2017, 64 (1), 1-10 Advance Publication doi: 10.1507/endocrj.EJ16-0338 Original

Anti-diabetic action of all-trans retinoic acid and the orphan G coupled GPRC5C in pancreatic β-cells

Stefan Amisten1), 2) *, Israa Mohammad Al-Amily3) *, Arvind Soni3), Ross Hawkes2), Patricio Atanes2), Shanta Jean Persaud2), Patrik Rorsman1), 4) and Albert Salehi3), 4)

1) The Oxford Centre for Diabetes, Endocrinology & Metabolism, University of Oxford, Oxford, UK 2) Diabetes Research Group, Division of Diabetes & Nutritional Sciences, King’s College London, Faculty of Life Sciences & Medicine, London, UK 3) Department of Clinical Science, UMAS, Clinical Research Center, University of Lund, Malmö, Sweden 4) Department of Neuroscience and Physiology, Metabolic Research Unit, University of Gothenburg, Gothenburg, Sweden

Abstract. Pancreatic islets express high levels of the orphan G-protein coupled receptor C5C (GPRC5C), the function of which remains to be established. Here we have examined the role of GPRC5C in the regulation of insulin secretion and β-cell survival and proliferation using human and mouse pancreatic islets. The expression of GPRC5C was analysed by RNA-sequencing, qPCR, western blotting and confocal microscopy. Insulin secretion and cell viability were determined by RIA and MTS assays, respectively. GPRC5C mRNA expression and protein level were reduced in the islets from type-2 diabetic donors. RNA sequencing in human islets revealed GPRC5C expression correlated with the expression of controlling , cell survival and proliferation. A reduction in Gprc5c mRNA and protein expression was observed in islets isolated from old mice (>46 weeks of age) compared to that in islets from newborn (<3 weeks) mice. Down-regulation of Gprc5c led to both moderately reduced glucose-stimulated insulin release and also reduced cAMP content in mouse islets. Potentiation of glucose-stimulated insulin secretion concomitant with enhanced islet cAMP level +2 by all-trans retinoic acid (ATRA) was attenuated upon Gprc5c-KD. ATRA also increased [Ca ]i in Huh7-cells. Gprc5c over expression in Huh7 cells was associated with increased ERK1/2 activity. Gprc5c-KD in clonal MIN6c4 cells reduced cell proliferation and in murine islets increased apoptosis and the sensitivity of primary islet cells to a cocktail of pro-apoptotic cytokines. Our results demonstrate that agents activating GPRC5C represent a novel modality for the treatment and/or prevention of diabetes by restoring and/or maintaining functional β-cell mass.

Key words: Orphan GPCR, Diabetes, Insulin release, RAIG2, RAIG3

PANCREATIC β-cells secrete insulin in response to apoptosis exceeds the normal physiological rate [1-7]. various metabolic, nervous and hormonal factors and These data echo clinical observations indicating that a play a pivotal role in systemic glucose homeostasis [1, sustained increase in plasma glucose levels is associ- 2]. Both impaired glucose-stimulated insulin secretion ated with impaired nutrient-stimulated insulin secre- (GSIS) and loss of pancreatic β-cell mass are hallmarks tion in diabetes-prone individuals [3, 6-9]. of type-2 diabetes (T2D). However, the impairment of Evidence has been presented that daily consumption GSIS is a progressive disorder and is generally consid- vitamin A improves pancreatic β-cell function and pre- ered to be a multistep process [3-6]. Pancreatic β-cells vents or delays the transition from pre-diabetes to frank are very sensitive to pathophysiological stressors (such T2D [10]. There are also data implicating vitamin A in as increased levels of reactive oxygen species, cyto- pancreatic islet development, formation, and function kines and metabolites) and, according to one prevalent and that postnatal restriction of vitamin A results glu- view, β-cell dysfunction occurs once the rate of β-cell cose intolerance in adult rodents [11-13]. Many effects of vitamin A are mediated by its oxi- Submitted Jul. 13, 2016; Accepted Nov. 21, 2016 as EJ16-0338 dised metabolite all-trans retinoic acid (ATRA), which Released online in J-STAGE as advance publication Feb. 22, 2017 has been shown to regulate the of many Correspondence to: Albert Salehi, Department of Clinical Science, UMAS, University of Lund, Malmö, Sweden. genes involved in cell growth and differentiation [11, E-mail: [email protected] 14], effects that require at least 2-4 hours to occur [11- * These authors contributed equally to this work. 13]. However, the finding that ATRA influence insu- ©The Japan Endocrine Society

Endocrine Journal Advance Publication 2 Amisten et al. lin secretion during acute exposure (<1 h) suggests purity of the islets preparation was 70-90%. For the that it also signals via a non-genomic mechanism that qPCR measurements, clean islets without any detect- is independent of transcriptional regulation induced by able attachment of exocrine tissue were selected nuclear retinoid acid receptors [11-13]. under a stereomicroscope. Work on human pancreatic We recently reported that the membrane-bound islets was approved by the ethical committees at the G-protein coupled receptor GPRC5C is highly Universities of Uppsala and Lund, Sweden. expressed in pancreatic islets [15]. We have now investigated the expression of GPRC5C in human and Chemicals mouse pancreatic islets and studied the consequences Collagenase (CLS 4) was from Sigma, and fatty acid- of GPRC5C down-regulation on glucose-stimulated free bovine serum albumin (BSA) was from Boehringer insulin secretion, insulin content, cell proliferation Mannheim. Polyclonal rabbit anti-GPRC5C and HRP- and islet cell viability. Our data indicate that GPRC5C conjugated goat anti-rabbit IgG were from Santa Cruz plays a critical role in the function, survival and pro- Biotechnologies. Cy2-conjugated anti-rabbit IgG and liferation of pancreatic β-cells and mediates the acute Cy5-conjugated anti-guinea pig IgG were from Jackson (non-genomic) effects of ATRA. Immunoresearch Laboratories Inc. The insulin radio- immunoassay was from Millipore. All other chemicals Materials and Methods were from Merck AG or Sigma. Gprc5c shRNA and scrambled control lentiviral particles were obtained Animals from Santa Cruz Biotechnology. Female mice of the NMRI strain (B&K, Sollentuna, Sweden), weighing 25-30 g, were used for the experi- Expression of GPRC5A, GPRC5B, GPRC5C and ments on primary islets. The mice were fed a standard GPCR5D in pancreatic islets pellet diet (B&K) and tap water ad libitum. All animals The expression of GPRC5A-D in human and were housed in metabolic cages with constant temper- mouse islets was quantified by quantitative real-time ature (22ºC) and 12-h light/dark cycles. All experi- PCR (qPCR) using Qiagen’s QuantiFast qPCR kit mental protocols and all procedures involving animals and QuantiTect primers as described elsewhere [17]. were approved by the local animal welfare committee In order to mitigate any potential effects caused by (Lund, Sweden). changes in house keeping expression in islets from non-diabetic and T2D human donors, the mRNA Isolation of mouse pancreatic islets expression data was normalised against three separate Preparation of mouse pancreatic islets was per- housekeeping genes (GAPDH, HPRT and PPAI). formed by retrograde injection of a collagenase solu- tion via the biliopancreatic duct. The islets were iso- Expression of GPRC5C in human pancreatic islets lated and handpicked under a stereomicroscope at analysed by RNA-sequencing room temperature [16]. Total RNA was isolated from pancreatic islets from 89 human cadaveric donors, RNA-sequencing sample Human pancreatic islets preparation was performed using Illumina’s TruSeq Isolated human pancreatic islets from nondiabetic RNA Sample Preparation Kit. The resulting librar- and diabetic male and female donors were provided ies were quality controlled on a 2200 Tapestation by the Nordic network for clinical islet transplan- (Agilent Technologies) before combining 6 samples tation (Professor O. Korsgren, Uppsala University, into one pool for sequencing on one lane on a Flow Sweden). Donor characteristics are summarised in cell sequenced on a HiSeq 2000 (Illumina). The out- Supplementary Table 1. Prior to the experiments, put reads were aligned to the human reference genome the human islets were cultured for 1-5 days in CMRL (hg19) with STAR [18, 19] and feature counting was 1066 (ICN Biomedicals) supplemented with 10 mmol/l done using feature Counts [20]. Raw data was nor- HEPES, 2 mmol/l L-glutamine, 50 μg/ml gentamicin, malised using trimmed mean of M-values and pre- 0.25 μg/ml fungizone (Gibco), 20 μg/ml ciprofloxa- sented as Fragments Per Kilobase of Exon Per Million cin (Bayer Healthcare) and 10 mmol/l nicotinamide Fragments Mapped (FPKM) or transformed using at 37°C and gassed with 5% CO2. Upon arrival, the into log2 counts per million using the voom-func-

Endocrine Journal Advance Publication GPRC5C, retinoic acid and β-cell 3

tion (edgeR/limma R-packages). Human islet genes including 100 μmol/l IBMX. After the incubation, the strongly associated (Spearman’s rho>0.8) with the islets were stored in HCl (100 mmol/l) for 5 min at expression of GPRC5C transcript were identified room temperature followed by sonication. cAMP was using the “Cor.text” function in the statistical soft- measured using a direct cAMP ELISA kit (Enzo Life ware R and uploaded into IPA and subjected to an ‘IPA Sciences) according to the manufacturer’s instructions. Core Analysis’. Genes correlated with the expression The protein concentrations of the islet lysates were of GPRC5C in human islets and a range of annotated measured by a BCA kit (Thermo Fisher Scientific Inc.). cellular functions known to impact β-cell function and diabetes (cell death and survival, cell cycle, carbohy- Gprc5c transfection and detection of associated signal drate metabolism, lipid metabolism, glucose tolerance, Signalling pathways regulated by the orphan recep- hormone secretion, hyperglycemia, quantity of islet tor GPRC5C were identified using a Cignal 45-Pathway cells, T2D risk, transport of vesicles) were then identi- Reporter Array (Qiagen) in Huh7 cells, as the Huh7 fied from the ‘Biological Functions’ analysis of the IPA cell line is both easy to transfect and expresses - Core Analysis. tively high levels of native GPRC5C receptor, which suggests that all components required for GPRC5C Western blotting signaling are also present in the Huh7 cell line. MIN6 Lysates of islets (150 islets/vial; 20 µg of total pro- cells are not suitable for the Cignal Array assay, as tein/lane) were analysed by SDS–PAGE using 7.5% they cannot be transfected at sufficient efficiency for SDS-polyacrylamide gels (Bio-Rad), transferred to the Cignal array assay. Briefly, a Huh7 cell line sta- nitrocellulose membranes (Bio-Rad) and blocked for bily over-expressing the GPRC5C receptor (+~400% 1 h at room temperature in 5% (weight/vol) fat-free GPRC5C expression as measured by qPCR) was gen- milk protein. The expression of mouse and human erated by transfecting Huh7 cells with a pcDNA6.2/ GPRC5C receptor protein relative β-actin was deter- V5-DEST plasmid (Life Technologies) encoding mined using a rabbit-raised polyclonal anti-human the full-length ORF of the human GPRC5C receptor and mouse Gprc5c antibody (1:500) and a rabbit (Source Boscience). Signalling pathways activated or anti-β-actin antibody (Sigma) (1:200) that were incu- inhibited by GRC5C over-expression were then iden- bated with the membrane for 1 h at room temperature tified following the manufacturer’s instructions. For in Tris-buffered saline-Tween 20 (TBST) buffer with validation of stimulation of MAPK signaling (i.e. 0.01 % (v/v) Tween 20. After several washes in TBST ERK1/2 phosphorylation) in GPRC5C over express- buffer, blots were probed with an HRP-conjugated sec- ing Huh7 cells, we used a rabbit anti-phospho-p44/42 ondary antibody (Life Technologies) (1:2,000). For primary antibody (Cell Signalling Technologies), and a the comparison of Gprc5c protein expression between rabbit anti-GAPDH antibody (New England Biolabs) newborn (<3 weeks) and aged mice (>46 weeks), was used to label the loading control protein GAPDH. islets (150 islets/vial) were pooled from 5-10 mice in Both antibodies were visualized with help of an RDye each experiment, due to the low yield of islets from the 800CW conjugated goat anti-rabbit secondary anti- newborn mice. body (LI-COR Biosciences Ltd, UK). Densitometry was performed on 4 independent experiments. The - Confocal microscopy ative expression of each band was quantified and nor- The co-expression of Gprc5c with insulin was deter- malized to the GAPDH as loading control. mined by immunohistochemistry using Zeiss confocal microscopy software (Zen lite 2012) as described else- Effects of Ca2+ mobilisation by ATRA in GPRC5C where [8, 9] with the polyclonal antibodies targeting over-expressing Huh7 cells insulin (1:400) and GPRC5C (1:200) (see above). The effects on Ca2+ mobilisation by Huh7 cells over-expressing the GPRC5C receptor were probed cAMP measurement using FLIPR. Briefly, 100,000 Huh7 cells/well were Mouse islets with either normal or Gprc5c-KD were plated out in black walled, clear bottomed plates and incubated for 60 min at 1 or 16.7 mmol/l glucose in the left for 48 hours to recover in a humidified incubator absence or presence of RA (5 μmol/l). Each incuba- at 37°C with 5% CO2. On the day of experimentation tion vial contained 50 islets in 1.0 ml of KRB-buffer media was removed and cells were loaded with 50 µl

Endocrine Journal Advance Publication Endocrine Journal Advance Publication 4 Amisten et al. of dye buffer (assay buffer with 2.5 µM Fura-2-AM 6 days after the recovery period. Briefly, the MIN6c4 and 1 mM probenecid) for 1 hour at 37°C before being cells were washed three times with PBS, pulsed with replaced with 100 µl of assay buffer (10 mM HEPES, 1 µCi/well of [methyl-3H] thymidine (Amersham) and 170 mM NaCl, 5 mM KCl, 2 mM CaCl2, 1 mM MgCl2, incubated for an additional 20 h in complete RPMI 10 mM glucose, pH7.4). ATRA (10 µM) or the positive medium. The cells were harvested onto glass fibre fil- control ATP (10 µM), was added to the cell plate and ters using a FilterMate harvester (Perkin Elmer). The calcium flux was measured during a 5 minute period filters were air-dried, and bound radioactivity measured using a Flexstation 3 instrument. Ratiometric fluores- using a liquid scintillation counter (Perkin Elmer). In cence values were captured and the effects of calcium some experiments proliferation was measured by BrdU flux assessed. colorimetric assay according to the manufacturer’s instructions (Roche, Mannheim, Germany). Down-regulation of Gprc5c Freshly isolated mouse islets were cultured in 1 ml Cell viability and apoptosis RPMI 1670 medium (Statens Veterinärmedicinska After treatment with Gprc5c shRNA lentiviral par- Anstalt, Sweden) for 36 h in the presence of a pool ticles (see above), mouse islets were dispersed into of five different Gprc5c shRNA lentiviral particles single cells using Ca2+ free medium. The islet cells or scrambled control lentiviral particles according were then cultured with or without a cocktail of to the manufacturer’s recommendations (Santa Cruz pro-apoptotic cytokines (IL-1β, 100 ng/ml; TNFα, 125 Biotechnology). After transfection, the islets were ng/ml; and INFγ, 125 ng/ml) for 24 h in RPMI 1640 washed, supplied with fresh RPMI 1670 medium and with 5 mmol/l glucose and 10% FSB supplemented allowed to recover for 12 h under cell culture condi- with 5 μmol/l ATRA. Measurements of cell viability tions. The islets were then washed again and assayed and detection of apoptosis were performed using the for insulin secretion in the absence or presence of MTS reagent kit (Promega) and Cell Death Detection test agents as described elsewhere [9]. The degree ELISAplus kit (Roche) according to the manufactur- of target-gene-specific down-regulation of Gprc5c er’s instructions. gene and protein expression was determined by qPCR and western blotting as described above. In addi- Statistics tion, glucose-stimulated insulin secretion in relation Results are expressed as means ± SEM for the indi- to total insulin and protein content was investigated in cated number of observations or illustrated by an MIN6c4 cells following down-regulation of Gprc5c. observation representative of a result obtained from different experiments (western blots and confocal Cell proliferation microscopy). Probability levels of random differences Clonal mouse insulin-secreting Min6c4 cells were were determined by Student’s t-test or, where appli- seeded at 10 000 cells/well into 48-well plates in cable, the analysis of variance followed by Tukey- Dulbecco’s modified Eagle’s medium (DMEM) (Life Kramers’ multiple comparisons test. A p value <0.05 Technologies) with GlutaMaxTM (Gibco, USA) con- was considered significant. taining 4.5 g/l glucose and supplemented with 15% fetal calf serum, 50 μg/l streptomycin (Gibco), 75 mg/l peni- Results cillin sulphate (Gibco) and 5 μl/ml β-mercaptoethanol (Sigma). The cells were transfected with lentiviral Expression of the orphan receptors GPRC5C and particles targeting Gprc5c as described above. After its correlation with other gene transcripts in human transfection and a 12 h recovery period, the plates were pancreatic islets incubated for 1-6 days at 37°C with 5% ambient CO2. qPCR was used to verify that GPRC5C is one of the The cells (in individual wells) were harvested by tryp- most abundantly expressed GPCR mRNAs in human sination and counted after 2-6 days in tissue culture islets (Fig. 1A). Similar results were obtained when using a Bürcker chamber. the mRNA expression of GPRC5C was determined The effects of Gprc5c down-regulation on cell pro- using RNA-sequencing (n=89 human pancreatic islet liferation were also studied by measuring thymidine preparations, data not shown). A marked and signif- incorporation in the DNA of replicating cells 12 h or icant correlation between GPRC5C and 1,194 gene

Endocrine Journal Advance Publication GPRC5C, retinoic acid and β-cell 5

ABGPR56 (ADGRG1) 0.16 0.08 GPR119 0.04 GPR40 (FFAR1) Carbohydrate/lipid metabolism (159) Cell death and 0.03 GLP1R survival (298) 0.02 GPRC5C Cell cycle (71) 0.01 GCGR Mean expression/GAPDH 0 Mean expression of 293 GPCRs in human islets Total number (correlated transcripts 528)

Fig. 1 Expression of 293 GPCRs in human pancreatic islet (A) Mean expression level of 293 GPCRs performed by qPCR in isolated human islets (n=3-4 donors) illustrating that GPRC5C is one of the most abundantly expressed GPCRs in human islets. (B) Pie chart of functional annotations of the 388 human islet genes strongly associated with the mRNA expression of GPRC5C: 56.4% of the genes with known function (298 genes) regulate cell death and survival, 13.4% (71 genes) regulate cell cycle and 30.1% (159 genes) regulate carbohydrate and lipid metabolism. Please note that several genes were annotated to more than one cellular function category. The remaining 806 strongly associated genes have as yet unknown functions relating to β-cell function and diabetes. Data are based on 89 individual human islet preparations.

transcripts from human pancreatic islets was found noreactivity was also detected in insulin- negative islet (Supplementary Table 2, Pearson Correlation, ~p<1.1 cells (Fig. 2B). × 10-23, r>0.8). Ingenuity Pathways Analysis (IPA) revealed that 806 (67.5%) of the genes correlated with Expression of the orphan receptors GPRC5A, GPRC5C mRNA expression have no known functions GPRC5B, GPRC5C and GPRC5D in mouse and relevant for islet physiology or diabetes, whereas the human pancreatic islets remaining 388 genes (32.5%) are involved in regulat- We also compared the expression levels of all mem- ing (one or several) cellular functions associated with bers of the GPRC5 family with that of GLP1R, a GPCR islet physiology and diabetes: cell death and survival with a well-documented function within the pancre- (298 genes), cell cycle (71 genes), carbohydrate/lipid atic islet [21]. The qPCR data showed that GPRC5C metabolism (159 genes) (Supplementary Table 2 and is expressed at levels similar to GLP1R in both mouse Fig. 1B). No genes with known associations to glu- and human islets (Fig. 2C and 2D). We also found cose tolerance, hormone secretion, hyperglycemia, that the mRNA expression of GPRC5C relative to the quantity of islet cells, T2D risk or transport of vesi- house-keeping gene GAPDH is roughly ~100% higher cles were associated with the expression of GPRC5C in human islets than in mouse islets. The qPCR data in human islets. also revealed a very low expression of GPRC5A and GPRC5D in mouse and human islets (Fig. 2C and 2D). Gprc5c protein expression in different mouse tissues We also compared GPRC5C expression by qPCR and and pancreatic islets western blots in islet obtained from non-diabetic (ND; Protein expression of Gprc5c relative β-actin was n=15) and type-2 diabetic (T2D; n=18) donors. Both determined in mouse brain, lung, heart, liver, kidney GPRC5C mRNA expression and protein level were and pancreatic islets. Although Gprc5c receptor pro- lower in T2D compared to ND islets (Fig. 2E and 2F). tein was detected in all the examined tissue homoge- Several studies have implicated Gprc5c in develop- nates, it was expressed at particularly high levels rela- ment/organogenesis [14, 22, 23]. We therefore com- tive to β-actin in the heart and pancreatic islets (Fig. pared the mRNA expression of Gprc5c in islets of 2A). Next, we determined the cellular distribution of newborn (<3 weeks) and older mice (>46 weeks), and the Gprc5c protein in isolated mouse islets by confo- found that the Gprc5c expression was 3-fold higher cal microscopy. As shown in Fig. 2F, 80±3% of the in islets from newborn compared to older mice (Fig. Gprc5-positive cells also contained insulin (Zen 2012 2G), a finding that was confirmed by western blot software, Zeiss Microscope). However, Gprc5c immu- analysis (Fig. 2H).

Endocrine Journal Advance Publication Endocrine Journal Advance Publication 6 Amisten et al.

A Tissue expression (mouse) B Pancreatic islet (mouse) Gprc5c β-actin

Brain Lung Heart Liver Kidney Islets Gprc5c Insulin Merge

CD0.10 qPCR (mouse) qPCR (human)

0.05 0.04 0.04 0.03 0.02

mRNA/GAPDH 0.02 0.01 mRNA/GAPDH 0 0

Gprc5a Gprc5d Gprc5c Gprc5d Glp1R GLP1R GPRC5AGPRC5BGPRC5CGPRC5D EFqPCR (human) Western blot (human) GPRC5C

1.5 0.15 β-actin

1.0 * 0.10 * 0.5 0.05 Protein (%) GPRC5 mRNA 0 0 T2DND T2DND

G qPCR (mouse) H Western blot (mouse) 400 1.5 GPRC5C

β-actin 300 1.0 200 * * * 0.5 100 Intensity (fold change) 0 0 Gprc5c/Gapdh (% of adult islets) Newborn Adult Newborn Adult

Fig. 2 CPRC5C expression in mouse and human pancreatic islets (A) Gprc5c expression in different mouse tissues analysed by western blotting. A representative image of Gprc5c protein expression in mouse brain, lung, heart, liver, kidney and pancreatic islets. For comparison, the protein expression of β-actin in each tissue is also shown (B) Confocal microscopy of mouse islets labelled for Gprc5c (green fluorescence) and insulin (red fluorescence) and the overlay of the two (yellow). Scale bar: 20 μm. (C) Expression of Gprc5a, Gprc5b, Gprc5c, Gprc5d and Glp1r mRNA relative to Gapdh mRNA analysed by qPCR in isolated mouse pancreatic islets (n=6 in each group). (D) Comparison of mRNA expression of GPRC5A, GPRC5B, GPRC5C, GPRC5D and GLP1R analysed by qPCR in human islets relative to GAPDH (n=3 donors). (E) mRNA expression of GPRC5C (n=15 in each group) normalised to three separate housekeeping genes (GAPDH, HPRT and PPAI) in human islets from non-diabetic (Non-Diab) and type-2 diabetic (Diab) islet donors. (F) Representative western blot of GPRC5C protein expression in human islets from non-diabetic and diabetic donors. Mean band intensities (means ± SEM) are presented. The blot for the endogenous control protein β-actin is also shown (n=4). (G) Gprc5c mRNA expression relative to Gapdh mRNA in newborn mouse islets plotted relative the mRNA expression of Gprc5c in adult mouse islets (n=5-6 in each group). (H) Fold change differences in intensity of western blot bands showing Gprc5c protein expression relative to β-actin in isolated islets of newborn (<3 weeks) and adult (>46 weeks) mice analysed by qPCR (n=10-16 in each group). * p<0.05, ** p<0.01.

Endocrine Journal Advance Publication GPRC5C, retinoic acid and β-cell 7

Gprc5c activation and down-stream signalling path- overexpressing GPRC5C and, as shown in Fig. 3B, 2+ ways an oscillatory amplification of [Ca ]i as expressed Since Gprc5c was first identified as a retinoic as relative fluorescent units (RFU) was recorded acid-induced gene we next examine the impact of when ATRA (5 μmol/l) was present. In the following ATRA on cAMP content after down-regulation of experiment, we also studied the impact of GPRC5C 2+ Gprc5c in mouse islets and also on [Ca ]i using over-expression on signaling pathways by measur- Huh7-cells overexpressing Gprc5c. A marked reduc- ing the activation or inhibition of pathway specific tion in cAMP production was observed when islets transcription factors in in Huh7-cells overexpress- were incubated for 60 min at high glucose (16.7 ing GPRC5C. Fig. 3C shows that over-expression mmol/l) compared with low glucose (1 mmol/l) fol- of GPRC5C is associated with increased activity of lowing Gprc5c-KD (p<0.0005) (Fig. 3A). While transcription factors (Cell cycle pathway), SRE ATRA (5 μmol/l) greatly potentiated glucose-induced (MAPK/ERK pathway), NFAT (PKC/Ca2+ path- augmentation of islet cAMP content, no effect was way), PAX6, HNF4 and no significant effect on the seen in the Gprc5c-KD islets (Fig. 3A). PPAR or STAT3, while the pro-inflammatory NFκB 2+ To examine the possible impact of ATRA on [Ca ]i is markedly suppressed signal- in relation to Gprc5c expression, we used Huh7-cells ing from. GPRC5C over-expression in Huh7-cells

A Scramble B ATRA C E2F 10 1.42 SRE * NFAT 8 * * PAX6 6 Gprc5c-KD 1.40 * HNF4 4 * * (RFU: 520 nm) i PPAR ] 2+

cAMP content STAT3 2 * 1.38 Transcription factor (pmol/mg protein) [Ca NFκB 0 1G 1G 0 20 40 60 80 100 120 -2 -1 0 12 16.7G 16.7G Time (sec) Fold change (Log2) GPRC5C over-expressing/wt cells 16.7G+RA 16.7G+RA

wt GPRC5C-OE D E ERK1/2 400 GAPDH * 8 * * 300 6 * Wt 200 4 GPRC5C-OE 100 2 * * Cell proliferation (%) 0 0 Phosphorylation/GAPDH ERK1 ERK2 Wt GPRC5C-OE

Fig. 3 CPRC5C activation and down stream signalling molecules (A) The effect of Gprc5c knock-down on isolated mouse islet cAMP content incubated at low (1 mmol/l) or high (16.7 mmol/l) glucose in the presence or absence of ATRA (5 μmol/l) for 60 min. (B) Addition of ATRA (10 µM) causes increased Ca2+ oscillations in Huh7 cells over-expressing the GPRC5C receptor as measured by FLIPR. (C) Signalling pathways associated with cell cycle progression (E2F), MAPK/ERK (SRE), Ca2+ mobilization (NFAT) and β-cell function (PAX6, HNF4) are more active in cells over-expressing the GPRC5C receptor, whereas the pro-inflammatory transcription factor NFκB is inhibited. GPRC5C over-expression did not affect STAT3 and PPAR transcription factor activity. Data presented as log2 fold change of the mean of transcription factor activity in GPRC5C over-expressing Huh7 cells vs. wt Huh7 cells. (D) Western blots showing phosphorylated ERK1/2 (pERK1/2) relative to GAPDH in wt or HUH7-cells transfected with GPRC5C and densitometry analysis of pERK1/2 bands relative to GAPDH. Mean ± SEM are presented for 4 observations in each group. (E) The proliferative effect on GPRC5C over-expressing Huh7 cells vs. wt were evaluated by BrdU incorporation. Dashed line denotes wt cell proliferation in the absence of FBS. Data are means ± SEM of 4 independent experiments in duplicate. * p<0.05, ** p<0.01, *** p<0.001.

Endocrine Journal Advance Publication Endocrine Journal Advance Publication 8 Amisten et al. was associated with an increased phosphorylation mRNA (-87±4% relative to control islets (scrambled of ERK1/2 (Fig. 3D) and also with an increased cell siRNA; p<0.001; Fig. 4B)), and a similar reduction of proliferation as measured by BrdU incorporation Gprc5c protein expression (Fig. 4C). (Fig. 3E). Since Gprc5c was first identified as a retinoic acid-induced gene and shares some sequence simi- The impact of Gprc5c down-regulation on insulin larity with the metabotropic glutamate receptor fam- secretion in mouse pancreatic islets ily [22], we compared the effects on insulin secre- Insulin secretion from intact mouse islets was mea- tion by exposing islets for 1h to ATRA (0-1,000 sured at either low (1 mmol/l) or high (20 mmol/l) glu- μmol/l) and L-glutamate (50 and 1,000 μmol/l) at cose from islets treated with lentiviral particles con- 8.3 mmol/l glucose (a glucose concentration chosen taining either scrambled (control) or Gprc5c-specific to facilitate the detection of potentiating effects). In shRNAs. Down-regulation of Gprc5c did not affect control islets, ATRA stimulated insulin secretion at basal insulin secretion but moderately (-30%) reduced low concentrations (≤5 μM) whereas no effect was insulin secretion evoked by 20 mmol/l (Fig. 4A). The observed at higher ATRA concentrations (Fig. 4D). In efficiency of silencing was determined by qPCR and Gprc5c-deficient islets, no stimulatory effect of ATRA western blot. Lentivirally delivered shRNAs target- was observed at low concentrations and at higher con- ing Gprc5c resulted in reduced expression of Gprc5c centrations (>500 μmol/l) ATRA inhibited insulin

ABGlucose qPCR C Western blot 1.5 *** Scramble GPRC5C Gprc5c (KD) 1.6 8 β-actin 1.0 * 1.2 6

0.8 4 0.5 0.4 2 *

* Densitometry * * (GPRC5C/β-actin) 0 0 0

Insulin release (ng/islet per h) 1.0 20 Scramble Gprc5c Scramble GPRC5C Expression relative GAPDH Glucose (mmol/l) (KD) (KD)

D Retinoic acid E Glutamate 2.5 Scramble Gprc5c (KD) 1.5 Scramble 2.0 Gprc5c (KD) a 1.5 1.0 1.0 * * * 0.5 * 0.5

Insulin (ng/islet per h) *

* Insulin (ng/islet/ h) 0 0 0 0.5 550 500 1,000 8.3G 8.3G 8.3G Retinoic acid (μmol/l) Glutam (μmol/l) – 50 1,000

Fig. 4 CPRC5C down-regulation and pancreatic islet function (A) The effect of down-regulation of Gprc5c on insulin secretion from isolated islets at low (1 mmol/l) and high (20 mmol/l) glucose. (B) Efficiency of Gprc5c mRNA down-regulation analysed by qPCR in scrambled control and lentivirus-delivered Gprc5c-down-regulated islets. (C) Representative western blots and densitometric analysis of the western blot bands of Gprc5c protein expression relative the endogenous control protein β-actin in scrambled control islets and in islets treated with Gprc5c specific shRNAs. D( ) Concentration-dependent effects of retinoic acid and glutamate (E) on insulin secretion from isolated mouse islets at 8.3 mmol/l glucose in control islets and after down-regulation of Gprc5c (n=6-8 in each group). * p<0.05; ** p<0.01. a p<0.01 (ATRA 5 μmol/l vs. 8.3 mmol/l glucose alone in control islets).

Endocrine Journal Advance Publication GPRC5C, retinoic acid and β-cell 9

secretion by ~25% (Fig. 4D). By contrast, no impact mmol/l) was decreased (-26%) following Gprc5c of Gpcr5c-depletion on insulin secretion induced by down-regulation (Fig. 5C-D). glutamate was observed (Fig. 4E). To study the effect of ATRA and Gprc5c on cell via- bility (i.e. measuring the reductive capacity of islet The impact of Gprc5c on cell proliferation and cyto- cells) and apoptosis, we used mouse primary islet cells, kine-induced apoptosis since these cells do not proliferate in vitro, facilitat- The human islet RNA-sequencing data highlighted ing the analysis of cell viability and apoptosis. After a correlation between GPRC5C and genes influenc- down-regulation of Gprc5c, the islets were dispersed ing cell proliferation and apoptosis. To investigate into single cells and cultured with or without a cock- the effects of GPRC5C down-regulation on β-cell tail of the cytokines IL-1β, TNFα and INFγ (IL-1β, 100 proliferation, we used mouse Min6c4 insulinoma cells ng/ml; TNFα, 125 ng/ml; and INFγ, 125 ng/ml) in the rather than primary mouse β-cells as the latter exhibit presence of ATRA (5 μmol/l). Compared to the scram- a low rate of cell division [5]. As shown in Fig. 5A, bled controls, where the cytokine-mediated reduction there was a significant reduction (~30-50 %) in total in cell viability was approximately 30%, the sensitiv- Min6c4 cell numbers at days 4 and 6 when Gprc5c ity to the cytokine cocktail was exacerbated follow- had been down-regulated (p<0.05). This effect was ing Gprc5c down-regulation and cell viability was further assessed by [3H]-thymidine incorporation. reduced by >70% (p<0.001 vs. control cells; Fig. 6A). While there was no immediate (at day 0) significant We next investigated the impact of ATRA on islet cells effect on MIN6c4 cell proliferation following Gprc5c apoptosis after down-regulation of Gprc5c using iso- down-regulation, there was a marked reduction in thy- lated murine islets cultured for 72h at 5 mmol/l glucose midine-incorporation at day 6, indicating a reduction ± ATRA (5 μmol/l). The preventive effect of ATRA in MIN6c4 cell replication (p<0.05) (Fig. 5B). on islet cell apoptosis was diminished when Gprc5c We also studied the effect of Gprc5c on glucose- was down-regulated (Fig. 6B). We ascertained also by stimulated insulin secretion (GSIS) in Min6c4 cells qPCR that Gprc5c down-regulation did not have any after on day 0, when cell number was not affected. off-target effects on the expression of the nuclear recep- While basal (at 1 mM glucose) insulin release was tors for retinoid acid (i.e. RARα, RARβ and RARγ; not affected, that evoked by high glucose (16.7 Fig. 6C).

AB50 Scramble After 24h After 6 days CD Gprc5c (KD) 60 0.4 30 40 0.3 30 40 20 * * 20 0.2

* Insulin 10 20 10 * 0.1 * Cell number (× 10 000)

0 Insulin (fold increased) * (1,000 cpm/mg protein) Thymidine incorporation 0 00 0246 (released/content/mg protein) Culture period (day) Scramble Scramble Scramble Gprc5c-KD Gprc5c-KD Gprc5c-KD 1G (Scramble) 1G (Gprc5c-KD)16.7G (Scramble) 16.7G (Gprc5c-KD)

Fig. 5 The effect of Gprc5c down-regulation on cell proliferation (A) The effect of Gprc5c down-regulation on Min6c4 cell relative cell numbers counted at time 0 (after 36 h shRNAs treatment and 12 h recovery period) followed by culture and counting of the cells after 2, 4 and 6 days compared with Min6 cells exposed to scrambled control shRNAs. (B) The impact of Gprc5c down-regulation on [3H]thymidine-incorporation in Min6c4 cells direct after down-regulation period or at day 6 (n=10-12 in each group). (C) The effect of Gprc5c down-regulation (36h + 12h recovery period) on glucose-stimulated insulin release in Min6c4 cells (n=6 in each group). (D) Fold change reduction in glucose-stimulated insulin secretion from (C) (n=6 in each group). * p<0.05, ** p<0.01.

Endocrine Journal Advance Publication Endocrine Journal Advance Publication 10 Amisten et al.

AB120 4 C 1.5 *** 1.2 3 * 80 * * 0.9 2 0.6 40 * Apoptosis * * Cell viability (%) * 1 * * * 0.3 * (fold change/mg protein)

0 0 Expression relative GAPDH 0 ScrGprc5c-KD Rarα Rarβ Rarγ Gprc5c Scramble Gprc5c (KD) 5G Scramble Scramble + Cytok 5G + RA Gprc5c (KD) Gprc5c (KD) + Cytok RA

Fig. 6 The impact of Gprc5c down-regulation on cell viability (A) The effect of retinoic acid on cell viability in the absence or presence of a mixture of pro-apoptotic cytokines (IL-1β + INFγ + TNFα) in scrambled control (white columns) and Gprc5c down-regulated (black columns) dispersed mouse islet cells. The islets had been previously cultured for 24h with indicated agents (n=4-6). (B) The impact of ATRA (5 μmol/l) on isolated mouse pancreatic islet cell apoptosis after 24h culture at 5 mmol/l glucose following knock-down of Gprc5c in the islets (n=6-8 in each group). (C) The influence of Gprc5c down-regulation on the expression of nuclear receptors of retinoic acid (Rarα, Rarβ, Rarγ) relative to GAPDH in mouse pancreatic islets analyzed by qPCR after down-regulation of Gprc5c by shRNA treatment (n=6-8). ** p<0.01, *** p<0.001.

Discussion ATRA which are markedly dependent on the expres- sion levels of Gprc5c. Our results also show that both We have recently shown that two members of the metabolic and receptor-mediated cAMP production GPRC5 family (GPRC5B and GPRC5C) are highly is disturbed after Gprc5c-KD, indicating both a posi- expressed in both mouse and human islets [9, 15]. tive tonic activity of the GPRC5C on β-cell function While activation of GPRC5B leads to reduced β-cell and also a role for GPRC5C in rapid, non-genomic function in terms of a reduced viability and a reduced effects of ATRA in this regards, although consider- secretory capacity [9], the functional role of GPRC5C able research still needs to be conducted to ascertain in β-cells has not been investigated to date. Using a whether GPRC5C is the sole receptor accounting for range of techniques, we report here that GPRC5C plays the observed results. The importance of the tonic activ- an important role in β-cell function and survival. Our ity of the GPRC5C receptor, which might explain the RNA sequencing and qPCR data confirm that GPRC5C anti-apoptotic properties GPRC5C, is further con- is highly expressed in human islet cells and at levels firmed by the discovery of a significant increase in the comparable to that of the GLP-1 receptor (GLP1R). activity of several transcriptional factors with benefi- Interestingly, the GPRC5C mRNA expression is cial impact on β-cell function, as well as a strong inhi- approximately 100% higher in human islets compared bition of NFκB activity. As we show here for the first to mouse islets, suggesting that GPRC5C might have time, overexpression of GPRC5C was associated with an even stronger influence on β-cell function in human an increased activity of ERK1/2. It is well known that islets than in mouse islets. The lack of detectable levels ERK1/2 signalling pathway is crucial in cell survival, of other members of the GPRC5 family (GPRC5A and proliferation and differentiation [8]. GPRC5D) argue that these receptors are not involved Our insulin secretion experiments suggest that in pancreatic islet function [24]. Gprc5c has a moderate stimulatory effect on insu- The mechanism by which ATRA modulates pancre- lin secretion in mouse islets and that it might inter- atic β-cell function is not known in details. The pres- act with ATRA rather than with glutamate, previously ent study reveals novel aspects of the interacting role proposed to be the natural ligand of GPRC5B [9]. A 2+ of cAMP and [Ca ]i as effectors of the rapid action of comparison of the responses to ATRA before and after

Endocrine Journal Advance Publication GPRC5C, retinoic acid and β-cell 11

silencing of Gprc5c indicates that whereas low (stimu- during the fetal/neonatal stage and it is known to play latory) ATRA concentrations (<5 μmol/l) may act via an important role in embryogenesis [14, 23, 26], it the GPRC5C receptor, higher (inhibitory) concentra- seems likely that Gprc5c influences cell prolifera- tions of ATRA utilise mechanisms not involving this tion and thereby contributes to the appropriate expan- receptor. In addition, both glucose-induced insulin sion and maintenance of functional β-cell mass. It is secretion and β-cell viability and proliferation were therefore of interest that the levels of GPCR5C pro- reduced in islets with reduced GPRC5C expression. tein were reduced in islets from organ donors with Importantly, glutamate remained capable of potenti- type-2 diabetes, which may contribute to the reduc- ating glucose-stimulated insulin secretion, indicating tion in β-cell mass and insulin secretion associated that GPRC5C is not required for modulatory effects of with type-2 diabetes [4-6]. These observations there- glutamate on the mouse pancreatic β-cell. fore raise the exciting possibility that agents targeting The strong correlation between GPRC5C and genes GPRC5C may represent a novel therapeutic means of implicated in cell death/apoptosis and cell cycle con- increasing the functional β-cell mass and also mod- trol in human islets supports our experimental data estly boost insulin secretion. demonstrating an anti-apoptotic and pro-prolifera- tive effect of Gprc5c activation in β-cells, as silenc- Acknowledgements ing of Gprc5c in mouse islets and MIN6 cells was markedly associated with pancreatic β-cells apopto- We thank Britt-Marie Nilsson and Anna Maria sis and reduced MIN6 cell proliferation. This is in Ramsay for skilled technical assistance and Disa good agreement with our results from Huh7 cells Dahlman for the help with cell counting. We also where overexpression of GPRC5C is associated with thank the Lund University Diabetes Centre (LUDC) increased expression of E2F, a transcription factor for the provided tissues and information. Supported linked to the cell cycle, and with a potent inhibition of by grants from Novo Nordisk Foundation, Diabetes & NFκB, a transcription factor associated with inflam- Wellness Foundation, Öresund Diabetes Academy and mation and apoptosis. Collectively, these data indi- the EFSD/Lilly Research Fellowship program. SA is cate that GPRC5C represents an important regulator a Diabetes UK RD Lawrence Fellow (11/0004172). of β-cell survival and proliferation, and also moder- PR is a Wallenberg Scholar (funded by the Knut and ately stimulates insulin secretion. Additional evidence Alice Wallenbergs Stiftelse), a recipient of a Swedish for an important role of GPRC5C in β-cell expansion Research Council International Recruitment Grant and is provided by the observation that Gprc5c is more holds a Wellcome Trust Senior Investigator Award. abundantly expressed in pancreatic islets of newborn mice (<3 weeks old) compared to older (>46 weeks) Duality of Interest mice. It is known that rodent β-cells in fetal and neo- natal islets are not fully developed and continue to dif- The authors declare that there is no duality of inter- ferentiate into functional β-cells almost up to 4 weeks est associated with this research paper. after birth [5, 25]. Since Gprc5c is highly expressed

Supplementary Table 1 Characteristics of non-diabetic (ND) and type-2 diabetic (T2D) organ donors Gender Age BMI Hb1Ac (%)

Male (ND) (82) 55.7 ± 1.5 23.7 ± 0.2 5.5 ± 0.04

Female (ND) (44) 58 ± 1.2 23.2 ± 0.34 5.6 ± 0.06

Male (T2D) (22) 62.6 ± 1.9 26.5 ± 0.9 *** 6.95 ± 0.26 *** Male ND vs. male T2D

Female (T2D) (13) 59.8 ± 3.1 30.2 ± 1.0 *** 6.82 ± 0.16 *** Female ND vs. female T2D The age, sex, BMI and Hb1Ac are shown. Pancreatic islets from donors with HbA1c <6.2% and BMI of 18.5 to 25.5 were consider non-diabetic (ND) and the islets from donors with the Hb1Ac >6.2% or history of diabetes considered as diabetic islets in our study. *** p<0.001 for male ND vs. male T2D and female ND vs. female T2D.

Endocrine Journal Advance Publication Endocrine Journal Advance Publication 12 Amisten et al.

Supplementary Table 2 Association (rho>0.80) of human islet genes known to regulate carbohydrate and lipid metabolism (CLM), cell cycle (CC) and cell death and survival (CDS) with the expression in human islets of GPRC5C Symbol Annotated function rho COMT CLM, CDS 0.95 MBOAT7 CLM, CDS 0.90 PAFAH1B3 CDS 0.88 GNB2 CDS 0.95 VAC14 CLM, CDS 0.90 DEDD2 CDS 0.88 SHARPIN CDS 0.95 PRDX5 CDS 0.90 RBCK1 CC, CDS 0.88 PKP3 CDS 0.94 LTBR CDS 0.90 FASN CLM, CC, CDS 0.88 NELFB CC, CDS 0.94 PACS2 CDS 0.90 HYAL2 CLM, CDS 0.88 GADD45GIP1 CDS 0.94 BCAR1 CDS 0.90 EIF4EBP1 CLM, CC, CDS 0.88 MAP3K11 CC, CDS 0.94 CLCN7 CDS 0.90 JUP CDS 0.88 FZR1 CC, CDS 0.94 ARHGDIA CLM, CDS 0.90 HMG20B CC 0.88 CACFD1 CDS 0.94 CLDN3 CDS 0.90 EHD1 CLM, CDS 0.88 STUB1 CDS 0.94 ARHGEF1 CDS 0.90 TMEM132A CDS 0.88 NINJ1 CDS 0.94 IGFBP2 CLM, CDS 0.90 VPS18 CC 0.88 NACC1 CDS 0.93 MGMT CLM, CC, CDS 0.90 ERF CDS 0.88 NAGLU CLM, CDS 0.93 MYH14 CC 0.90 BAX CLM, CC, CDS 0.88 FURIN CDS 0.93 TRIP6 CDS 0.90 GRINA CLM, CDS 0.88 PAK4 CDS 0.93 SIGIRR CLM, CDS 0.90 PINK1 CDS 0.87 TSPO CLM, CDS 0.93 PIM3 CDS 0.90 CTBP1 CDS 0.87 KATNB1 CDS 0.93 MAP2K7 CC, CDS 0.90 STK11 CLM, CC, CDS 0.87 NR1H2 CLM, CDS 0.92 DVL1 CDS 0.90 MTG2 CDS 0.87 SIRT6 CLM, CDS 0.92 HSPB1 CDS 0.90 CC2D1A CDS 0.87 GRK6 CDS 0.92 GNA11 CLM, CDS 0.90 WDR81 CDS 0.87 LMNA CLM, CC, CDS 0.92 LRP5 CLM, CDS 0.90 OSGIN1 CC, CDS 0.87 XAB2 CDS 0.92 HSF1 CC, CDS 0.90 ADRBK1 CLM, CDS 0.87 SYNGR2 CDS 0.92 MRPL41 CDS 0.90 LRPAP1 CLM, CDS 0.87 GNPTG CDS 0.92 ISG15 CDS 0.89 PLEC CDS 0.87 APRT CDS 0.92 FASTK CDS 0.89 BOK CDS 0.87 SCYL1 CDS 0.92 DPM3 CLM, CDS 0.89 HDAC11 CC, CDS 0.87 ATP13A2 CDS 0.92 CST3 CDS 0.89 AGAP3 CDS 0.87 CDC34 CDS 0.92 ADRM1 CDS 0.89 SHISA5 CDS 0.87 MPG CLM, CC, CDS 0.92 LGALS3BP CDS 0.89 RXRA CLM, CC, CDS 0.87 MDK CDS 0.92 IRAK1 CLM, CDS 0.89 AKT1S1 CDS 0.87 KIFC3 CDS 0.92 ST14 CDS 0.89 COL18A1 CLM, CDS 0.87 ATAD3A CDS 0.92 HSPBP1 CDS 0.89 SDF2L1 CDS 0.87 NAA38 CDS 0.92 SLC9A3R2 CLM, CDS 0.89 JUNB CC, CDS 0.87 BSG CLM, CDS 0.92 LRWD1 CC 0.89 ZYX CDS 0.87 GPX1 CLM, CDS 0.91 TICAM1 CDS 0.89 TMEM259 CDS 0.87 FKBP8 CDS 0.91 ELL CLM, CDS 0.89 EMD CDS 0.87 MAP2K2 CDS 0.91 MXD4 CC 0.89 ABCD1 CLM, CDS 0.87 TNIP2 CDS 0.91 BRAT1 CDS 0.89 TFPT CDS 0.87 MVP CLM, CDS 0.91 SLC25A10 CDS 0.89 UNC119 CDS 0.87 GPC1 CLM, CC, CDS 0.91 RUVBL2 CC, CDS 0.89 RRAS CDS 0.86 PHLDA3 CDS 0.91 GATAD2A CDS 0.89 LZTS2 CDS 0.86 DPP7 CDS 0.91 ARAF CC, CDS 0.89 AATK CDS 0.86 SLC4A2 CDS 0.91 EEF1D CDS 0.88 MAFK CDS 0.86 PSMB10 CDS 0.91 INTS1 CDS 0.88 H2AFX CDS 0.86 TADA3 CDS 0.91 MAP1S CDS 0.88 LIPE CLM, CDS 0.86 PNPLA2 CLM, CDS 0.91 SCRIB CDS 0.88 MAP3K10 CDS 0.86 ARSA CLM, CDS 0.91 UBE2S CDS 0.88 ZBTB17 CDS 0.86 DDAH2 CLM, CDS 0.91 PVRL2 CDS 0.88 TCF3 CC, CDS 0.86 NLRX1 CDS 0.91 TRAF2 CC, CDS 0.88 ORAI1 CDS 0.86 PKN1 CDS 0.91 CC, CDS 0.88 BIN1 CDS 0.86 GRN CDS 0.91 NME3 CDS 0.88 TRAF7 CDS 0.86 AGPAT2 CLM, CDS 0.91 PRKD2 CDS 0.88 NME4 CLM, CDS 0.86 PIN1 CC, CDS 0.91 BOP1 CC 0.88 PML CLM, CC, CDS 0.86

Endocrine Journal Advance Publication GPRC5C, retinoic acid and β-cell 13

Supplementary Table 2 Cont. Symbol Annotated function rho OAZ1 CDS 0.86 MEN1 CLM, CC, CDS 0.84 NCOR2 CLM, CDS 0.82 ID3 CC, CDS 0.86 GLTSCR2 CDS 0.84 CD14 CLM, CDS 0.82 SUN2 CC, CDS 0.86 RHOG CDS 0.84 NFIC CDS 0.82 CDC37 CDS 0.86 ENDOG CDS 0.84 EHMT1 CC, CDS 0.82 CYBA CDS 0.86 HIGD2A CDS 0.84 INS CLM, CC, CDS 0.82 REPIN1 CDS 0.86 GLIS2 CDS 0.84 CTRB2 CDS 0.82 ID1 CC, CDS 0.86 PDXK CDS 0.84 SOD3 CLM, CDS 0.82 RABL6 CDS 0.86 CEBPB CLM, CC, CDS 0.84 BBC3 CLM, CDS 0.82 MTA1 CDS 0.86 POLD4 CC 0.84 INPP5E CLM, CDS 0.82 C1orf86 CC 0.86 USE1 CDS 0.84 DYRK1B CC 0.82 FGFR4 CLM, CDS 0.86 PLEKHF1 CDS 0.84 CDK9 CDS 0.82 BRMS1 CLM, CDS 0.86 RAD9A CC, CDS 0.84 ACHE CLM, CDS 0.82 CSK CDS 0.86 ZBTB7A CLM, CDS 0.84 IER3 CC, CDS 0.81 LAMA5 CDS 0.86 AXIN1 CDS 0.84 MAP1LC3A CDS 0.81 PIAS4 CDS 0.86 GSTP1 CLM, CDS 0.84 ATN1 CDS 0.81 BCL2L12 CDS 0.86 TRADD CDS 0.84 C1orf159 CDS 0.81 HPN CLM, CDS 0.86 SIRT7 CDS 0.84 PPP1R9B CDS 0.81 ADAM15 CDS 0.86 TNFRSF14 CC, CDS 0.83 PIEZO1 CDS 0.81 CTSD CLM, CDS 0.86 NACC2 CDS 0.83 RPS19 CDS 0.81 DPP9 CDS 0.86 KEAP1 CDS 0.83 SLC4A3 CDS 0.81 AKT2 CLM, CDS 0.85 SIVA1 CDS 0.83 PTOV1 CC 0.81 LONP1 CDS 0.85 FAU CDS 0.83 NAPA CDS 0.81 VPS28 CDS 0.85 COMMD4 CDS 0.83 COX8A CDS 0.81 CAPN10 CLM, CDS 0.85 A4GALT CLM, CDS 0.83 MCOLN1 CDS 0.81 NR0B2 CLM, CDS 0.85 NCK2 CDS 0.83 SREBF1 CLM, CC, CDS 0.81 ZNF512B CDS 0.85 PDLIM4 CDS 0.83 ADAM8 CDS 0.81 HNF1A CLM, CC, CDS 0.85 POMC CLM, CC, CDS 0.83 RAC3 CDS 0.81 BRF1 CDS 0.85 NOC2L CDS 0.83 IDUA CLM, CDS 0.81 PELP1 CDS 0.85 ABCB8 CDS 0.83 NSMF CDS 0.81 BRD1 CDS 0.85 RARA CLM, CC, CDS 0.83 FBXL15 CC 0.81 PLCD3 CDS 0.85 TNFSF12 CLM, CDS 0.83 HSPG2 CDS 0.81 CIZ1 CC 0.85 PKD1 CLM, CC, CDS 0.83 HDAC5 CC, CDS 0.81 HIRA CC 0.85 DOT1L CC, CDS 0.83 MAPK7 CC, CDS 0.81 HS1BP3 CDS 0.85 TSC2 CLM, CC, CDS 0.83 GDF15 CLM, CC, CDS 0.80 TNFRSF12A CLM, CDS 0.85 TFAP4 CC, CDS 0.83 UBE2M CDS 0.80 FOXK2 CDS 0.85 BAP1 CC, CDS 0.83 RPS6KB2 CLM, CDS 0.80 TRPM4 CDS 0.85 DVL2 CDS 0.83 FDXR CLM, CDS 0.80 TRIM28 CDS 0.85 MVK CLM, CDS 0.83 THOC6 CDS 0.80 RPTOR CLM, CC, CDS 0.85 EIF3G CDS 0.83 MT2A CDS 0.80 GPX4 CLM, CDS 0.85 BCL3 CC, CDS 0.83 DAB2IP CLM, CDS 0.80 IRF7 CDS 0.85 JAG2 CDS 0.83 RAPGEF1 CDS 0.80 DNAJC5 CDS 0.85 FBXO2 CDS 0.83 CDR2L CDS 0.80 HLA-F CDS 0.85 DPEP1 CDS 0.83 NRTN CDS 0.80 HIP1R CDS 0.85 IMPDH1 CDS 0.83 PLIN3 CLM, CDS 0.80 GADD45G CC, CDS 0.85 BRD4 CC, CDS 0.83 MAD1L1 CDS 0.80 AGRN CDS 0.84 CAMK2N2 CDS 0.83 MNT CDS 0.80 TXNRD2 CDS 0.84 KDM6B CDS 0.82 CD151 CDS 0.80 SPHK1 CLM, CC, CDS 0.84 TFEB CLM, CDS 0.82 CBX4 CDS 0.80 ADCK3 CDS 0.84 KLHDC8B CC 0.82 Bonferoni-corrected p values for all listed POR CLM, CDS 0.84 FAAH CLM, CDS 0.82 genes < 1e-23. SLC12A7 CDS 0.84 C9orf114 CDS 0.82 PPM1F CDS 0.84 SPHK2 CLM, CC, CDS 0.82 KDM4B CC 0.84 KMT2B CLM, CDS 0.82

Endocrine Journal Advance Publication Endocrine Journal Advance Publication 14 Amisten et al.

References 1. Lundquist I, Panagiotidis G, Salehi A (1996) Islet acid stellate cell activation: implications for treatment of glucan-1,4-alpha-glucosidase: a putative key enzyme pancreatic fibrosis. Gut 55: 79-89. in nutrient-stimulated insulin secretion. Endocrinology 14. Gudas LJ (2013) Retinoids induce stem cell differen- 137: 1219-1225. tiation via epigenetic changes. Semin Cell Dev Biol 24: 2. Ashcroft FM, Rorsman P (2012) Diabetes mellitus and 701-705. the beta cell: the last ten years. Cell 148: 1160-1171. 15. Amisten S, Salehi A, Rorsman P, Jones PM, Persaud 3. Muhammed SJ, Lundquist I, Salehi A (2012) Pancreatic SJ (2013) An atlas and functional analysis of G-protein beta-cell dysfunction, expression of iNOS and the coupled receptors in human islets of Langerhans. effect of phosphodiesterase inhibitors in human pancre- Pharmacol Ther 139: 359-391. atic islets of type 2 diabetes. Diabetes Obes Metab 14: 16. Gotoh M, Maki T, Kiyoizumi T, Satomi S, Monaco AP 1010-1019. (1985) An improved method for isolation of mouse pan- 4. Del Guerra S, Lupi R, Marselli L, Masini M, Bugliani creatic islets. Transplantation 40: 437-438. M, et al. (2005) Functional and molecular defects of 17. Amisten S (2012) A rapid and efficient platelet puri- pancreatic islets in human type 2 diabetes. Diabetes 54: fication protocol for platelet gene expression studies. 727-735. Methods Mol Biol 788: 155-172. 5. Weir GC, Aguayo-Mazzucato C, Bonner-Weir S (2013) 18. Trapnell C, Pachter L, Salzberg SL (2009) TopHat: dis- beta-cell dedifferentiation in diabetes is important, but covering splice junctions with RNA-Seq. Bioinformatics what is it? Islets 5: 233-237. 25: 1105-1111. 6. Weir GC, Bonner-Weir S (2013) Islet beta cell mass in 19. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski diabetes and how it relates to function, birth, and death. C, et al. (2013) STAR: ultrafast universal RNA-seq Ann N Y Acad Sci 1281: 92-105. aligner. Bioinformatics 29: 15-21. 7. Mosen H, Ostenson CG, Lundquist I, Alm P, 20. Liao Y, Smyth GK, Shi W (2014) featureCounts: an effi- Henningsson R, et al. (2008) Impaired glucose-stimu- cient general purpose program for assigning sequence lated insulin secretion in the GK rat is associated with reads to genomic features. Bioinformatics 30: 923-930. abnormalities in islet nitric oxide production. Regul 21. De Marinis YZ, Salehi A, Ward CE, Zhang Q, Pept 151: 139-146. Abdulkader F, et al. (2010) GLP-1 inhibits and adrena- 8. Meidute Abaraviciene S, Muhammed SJ, Amisten S, line stimulates glucagon release by differential modula- Lundquist I, Salehi A (2013) GPR40 protein levels are tion of N- and L-type Ca2+ channel-dependent exocy- crucial to the regulation of stimulated hormone secre- tosis. Cell Metab 11: 543-553. tion in pancreatic islets. Lessons from spontaneous obe- 22. Robbins MJ, Michalovich D, Hill J, Calver AR, sity-prone and non-obese type 2 diabetes in rats. Mol Medhurst AD, et al. (2000) Molecular cloning and char- Cell Endocrinol 381: 150-159. acterization of two novel retinoic acid-inducible orphan 9. Soni A, Amisten S, Rorsman P, Salehi A (2013) G-protein-coupled receptors (GPRC5B and GPRC5C). GPRC5B a putative glutamate-receptor candidate Genomics 67: 8-18. is negative modulator of insulin secretion. Biochem 23. Rochette-Egly C (2015) Retinoic acid signaling and Biophys Res Commun 441: 643-648. mouse embryonic stem cell differentiation: Cross talk 10. Dakshinamurti K, Bagchi RA, Abrenica B, Czubryt MP between genomic and non-genomic effects of RA. (2015) Microarray analysis of pancreatic gene expres- Biochim Biophys Acta 1851: 66-75. sion during biotin repletion in biotin-deficient rats. Can 24. Draghici S, Khatri P, Eklund AC, Szallasi Z (2006) J Physiol Pharmacol 93: 1103-1110. Reliability and reproducibility issues in DNA microar- 11. Driscoll HK, Chertow BS, Jelic TM, Baltaro RJ, ray measurements. Trends Genet 22: 101-109. Chandor SB, et al. (1996) Vitamin A status affects 25. Jermendy A, Toschi E, Aye T, Koh A, Aguayo- the development of diabetes and insulitis in BB rats. Mazzucato C, et al. (2011) Rat neonatal beta cells lack Metabolism 45: 248-253. the specialised metabolic phenotype of mature beta 12. Matthews KA, Rhoten WB, Driscoll HK, Chertow BS cells. Diabetologia 54: 594-604. (2004) Vitamin A deficiency impairs fetal islet develop- 26. Mizoguchi Y, Moriya M, Taniguchi D, Hasegawa A ment and causes subsequent glucose intolerance in adult (2014) Effect of retinoic acid on gene expression pro- rats. J Nutr 134: 1958-1963. files of bovine intramuscular preadipocytes during adi- 13. McCarroll JA, Phillips PA, Santucci N, Pirola RC, pogenesis. Anim Sci J 85: 101-111. Wilson JS, et al. (2006) Vitamin A inhibits pancreatic

Endocrine Journal Advance Publication