Diabetes Volume 66, December 2017 3013 Glucose-Induced Changes in Gene Expression in Human Pancreatic Islets: Causes or Consequences of Chronic Hyperglycemia Emilia Ottosson-Laakso,1 Ulrika Krus,1 Petter Storm,1 Rashmi B. Prasad,1 Nikolay Oskolkov,1 Emma Ahlqvist,1 João Fadista,2 Ola Hansson,1 Leif Groop,1,3 and Petter Vikman1 Diabetes 2017;66:3013–3028 | https://doi.org/10.2337/db17-0311 Dysregulation of gene expression in islets from patients In patients with type 2 diabetes (T2D), islet function de- with type 2 diabetes (T2D) might be causally involved clines progressively. Although the initial pathogenic trigger in the development of hyperglycemia, or it could develop of impaired b-cell function is still unknown, elevated glu- as a consequence of hyperglycemia (i.e., glucotoxicity). cose levels are known to further aggravate b-cell function, a To separate the genes that could be causally involved condition referred to as glucotoxicity, which can stimulate in pathogenesis from those likely to be secondary to hy- apoptosis and lead to reduced b-cell mass (1–5). Prolonged perglycemia, we exposed islets from human donors to exposure to hyperglycemia also can induce endoplasmic re- ISLET STUDIES normal or high glucose concentrations for 24 h and ana- ticulum (ER) stress and production of reactive oxygen spe- fi lyzed gene expression. We compared these ndings with cies (6), which can further impair islet function and thereby gene expression in islets from donors with normal glucose the ability of islets to secrete the insulin needed to meet the tolerance and hyperglycemia (including T2D). The genes increased demands imposed by insulin resistance and obe- whose expression changed in the same direction after sity (7). Although these changes are likely to contribute to short-term glucose exposure, as in T2D, were considered most likely to be a consequence of hyperglycemia. Genes deterioration of islet function in patients with manifest whose expression changed in hyperglycemia but not disease, they are less likely to explain the development of after short-term glucose exposure, particularly those T2D in individuals with normoglycemia. In previous studies, we analyzed the gene expression that also correlated with insulin secretion, were consid- fi ered the strongest candidates for causal involvement pro le in individuals with chronically elevated glucose as in T2D. For example, ERO1LB, DOCK10, IGSF11,and measured by elevated HbA1c levels (8,9). However, these PRR14L were downregulated in donors with hyperglyce- studies could not demonstrate whether the changes in gene mia and correlated positively with insulin secretion, expression are the cause or the consequence of hyperglyce- suggesting a protective role, whereas TMEM132C was mia.Onewaytoaddressthisquestionistocomparegene upregulated in hyperglycemia and correlated negatively expression in islets chronically exposed to hyperglycemia with insulin secretion, suggesting a potential pathogenic (prediabetes or diabetes) with gene expression changes after role. This study provides a catalog of gene expression short-term exposure to hyperglycemia, with the assump- changes in human pancreatic islets after exposure to tion that gene expression changes seen in islets from pa- glucose. tients with T2D but not after short-term hyperglycemia are the cause rather than the consequence of hyperglycemia (i.e., contributing to the pathogenesis of T2D). Thus, we The function of pancreatic islets is critical for maintaining performed RNA sequencing of human islets incubated at glucose homeostasis, and dynamic changes of gene expres- physiological (5.5 mmol/L) and high (18.9 mmol/L) glucose sion is part of the islets’ response to blood glucose changes. concentrations and compared the glucose-regulated genes 1Lund University Diabetes Centre, Department of Clinical Sciences, Lund Univer- This article contains Supplementary Data online at http://diabetes sity, Malmö, Sweden .diabetesjournals.org/lookup/suppl/doi:10.2337/db17-0311/-/DC1. 2 Department of Epidemiology Research, Statens Serum Institut, Copenhagen, © 2017 by the American Diabetes Association. Readers may use this article as Denmark long as the work is properly cited, the use is educational and not for profit, and the 3 Finnish Institute of Molecular Medicine, University of Helsinki, Helsinki, Finland work is not altered. More information is available at http://www.diabetesjournals Corresponding author: Emilia Ottosson-Laakso, [email protected]. .org/content/license. Received 12 March 2017 and accepted 30 August 2017. 3014 Glucose-Induced Changes in Gene Expression Diabetes Volume 66, December 2017 with the gene expression profile seen in islets from hyper- annotation to guide the alignment, and featureCounts (15) glycemic donors (8). was used to count the number of reads aligned to the genes. Samples with ,10 million counts and genes with ,2counts RESEARCH DESIGN AND METHODS per million (CPM) in $10 samples were excluded from Donors and Islet Culture further analysis. The raw counts were transformed to log2 Human islets of Langerhans were obtained from the CPM, and the mean-variance trend was identified by using Human Tissue Laboratory (Lund University), which is funded mean-variance modeling at the observational level (voom) (16) by the Excellence Of Diabetes Research in Sweden (EXODIAB) to allow for linear modeling after batch correction with network (www.exodiab.se/home) in collaboration with The ComBat (17). Differential expression was analyzed by using Nordic Network for Clinical Islet Transplantation Program the paired data for 31 pairs of normoglycemic donor islets # (www.nordicislets.org). All islet donors had given consent (HbA1c 6% [42 mmol/mol]) and 14 pairs of hyperglycemic for donation of organs for medical research, and the proce- donor islets (HbA1c .6% [42 mmol/mol]) separately by dures were approved by the ethics committee at Lund Univer- fitting a linear model with linear models for microarray sity (Malmö, Sweden; permit number 2011263). The islets and RNA sequencing data (limma) (18). To exclude that were prepared from cadaver donors by using enzymatic diges- the number of genes responding to glucose was not due to tion and density gradient separation. Islet preparation purity differences in group size, we permuted 14 normoglycemic and count were determined as described previously (10). Be- samples 1,000 times and calculated the mean number of tween 3,000 and 5,000 islets each from 45 donors were di- changed genes by using the R function sample (19). vided into two pools that were incubated for 24 h in CMRL Linear modeling also was used to analyze differential gene ex- 1066 medium (ICN Biomedicals) containing 5.5 mmol/L or pression (adjusted for age, sex, and days in culture) in untreated 18.9 mmol/L glucose at 37°C. The study design is described islets from 81 donors with normoglycemia (HbA1c ,6% [42 A in Fig. 1 , and the clinical characteristics of the donors can mmol/mol]), 19 donors with hyperglycemia (HbA1c ,6.5% be found in Supplementary Table 1. RNA from the islets [48 mmol/mol]), and 16 donors with T2D (HbA1c $6.5% was extracted by using the miRNeasy Mini Kit (QIAGEN). [48 mmol/mol]), including the samples in the glucose-treated RNA quantity and integrity was assessed with an ND-1000 data set and overlapping a previous study by us (8). The HbA1c spectrophotometer (NanoDrop) and on a 2100 Bioanalyzer cutoffs for the groups were chosen to reflect the increased – or a 2200 TapeStation instrument (Agilent Technologies). risk seen in individuals with HbA1c 6.0 6.5% and to follow the recommendations by the International Expert Commit- Sample Preparation and Sequencing tee Report on the Role of the A1C Assay in the Diagnosis With RNA sequencing, we performed a truly global analysis of Diabetes (20). without a priori assumptions (11). A total of 1-mg high-quality RNA (RNA integrity number $8)wasusedasinputforthe The Effect of Coding Variants on In Vivo Insulin Secretion TruSeq RNA Library Preparation Kit (Illumina). The resulting To test the effect of gene expression on insulin secretion libraries were quality controlled on a 2200 TapeStation in vivo in humans, we used exome genotype array data to instrument before sequencing on a HiSeq 2000 system (Illu- find potential loss-of-function and gain-of-function variants mina) for an average depth of 32.4 mol/L paired-end reads in filtered genes. Genotypes with a frequency .5% from (2 3 100 base pairs). the human Infinium Exome-24 v1.1 array (Illumina) Insulin Secretion were obtained from the Prevalence, Prediction and Preven- Insulin secretion capacity of the islet preparations was tion-Botnia study (21), and association with corrected insulin evaluated by stimulatory index (SI). Twenty handpicked response (CIR) was performed, adjusted for age, sex, and P , islets were perifused with low glucose (1.67 mmol/L) for BMI, in 3,720 samples; unadjusted 0.05 was considered fi 42 min, high glucose (20 mmol/L) for 48 min, and then low nominally signi cant. The single nucleotide polymorphisms glucose again. Fractions were collected at 6-min intervals, and (SNPs) in coding parts of the gene were considered as po- the secreted insulin was measured by ELISA. SI was defined tential loss-of-function/gain-of-function variants. as the ratio between the areas under the curve calculated for Expression Quantitative Trait Loci Analysis in the Human the low and high glucose concentrations (12). SI was used for Islets all estimates of insulin secretion in the in vitro experiments. Islet expression