown failure of nonmetabolizable glucose analogs to activate transcription, that the effects of glucose were blocked by diazoxide, and that KCl in the presence of a PI 3-kinase inhibitor showed no activation of genes (Fig. 3). Another limitation of the current study was that the levels of expression of genes activated by glucose and insulin were for the most part in the 1.3- to 1.5-fold range. However, multiple determinations of fold change by microarray resulted in small variances of measurements, as seen in Fig. 1C, and several of the glucose and insulin response genes were validated by other means (Table 3). Although we believe that in general these 30–50% changes in gene transcriptional rates are significant, rather than speculating on the importance of these modest fold changes in gene expression, we found that the main conclusion of the study is that common signaling pathways are similarly activated by two distinctive important ␤-cell stimuli: glucose and insulin. Not all ␤-cell early response genes such as c-myc and c-fos were observed in the present study (7,34). These genes were not included in the original EPCon arrays, presumably because of the low abundance of those transcripts. Despite those omissions, our current studies are confirming and extending the previous findings, such as Egr1, and highlighting the use of microarray to assess similarities of treatments (e.g., glucose and insulin) and to identify signaling pathways activated by those treatments. The microarray results with insulinoma cells are limited to ␤-cells adapted to culture, yet isolated islets are a mixture of several cell types and are further complicated by this issue. Isolation of pure ␤-cells from isolated islets raises additional concerns regarding the conditions of preparation, so every method has its limitations. The results of our experiments in insulinoma cells can now be used to measure responses to primary islets in culture,

FIG. 5. Schematic diagram of potential models for signal transduction pathways leading to gene regulation in pancreatic ␤-cells. A: Signaling

by glucose through inhibition of the KATP channel and depolarization results in insulin-independent activation of a small number of genes, whereas a larger overlapping set of genes are activated by glucose and insulin through independent pathways. B: Glucose and insulin activate a large overlapping set of genes by converging on a common pathway(s). C: Glucose activates gene transcription indirectly, by ﬁrst stimulating insulin secretion that subsequently acts through its receptor to activate gene transcription (43).

1506 DIABETES, VOL. 53, JUNE 2004 M. OHSUGI AND ASSOCIATES both rodent and human. Importantly, our microarray data 3-kinase inhibitor almost completely eliminated gene ex- now identify dozens of new early response genes, some pression induced by both stimuli (Fig. 3). A third possibil- known and some only described in the databases as ESTs. ity is that the major effect of glucose/depolarization on Depolarization is the major component of glucose- ␤-cell gene transcription is through an autocrine/paracrine regulated early transcriptional activation. Remark- effect of glucose-stimulated insulin secretion (Fig. 5C) ably, Ͼ90% of the effects of glucose on gene transcription (41–43). It has been difficult to test the autocrine/paracrine were mimicked by KCl-induced depolarization (Fig. 2B). effects of insulin, because MIN6 cells are immersed in a Although Ca2ϩ influx after glucose-induced depolarization large amount of insulin and anti-insulin receptor antibody is a critical mechanism resulting in insulin secretion, the studies have been inconclusive (M.O., E.B.-M., M.A.P., role of depolarization in glucose-mediated gene expres- unpublished data). To test the third hypothesis, experi- sion has not been fully evaluated. Only few genes were ments are currently underway to silence insulin receptor shown to be regulated via depolarization in previous gene expression with small interfering RNA (44) in MIN6 studies of early response genes, but the results of this cells. study demonstrated that the majority of the glucose- regulated genes are activated through depolarization. Con- ACKNOWLEDGMENTS sidering the results of previous studies performed on This work was supported in part by National Institutes of limited numbers of genes (10,29), this is likely to result in Health Grants DK16746, DK56954, and DK99007 (to ϩ activation of Ca2 -regulated kinases leading to phosphor- M.A.P.) and the Washington University Diabetes Research ylation and activation of transcription factors. Although and Training Center. signaling events leading to transcriptional activation or We gratefully acknowledge the D. Melton lab (Harvard suppression after depolarization remain to be further University) and K. Kaestner and C. Stoeckert labs (Univer- defined, the current studies indicated that the PI 3-kinase sity of Pennsylvania), as well as Ellen Ostlund, Jessica pathway seems to be involved (Fig. 3). It remains to be Murray, Sandy Clifton, Hiroshi Inoue, Chris Sawyer, Mike determined whether PI 3-kinase activation occurs via an Heinz, Elaine Mardis, and other members of the Genome autocrine/paracrine effect of insulin. Sequencing Center for their work with EPCon and mi- Activation of a common set of genes by glucose and croarrays. We would also like to thank Gary Stormo for insulin. The observation that glucose and insulin induce helpful suggestions and Gary Skolnick for preparation of an overlapping set of genes is interesting in light of recent the manuscript. data that provide evidence for an important role of insulin in the regulation of gene transcription in ␤-cells. Insulin REFERENCES stimulation of ␤-cells in vitro results in transcriptional 1. Deeney JT, Prentki M, Corkey BE: Metabolic control of beta-cell function. induction of the preproinsulin, liver-type pyruvate kinase, Semin Cell Dev Biol 11:267–275, 2000 and acetyl-CoA carboxylase I genes by activation of the 2. Shalev A, Pise-Masison CA, Radonovich M, Hoffmann SC, Hirshberg B, Brady JN, Harlan DM: Oligonucleotide microarray analysis of intact human insulin receptor substrate/PI 3-kinase pathway (35–37). pancreatic islets: identification of glucose-responsive genes and a highly Mouse models (38–40) clearly indicated the importance of regulated TGFbeta signaling pathway. Endocrinology 143:3695–3698, 2002 the roles of the insulin receptor (␤IRKO mouse) and the 3. Webb GC, Akbar MS, Zhao C, Steiner DF: Expression profiling of pancre- downstream targets of the insulin-signaling pathway atic beta cells: glucose regulation of secretory and metabolic pathway ␤ genes. Proc Natl Acad SciUSA97:5773–5778, 2000 (IRS2) in islet -cell survival and function, but its precise 4. Minami K, Yano H, Miki T, Nagashima K, Wang CZ, Tanaka H, Miyazaki JI, mechanisms are still elusive. Seino S: Insulin secretion and differential gene expression in glucose- Regarding the GO functions of the genes commonly responsive and -unresponsive MIN6 sublines. Am J Physiol Endocrinol regulated by glucose and insulin, there was a high propor- Metab 279:E773–E781, 2000 5. Flamez D, Berger V, Kruhoffer M, Orntoft T, Pipeleers D, Schuit FC: Critical tion of regulated genes that belonged to catalytic and role for cataplerosis via citrate in glucose-regulated insulin release. binding activity classes, and there were lower proportions Diabetes 51:2018–2024, 2002 in the other categories. The differences in the number of 6. Lilla V, Webb G, Rickenbach K, Maturana A, Steiner DF, Halban PA, genes in these GO function categories differed slightly Irminger JC: Differential gene expression in well-regulated and dysregu- from the distribution of genes on the entire array (data not lated pancreatic beta-cell (MIN6) sublines. Endocrinology 144:1368–1379, 2003 shown), although the array as a whole is not representa- 7. Susini S, Roche E, Prentki M, Schlegel W: Glucose and glucoincretin tive of the MIN-6 cells, because it comprises genes ex- peptides synergize to induce c-fos, c-jun, junB, zif-268, and nur-77 gene pressed in all cells of the pancreas. Further, it would be expression in pancreatic beta(INS-1) cells. FASEB J 12:1173–1182, 1998 speculative to suggest that these categories of biological 8. Sheng M, McFadden G, Greenberg ME: Membrane depolarization and calcium induce c-fos transcription via phosphorylation of transcription functions represent true changes at the cellular level factor CREB. Neuron 4:571–582, 1990 considering the small numbers of genes in each GO 9. Bernal-Mizrachi E, Wen W, Srinivasan S, Klenk A, Cohen D, Permutt MA: category. Activation of Elk-1, an Ets transcription factor, by glucose and EGF The broadly overlapping set of genes common to both treatment of insulinoma cells. Am J Physiol Endocrinol Metab 281:E1286– glucose and insulin treatment in the current studies sug- E1299, 2001 10. Bernal-Mizrachi E, Wice B, Inoue H, Permutt MA: Activation of serum gested at least three possible explanations. One is that response factor in the depolarization induction of Egr-1 transcription in these two islet ␤-cell stimuli act through independent pancreatic islet beta-cells. J Biol Chem 275:25681–25689, 2000 pathways that finally converge on the same gene promot- 11. Lingohr MK, Buettner R, Rhodes CJ: Pancreatic beta-cell growth and ers (Fig. 5A). Another is that the two stimuli act on two survival: a role in obesity-linked type 2 diabetes? Trends Mol Med 8:375–384, 2002 different pathways that converge on a common pathway 12. Scearce LM, Brestelli JE, McWeeney SK, Lee CS, Mazzarelli J, Pinney DF, that ends by activating the same gene promoters (Fig. 5B). Pizarro A, Stoeckert CJ Jr, Clifton SW, Permutt MA, Brown J, Melton DA, This latter alternative seems the most likely because the PI Kaestner KH: Functional genomics of the endocrine pancreas: the pan-

DIABETES, VOL. 53, JUNE 2004 1507 GENE REGULATION IN PANCREATIC ␤-CELLS

creas clone set and PancChip, new resources for diabetes research. 29. Wice BM, Bernal-Mizrachi E, Permutt MA: Glucose and other insulin Diabetes 51:1997–2004, 2002 secretagogues induce, rather than inhibit, expression of Id-1 and Id-3 in 13. Bernal-Mizrachi E, Cras-Meneur C, Ohsugi M, Permutt MA: Gene expres- pancreatic islet beta cells. Diabetologia 44:453–463, 2001 sion profiling in islet biology and diabetes research. Diabete Metab Res Rev 30. Zinszner H, Kuroda M, Wang X, Batchvarova N, Lightfoot RT, Remotti H, 19:32–42, 2003 Stevens JL, Ron D: CHOP is implicated in programmed cell death in 14. Kaestner KH, Lee CS, Scearce LM, Brestelli JE, Arsenlis A, Le PP, Lantz KA, response to impaired function of the endoplasmic reticulum. Genes Dev Crabtree J, Pizarro A, Mazzarelli J, Pinney D, Fischer S, Manduchi E, 12:982–995, 1998 Stoeckert CJ Jr, Gradwohl G, Clifton SW, Brown JR, Inoue H, Cras-Meneur 31. Diamond RH, Du K, Lee VM, Mohn KL, Haber BA, Tewari DS, Taub R: C, Permutt MA: Transcriptional program of the endocrine pancreas in mice Novel delayed-early and highly insulin-induced growth response genes: and humans. Diabetes 52:1604–1610, 2003 identification of HRS, a potential regulator of alternative pre-mRNA 15. Miyazaki J, Araki K, Yamato E, Ikegami H, Asano T, Shibasaki Y, Oka Y, splicing. J Biol Chem 268:15185–15192, 1993 Yamamura K: Establishment of a pancreatic beta cell line that retains 32. Kido Y, Nakae J, Accili D: Clinical review 125: the insulin receptor and its glucose-inducible insulin secretion: special reference to expression of cellular targets. J Clin Endocrinol Metab 86:972–979, 2001 glucose transporter isoforms. Endocrinology 127:126–132, 1990 33. Yang YH, Speed T: Design issues for cDNA microarray experiments. Nat 16. Manduchi E, Scearce LM, Brestelli JE, Grant GR, Kaestner KH, Stoeckert Rev Genet 3:579–588, 2002 CJ Jr: Comparison of different labeling methods for two-channel high- 34. Yamashita S, Tobinaga T, Ashizawa K, Nagayama Y, Yokota A, Harakawa density microarray experiments. Physiol Genomics 10:169–179, 2002 S, Inoue S, Hirayu H, Izumi M, Nagataki S: Glucose stimulation of 17. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ: Basic local alignment protooncogene expression and deoxyribonucleic acid synthesis in rat islet search tool. J Mol Biol 215:403–410, 1990 cell line. Endocrinology 123:1825–1829, 1988 18. Quackenbush J: Microarray data normalization and transformation. Nat 35. Leibiger IB, Leibiger B, Moede T, Berggren PO: Exocytosis of insulin Genet 32 (Suppl.):496–501, 2002 promotes insulin gene transcription via the insulin receptor/PI-3 kinase/ 19. Yang IV, Chen E, Hasseman JP, Liang W, Frank BC, Wang S, Sharov V, p70 s6 kinase and CaM kinase pathways. Mol Cell 1:933–938, 1998 Saeed AI, White J, Li J, Lee NH, Yeatman TJ, Quackenbush J: Within the fold: assessing differential expression measures and reproducibility in 36. da Silva Xavier G, Varadi A, Ainscow EK, Rutter GA: Regulation of gene microarray assays. Genome Biol 3:research0062, 2002 expression by glucose in pancreatic beta-cells (MIN6) via insulin secretion Ј 20. Chen Y, Kamat V, Dougherty ER, Bittner ML, Meltzer PS, Trent JM: Ratio and activation of phosphatidylinositol 3 -kinase. J Biol Chem 275:36269– statistics of gene expression levels and applications to microarray data 36277, 2000 analysis. Bioinformatics 18:1207–1215, 2002 37. Andreolas C, da Silva Xavier G, Diraison F, Zhao C, Varadi A, Lopez- 21. Woo AL, Gildea LA, Tack LM, Miller ML, Spicer Z, Millhorn DE, Finkelman Casillas F, Ferre P, Foufelle F, Rutter GA: Stimulation of acetyl-CoA FD, Hassett DJ, Shull GE: In vivo evidence for interferon-gamma-mediated carboxylase gene expression by glucose requires insulin release and sterol homeostatic mechanisms in small intestine of the NHE3 Naϩ/Hϩ ex- regulatory element binding protein 1c in pancreatic MIN6 ␤-cells. Diabetes changer knockout model of congenital diarrhea. J Biol Chem 277:49036– 51:2536–2545, 2002 49046, 2002 38. Kulkarni RN, Bruning JC, Winnay JN, Postic C, Magnuson MA, Kahn CR: 22. Dunckley T, Lukas RJ: Nicotine modulates the expression of a diverse set Tissue-specific knockout of the insulin receptor in pancreatic beta cells of genes in the neuronal SH-SY5Y cell line. J Biol Chem 278:15633–15640, creates an insulin secretory defect similar to that in type 2 diabetes. Cell 2003 96:329–339, 1999 23. Ferrante AW Jr, Thearle M, Liao T, Leibel RL: Effects of leptin deficiency 39. Kubota N, Tobe K, Terauchi Y, Eto K, Yamauchi T, Suzuki R, Tsubamoto Y, and short-term repletion on hepatic gene expression in genetically obese Komeda K, Nakano R, Miki H, Satoh S, Sekihara H, Sciacchitano S, Lesniak mice. Diabetes 50:2268–2278, 2001 M, Aizawa S, Nagai R, Kimura S, Akanuma Y, Taylor SI, Kadowaki T: 24. Sturn A, Quackenbush J, Trajanoski Z: Genesis: cluster analysis of microar- Disruption of insulin receptor substrate 2 causes type 2 diabetes because ray data. Bioinformatics 18:207–208, 2002 of liver insulin resistance and lack of compensatory beta-cell hyperplasia. 25. Morrison TB, Weis JJ, Wittwer CT: Quantification of low-copy transcripts Diabetes 49:1880–1889, 2000 by continuous SYBR Green I monitoring during amplification. Biotech- 40. Withers DJ, Gutierrez JS, Towery H, Burks DJ, Ren JM, Previs S, Zhang Y, niques 24:954–958, 960, 962, 1998 Bernal D, Pons S, Shulman GI, Bonner-Weir S, White MF: Disruption of 26. Wittwer CT, Herrmann MG, Moss AA, Rasmussen RP: Continuous fluores- IRS-2 causes type 2 diabetes in mice. Nature 391:900–904, 1998 cence monitoring of rapid cycle DNA amplification. Biotechniques 22:130– 41. Rothenberg PL, Willison LD, Simon J, Wolf BA: Glucose-induced insulin 131, 134–138, 1997 receptor tyrosine phosphorylation in insulin-secreting ␤-cells. Diabetes 27. Ishihara H, Asano T, Tsukuda K, Katagiri H, Inukai K, Anai M, Kikuchi M, 44:802–809, 1995 Yazaki Y, Miyazaki JI, Oka Y: Pancreatic beta cell line MIN6 exhibits 42. Xu GG, Rothenberg PL: Insulin receptor signaling in the ␤-cell influences characteristics of glucose metabolism and glucose-stimulated insulin insulin gene expression and insulin content: evidence for autocrine secretion similar to those of normal islets. Diabetologia 36:1139–1145, beta-cell regulation. Diabetes 47:1243–1252, 1998 1993 43. Leibiger IB, Leibiger B, Berggren PO: Insulin feedback action on pancre- 28. Jin K, Mao XO, Eshoo MW, del Rio G, Rao R, Chen D, Simon RP, Greenberg atic beta-cell function. FEBS Lett 532:1–6, 2002 DA: cDNA microarray analysis of changes in gene expression induced by 44. Brummelkamp TR, Bernards R, Agami R: A system for stable expression of neuronal hypoxia in vitro. Neurochem Res 27:1105–1112, 2002 short interfering RNAs in mammalian cells. Science 296:550–553, 2002

1508 DIABETES, VOL. 53, JUNE 2004