l e t t e r s

TCF1 links GIPR signaling to the control of function and survival

Jonathan E Campbell1, John R Ussher1,7, Erin E Mulvihill1, Jelena Kolic2,3, Laurie L Baggio1, Xiemen Cao1, Yu Liu1, Benjamin J Lamont1,7, Tsukasa Morii1,7, Catherine J Streutker4, Natalia Tamarina5, Louis H Philipson5, Jeffrey L Wrana1, Patrick E MacDonald2,3 & Daniel J Drucker1,6

The glucagon-like peptide-1 (GLP-1) receptor and the glucose- and survival1, rendering them attractive targets for the treatment dependent insulinotropic polypeptide (GIP) receptor transduce of type 2 diabetes (T2D). Indeed, the prevention of incretin degrada- nutrient-stimulated signals to control beta cell function1. tion by inhibiting dipeptidyl peptidase-4 (DPP4)5 and the augmenta- Although the GLP-1 receptor (GLP-1R) is a validated drug tion of incretin receptor signaling using GLP-1R agonists or emerging target for diabetes1, the importance of the GIP receptor co- and tri-agonists that target this pathway6,7 represent established (GIPR) for the function of beta cells remains uncertain2–4. and investigational strategies for the treatment of T2D. We demonstrate that mice with selective ablation of GIPR in The receptors for GLP-1 and GIP are highly related in structure; beta cells (MIP-Cre:GiprFlox/Flox; Gipr−/−βCell) exhibit lower both are regulated by transcription factor 7-like 2 (TCF7L2)8 in islets levels of meal-stimulated insulin secretion, decreased and control beta cell function and survival through cAMP-dependent expansion of adipose tissue mass and preservation of insulin pathways. Although genetic variation within the coding region of the sensitivity when compared to MIP-Cre controls. Beta cells from GLP1R has been linked to differences in fasting glucose levels Gipr−/−βCell mice display greater sensitivity to apoptosis and and in beta cell function in humans9, interpretation of the impor- markedly lower islet expression of T cell–specific transcription tance of the GIPR for beta cell function is confounded by genetic and factor-1 (TCF1, encoded by Tcf7), a not previously physiological data in humans and in mice and rats that suggest roles characterized in beta cells. GIP, but not GLP-1, promotes beta for the GIPR in the regulation of adipose-tissue accretion, body cell Tcf7 expression via a cyclic adenosine monophosphate weight and insulin sensitivity2–4,10. (cAMP)-independent and extracellular signal–regulated kinase GLP-1R agonists are increasingly used for the treatment of T2D (ERK)-dependent pathway. Tcf7 (in mice) or TCF7 (in humans) and obesity and are under investigation for the treatment of type 1 levels are lower in islets taken from diabetic mice and in diabetes1,11. By contrast, there is less interest in the GIPR as a drug humans with type 2 diabetes; knockdown of TCF7 in human target because of reports of defective GIP action in people with and mouse islets impairs the cytoprotective responsiveness to diabetes who are severely hyperglycemic12. Nevertheless, a brief GIP and enhances the magnitude of apoptotic injury, whereas period of insulin therapy markedly restores GIP responsiveness in restoring TCF1 levels in beta cells from Gipr−/−βCell mice lowers subjects with T2D (ref. 13), and GIPR activation is a key component of the number of apoptotic cells compared to that seen in MIP- the action of several novel co- and tri-agonists that are under investi- Cre controls. Tcf7−/− mice show impaired insulin secretion, gation as potential treatments for diabetes and obesity6,7. Accordingly, deterioration of glucose tolerance with either aging and/or to delineate the importance of GIPR signaling in beta cells inde- high-fat feeding and increased sensitivity to beta cell injury pendently of its potentially confounding actions in extrapancreatic relative to wild-type (WT) controls. Hence the GIPR-TCF1 axis tissues, we mated MIP-CreERT mice with GiprFlox/Flox mice to generate represents a potential therapeutic target for preserving both the mice with conditional and selective inactivation of Gipr in adult beta function and survival of vulnerable, diabetic beta cells. cells (Gipr−/−βCell) (Supplementary Fig. 1a). Levels of Gipr mRNA transcripts were 90% lower in the islets of Gipr−/−βCell mice than in The ingestion of nutrients triggers the secretion of multiple gut peptides, those of MIP-Cre mice, GiprFlox/Flox mice and WT mice (Fig. 1a and including GLP-1 and GIP, both of which are incretin hormones that Supplementary Fig. 1b), whereas Gipr expression in adipose tissue amplify insulin secretion. GLP-1 and GIP also control beta cell growth was not perturbed (Supplementary Fig. 1b). GIP robustly reduced

1Lunenfeld-Tanenbaum Research Institute, Mt. Sinai Hospital, Toronto, Ontario, Canada. 2Department of Pharmacology, University of Alberta, Edmonton, Alberta, Canada. 3Alberta Diabetes Institute, University of Alberta, Edmonton, Alberta, Canada. 4St. Michael’s Hospital, University of Toronto, Toronto, Ontario, Canada. 5Department of Medicine, Kovler Diabetes Center, University of Chicago, Chicago, USA. 6Department of Medicine, University of Toronto, Toronto, Ontario, Canada. 7Present addresses: Faculty of Pharmacy and Pharmaceutical Sciences, University of Alberta, Edmonton, Alberta, Canada (J.R.U.); Department of Medicine, University of Melbourne, Melbourne, Australia (B.J.L.); Akita University Graduate School of Medicine, Akita, Japan (T.M.). Correspondence should be addressed to D.J.D. ([email protected]). Received 24 August; accepted 26 October; published online 7 December 2015 ; doi:10.1038/nm.3997

nature medicine advance online publication  l e t t e r s

G 0 min a G c 0 min 0 min b G G d 1,500 10 min Veh DPP4i 6.7 mM nM GIP 10 min 10 min 1,000 1 Veh DPP4i 25 * 2.5 30 min 6.7 mM 10 2.7 mM 1 2.7 mM + * 500

) 2.5 20 * 2.0 1.5 0.15 * 20 400 2.5 * 20 500 0 * MIP-Cre 300 * 400 * * 200 * 300 2.0 * 15 1.5 Gipr –/–�Cell 15 2.0 15 200 ession 1.0 0.10 100 100 1.5 0 0 10 1.0

elease 1.5 xpr 10 tal insulin) 10 mRNA 1.0 5 MIP-Cre 0.5 Glucose (mM) Insulin (ng/ml) to

e 1.0 0.5 0.05 Gipr –/–�Cell 5 0.5

5 Insulin (ng/ml) Gipr 0.5 0 0 Glucose (mM) Glucose (mM) Insulin r Insulin (ng/ml) elativ * (% of 0 0 30 60 90 120 e ell 0 0 C (R 0 0 0 -Cr � h Time (min) –/– e ell 0 8 16 24 32 –30 0 30 60 90 120 h –30 0 30 60 90 120 Ve MIP C -Cr � Ve DPP4i Gipr –/– Time (min) Time (min) DPP4i Time (min) MIP-Cre MIP Gipr –/–�Cell Gipr 8,000 * 6,000 * 2,000 0 min 8,000 2.0 100 e f MIP-Cre h 4,000 20 30 1,500 150 6,000 2,000 * 1,000 5 10 min 4,000 40 Gipr –/–�Cell 0 500 30 min 2,000 1.5 75 15 0 4 0 20 * 100 30 3 * Glucose 50 10 20 1.0 2 (% of baseline) at mass Glucose at mass 10 50

MIP-Cre F F MIP-Cre MIP-Cre 25 5 1 10

Glucose (mM) 0.5 (% of baseline) –/– Cell

–/– Cell Insulin (ng/ml) � Gipr � Gipr –/–�Cell Gipr Insulin (ng/ml) 0 30 60 90 (% of body weight) (% of body weight) 0 0 0 0 0 Time (min) e ell 0 8 10 12 14 16 18 0 30 60 90 120 C 0 30 60 90 8 14 20 26 -Cr � –/– 0 7 14 Age (weeks) Time (min) MIP r Time (min) Age (weeks) Gip Pellet (mg) MIP-Cre MIP-Cre Gipr –/–�Cell Gipr –/–�Cell g MIP-Cre MIP-Cre MIP-Cre MIP-Cre i 15 –/–�Cell 4 –/–�Cell 15 –/–�Cell –/–�Cell Gipr Gipr Gipr Gipr 20 20 20 1,200 20 1,200 3 2 * 900 * 900 10 10 15 15 15 600 15 600 * 300 300 2 0 0 1 10 10 10 10 5 5 at mass 1 at mass F F Glucose (mM) Insulin (ng/ml) Glucose (mM) Insulin (ng/ml) 5 MIP-Cre 5 MIP-Cre 5 5 Glucose (mM)

–/–�Cell Glucose (mM) 0 0 0 0 Gipr Gipr –/–�Cell (% of body weight) (% of body weight) 0 0 0 0 0 0 0 0 :0 :0 :0 :0 4:00 4:00 4:00 4:00 8 10 12 14 16 18 8 10 12 14 16 18 0 30 60 90 120 0 30 60 90 120 07:0008:00 11 1 07:0008:00 11 1 07:0008:00 11 1 07:0008:00 11 1 Time of day (h) Time of day (h) Time of day (h) Time of day (h) Age (weeks) Age (weeks) Time (min) Time (min) +

−/− βCell G Figure 1 The phenotype of Gipr mice. (a) Gipr expression in mouse islets j MIP-Cre 0 min G G 6.7 mM G –/–�Cell 30 min 1 (n = 5). (b) Insulin release from mouse islets from 18-week-old mice on LFD Gipr 6.7 mM 2.7 mM 1 2.7 mM 1 nM Ex4 30 2.5 * 0.15 (n = 7). G, glucose. (c) Oral glucose tolerance test (OGTT) and plasma insulin MIP-Cre * ** 2.0 levels in 12-week-old (two left panels) and −/− βCell (two right panels) –/–�Cell MIP-Cre Gipr 20 0.10 Gipr * 1.5 * elease

mice administered vehicle (veh) or sitagliptin (a dipeptidyl peptidase-4 inhibitor; tal insulin)

1.0 to DPP4i) (MIP-Cre, n = 5; Gipr−/−βCell, n = 9). (d) Glycemic (left) and insulin 10 * 0.05 0.5 Insulin (ng/ml) Insulin r Glucose (mM) (middle) response during OGTT, and glycemic (bottom) response during an insulin 0 0 (% of 0 tolerance test (ITT), all in 18-week-old LFD mice ( = 7). ( ) Magnetic resonance e ell 0 8 16 24 32 n e 0 30 60 90 120 C -Cr � Time (min) –/– Time (min) imaging (MRI) assessment in LFD-fed mice (n = 7). (f) Glycemic and insulin (two left MIP r panels) responses during OGTT, glycemic (middle right) response during ITT; MRI assessment Gip (right) in 18-week-old HFD-fed mice (MIP-Cre, n = 9; and Gipr−/−βCell, n = 13). (g) Glycemic and insulin (left) responses to re-feeding in 10-week-old LFD-fed mice (MIP-Cre, n = 6; Gipr−/−βCell, n = 9). Glycemic and insulin (right) responses to refeeding in 10-week-old HFD-fed animals (MIP-Cre, n = 8; Gipr−/−βCell, n = 13). (h) Plasma insulin after a 16 h fast, after sham or after insulin pellet (7 or 14 mg) surgery (n = 3, 4 and 4, respectively). (i) MRI analysis (two left panels) after sham (left) or insulin (middle-left) pellet surgery. Glycemic levels (two right panels) during OGTT in 18-week-old mice without (middle-right) or with insulin (right) pellet (n = 3, 5, 3 and 6, respectively). (j) Glycemic (left) and insulin (middle) responses during an intraperitoneal glucose tolerance test (IPGTT) in 18-week-old HFD-mice given Ex4 (MIP-Cre, n = 9; Gipr−/−βCell, n = 13). Insulin secretion (right) from mouse islets isolated from 18-week-old LFD-fed mice (n = 7). *P < 0.05 versus control, or as indicated. Data are expressed as mean ± s.e.m. Statistical tests used: t-test in a and AUC insets (c,d,f,h,i); two-way analysis of variance (ANOVA) for b–g and i,j. glycemic excursion after glucose challenge in all three lines of control MIP-Cre control mice (Supplementary Fig. 1f–h). Unexpectedly, mice but did not lower glycemia in Gipr−/−βCell mice (Supplementary oral and intraperitoneal glucose tolerance and insulin sensitivity Fig. 1c). GIP also robustly stimulated insulin secretion in perifused were paradoxically better in 18-week-old Gipr−/−βCell mice than in islets from control mice, but not in those from Gipr−/−βCell mice MIP-Cre controls (Fig. 1d and Supplementary Fig. 1i). Gipr−/−βCell (Fig. 1b). Administration of a low dose of a DPP4 inhibitor modestly mice accumulated less white adipose tissue (WAT) with age and exhib- improved glucose tolerance and enhanced glucose-stimulated insu- ited reduced adipocyte size (Fig. 1e and Supplementary Fig. 2a), lin secretion in control mice but not in Gipr−/−βCell mice (Fig. 1c). despite having both similar expression levels of the that regulate For subsequent experiments, we used MIP-Cre mice as controls, adipose-tissue metabolism and comparable lipoprotein lipase activity minimizing any potential confounding effects associated with Cre- in adipose tissue (Supplementary Fig. 2b,c). driver lines in beta cells14,15. Notably, human growth hormone (hGH) We reasoned that GIPR-deficient beta cells produce less insulin release and expression of tryptophan hydroxylase-1 (Tph1) and -2 than do beta cells with normal GIPR levels, which, in turn, limits (Tph2) were similar in islets from MIP-Cre and from Gipr−/−βCell mice expansion of WAT depots, thereby improving insulin sensitivity16. (Supplementary Fig. 1d). Accordingly, we reassessed these phenotypes by using two independent Although mice with whole-body deletion of Gipr exhibit modestly strategies to restore insulin levels: (i) exposure to a high-fat diet impaired beta cell function10, 8-week-old Gipr−/−βCell mice fed a low-fat (HFD) and (ii) direct insulin replacement. A HFD stimulates GIP diet (LFD) showed normal oral glucose tolerance and glucose- secretion, which will expand adipose tissue mass, increase both resis- stimulated insulin levels (Supplementary Fig. 1e). Furthermore, tin expression and secretion and promote insulin resistance, thereby insulin sensitivity, weight gain, ambient glycemia, food intake indirectly enhancing insulin secretion even in the absence of GIPR and plasma incretin levels were all comparable in Gipr−/−βCell and action in beta cells17,18. HFD-fed Gipr−/−βCell mice exhibited weight

 advance online publication nature medicine l e t t e r s

a b c d e

6 * 2.0 MIP-Cre MIP-Cre 4 Veh 2.0 –/– Cell * Gipr β Gipr–/–βCell GIP 1.5 3 4 1.5 1.0 2 mRNA 1.0 mRNA 2 mRNA Tcf7 Positive cells 0.5 * 1 0.5 (% of total cells) * (relative expression) (relative expression) 0 0 0 (relative expression) 0 – STZ – – + + –/ WT Tcf7 Lef1 Tcf4 Tcf7 Lef1 Tcf4 Tcf7l1 Tcf7l2 Tcf7l1 Tcf7l2 Ctnnb1 Glp1r TCF1 f β-catenin binding region g h ATG ATG 1.5 40 Veh 4,000 40 3,000 * GIP 2,000 30 1.5 1.5 30 1,000 * * * 1.0 0 4,000

* 3,000 1.0 1.0 mRNA 5′ (1) 5′ (2) 3′ 20 * * 20

2,000 mRNA 0.9 kb Veh GIP mRNA * 1,000 0.5 1.3 kb 10 Veh 0.5 0.5

10 0 TCF7 Glucose (mM) Glucose (mM) Tcf7 GIP Gipr Veh GIP 0 0 0 0 (relative expression) 0 (relative expression) 0 30 60 90 120 0 30 60 90 120 (relative expression) b b Islets ThymusINS1 INS1EINS1832Thymus D Primers WT WT T2 Time (min) Time (min) db/d db/d ND lean 1.3 kb 5′(1) + 3′ ND obese TCF1 6.0 5.0 j 2.0 i * * * Lamin 0.9 kb 5′(2) + 3′ * 4.0 * A/C 1.5 4.0 3.0 pERK mRNA

mRNA 1.0

2.0 ERK TCF1 Figure 2 GIPR controls Tcf7 expression. 2.0

Tcf7 0.5 Tcf7 * * (a) TUNEL-positive β cells in mice treated 1.0 HSP90 protein levels (AU) (relative expression) with vehicle or STZ (left to right, n = 5, 3, 4 0 (relative expression) 0 h 0 h GIP – + + + + Ve GIP h and 4). (b) qPCR analysis of RNA from islets UO126 Ve GIP Ex4 Rp-cAMP – – + – – Ve GIP from 18-week-old LFD-fed mice (n = 7). (c) qPCR UO126 Forskolin UO126 – – – + – analysis of RNA from wild-type (WT) islets treated with GIP + UO126 SB203580 – – – – + GIP + UO126 PBS or GIP for 24 h (n = 6). (d) Tcf7 expression in islets from WT and Glp1r−/− mice (n = 4). (e) Representative image of immunohistochemistry for TCF1 expression in WT mouse pancreas. Scale bar, 50 µm (n = 3). (f) Schematic of the Tcf7 mRNA coding region (top) demonstrating multiple start codons. Representative image (bottom) of 1.3 kb and 0.9 kb Tcf7 PCR products from mouse islets and rat insulinoma cell (INS1) subclones INS1, INS1E and INS1832 (n = 2). (g) Glycemic response during an IPGTT in WT (left) and db/db (middle left) mice (n = 5). qPCR analysis for Gipr (middle right) and Tcf7 (right) mRNA in mouse islets (n = 5). (h) TCF7 expression in human islets (n = 5, 5 and 4). ND, non-diabetic. (i) Tcf7 expression in WT mouse islet RNA (left, n = 4) or INS1 832/32 β cells (right, n = 6). (j) Tcf1, Lamin A/C, phosphorylated or non-phosphorylated extracellular signal–regulated kinase (ERK) and heat shock protein 90 (HSP90) protein expression in INS1 832/32 β cells (n = 6). *P < 0.05 versus control or as indicated. Data are expressed as mean ± s.e.m. Statistical tests used: t-test for b–d,g (AUC insets); one-way analysis of variance (ANOVA) for h–j; two-way ANOVA for a,g. TUNEL, terminal deoxynucleotidyl transferase dUTP nick end labeling. gain, plasma incretin levels and glycemia similar to those seen in exhibited greater insulin secretion in response to the GLP-1R agonist MIP-Cre controls (Supplementary Fig. 2d), but they no longer exendin-4 (Ex4) in vivo and in isolated perifused islets ex vivo relative exhibited improved glucose tolerance, higher insulin sensitivity or to MIP-Cre mice (Fig. 1j). Conversely, the selective reduction of less adipose tissue mass as compared to MIP-Cre controls (Fig. 1f and GLP-1R signaling impaired glucose tolerance to a greater extent in Supplementary Fig. 2d). Gipr−/−βCell mice than in MIP-Cre controls (Supplementary Fig. 2g). Consistent with an essential role for the beta cell GIPR in nutrient- Similarly, individual patch-clamped beta cells from Gipr−/−βCell mice regulated insulin secretion, insulin levels were lower in LFD-fed did not respond to GIP yet exhibited enhanced sensitivity to Ex4 Gipr−/−βCell mice than in LFD-fed MIP-Cre mice during the re-feeding (Supplementary Fig. 2h), whereas beta cells from Gipr−/−βCell mice period after an overnight fast, despite showing comparable levels of demonstrated normal actin-depolymerization responses to high levels glycemia (Fig. 1g). In contrast, HFD feeding restored insulin profiles of either glucose or Ex4 but did not respond to GIP (Supplementary in Gipr−/−βCell mice to the levels observed in MIP-Cre mice (Fig. 1g). Fig. 2i). Gipr−/−βCell mice also exhibited enhanced sensitivity to We next implanted insulin pellets in Gipr−/−βCell mice to reverse the exogenous administration of the insulinotropic G protein–coupled relative hypoinsulinemia detected under LFD conditions (Fig. 1g). receptor (GPCR) 119 (GPR119) agonist AR231453 compared to The pellets raised insulin levels by ~1 ng per ml (Fig. 1h), an amount control mice (Supplementary Fig. 2j). similar to the difference in postprandial insulin values detected in Given that class B GPCRs modulate beta cell growth and sur- MIP-Cre compared to Gipr−/−βCell mice (Fig. 1g); insulin pellets vival1,22, we analyzed Gipr−/−βCell islets. Loss of the beta cell GIPR had no effect on body weight, glycemia (Supplementary Fig. 2e) reduced beta cell area, but islet size, islet number, granule morphology or food intake (data not shown). The normalization of postprandial and granule size were all unaffected (Supplementary Fig. 3a,b). insulinemia restored adiposity and insulin sensitivity to control levels The adaptive increase in beta cell mass after exposure to a HFD was in Gipr−/−βCell mice (Fig. 1i and Supplementary Fig. 2f). preserved in Gipr−/−βCell mice (Supplementary Fig. 3a), revealing that The relatively modest basal glucose-tolerance phenotype of the beta cell GIPR is dispensable for the adaptive response to insulin Gipr−/−βCell mice may reflect the upregulation of related nutrient- resistance. Because GIPR signaling controls beta cell survival23,24, sensitive beta cell signaling pathways19–21. Indeed Gipr−/−βCell mice we assessed the response to the beta cell toxin streptozotocin (STZ).

nature medicine advance online publication  l e t t e r s

−/− Figure 3 Phenotype of Tcf7 mice. WT WT 2,500 a –/– –/– b 1,500 * c Tcf7 Tcf7 1,000 WT WT (a) Glycemic (left) and insulin (right) response 1,500 –/– –/– * 4.0 25 500 2.5 Tcf7 Tcf7 25 1,000 ) * 0 150 during an OGTT in 18-week-old mice fed a LFD * 500 20 3.0 20 * * 2.0 * * 0 ( = 7 and 8). (b) Glycemic (left) and insulin 15 15 * 1.5 100 n * 2.0 10 10 1.0

(right) response during an OGTT in 12-week-old Insulin 50 1.0 WT Glucose 5 5 –/– 0.5 Insulin (ng/ml Glucose (mM) Glucose (mM) Tcf7 (fold increase) HFD-fed mice (n = 4 and 8). (c) Glycemic 0 0 0 0 (% of baseline) 0 response during an intraperitoneal insulin 0 30 60 90 120 010 0 10 0 30 60 90 120 010 0 10 0 30 60 90 120 tolerance test (ITT) in 12-week-old HFD-fed Time (min) Time (min) Time (min) Time (min) Time (min) mice (n = 5). (d) Insulin release from islets WT 1.5 d –/– e 50 Scrambled isolated from 18-week-old LFD-fed mice (n = 8). Tcf7 16.7 mM G 40 TCF7 siRNA 2.7 mM G 8.3 mM G 16.7 mM G 2.7 mM G +10 nM GIP 1.0 (e) TCF7 expression in beta cells from human 0.30 30 + veh mRNA * donors (top left) (n = 4 and 4). Cumulative 0.5 20 ** (relative

* expression) TCF7 capacitance during sequential depolarization in Cumulative 10 0 0 individual human beta cells exposed to veh (top 0.20 Capacitance (fF/pF) * right, n = 16 and 18), GIP (bottom left, n = 15 0 1 2 3 4 5 6 7 8 910 siRNA Depolarization number Scrambled and 15) or Ex4 (bottom right, n = 14 and 13). TCF7 (f) TUNEL-positive beta cells in mice treated 0.10 ** 50 Scrambled 50 Scrambled * * TCF7 siRNA * with veh or STZ ( = 5, 4, 4 and 5). (g) TUNEL- 40 * * 40 TCF7 siRNA n + 10 nM * * GIP * + 1 nM Ex4 positive beta cells in mice treated with STZ plus Insulin release (% of total) 30 * * 30 20 * 20 veh, GIP, or Ex4 (n = 4). (h) TUNEL-positive 0 Cumulative 10 Cumulative 10 staining in human beta cells (several hundred 0 10 20 30 40 0 0 capacitance (fF/pF) Time (min) capacitance (fF/pF) cells from four individual donors) treated with 0 1 2 3 4 5 6 7 8 910 0 1 2 3 4 5 6 7 8 910 low (5.5 mM) glucose (G) or high (25 mM) G Depolarization number Depolarization number plus interleukin 1-β (IL-1β). (i) Apoptotic index f Time (d) 0 1 2 3 4 5 g h i in mouse beta cells (n = 3 for MIP-Cre and 5 Time (d) –5–4 –3 –2 –1 0 1 2 3 4 5 STZ MIP-Cre STZ Takedown Takedown Scrambled −/− βCell (50 mg/kg) ) for Gipr ). *P < 0.05 versus control, or as (50 mg/kg) TCF7 siRNA Gipr–/–�Cell WT PBS/GIP/Ex4 )

) –/– (24 nmol/kg) * indicated. Data are expressed as mean ± s.e.m. 6 Tcf7 WT 4 * 30 * ) 8 –/– * 5 Tcf7 * Statistical tests used: t-test for e (top left) and + * * 3 * L 20 * 4 + 6 AUC insets (a,b); two-way ANOVA for a–i. 3 L 2 * 4 (fold increase 10 +

TUNE 2 * L 1 1 TUNE 2 / # of total cells ** Apoptosis index

(% of total cells 0 0

0 (# of apoptotic cells (% of total cells

More apoptotic beta cells were detected in 0 TUNE G G Ad-GFP + + –– STZ –– + + β −/−βCell STZ + + + + + + Ad-Tcf7-GFP – – + + islets from STZ-treated Gipr mice than GIP ––+ ––+ 5.5 mM 25 mM+ IL1- Thapsigargin – + – + in those from MIP-Cre controls (Fig. 2a), Ex4 – – + –– + whereas levels of mRNA transcripts for genes encoding either pro- exogenous GIP (Fig. 2g and Supplementary Fig. 4a–d). By contrast, survival GPCRs, components of the glucose-sensing machinery, the levels of Glp1r and Tcf7l2 mRNA transcripts were unchanged including GLUT2 (Slc2a2, which transports STZ), and mediators of in islet RNA from 6- or 12-week-old db/db mice (Supplementary proliferation or survival were not differentially expressed in Gipr−/−βCell Fig. 4e). Levels of TCF7 mRNA transcripts were lower in islets from islets (Supplementary Fig. 3c and Supplementary Table 1). obese, non-diabetic subjects and significantly lower in islets from As the Wnt target TCF7L2 regulates both incretin-receptor expres- subjects with T2D than in non-obese, non-diabetic control subjects sion and beta cell survival8,25, we assessed mRNA levels of β-catenin (Fig. 2h and Supplementary Fig. 4f). The GIP-dependent induction and members of the TCF family. Islet expression levels of catenin of Tcf7 expression was not mimicked by forskolin or by Ex4 and was (cadherin-associated protein), beta 1 (Ctnnb1), Tcf7l1, Tcf7l2 and Tcf4 not sensitive to the cAMP-inhibitor Rp-cAMP (Fig. 2i). By contrast, were not affected by a loss of GIPR in beta cells; however, mRNA the inhibition of ERK1/2 phosphorylation with the mitogen-activated transcripts for Tcf7 and for lymphoid enhancer binding factor 1 protein kinase inhibitor UO126 completely prevented the GIP-stimu- (Lef1), both of which encode not previously assigned func- lated increase in Tcf7 RNA (Fig. 2i) and Tcf1 protein levels (Fig. 2j). tional roles in beta cells, were markedly lower in Gipr−/−βCell islets Hence signaling via the GIPR, but not the GLP-1R, controls Tcf7/Tcf1 (Fig. 2b). Moreover, GIP directly increased Tcf7 expression in expression through a cAMP-independent and ERK1/2-dependent WT mouse islets ex vivo (Fig. 2c). Furthermore, although GLP-1R pathway in beta cells. and GIPR share cAMP-dependent pathways26 to regulate beta cell To identify a role for Tcf1 in beta cells, we studied Tcf7−/− mice. function and survival27, Tcf7 expression was not perturbed in islets Body weight and fat mass were similar in WT mice and Tcf7−/− mice from Glp1r−/− mice (Fig. 2d). maintained on a LFD or HFD (Supplementary Fig. 5a,b). Glucose Immunohistochemistry localized Tcf1 (the protein product of Tcf7) tolerance and plasma insulin levels were normal in 8-week-old Tcf7−/− to mouse islets (Fig. 2e and Supplementary Table 2). Tcf7 has been mice on a LFD (Supplementary Fig. 5c). However, intraperitoneal principally studied in immune cells, wherein alternative promoter glucose tolerance was profoundly impaired, and the insulin response utilization and RNA splicing28 gives rise to multiple RNA transcripts, to glucose was markedly attenuated, in older (18-week-old) Tcf7−/− including two major transcripts (~1.3 kb and ~0.9 kb) in the thymus mice on a LFD (Fig. 3a). Both oral and intraperitoneal glucose tol- (Fig. 2f). The dominant Tcf7 mRNA transcript detected in mouse erance were severely impaired after only 4 weeks of HFD feeding islets and beta cell lines was ~0.9 kb (Fig. 2f) and was absent in Tcf7−/− because of the defective upregulation of glucose-stimulated insulin islets (Supplementary Fig. 3d). GIP directly induced Tcf7 expression levels (Fig. 3b and Supplementary Fig. 5d), whereas sensitivity to in the rat insulin-secreting (INS-1) beta cell line, and in human islets exogenous insulin was comparable in WT and Tcf7−/− mice (Fig. 3c). (TCF7) (Supplementary Fig. 3e,f). Although GIP lowered glycemia in Isolated islets from Tcf7−/− mice displayed significantly impaired insu- WT control mice, 12-week-old, obese, diabetic db/db mice displayed lin release in response to glucose and GIP (Fig. 3d). Similarly to islets significantly lower islet levels of Tcf7 and Gipr mRNA transcripts and from Gipr−/−βCell mice, Tcf7−/− beta cells failed to respond to GIP yet did not lower their glucose levels or secrete insulin in response to demonstrated enhanced sensitivity to Ex4 (Supplementary Fig. 5e).

 advance online publication nature medicine l e t t e r s

Control a b c 2.5 Tcf7 912 1110 2.0 Tcf7 1.5 10 * 1.5 4 * * * * 8 * 3 * 1.5 * * 1.0 1.0 * 6 * mRNA mRNA

mRNA 1.0 mRNA 2 mRNA * 4 0.5 * 0.5 * Tcf7 Tcf7 Gipr 1 0.5 2 Glp1r (relative expression) (relative expression) (relative expression) (relative expression) 0 0 (relative expression) 0 0 0 Thapsigargin – + + Thapsigargin – + + Bcl2 Xiap Ins2 Hif1a siRNA Gipr siRNA –– + siRNA Glp1r siRNA –– + Bcl2l1 Birc2 Birc3 ScrambledGipr Scrambled Glp1r d e f g 8 wk 20 wk –/– –/– 1.5 1.5 1.5 10 7 7 A 60 WT 8 wk A * Tcf WT 20 wkTcf A 8 * 1.0 1.0 1.0 Cidea Pro-apoptotic mRN mRN Caspase 3 mRN * 6 40 Casp4 * C-caspase 3 Agr2 0.5 0.5 0.5 Pttg1 Pttg1 4 Chek1 Pttg1 TCF1 Prap1

20 (relative expression) (relative expression) (relative expression) * expression (AU) / # of total cells) Thapsigargin – + + + 2 Apoptosis index Cyr61 0 0 0 (# of apoptotic cells C-caspase 3 protein Msln –/– Cell Tcf7 cDNA ––– + 0 0 Pctp Anti-apoptotic WT 7 β WT Adora2b –/– db/db Thapsigargin – + + Thapsigargin – + + Tcf MIP-Cre Dkk3 Gipr Tcf7 cDNA – – + Tcf7 cDNA –– + Pttg1 Bank1 60 Ms4a1 h i * * Figure 4 TCF1 engages anti-apoptotic pathways in beta cells. Cxcl5 (a) (left) and (right) expression in MIN6 beta cells Tnf Gipr Tcf7 In ammation 1.5 Mst1r 40 (n = 6). (b) Glp1r (left) and Tcf7 (right) expression in MIN6 Tnfrs11b beta cells (n = 6). (c) qPCR analysis of anti-apoptotic genes Cxcl2 1.0 Lbp mRNA * in MIN6 beta cells (n = 4). Bcl, B cell lymphoma 2; Birc, Cd14 20 0.5 / # of total cells) baculoviral IAP repeat containing; Xiap, X-linked inhibitor Unc5cl Apoptosis index Pttg1 Tlr4 (# of apoptotic cells of apoptosis protein; Hif1a, hypoxia-inducible factor 1a. Tnfrs19 (d) Cleaved (C) caspase-3 expression in MIN6 beta cells (n = 6). (relative expression) 0 0 Thapsigargin – ++ + (e) Apoptotic index in MIN6 beta cells (n = 6). (f) Heat map z score siRNA demonstrating the most robust changes in the transcriptome of Tcf7 siRNA – – + + ScrambledTcf7 islets isolated from 8-week-old mice fed a LFD and 20-week-old –2 –1 0 1 2 Pttg1 cDNA –– – + mice fed a HFD (n = 2). (g) Pttg1 expression in RNA isolated from mouse islets from 20-week-old HFD-fed mice (left, n = 4), 18-week-old LFD-fed mice (middle, n = 7) and 12-week-old mice on standard rodent chow (right, n = 5). (h) Pttg1 expression in MIN6 beta cells (n = 6). (i) Apoptotic index in MIN6 beta cells (n = 6). *P < 0.05 vs control, or as indicated. Data are expressed as mean ± s.e.m. Statistical tests used: t-test for a (left), b (left), g,h; one-way analysis of variance (ANOVA) for a (right), b (right), c–e,i.

Furthermore, knockdown of TCF7 in human beta cells produced Knockdown of TCF7 in human islet cells increased the number of a defective exocytosis response to GIP but not to Ex4 (Fig. 3e). apoptotic cells in islets cultured either in low glucose alone or in 25 mM Consistent with findings in Gipr−/−βCell islets, Tcf7−/− islets showed of glucose with interleukin 1–β (Fig. 3h). Rescue of Tcf1 levels in no significant differences in their expression of genes encoding pro- dispersed islets from Gipr−/−βCell mice lowered apoptotic rates (after survival GPCRs, components of the glucose-sensing machinery, thapsigargin administration) to levels comparable to those detected mediators of proliferation or survival, or members of the TCF family in WT controls (Fig. 3i and Supplementary Fig. 7). (Supplementary Fig. 5f). Unlike findings for Gipr−/−βCell mice, both The induction of apoptosis by administration of thapsigargin LFD- and HFD-fed Tcf7−/− mice demonstrated normal glycemia enhanced Tcf7 expression in mouse insulinoma 6 (MIN6) beta cells, and insulin release in response to refeeding after an overnight fast which was significantly attenuated by knockdown of the Gipr, but (Supplementary Fig. 5g). However, histological analysis of Tcf7−/− not of the Glp1r (Fig. 4a,b). The transfection of MIN6 beta cells pancreata revealed an increase in beta cell mass after HFD feeding with cDNAs encoding the single TCF1 protein, Tcf7912 (the isoform compared to that in WT controls (Supplementary Fig. 5h), com- expressed in mouse islets), increased the expression of anti-apoptotic parable to findings in Gipr−/−βCell mice (Supplementary Fig. 3a). genes (Fig. 4c) and reduced thapsigargin-stimulated caspase-3 Furthermore, Tcf7−/− mice on either a LFD or a HFD demonstrated cleavage and apoptosis (Fig. 4d,e and Supplementary Fig. 8a). reduced glycemia in response to the administration of DPP4 inhibi- To further elucidate the consequences of TCF1-deficiency in beta tors or incretin-receptor agonists (Supplementary Fig. 6a–d). cells, we performed RNA-seq analysis on islets from young, meta- Thus, Tcf7−/− mice recapitulate many, but not all, of the metabolic bolically healthy (8-week-old, LFD) mice, older, obese (20-week-old, phenotypes that are evident in Gipr−/−βCell mice. HFD) Tcf7−/− mice and littermate control mice. Robust changes in Given that TCF1 controls thymocyte survival29, and that Gipr−/−βCell genes important for apoptosis and inflammation are detected in RNA mice demonstrate enhanced sensitivity to beta cell apoptosis, we from the islets of older Tcf7−/− mice (Fig. 4f). Notably, expression of assessed apoptotic injury in Tcf7−/− beta cells. Notably, Tcf7−/− islets pituitary tumor–transforming gene 1 (Pttg1), which encodes securin, exhibited a greater number of apoptotic beta cells after STZ adminis- a protein that regulates stability and DNA repair30, was tration than did WT islets (Fig. 3f). Furthermore, GIP reduced apop- virtually absent in islet RNA from older Tcf7−/− mice (Fig. 4g). Pttg1 tosis in the islets of STZ-treated WT mice but not in islets from Tcf7−/− expression was also significantly reduced both in RNA from Gipr−/−βCell mice (Fig. 3g). By contrast, Ex4 significantly reduced islet apoptosis islets (Fig. 4g) and in islet RNA (Fig. 4g) from 12-week-old, diabetic in both STZ-treated WT and STZ-treated Tcf7−/− mice (Fig. 3g). db/db mice that exhibit reduced expression of Gipr and Tcf7 (Fig. 2g).

nature medicine advance online publication  l e t t e r s

These results are consistent with a GIPR-TCF1-PTTG1 axis. Levels of Tcf7 mRNA transcripts are reduced in the pancreas from Finally, knockdown of Tcf7 mRNA transcripts in MIN6 beta cells HFD-fed rats36, and variation in the human TCF7 gene has been (Supplementary Fig. 8b) led to a significant reduction in Pttg1 linked to an increased risk for development of type 1 diabetes37,38; expression (Fig. 4h), whereas the restoration of Pttg1 expression whether these latter findings reflect TCF1-dependent disturbances attenuated the increased susceptibility to apoptosis in MIN6 beta of immune regulation, beta cell function or beta cell survival in sus- cells seen after Tcf7 knockdown (Fig. 4i). ceptible individuals requires further investigation. Moreover, the Taken together, these results greatly extend our understanding of importance of the TCF1 target, PTTG1, for beta cell function and incretin action and beta cell function and have direct translational survival has been independently shown in studies of older Pttg1−/− implications. Previous findings that Gipr−/− mice are protected against mice, which exhibited reduced beta cell mass, impaired beta cell diet-induced obesity and insulin resistance, coupled with genetic linkage function and increased beta cell apoptosis39,40. of variants in the human GIPR to body weight3, were attributed to Our current data establish that TCF1, acting through PTTG1, links GIP action in adipose tissue. The relatively modest differences in GIPR signaling to the control of insulin secretion, the survival of beta glucose tolerance arising from the selective loss of GIPR signaling cells and the adaptation of these cells to metabolic stress. Given that in beta cells is ameliorated in part by compensatory increases in the GIPR, but not GLP-1R, signaling controlled the TCF1-PTTG1 axis activity of functionally related beta cell GPCRs. Nevertheless, our data in beta cells, and that GIP responsiveness was rapidly restored after demonstrate that the selective attenuation of GIP action in beta cells a brief period of improved glucose control in human subjects with limits meal-related insulin release, indirectly reducing the expansion T2D13, our findings highlight the potential of targeting GIPR-TCF1- of adipose tissue mass and leading to improvements in insulin sen- PTTG1 signaling for the preservation of beta cell mass and the treat- sitivity and glucose tolerance. These findings support observations ment of diabetes. Collectively, our data imply that the development linking variation in the human GIPR gene with reduced insulin of GIP-based therapies may target novel pathways, independently secretion and decreased body mass index31, necessitating reconsidera- of GLP-1R signaling, thus linking nutrient-activated signals to the tion of the importance of direct versus indirect GIP action in beta cells control of beta cell function and survival. compared to that in adipocytes for the control of insulin sensitivity and adipose tissue mass. Methods Although GLP-1 and GIP exert their actions through structurally Methods and any associated references are available in the online and functionally related incretin receptors1,22, our findings that GIPR, version of the paper. but not GLP-1R, signaling controls Tcf7 expression independently of cAMP signaling identify divergent downstream signaling pathways Accession codes. RNA-seq data has been deposited with accession that link incretin-receptor signaling to beta cell survival. Indeed, the code GSE65361. loss of GLP-1R, but not of GIPR, signaling, impairs expression of 18 Note: Any Supplementary Information and Source Data files are available in the insulin receptor substrate 2 (Irs2) in mouse islets , and beta cells from online version of the paper. Glp1r−/− mice exhibit greater sensitivity to STZ-induced apoptotic injury than Gipr−/− beta cells24. Furthermore, GIP, but not GLP-1, con- Acknowledgments trols osteopontin expression in mouse islets31, whereas GLP-1R, but not The authors thank H. Bates for excellent technical assistance and B. Yusta for GIPR, signaling is essential for the adaptive islet response to pregnancy. thoughtful discussions and critical reading of the manuscript; H. Clevers for Tcf7−/− mice; and R. DiMarchi for critical reagents. This work in Toronto was Our identification of a GIPR-TCF1 axis uncovers a novel mechanistic supported by the Canada Research Chairs Program, the Banting and Best pathway linking differential incretin-receptor signaling to the control Diabetes Centre Novo Nordisk Chair in Incretin Biology, and Canadian Institute of beta cell mass and function, which has encouraging implications for Health Research (CIHR) grants 82700 and 123391 (DJD). We thank the for therapeutic strategies based on GIPR agonism. Human Organ Procurement and Exchange (HOPE) program and the Trillium Gift of Life Network (TGLN) for their assistance in obtaining pancreases and Considerable evidence from human genetic and physiological islets from human organ donors for research, and we thank J. Lyon (Alberta studies has linked variation within the human TCF7L2 gene to Diabetes Institute; http://www.bcell.org/isletcore.html) and T. Kin and impairment of beta cell function and an increased risk of developing A.M.J. Shapiro (University of Alberta, Clinical Islet Isolation Facility) for T2D32,33. Nevertheless, our understanding of how TCF7L2 controls human islet isolation. The Alberta Diabetes Foundation funded the human islet isolations. Work at the University of Alberta was funded by a grant from the beta cell function has been challenged by studies demonstrating that Canadian Diabetes Association (to P.E.M.). P.E.M. holds a Canada Research genetic inactivation of the Tcf7l2 gene in beta cells does not impair Chair in Islet Biology. Postdoctoral fellowship funding was provided by the beta cell function34, whereas enhanced Tcf7l2 expression (TCF4) Canadian Institute for Health Research (J.E.C., J.R.U., and E.E.M.), Banting and in liver activated a metabolic gene-expression program linked to Best Diabetes Centre (J.E.C.), the Canadian Diabetes Association (E.E.M.) and 34 Alberta Innovates Health Solutions (J.R.U.). D.J.D. is the main guarantor of this increased hepatic glucose output . This suggests that TCF4 may work and takes responsibility for all content. indirectly affect beta cell function through hepatic Wnt signaling. Tcf1 is deficient in the hepatocytes of diabetic mice, and restoring AUTHOR CONTRIBUTIONS Tcf7 expression reduces gluconeogenesis35, further positioning J.E.C. and D.J.D. designed and directed the study, analyzed data and wrote Tcf7 in the pathophysiology of diabetes. Collectively, our find- the manuscript. J.R.U., E.E.M., L.L.B. and P.E.M. contributed to the study ings demonstrated that Tcf7/Tcf1 is required for maintaining beta design and the preparation of the manuscript. J.E.C., J.R.U., E.E.M., J.K., L.L.B., X.C., B.J.L. and T.M. performed experiments. J.K. and P.E.M. carried out cell survival and that the disruption of Tcf7 (loss of Tcf1) unmasks experiments on human islets. Y.L. and J.L.W. provided assistance with the divergent mechanisms regulating the anti-apoptotic actions of GIPR RNA-seq data. N.T. and L.H.P. provided MIP-Cre mice. C.J.S. performed the as compared to GLP-1R signaling in mouse islets. In contrast to electron microscopy. All authors reviewed the manuscript and provided final conflicting data surrounding the role of TCF7L2 in beta cells, our approval for submission. latest data unequivocally reveal the importance of Tcf7/Tcf1, regu- COMPETING FINANCIAL INTERESTS lated by GIPR signaling, for the direct control of beta cell survival in The authors declare competing financial interests: details are available in the online murine and human islets. version of the paper.

 advance online publication nature medicine l e t t e r s

Reprints and permissions information is available online at http://www.nature.com/ 21. Pederson, R.A. et al. Enhanced glucose-dependent insulinotropic polypeptide reprints/index.html. secretion and insulinotropic action in glucagon-like peptide 1 receptor −/− mice. Diabetes 47, 1046–1052 (1998). 1. Campbell, J.E. & Drucker, D.J. Pharmacology physiology and mechanisms of incretin 22. Mayo, K.E. et al. International Union of Pharmacology. XXXV. The glucagon receptor hormone action. Cell Metab. 17, 819–837 (2013). family. Pharmacol. Rev. 55, 167–194 (2003). 2. Saxena, R. et al. Genetic variation in GIPR influences the glucose and insulin 23. Kim, S.J. et al. GIP stimulation of pancreatic beta-cell survival is dependent upon responses to an oral glucose challenge. Nat. Genet. 42, 142–148 (2010). phosphatidylinositol 3-kinase (PI3-K)/ protein kinase B (PKB) signaling, inactivation 3. Speliotes, E.K. et al. Association analyses of 249,796 individuals reveal 18 new of the Forkhead transcription factor Foxo1 and downregulation of bax expression. loci associated with body mass index. Nat. Genet. 42, 937–948 (2010). J. Biol. Chem. 280, 22297–22307 (2005). 4. Miyawaki, K. et al. Inhibition of gastric inhibitory polypeptide signaling prevents 24. Maida, A., Hansotia, T., Longuet, C., Seino, Y. & Drucker, D.J. Differential importance obesity. Nat. Med. 8, 738–742 (2002). of GIP versus GLP-1 receptor signaling for beta cell survival in mice. Gastroenterology 5. Mulvihill, E.E. & Drucker, D.J. Pharmacology, physiology and mechanisms of action 137, 2146–2157 (2009). of dipeptidyl peptidase-4 inhibitors. Endocr. Rev. 35, 992–1019 (2014). 25. Shu, L. et al. Transcription factor 7-like 2 regulates beta-cell survival and function 6. Finan, B. et al. Unimolecular dual incretins maximize metabolic benefits in rodents, in human pancreatic islets. Diabetes 57, 645–653 (2008). monkeys and humans. Sci. Transl. Med. 5, 209ra151 (2013). 26. Yusta, B. et al. GLP-1 receptor activation improves β-cell function and survival 7. Finan, B. et al. A rationally designed monomeric peptide triagonist corrects obesity following induction of endoplasmic reticulum stress. Cell Metab. 4, 391–406 and diabetes in rodents. Nat. Med. 21, 27–36 (2015). (2006). 8. Shu, L. et al. Decreased TCF7L2 protein levels in type 2 diabetes mellitus correlate 27. Li, Y. et al. Glucagon-like peptide-1 receptor signaling modulates beta cell apoptosis. with downregulation of GIP- and GLP-1 receptors and impaired beta-cell function. J. Biol. Chem. 278, 471–478 (2003). Hum. Mol. Genet. 18, 2388–2399 (2009). 28. Van de Wetering, M., Castrop, J., Korinek, V. & Clevers, H. Extensive 9. Wessel, J. et al. Low-frequency and rare exome chip variants associate with fasting alternative splicing and dual promoter usage generate Tcf-1 protein isoforms with glucose and type 2 diabetes susceptibility. Nat. Commun. 6, 5897 (2015). differential transcription control properties. Mol. Cell. Biol. 16, 745–752 10. Miyawaki, K. et al. Glucose intolerance caused by a defect in the entero-insular (1996). axis: a study in gastric inhibitory polypeptide receptor knockout mice. Proc. Natl. 29. Ioannidis, V., Beermann, F., Clevers, H. & Held, W. The beta-catenin–TCF-1 Acad. Sci. USA 96, 14843–14847 (1999). pathway ensures CD4+CD8+ thymocyte survival. Nat. Immunol. 2, 691–697 11. Meier, J.J. GLP-1 receptor agonists for individualized treatment of type 2 diabetes (2001). mellitus. Nat. Rev. Endocrinol. 8, 728–742 (2012). 30. Vlotides, G., Eigler, T. & Melmed, S. Pituitary tumor-transforming gene: physiology 12. Nauck, M.A. et al. Preserved incretin activity of glucagon-like peptide 1 [7–36 and implications for tumorigenesis. Endocr. Rev. 28, 165–186 (2007). amide] but not of synthetic human gastric inhibitory polypeptide in patients with 31. Lyssenko, V. et al. Pleiotropic effects of GIP on islet function involves osteopontin. type-2 diabetes mellitus. J. Clin. Invest. 91, 301–307 (1993). Diabetes 60, 2424–2433 (2011). 13. Højberg, P.V. et al. Four weeks of near-normalisation of blood glucose improves the 32. Grant, S.F. et al. Variant of transcription factor 7-like 2 (TCF7L2) gene confers risk insulin response to glucagon-like peptide-1 and glucose-dependent insulinotropic of type 2 diabetes. Nat. Genet. 38, 320–323 (2006). polypeptide in patients with type 2 diabetes. Diabetologia 52, 199–207 (2009). 33. Florez, J.C. et al. TCF7L2 polymorphisms and progression to diabetes in the 14. Brouwers, B. et al. Impaired islet function in commonly used transgenic mouse Diabetes Prevention Program. N. Engl. J. Med. 355, 241–250 (2006). lines due to human growth hormone minigene expression. Cell Metab. 20, 979–990 34. Boj, S.F. et al. Diabetes risk gene and Wnt effector Tcf7l2/TCF4 controls hepatic (2014). response to perinatal and adult metabolic demand. Cell 151, 1595–1607 15. Oropeza, D. et al. Phenotypic characterization of MIP-CreERT1Lphi mice with (2012). transgene-driven islet expression of human growth hormone. Diabetes 64, 35. Kaur, K. et al. Elevated hepatic miR-22–3p expression impairs gluconeogenesis by 3798–3807 (2015). silencing the Wnt-responsive transcription factor, Tcf7. Diabetes 64, 3659–3669 16. Mehran, A.E. et al. Hyperinsulinemia drives diet-induced obesity independently of (2015). brain insulin production. Cell Metab. 16, 723–737 (2012). 36. Columbus, J. et al. Insulin treatment and high-fat diet feeding reduces the expression 17. Chen, S., Okahara, F., Osaki, N. & Shimotoyodome, A. Increased GIP signaling of three Tcf genes in rodent pancreas. J. Endocrinol. 207, 77–86 (2010). induces adipose inflammation via a HIF-1alpha-dependent pathway and impairs 37. Noble, J.A. et al. A polymorphism in the TCF7 gene, C883A, is associated with insulin sensitivity in mice. Am. J. Physiol. Endocrinol. Metab. 308, E414–E425 type 1 diabetes. Diabetes 52, 1579–1582 (2003). (2015). 38. Erlich, H.A., Valdes, A.M., Julier, C., Mirel, D. & Noble, J.A. Evidence for association 18. Lamont, B.J. & Drucker, D.J. Differential anti-diabetic efficacy of incretin agonists of the TCF7 with type I diabetes. Genes Immun. 10 (suppl. 1), S54–S59 vs. DPP-4 inhibition in high fat fed mice. Diabetes 57, 190–198 (2008). (2009). 19. Ali, S., Lamont, B.J., Charron, M.J. & Drucker, D.J. Dual elimination of the glucagon 39. Wang, Z., Moro, E., Kovacs, K., Yu, R. & Melmed, S. Pituitary tumor transforming and GLP-1 receptors in mice reveals plasticity in the incretin axis. J. Clin. Invest. gene-null male mice exhibit impaired pancreatic beta cell proliferation and diabetes. 121, 1917–1929 (2011). Proc. Natl. Acad. Sci. USA 100, 3428–3432 (2003). 20. Pamir, N. et al. Glucose-dependent insulinotropic polypeptide receptor null mice 40. Chesnokova, V. et al. Diminished pancreatic beta-cell mass in securin-null mice is exhibit compensatory changes in the enteroinsular axis. Am. J. Physiol. Endocrinol. caused by beta-cell apoptosis and senescence. Endocrinology 150, 10 (suppl. 1) Metab. 284, E931–E939 (2003). 2603–2610 (2009).

Competing financial interests D.J.D. has been a consultant to Novo Nordisk Inc. and other companies that develop and/or sell incretin-based therapies, including Arisaph Pharmaceuticals Inc., Intarcia Therapeutics, Merck Research Laboratories, MedImmune, Receptos, Sanofi, Takeda and Transition Pharmaceuticals Inc. Neither D.J.D. nor his family members hold stock directly or indirectly in any of these companies.

EDITORIAL SUMMARY AOP: The details of the GIP signaling pathway are murky, but new data identify a downstream pathway involving Tcf7 that regulates beta cell survival and activity.

nature medicine advance online publication  ONLINE METHODS Insulin supplementation. 8-week-old mice had a single insulin pellet (7 or 14 mg, Animals. Male mice were used for all mouse studies and were maintained LinBit) inserted into the intrascapular region under isoflurane anesthesia by under a 12 h light/12 h dark cycle at constant temperature (23 °C) with free following the manufacturer’s protocol. access to food and water. All animal studies were approved by Mt. Sinai Hospital (Toronto) and the Toronto Centre for Phenogenomics animal-care committee. Apoptosis in vivo. Mice were treated with streptozotocin (Sigma, 50 mg per Animals were fed either a low-fat diet (10% kcal from fat; Research Diets, kg) for 5 consecutive days at 0800. Twenty-four hours after the final treat- D12450B) or high-fat diet (45% kcal from fat; Research Diets, D12451). To gen- ment, mice were euthanized and the pancreas was excised and immediately erate Gipr−/−βCell mice, MIPcreER transgenic mice (on a C57BL/6J background) immersed in 10% formalin. All histological analysis was performed in a expressing tamoxifen-inducible Cre driven by the mouse insulin promoter were blinded fashion. bred with floxed Gipr mice (GiprFlox/Flox), backcrossed 8 times to C57BL/6J background)41. Cre-induced inactivation of the Gipr gene was carried out via 5 Real time quantitative PCR. First-strand cDNA was synthesized from total consecutive daily intraperitoneal (i.p.) injections of tamoxifen (40 mg per kg) in RNA using the SuperScript III synthesis system (Invitrogen). Real-time PCR 6-week-old mice. Glp1r−/− and Tcf7−/− mice (both on C57BL/6J backgrounds) was carried out with the ABI Prism 7900 Sequence Detection System using have been previously described42,43. Tcf7−/− mice were generously provided by TaqMan Assays (Applied Biosystems). Relative mRNA tran- H. Clevers, db/db mice were purchased from Jackson laboratories (#000697). script levels were quantified with the 2−∆Ct method. PCR primers are shown For all animal experiments, the sample size required to achieve adequate power in Supplementary Table 1. was estimated on the basis of pilot work or previous experiments. When appro- priate, animals were randomly allocated to individual experimental groups. Qualitative PCR. Amplification of mouse Gipr cDNA was performed using the primer pairs 5′–CTG CTT CTG CTG CTG TGG T–3′ (forward primer) Peptides and reagents. Peptides were reconstituted in phosphate-buffered and 5′–CAC ATG CAG CAT CCC AGA–3′ (reverse primer). PCR was car- saline (PBS), aliquoted and stored at –80 °C. [D-Ala2]GIP (GIP) was from ried out using 35 cycles at an annealing temperature of 50 °C to generate Chi Scientific, Ex4 (exendin-4) was from California Peptide Research Inc. a 1.5 kb product. Amplification of the mouse Tcf7 isoforms was performed Plasmid constructs (Tcf7 and Pttg1) and siRNA (Gipr, Glp1r, Tcf7, TCF7) were using the common reverse primer 5′–CTA GAG CAC TGT CAT CGG–3′and from Origene. The GLP-1R antagonist (JANT-4)44 was a generous gift from two different 5′ (forward) primers targeting alternative start codons (1.3 kb R. DiMarchi, University of Indiana. product – ATG CCG CAG CTG GAC TCG; 0.9 kb product – ATG TAC AAA GAG ACT GTC TAC T). PCR was carried out using 35 cycles at an annealing Mouse islet isolation. Primary mouse islets were isolated as previously temperature of 56 °C. described45. Briefly, the pancreas was inflated via the pancreatic duct with collagenase type V (0.8 mg per ml), excised and digested for 10–15 min. The Western blot analysis. Thirty µg of total protein was separated by SDS-PAGE, digest was washed with cold RPMI (2 mM L-glutamine, 10 mM glucose, 0.25% transferred to nitrocellulose membranes and blocked in 5% milk in PBS-T for 1 h BSA, 100 U/ml penicillin, and 100 µg/ml streptomycin), and the islets were before incubation in primary antibody overnight at 4 °C. Immunoblots were separated using a Histopaque gradient. Individual islets were handpicked and visualized with the enhanced chemiluminescence Western blot detection kit either immersed in TRI Reagent for subsequent mRNA isolation or allowed to (PerkinElmer) and quantified with Carestream Molecular Imaging Software recover overnight in RPMI with 10% FBS for experiments ex vivo. (Kodak). Primary antibodies are shown in Supplementary Table 2.

Primary mouse islet insulin secretion. After overnight incubation, 75–80 Human islet isolation. Primary human islets were isolated as previously medium-sized islets were handpicked into 0.275 ml chambers containing KRB described6 at the Alberta Diabetes Institute Islet Core (http://www.bcell.org/ buffer (135 mM NaCl, 3.6 mM KCl, 1.5 mM CaCl2, 0.5 mM NaH2PO4, 0.5 mM isletcore.html) and the Clinical Islet Isolation Facility at the University of Alberta MgSO4, 5 mM Hepes, 5 mM NaCO3, 0.1% BSA pH 7.5). Islets were perifused and cultured in low glucose (5.5 mM) DMEM with L-glutamine, 110 mg per for 1 h in KRB with 2.7 mM glucose at a flow rate of 200 µl per min using the L sodium pyruvate, 10% FBS and 100 U/ml penicillin/streptomycin. We studied Biorep Perifusion system. After this equilibration period, islets were peri- islets from 15 non-diabetic, donors (age: 57 ± 11 years; HbA1c: 5.7 ± 1.4 fused at 8 min intervals in experimental media (KRB plus various conditions), BMI range 19–39) and 4 T2D donors (age: 63 ± 5 years; HbA1c: 6.6 ± 0.9; then collected and lysed in acid ethanol for total insulin measurements. Insulin BMI range- 29–37). concentrations were determined by radioimmunoassay (Millipore), and insulin secretion was expressed as a percent of total insulin. Capacitance measurement. Islets were dispersed in calcium-free dissociation buffer in 35 mm dishes and incubated overnight in RPMI containing 11mM Islet hGH release. Isolated islets were incubated in batches of 100 at 37 °C (mouse) or DMEM containing 5.5 mM (human) glucose. GIP (10 nM), Ex4 (1 nM) 7 in HEPES Krebs buffer containing 20 mM glucose, as previously described . or vehicle (H2O) was added to each dish 1 h before patch clamping, using After 1 h, the buffer was removed and assayed for hGH release using a hGH the standard whole-cell technique with the sine + DC lock-in function of an ELISA (Invitrogen). EPC10 amplifier and Patchmaster software (HEKA Electronics). Experiments were performed as described previously7 at 32–35 °C using an extracellular Glucose- and insulin-tolerance tests. Oral and intraperitoneal glucose-tolerance bath solution (118 mM NaCl, 20 mM TEA, 5.6 mM KCl, 1.2 mM MgCl2, tests (GTTs) were performed in mice fasted for 5 h (0700–1200) using a glu- 2.6 mM CaCl2, 5 mM glucose, 5 mM Hepes, pH – 7.4) and pipette solution cose dose of 1.5 g per kg. During IPGTT, mice were i.p. injected with either (125 mM CsGlutamate, 10 mM CsCl, 10 mM NaCl, 1 mM MgCl2, 0.05 mM GIP (4 nmol per kg), Ex4 (0.3 nmol per kg), or saline (veh), 10 min before EGTA, 5 mM Hepes, 3 mM MgATP, 0.1 mM cAMP, pH – 7.15). Capacitance glucose administration. For tests using a DPP4 inhibitor, sitagliptin (Merck, responses to a train of 10 depolarizations from −70 to 0 mV at 1 Hz were 40 µg per mouse) was given orally 30 min before glucose. For both OGTT normalized to initial cell size and expressed as femtofarad per picofarad and IPGTT, blood was collected at 0, 10 and 30 min in capillary tubes coated (fF/pF). β cells were identified using positive insulin immunostaining. with 10% (vol/vol) TED (500,000 IU/ml Trasylol; 1.2 mg/ml EDTA; and 0.1 mM diprotin A) and plasma separated by centrifugation at 4 °C and stored RNA-seq. Total RNA from isolated islets was extracted using TRI Reagent. The at −80 °C. Insulin-tolerance tests (ITTs) were performed in mice fasted for 5 h yield and quality of total RNA was assessed using Nanodrop (Thermo Fisher) (0700–1200) using an insulin dose of 0.7 U/kg (Humalog, Lilly). and BioAnalyzer (Agilent), respectively. Ribosomal RNA was removed using a bead-based hybridization kit (RiboZero, Epicentre) and cDNA libraries were Plasma hormone analysis. Insulin (Alpco Diagnostics) and total GIP prepared using the Illumina TruSeq RNA sample preparation kit. The quality (Linco) levels were analyzed by ELISA. Total GLP-1 levels were measured by and concentration of libraries were assessed using BioAnalyzer and qPCR, immunoassay (Mesoscale). respectively. The libraries were loaded as two indexed samples per lane on an

nature medicine doi:10.1038/nm.3997 Illumina HiSeq 2000. Raw sequenced reads were obtained in fastq format, and (Sigma, Oakville, ON, Canada), as indicated. Cell-death assays were performed mapped onto the mouse genome (mm9) using Tophate1.4.1, and then analyzed on dispersed human islet cells by use of the In situ Cell Death Detection Kit TMR using a custom R-based pipeline to calculate gene-expression profiles using Red (Roche, Mannheim, Germany), using TUNEL technology, according to the ENSEMBL annotation for coding genes. The number of reads mapped onto manufacturer’s directions. Images were obtained using a Zeiss AxioObserver the gene was counted regardless of transcription isoform and normalized to Z1 with a Zeiss-Colibri light source at 488 nm and 594 nm, a × 40/1.3 NA total mapped reads to obtain transcript union Read Per Million total reads lens, and an AxioCam HRm camera. Images were acquired in Axiovision 4.8 (truRPMs). The data have been deposited in NCBI’s Gene Expression Omnibus software (Carl Zeiss MicroImaging, Göttingen, Germany) and analyzed using and are accessible through GEO Series accession number GSE65361. ImageJ software (National Institutes of Health). Cell death was determined as ((#TUNEL+/Alexa488+) / (#Alexa488+)) and expressed as a fold increase over Cell-line culture. INS1823/32 cells were a generous gift from C. Newgard, control, unstimulated conditions (5.5mM glucose, scrambled siRNA). Duke University. Cells were grown in RPMI (10% FBS, 1% P/S). For GIP exper- iments, cells were starved in RPMI (1% FBS, 1% P/S) for 3 h. MIN6 cells (from Statistical analysis. All values are presented as mean ± s.e.m. Statistical analy- ATCC) were grown in DMEM (20% FBS, 1% P/S). siRNA knockdown experi- sis was performed using GraphPad Prism 5.0. The appropriate t-test, one- ments were performed by following the manufacturer’s instructions (Origene). way analysis of variance (ANOVA), or two-way ANOVA was completed using All cell lines were previously tested for mycoplasma contamination. P < 0.05 to signify significant differences. Bonferroni post-hoc analysis was performed where appropriate. All data was assessed to ensure normal dis- Apoptosis assays. MIN6 beta cells. After 24 h exposure to thapsigargin (5 uM, tribution and equal variance between groups, using GraphPad Prism 5.0. Sigma) in culture media (DMEM, 20% FBS, 1% P/S), apoptosis was analyzed Prior to the experiment, it was determined that individual data points would using a MitoPT assay (ImmunoChemistry Technologies) by following the be excluded if their value was greater than 2 × SD from the mean, in an kit’s instructions. Briefly, cells were exposed to 100 nM MitoPT for 10 min, experiment with a sample size greater than seven. washed 1 × with PBS and then detached from the plate with trypsin. An aliquot of cells was visualized on a microscope slide. Total cell number was counted using bright field and tetramethylrhodamine methyl ester (TMRM) uptake was visualized with a 545 nM filter. Apoptotic cells were calculated as 41. Wicksteed, B. et al. Conditional gene targeting in mouse pancreatic ss-cells: analysis (total cells – TMRM positive cells)/Total cells × 100%. of ectopic Cre transgene expression in the brain. Diabetes 59, 3090–3098 Mouse islets. Dispersed islets were transduced with Ad-GFP or Ad-Tcf7-GFP (2010). at an MOI of 10 to achieve a >90% induction rate. After transduction, cells were 42. Scrocchi, L.A. et al. Glucose intolerance but normal satiety in mice with a null mutation in the glucagon-like peptide receptor gene. Nat. Med. 2, 1254–1258 exposed to control or thapsigargin (100 uM) for 72 h. Apoptosis was assessed (1996). using a MitoPT assay. 43. Verbeek, S. et al. An HMG-box-containing T-cell factor required for thymocyte Human islets. Dispersed human cells were transfected with human siTCF7 differentiation. Nature 374, 70–74 (1995). or siScram control duplexes (OriGene, Rockville, MD) and an Alexa488 labeled 44. Patterson, J.T. et al. A novel human-based receptor antagonist of sustained action reveals body weight control by endogenous GLP-1. ACS Chem. Biol. 6, 135–145 negative control siRNA (Qiagen, Toronto, ON), using Dharmafect (Thermo (2011). Scientific, Ottawa, ON, Canada). 24 h post-transfection culture medium was 45. Lamont, B.J. et al. Pancreatic GLP-1 receptor activation is sufficient for incretin changed to fresh medium containing glucose and/or human recombinant IL1-β control of glucose metabolism in mice. J. Clin. Invest. 122, 388–402 (2012).

doi:10.1038/nm.3997 nature medicine Journal: Nature Medicine

Article Title: TCF1 links GIPR signaling to control of  cell function and survival Corresponding Daniel J. Drucker, [email protected] Author:

Supplementary Item Title or Caption & Number Supplementary Figure Generation of Gipr–/–βCell mice. 1 Supplementary Figure Phenotype of Gipr–/–βCell mice. 2 Supplementary Figure Characterization of Gipr–/–βCell islets. 3 Supplementary Figure Diabetic db/db mice fail to lower glucose in response to 4 exogenous GIP Supplementary Figure Phenotype of Tcf7–/– mice 5 Supplementary Figure Glycemic and insulin response to incretin receptor agonists 6 and DPP4 inhibition in Tcf7–/– mice. Supplementary Figure Representative images of cells following adenoviral 7 transduction and apoptotic index following treatment with or without thapsigargin. Supplementary Figure Representative image of apoptotic index and Tcf7 knockdown 8 in MIN6  cells Supplementary Table 1 List of real time primers Supplementary Table 2 List of antibodies Supplementary Methods

Nature Medicine: doi:10.1038/nm.3997 Nature Medicine: doi:10.1038/nm.3997 Supplementary Figure 1 Figure Supplementary a c h e g d F A Glucose (mM) C Total GIP (pg/ml)1200 Mass (g) Glucose (mM) hGH release (pg/islet) 10 20 30 20 10 25 30 15 20 5 10 M 1.0 2.0 3.0 300 600 900 1.65 kb ’ 0 0 5 0 I P 0 C 1 8 0

0 0 P R 1 kb

Blood Glucose (mM) Blood Glucose (mM) r * E o 1 1 2 2 3 3 H 1 1 2 2 3 3 H m 10 0 5 0 5 0 5 0 5 S Floxed S 0 5 - 0 5 0 5 0 5 * P - 2 o P 2 Gipr 9 Gipr MIP- MIP- 9 Age ( Age t 30 0 30 10 min min 10 0 W 0 min 0 min * 0 e * 0 C E Time (min) Time (min) r 12 R R 7 T 0 0 –/– –/– * E Cre Cre Cre Cre WT wk 8 14 60 60 Gipr

*

2 + cell cell β cell cell β 2 C

0 4 0 R ) 12 - 16 O * T T E * H

i 4 i - 4 m m T 90 90 15 0 a 0 1,000 1,500 2,000 * 18 m e 1,000 1,500 e 500 3 o

500

x ( ’ 6 ( 0 6 i m m f 0 0 e 0 E 120 120 n Veh i R i n n 8 )

8 Tph1 mRNA ) Glucose (mM) 0 5 0

Total GLP-1 (pg/ml) GIP C ’ *

10 20 30 (relative expression) 9 6 7 5 8 4 R 1 1 D P 15 10 1 D P 0 8 E 0 0 5 B A 0 B A KO 0 0 S - S - G G 10 1 1 I I 2 2 E P P Gipr 10 min min 10 Gipr 0 min 0 min MIP- Age ( Age 0 0 R

* Glucose (mM)

12 Insulin (ng/ml) 30 20 10 7 0.5 1.0 1.5 2.0 2.5 0

–/– AUC (Glucose x Time) AUC (Glucose x Time) 0 1 1 2 2 1 1 2 2 0 Cre Cre 15 5 5 0 5 0 5 0 5 0 5 * wk 14 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 cell cell β 3 * ’ ) * 16 P P * 30 B B * * b * S S Time (min)

Gipr Gipr mRNA 18 D D P A A (relative expression) = * 60 0 0.0 - - 1.5 0.5 1.0 2.0 * G G . 30 min min 30 min 10 0 min 1 Flox I I * 0 P P i B Mass (g/day) D

Glucose (mM) Tph2 mRNA / Flox 90 4 2 3 1 0 20 10 30 (relative expression) 1,000 1,500 2,000 0 500 0.5 1.5 2.5 1.0 2.0 Relative mRNA levels

0 0 1 1 2 0 0 0 . . . . . Blood Glucose (mf M) Blood Glucose (mM) 0 5 0 5 0 120 1 1 2 2 3 3 1 1 2 2 3 3 Glucose Veh 0 5 0 5 0 5 0 5 0 5 0 5 0 5 0 5 - - 2 2 Gipr

MIP- 30 0 0 F * (%Baseline) B GIP 100 Time (min) W * Islets Western Blot L 25 75 50 C 0 T 0 0 X –/– K Cre Cre 0 * 60 * O

2 2 cell cell β 0 0 T Group – p<0.05 – p<0.05 Group T I Gipr C MIP- s

i Glucose (mM) 4 i * 4 m * m 20 10 30 90 R 0 0 l Time (min) 30 0 e e 2,000 1,000 1,500 e E

500

0 –/– ( 6 t ( 6 Cre Cre m m 0 0 0 *

120 i i cell cell β n G n * 8 ) 8

) Gipr mRNA 30 F 0 60 0 * i Time (min) L p

MIP- (relative expression) 1 0.5 1.0 1.5 X * 1 D P D P 0 r 0 0 B A 0 B A 0 60 S - S - E * G 1 G 1 2 Cre I 2 I 90 P B P 0 0 C K 90 *

AUC (Glucose x Time) AUC (Glucose x Time) 1,000 1,500 2,000 1 1 2 2 1 1 2 2 O 5 5 0 5 0 5 0 5 0 5 500 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 E E 120 Veh Islets Confocal IHC P P x x B B o o S S

n n

GIP 1 4 * D D 2 - 5 A A * - - - 1 G G I I 3 P P Glucose (mM) 20 10 30 0 0 Veh GIP 30 Gipr Time (min)

60 –/–

cell cell β 90 1,000 1,500 2,000 500 0 120 Veh

GIP Supplementary Figure 2

b a ) c 30 2 5,000 2.0 MIP-Cre 2.0 MIP-Cre 1.5 MIP-Cre Gipr–/– βcell Gipr–/– βcell Gipr–/– βcell 4,000 20 1.5 1.5 * * 3,000 1.0 1.0

1.0 /ml/min) 2,000 * mRNA mRNA 10 mRNA * 0.5 * 0.5

Mass(g/kg) 0.5 LPL Activity Activity LPL * 1,000 nmol * ( 0 0 0 0 0 Adipocytesize (um (relativeexpression) (relativeexpression)

d 2,500 50 1,200 40 12 40 2,000 0 min 0 min 1,500 10 min /ml) 10 min 40 /ml) * 10 30 1,000 900 pg 30 500 30 pg * * * 0 600 20 8 20 20

Mass(g) MIP-Cre 300 10 6 MIP-Cre 10 MIP-Cre 10 –/– βcell –/– βcell –/– βcell Gipr Glucose(mM) Gipr Gipr Glucose(mM) 0 Total GIP ( 0 0 4 0 8 12 16 20 24 GLP-1( Total 8 12 16 20 24 0 30 60 90 120 Age (wk) Age (wk) Time (min) e 35 35 15 15 Surgery Surgery 30 30 10 10 25 25 5 5 Mass(g) 20 MIP-Cre Mass(g) 20 MIP-Cre MIP-Cre MIP-Cre Glucose(mM) Gipr–/– βcell Gipr–/– βcell Glucose(mM) Gipr–/– βcell Gipr–/– βcell 15 15 0 0 0 10 20 30 40 0 10 20 30 40 0 1 2 10 20 30 0 1 2 10 20 30 Day Day Day Day f 6,000 * 6,000 4,000 g 125 4,000 40 4,000 100 2,000 2,000 * 0 100 0 30 * * 75 3,000 75 20 2,000 50 50c 25 MIP-Cre 25 MIP-Cre 10 MIP-Cre 1,000 Gipr–/– βcell –/– βcell –/– βcell Group – p<0.05 Gipr Glucose(mM) Gipr Group – p<0.05 0 0 0 min)AUCx (mM 0 Glucose(%Baseline) 0 30 60 90 Glucose(%Baseline) 0 30 60 90 0 30 60 90 120 Time (min) Time (min) Time (min) h 70 70 MIP-Cre * 70 MIP-Cre * * –/– βcell * 60 MIP-Cre 60 Gipr–/– βcell * 60 Gipr * /pF) /pF) * /pF) –/– βcell /pF) * fF

fF 50 50 Gipr fF 50 + 10 nM * + 1 nM 40 + Veh 40 * 40 30 30 GIP * 30 Ex4 Cumulative Cumulative 20 Cumulative 20 20 capacitance(

10 10 capacitance( 10 capacitance( Group – p<0.05 Group – p<0.05 2.80 mM Glucose 16.7 mM Glucose Ex-4 (15min) 0 GIP (15min) LatrunculinB 0 2.80 mM 1 Glucose 2 3 416.7 5 mM6 Glucose7 8 9 10Ex-4 (15min) 0 GIP1 (15min)2 3 4 5 LatrunculinB6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 Depolarization Number Depolarization Number Depolarization Number

2.8 mM Glucose 16.7 mM Glucose Ex-4 (15min) GIP (15min) LatrunculinB i WT 2.8 mM G 16.7 mM G 1 nM Ex-4 10 nM GIP

WT 1.5 MIP-Cre Gipr–/– βcell j WT 2,000 MIP-Cre Actin Intensity 1.0 30 1,500 * 1,000 Actin Intensity * 500 20 2.8 mM Glucose 16.7 mM Glucose Ex-4 (15min) GIP (15min) LatrunculinB 0 2.8 mM Glucose 16.7 mM Glucose Ex-4 (15min) GIP (15min) LatrunculinB Actin Intensity 0.5 Acn Intensity 10 MIP-Cre GIP-/- β cell 2.8 mM Glucose 16.7 mM Glucose Ex-4 (15min) GIP (15min) –/– βcell

LatrunculinB Glucose(mM) Gipr

GIP-/- Group – p<0.05 –/–

-cell 0 0.0 Average Peak Intensity (AU) -cell 0 30 60 90 120 Gipr Time (min) GIP-/- -cell Actin Intensity Acn Intensity Actin Intensity Actin Intensity WT

) GIP-/- 1.2 W-cellT ) AU GIP-/- ( 1.2 -cell y AU t i

( ***

s y n t

i ***

e WT s t ) Nature Medicine: doi:10.1038/nm.3997n 0.1.82 GIP-/- e

t -cell in In AU

t 0.8 (

c y in In t A t i *** c s n A

e eak t 0.4

P 0.8

eak e 0.4 in In g P t

a c e r A g

a ve r A

eak 0.0 ve 0.4 P A 0.0 Glucose (mM)e 2.8 16.7 2.8 2.8 2.8 g a

Glucose (mM)r 2.8 16.7 2.8 2.8 2.8 Ex-4 (1 nM) - - + - - ve

Ex-4 (1 nM)A - - + - - GIP (10 nM) 0.0 - - - + - GlucoseGIP (10 nM)(mM) 2.8- 16.7- 2.8- 2.8+ 2.8- LatrunculinB (10 M) - - - - + Ex-4LatrunculinB (1 nM) (10 M) -- -- +- -- -+ GIP (10 nM) - - - + - LatrunculinB (10 M) - - - - + Supplementary Figure 3

a 2.0 20,000 3x106 ) 2 1.5 15,000 * 2x106 1.0 10,000 pancreas) 2 1x106

βcell area 0.5 * 5,000

Islet numberIslet Islet sizeIslet (um (% of pancreas)(%of

0 0 (#/um 0

b MIP-Cre Gipr–/– βcell 0.3

0.2

0.1

Granulesize (um) 0

c 2.0 2.0 MIP-Cre MIP-Cre 2.0 MIP-Cre 2.0 MIP-Cre Gipr–/– βcell Gipr–/– βcell Gipr–/– βcell Gipr–/– βcell 1.5 1.5 1.5 1.5 1.0 1.0 1.0 1.0 mRNA mRNA mRNA mRNA mRNA 0.5 0.5 mRNA 0.5 0.5 * 0 0 0 0 (relativeexpression) (relativeexpression) (relativeexpression) (relativeexpression)

d

0.9 kb

e Tcf7 f Human Islets GIP - + - + - + - + - + - + - + 1.4 p=0.052 0.9 kb 1.3 1.2 Cycles: 28 34 mRNA 1.1

TCF7 1.0 0.9 (relativeexpression) VEH GIP

Nature Medicine: doi:10.1038/nm.3997 Supplementary Figure 4

4,000 a * b c 3,000 60 80 4,000 40 2,000 * 40 Veh * 3,000 ) * ) 1,000 GIP 2,000 * * 0 60 30 1,000 30 mM 40 * * * mM * * * 0 * * 40 20 * 20 * 20 * * Fat Mass Fat 20 10 10 Veh Glucose( Glucose( Bodyweight (g) GIP 0 body(%of weight) 0 0 0 0 30 60 90 120 0 30 60 90 120 Time (min) Time (min)

d 6 wk WT 6 wk db/db

30 50 30 50 25 40 * 25 30 20 20 30 6 15 10 15 20 *

2 (mM) 10 10 10 4 Insulin(ng/ml) Glucose(mM) Glucose(mM) Insulin(ng/ml) 5 0 5 0 t10 – t0 2 0 10 0 10 0 10 0 10 Time (min) Time (min) Time (min) Time (min) 0 -2 12 wk WT 12 wk db/db

DeltaGlucose -4 30 50 30 50 25 30 25 40 20 20 30 15 10 15 20 2 10 10 10 Glucose(mM) Glucose(mM) Insulin(ng/ml) Insulin(ng/ml) 5 0 5 0 0 10 0 10 0 10 0 10 Time (min) Time (min) Time (min) Time (min)

e Glp1r Tcf7l2 f 1.5 1.5 40

) * 2 35 1.0 1.0 30

mRNA mRNA 0.5 mRNA 0.5 25 BMI (kg/m BMI 0 0 20 (relativeexpression) (relativeexpression)

Nature Medicine: doi:10.1038/nm.3997 Supplementary Figure 5

c a b 800 WT 1,000 2.0 800 20 600 –/– 20 60 50 Tcf7 * 600 400 400 40 15 200 1.5 * 15 200 0 0 40 30 10 1.0 10 20 p=0.06 20 Fat mass Fat 10 5 WT 0.5 5 WT

Insulin(ng/ml) –/– Tcf7–/– Glucose(mM) Tcf7 Glucose(mM) Body Weight (g) BodyWeight

0 body(%of weight) 0 0 0.0 0 0 30 60 90 120 0 10 0 10 0 30 60 90 120 Time (min) Time (min) Time (min) d 30 e 100 80 WT 100 WT * –/– –/– ) WT Tcf7 Tcf7 * /pF) /pF) /pF) 80 –/– /pF) 60 80 fF fF 20 Tcf7 fF mM 60 + 10 nM 60 + 1 nM 40 GIP Ex4 2,000 * 40 + Veh 40 10 * Cumulative Cumulative WT Cumulative 1,000 20 20 20 Tcf7-–/– Glucose( 0 capacitance( capacitance( 0 0 capacitance( 0 0 0 30 60 90 120 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 Time (min) Depolarization Number Depolarization Number Depolarization Number f 2.0 2.0 8.0 WT WT WT –/– –/– –/– 1.5 Tcf7 1.5 Tcf7 6.0 Tcf7 1.0 1.0 4.0 mRNA mRNA mRNA mRNA mRNA mRNA 0.5 0.5 2.0 0 0 0 (relativeexpression) (relativeexpression) (relativeexpression)

100 2.0 WT WT Tcf7–/– Tcf7–/– 1.5 60 40 1.0 2 mRNA mRNA mRNA mRNA 0.5 1 0 0 (relativeexpression) (relativeexpression)

g 15 Refed 1.5 Refed 15 Refed 4 Refed 3 10 1.0 10 2 5 0.5 5 WT WT WT 1 WT Insulin(ng/ml) –/– Insulin(ng/ml) –/– –/– –/– Glucose(mM) Glucose(mM) Tcf7 Tcf7 Tcf7 Tcf7 0 0 0 0

Time of Day (h) Time of Day (h) Time of Day (h) Time of Day (h)

h * 3 p=0.06 15,000 * )

2 * 2.0x106 2 10,000 *

* Pancreas) 6 2 1.0x10 1 5,000 βcell area Islet numberIslet Islet sizeIslet (um (% of pancreas)(%of

0 0 (#/um 0

Nature Medicine: doi:10.1038/nm.3997 Supplementary Figure 6

Low-Fat Diet: a WT Tcf7–/–

25 25 H2O 3 0 min H2O 3 0 min 800

20 DPP4i 10 min 20 DPP4i 10 min 600 15 2 15 2 * 400 10 * 10 * 1 1 * 5 5 200 Glucose(mM) Insulin(ng/ml) Insulin(ng/ml) Glucose(mM) AUC (mM x min)AUCx (mM 0 0 0 0 0 - -30 0 30 60 90 120 H2O DPP4i -30 0 30 60 90 120 H2O DPP4i DPP4i - + + Time (min) Time (min)

b IPGTT IPGTT IPGTT + PBS + GIP + Ex4 PBS GIP Ex4 1500 20 20 20 2.5 * 1000 1500 0 min 1500 * 15 500 15 1000 15 2.0 10 min 1000 * * 0 500 1.5 * 500 10 10 0 10 * 0 1.0 5 5 WT 5 WT WT 0.5 Insulin(ng/ml)

Glucose(mM) –/– Glucose(mM) Tcf7–/– Glucose(mM) Tcf7–/– Tcf7 0 0 0 0 0 30 60 90 120 0 30 60 90 120 0 30 60 90 120 Time (min) Time (min) Time (min)

High-Fat Diet: c WT Tcf7–/–

25 25 2000 15 0 min * 15 0 min * 20 * * 10 min 10 min p=0.053 20 * 1500 * 10 15 * 10 15 1000 * 10 * 10 5 5 * H O 500 5 2 5 H2O Insulin(ng/ml) Glucose(mM) Insulin(ng/ml) Glucose(mM) DPP4iTreatment – p<0.05 DPP4iTreatment – p<0.05 min)AUCx (mM 0 0 0 0 0 -30 0 30 60 90 120 H2O DPP4i -30 0 30 60 90 120 H2O DPP4i DPP4i - - + + Time (min) Time (min)

d IPGTT IPGTT IPGTT 1,500 * + PBS 1,000 + GIP + Ex4 PBS GIP Ex4 40 40 1,500 40 500 1,500 15 1,000 0 min 0 1,000 10 min 30 * 30 500 30 500 10 * * 0 * * 20 20 20 0 5 10 10 10 WT WT WT Glucose(mM) Glucose(mM) –/– –/– Insulin(ng/ml) Tcf7 Group – p<0.05 Glucose(mM) Tcf7–/– Tcf7 0 0 0 0 0 30 60 90 120 0 30 60 90 120 0 30 60 90 120 Time (min) Time (min) Time (min)

Nature Medicine: doi:10.1038/nm.3997 Supplementary Figure 7

a b Bright field GFP Tcf1

Ad-Gfp

50 kDa

Ad-Tcf7-Gfp 35 kDa

c MIP-Cre: PBS + PBS + Thapsigargin + Thapsigargin + Ad-Gfp Ad-Tcf7 Ad-Gfp Ad-Tcf7 Bright field TMRM

Gipr–/– βcell: PBS + PBS + Thapsigargin + Thapsigargin + Ad-Gfp Ad-Tcf7 Ad-Gfp Ad-Tcf7 Bright field TMRM

Nature Medicine: doi:10.1038/nm.3997 Supplementary Figure 8

a Bright field TMRM

Control

Thapsigargin

Thapsigargin + TCF1

b MIN6 β Cells 1.5

1.0

mRNA mRNA 0.5 * Tcf7

(RelativeExpression) 0.0 Supplementary Table 1. List of Real-Time Primers

Gene ABI Catalog # Amplicon Length Mouse Adipoq Mm00456425_m1 74 Akt1 Mm01331626_m1 69 Bcl2 Mm00477631_m1 85 Bcl2l1 Mm00437783_m1 65 Birc2 Mm00431811_m1 90 Birc3 Mm00431800_m1 61 Creb1 Mm00501607_m1 106 Ctnnb1 Mm00483039_m1 77 Dgat1 Mm00515643_m1 66 Dgat2 Mm01273905_m1 74 Egf Mm00438696_m1 96 Egr1 Mm00656724_m1 182 Ffar1 Mm00809442_m1 147 Ffar2 Mm02620654_s1 147 Ffar3 Mm02621638_s1 140 Ffar4 Mm00725193_m1 79 Gck Mm00439129_m1 69 Gipr Mm01316344_m1 92 Glp1r Mm01351008_m1 94 Gpr119 Mm00731497_s1 63 Hif1a Mm00468869_m1 75 Igf1 Mm00439561_m1 69 Igf1r Mm00802831_m1 106 Igfr2 Mm00439576_m1 64 Ins2 Mm00731595_gH 99 Irs2 Mm03038438_m1 63 Kcnj11 Mm00440050_m1 129 Lef1 Mm00550265_m1 84 Lep Mm004346579_m1 73 Lipe Mm00495359_m1 70 Lpl Mm00434770_m1 77 Mapk14 Mm00442497_m1 56 Mapk3 Mm01973540_g1 98 Neurod1 Mm01280117_m1 81 Pdx1 Mm00435565_m1 74 Pparg Mm01184322_m1 101 Ppia Mm02342430_g1 148 Pttg1 Mm00479224_m1 72 Rapgef4 Mm00517216_m1 67 Retn Mm00445641_m1 80 Serpine1 Mm00435860_m1 60 Slc2a2 Mm00446224_m1 83 Tcf4 Mm00443210_m1 98 Tcf7 Mm00493445_m1 91 Tcf7l1 Mm01188711_m1 78 Tcf7l2 Mm00501505_m1 85 Tph1 Mm01202614_m1 103 Tph2 Mm00557717_m1 82 Tnf Mm00443258_m1 81 Xiap Mm00776505_m1 144 Rat Tcf7 Rn01463535_m1 68 Ppia Rn00690933 149 Human TCF7 Hs00175273_m1 105 GAPDH Hs99999905 122

Nature Medicine: doi:10.1038/nm.3997 Supplementary Table 2 List of Antibodies

Protein Company Catalog # TCF1 Cell Signaling Technologies 2206 Lamin A/C Cell Signaling Technologies 2032 phosphoERK Cell Signaling Technologies 9101 ERK Cell Signaling Technologies 9102 Cleaved Caspase 3 Cell Signaling Technologies 9661 Caspase 3 Cell Signaling Technologies 9662 HSP90 BD Biosciences 610418

All Cell signaling technologies antibodies are profiled on antibodypedia.

Nature Medicine: doi:10.1038/nm.3997 Supplemental Information

TCF1 links GIPR signaling to the control of  cell function and survival

Jonathan E Campbell1, John R Ussher1,7, Erin E Mulvihill1, Jelena Kolic2,3, Laurie L Baggio1, Xiemen Cao1, Yu Liu1, Benjamin J Lamont1,7, Tsukasa Morii1,7, Catherine J Streutker4, Natalia Tamarina5, Louis H Philipson5, Jeffrey L Wrana1, Patrick E MacDonald2,3 & Daniel J Drucker1,6

Supplemental Figure 1 Generation of Gipr–/–βCell mice. (a) Schematic illustrating the generation of Gipr–/–βCell mice. (b) Gipr expression in mouse islets (left, n=3, 5, 3, and 5) and epididymal adipose tissue (right, n=7). (c) Intraperitoneal glucose tolerance test (IPGTT) in 10-week-old mice fed a LFD (n=8, 4, 7, and 5). (d) hGH release (left) from mouse islets in culture, Tph1 (middle) and Tph2 (right) expression in mouse islets (n=5, 5, and 6). (e) Glycemic (left) and insulin (right) response during an oral glucose tolerance test (OGTT) in 10-week-old mice fed a LFD (n=7). (f) Glycemic response during an intraperitonel insulin tolerance test in 8-week-old mice fed a LFD (n=4 and 6). (g) Body weight (left), fed glucose values (middle), and average food intake (right) in mice fed a LFD (n=7). (h) Total GIP (left) and GLP-1 (right) plasma concentration during an OGTT in mice fed a LFD (n=7). (i) IPGTT in 18-week-old mice fed a LFD (n=7). *P<0.05 vs control, or as indicated. Data are expressed as Mean± SEM. Statistical tests used: t-test – for g (right) and AUC inset (c, e and i); 1-way analysis of variance (ANOVA) – for b and d; 2-way ANOVA - for c, and e-i. LFD – low fat diet, GIP – glucose-dependent insulinotropic polypeptide, GLP-1 – glucagon-like peptide 1.

Supplemental Figure 2 Phenotype of Gipr–/–βCell mice. (a) Adipose depot weights (left) and epididymal adipocyte size (right) from 20-week-old mice (n=7). (b) qPCR analysis of genes important for adipose tissue metabolism (n=7). (c) Lipoprotein lipase activity in epididymal adipose depots from 20-week-old mice (n=7). (d) (left to right) Body weight (n=9 and 13), plasma total GIP concentrations (n=9), plasma total GLP-1 concentrations (n=9), fed glycemia (n=9 and 13), and intraperitoneal glucose tolerance test (IPGTT) (n=9 and 13) in mice fed a HFD. (e) Body weight (two left panels) and glycemia (two right panels) following sham (blue and red) or insulin pellet (purple and brown) surgery (n=3, 5, 3, and 6; MIP-Cre (sham), Gipr–/–βCell (sham), MIP-Cre (pellet), and Gipr–/–βCell (pellet)). (f) Glycemic response to insulin tolerance test in 18-week-old mice following sham (left) or insulin pellet (right) surgery (n=3, 5, 3, and 6; MIP-Cre (sham), Gipr–/–βCell (sham), MIP-Cre (pellet), and Gipr–/–βCell (pellet)). (g) Glycemic response during an oral glucose tolerance test in 16-week-old mice fed a HFD and given a GLP-1R antagonist (n=9). (h) Cumulative capacitance in isolated  cells determined by whole cell patch clamping , treated with vehicle (left panel), GIP (middle panel) or exendin-4 (right panel) (i) Actin intensity in individual  cells stimulated with low glucose (n=55 and 42, MIP- Cre and Gipr–/–βCell) 16.7 mM glucose (n=48 and 51, MIP-Cre and Gipr–/–βCell), 1 nM Ex4 (n=46 and 44, MIP-Cre and Gipr–/–βCell), 10 nM GIP (n=47 and 44, MIP-Cre and Gipr–/– βCell), or 10 uM latrunculin B (LatB) (n=42 and 32, MIP-Cre and Gipr–/–βCell). (j) IPGTT in 16-week-old mice given a GPR119 agonist (n=8 and 13). *P<0.05 vs control. Data are expressed as Mean± SEM. Statistical tests used: t-test – for a-c and i, and AUC (d, f and

Nature Medicine: doi:10.1038/nm.3997 g); 2-way analysis of variance (ANOVA) – for d-g. GIP – glucose-dependent insulinotropic polypeptide, GLP-1 – glucagon-like peptide 1, HFD – high fat diet.

Supplemental Figure 3 Characterization of Gipr–/–βCell islets. (a) Histological analysis of  cell area (left), islet size (middle), and islet number (right) in samples from mice fed a LFD (red and blue, n=7 and 7) or HFD (pink and light blue, n=9 and 13). (b) Electron microscopy showing insulin granular size in pancreata samples from mice fed a LFD (n=7). (c) qPCR analysis of RNA from isolated islets from 20-week-old mice fed a LFD (n=7). (d) Representative image of Tcf7 expression in RNA from thymus and islets from 8-week-old mice fed a LFD. (e) Representative image of Tcf7 expression in islets treated with Veh (-) or GIP (+) following sequential PCR cycling (n=3). (f) TCF7 expression in human islets treated with Veh or GIP (n=4). *P<0.05 vs control. Data are expressed as Mean± SEM. Statistical tests used: t-test – for b, c and f; 2-way analysis of variance (ANOVA) – for a. Veh – vehicle, GIP – glucose-dependent insulinotropic polypeptide. LFD – low fat diet, HFD – high fat diet.

Supplemental Figure 4 Diabetic db/db mice fail to lower glucose in response to exogenous GIP. (a) Body weight (n=5). (b) Body adiposity determined by MRI (n=5). (c) Glycemic response during an intraperitoneal glucose tolerance test in 6-week-old WT (left) and db/db (right) mice (n=5). (d) Glycemic response to exogenous GIP in random fed mice (n=5). (e) Glp1r (left) and Tcf7l2 (right) expression in RNA from mouse islets (n=5). (f) Body mass index in human subjects (n=5, 5, and 4). *P<0.05 vs control, or as indicated. Data are expressed as Mean± SEM. Statistical tests used: t-test – for b, c and f; 2-way analysis of variance (ANOVA) – for a. Veh – vehicle, GIP – glucose-dependent insulinotropic polypeptide, ND – non-diabetic, T2D – type 2 diabetic.

Supplemental Figure 5 Phenotype of Tcf7–/– mice. (a) Body weight in 18-week-old mice fed a LFD (dark blue and dark grey, n=7 and 5) or HFD (light blue and light grey, n=10 and 7). (b) Body adiposity determined by MRI in mice fed a LFD (dark blue and dark grey) or HFD (light blue and light grey) (n=7, 5, 10, and 7). (c) Glycemic (left) and insulin (middle) response during an oral glucose tolerance test (OGTT) and glycemic (right) response during an intraperitoneal glucose test (IPGTT) in 8-week-old mice on a LFD (n=7 and 8). (d) Glycemic response during an IPGTT in 12-week-old mice on a HFD (n=10 and 7). (e) Cumulative capacitance in individual  cells stimulated with Veh (left, n=18 and 20), GIP (middle, n=18 and 18), or Ex4 (right, n=15 and 17). (f) qPCR analysis of RNA from islets from 18-week-old mice fed a LFD (n=4). (g) Glycemic and insulin response during a fasting-refeeding test in 10-week-old mice fed a LFD (two left panels, n=7 and 5) or HFD (two right panels, n=7 and 10). (h) Histological analysis of  cell area (left), islet size (middle), and islet number (right) in samples from mice fed a LFD (dark blue and grey, n=7 and 5) or HFD (light blue and light grey, n=10 and 7). *P<0.05 vs control, or as indicated. Data are expressed as Mean± SEM. Statistical tests used: t-test – for f and AUC inset (c and d); 2-way analysis of variance (ANOVA) – for a-e, g, h. LFD- low fat diet, HFD – high fat diet, Veh – vehicle, GIP – glucose-dependent insulinotropic polypeptide, Ex4 – exendin-4.

Nature Medicine: doi:10.1038/nm.3997 Supplemental Figure 6 Glycemic and insulin response to incretin receptor agonists and DPP4 inhibition in Tcf7–/– mice. (a) Glycemic and insulin response during an oral glucose tolerance test (OGTT) in 12-week-old mice fed a LFD (n=5). (b) Glycemic and insulin response during an intraperitoneal glucose tolerance test (IPGTT) in 12-week-old mice fed a LFD (n=7 and 5, WT and Tcf7–/–). (c) Glycemic and insulin response during an OGTT in 12-week-old mice fed a HFD (n=9). (b) Glycemic and insulin response during and intraperitoneal glucose tolerance test (IPGTT) in 12-week-old mice fed a HFD (n=10 and 8 WT and Tcf7–/–). *P<0.05 vs control, or as indicated. Data are expressed as Mean± SEM. Statistical test used: 2-way analysis of variance (ANOVA). LFD- low fat diet, HFD – high fat diet, DPP4i – dipeptidyl peptidase 4 inhibitor, Veh – vehicle, GIP – glucose- dependent insulinotropic polypeptide, Ex4 – exendin-4.

Supplemental Figure 7 Representative images of adenoviral transduction and apoptotic index. (a) Representative image of bright field and GFP (495 nm) in mouse  cells. (b) Tcf1 protein expression in baby hamster kidney (BHK) fibroblast cells and mouse thymus. (c) Representative images of bright field and TMRM (545 nm) in mouse  cells.

Supplemental Figure 8 Representative image of apoptotic index and Tcf7 knockdown in MIN6  cells. (a) Representative images of bright field and TMRM (545 nm) in MIN6  cells. (b) Tcf7 expression in MIN6  cells (n=6). *P<0.05 vs control, or as indicated. Data are expressed as Mean± SEM. Statistical test used: t-test – for b.

Nature Medicine: doi:10.1038/nm.3997

Online Methods

Animals. Male mice were used for all mouse studies and were maintained under a 12 h light/12 h dark cycle at constant temperature (23 °C) with free access to food and water. All animal studies were approved by Mt. Sinai Hospital (Toronto) and the Toronto Centre for Phenogenomics animal-care committee. Animals were fed either a low-fat diet (10% kcal from fat; Research Diets, D12450B) or high-fat diet (45% kcal from fat; Research Diets, D12451). To generate Gipr–/–Cell mice, MIPcreER transgenic mice (on a C57BL/6J background) expressing tamoxifen-inducible Cre driven by the mouse insulin promoter were bred with floxed Gipr mice (GiprFlox/Flox), backcrossed 8 times to C57BL/6J background)41. Cre-induced inactivation of the Gipr gene was carried out via 5 consecutive daily intraperitoneal (i.p.) injections of tamoxifen (40 mg per kg) in 6-week- old mice. Glp1r–/– and Tcf7–/– mice (both on C57BL/6J backgrounds) have been previously described42,43. Tcf7–/– mice were generously provided by H. Clevers, db/db mice were purchased from Jackson laboratories (#000697). For all animal experiments, the sample size required to achieve adequate power was estimated on the basis of pilot work or previous experiments. When appropriate, animals were randomly allocated to individual experimental groups.

Peptides and reagents. Peptides were reconstituted in phosphate-buffered saline (PBS), aliquoted and stored at –80 °C. [D-Ala2]GIP (GIP) was from Chi Scientific, Ex4 (exendin-4) was from California Peptide Research Inc. Plasmid constructs (Tcf7 and Pttg1) and siRNA (Gipr, Glp1r, Tcf7, TCF7) were from Origene. The GLP-1R antagonist (JANT-4)44 was a generous gift from R. DiMarchi, University of Indiana.

Mouse islet isolation. Primary mouse islets were isolated as previously described45. Briefly, the pancreas was inflated via the pancreatic duct with collagenase type V (0.8 mg per ml), excised and digested for 10–15 min. The digest was washed with cold RPMI (2 mM L-glutamine, 10 mM glucose, 0.25% BSA, 100 U/ml penicillin, and 100 μg/ml streptomycin), and the islets were separated using a Histopaque gradient. Individual islets were handpicked and either immersed in TRI Reagent for subsequent mRNA isolation or allowed to recover overnight in RPMI with 10% FBS for experiments ex vivo.

Primary mouse islet insulin secretion. After overnight incubation, 75–80 medium-sized islets were handpicked into 0.275 ml chambers containing KRB buffer (135 mM NaCl, 3.6 mM KCl, 1.5 mM CaCl2, 0.5 mM NaH2PO4, 0.5 mM MgSO4, 5 mM Hepes, 5 mM NaCO3, 0.1% BSA pH 7.5). Islets were perifused for 1 h in KRB with 2.7 mM glucose at a flow rate of 200 μl per min using the Biorep Perifusion system. After this equilibration period, islets were perifused at 8 min intervals in experimental media (KRB plus various conditions), then collected and lysed in acid ethanol for total insulin measurements. Insulin concentrations were determined by radioimmunoassay (Millipore), and insulin secretion was expressed as a percent of total insulin.

Nature Medicine: doi:10.1038/nm.3997

Islet hGH release. Isolated islets were incubated in batches of 100 at 37 °C in HEPES Krebs buffer containing 20 mM glucose, as previously described7. After 1 h, the buffer was removed and assayed for hGH release using a hGH ELISA (Invitrogen).

Glucose- and insulin-tolerance tests. Oral and intraperitoneal glucose-tolerance tests (GTTs) were performed in mice fasted for 5 h (0700–1200) using a glucose dose of 1.5 g per kg. During IPGTT, mice were i.p. injected with either GIP (4 nmol per kg), Ex4 (0.3 nmol per kg), or saline (veh), 10 min before glucose administration. For tests using a DPP4 inhibitor, sitagliptin (Merck, 40 μg per mouse) was given orally 30 min before glucose. For both OGTT and IPGTT, blood was collected at 0, 10 and 30 min in capillary tubes coated with 10% (vol/vol) TED (500,000 IU/ml Trasylol; 1.2 mg/ml EDTA; and 0.1 mM diprotin A) and plasma separated by centrifugation at 4 °C and stored at –80 °C. Insulin-tolerance tests (ITTs) were performed in mice fasted for 5 h (0700–1200) using an insulin dose of 0.7 U/kg (Humalog, Lilly).

Plasma hormone analysis. Insulin (Alpco Diagnostics) and total GIP (Linco) levels were analyzed by ELISA. Total GLP-1 levels were measured by immunoassay (Mesoscale).

Insulin supplementation. 8-week-old mice had a single insulin pellet (7 or 14 mg, LinBit) inserted into the intrascapular region under isoflurane anesthesia by following the manufacturer’s protocol.

Apoptosis in vivo. Mice were treated with streptozotocin (Sigma, 50 mg per kg) for 5 consecutive days at 0800. Twenty-four hours after the final treatment, mice were euthanized and the pancreas was excised and immediately immersed in 10% formalin. All histological analysis was performed in a blinded fashion.

Real time quantitative PCR. First-strand cDNA was synthesized from total RNA using the SuperScript III synthesis system (Invitrogen). Real-time PCR was carried out with the ABI Prism 7900 Sequence Detection System using TaqMan Gene Expression Assays (Applied Biosystems). Relative mRNA transcript levels were quantified with the 2–Ct method. PCR primers are shown in Supplementary Table 1.

Qualitative PCR. Amplification of mouse Gipr cDNA was performed using the primer pairs 5–CTG CTT CTG CTG CTG TGG T–3 (forward primer) and 5–CAC ATG CAG CAT CCC AGA– 3reverse primer). PCR was carried out using 35 cycles at an annealing temperature of 50 °C to generate a 1.5 kb product. Amplification of the mouse Tcf7 isoforms was performed using the common reverse primer 5–CTA GAG CAC TGT CAT CGG–3and

Nature Medicine: doi:10.1038/nm.3997

two different 5 (forward) primers targeting alternative start codons (1.3 kb product – ATG CCG CAG CTG GAC TCG; 0.9 kb product – ATG TAC AAA GAG ACT GTC TAC T). PCR was carried out using 35 cycles at an annealing temperature of 56 °C.

Western blot analysis. Thirty μg of total protein was separated by SDS-PAGE, transferred to nitrocellulose membranes and blocked in 5% milk in PBS-T for 1 h before incubation in primary antibody overnight at 4 °C. Immunoblots were visualized with the enhanced chemiluminescence Western blot detection kit (Perkin Elmer) and quantified with Carestream Molecular Imaging Software (Kodak). Primary antibodies are shown in Supplementary Table 2.

Human islet isolation. Primary human islets were isolated as previously described6 at the Alberta Diabetes Institute Islet Core (www.bcell.org/isletcore.html) and the Clinical Islet Isolation Facility at the University of Alberta and cultured in low glucose (5.5 mM) DMEM with L- glutamine, 110 mg per L sodium pyruvate, 10% FBS and 100 U/ml penicillin/streptomycin. We studied islets from 15 non-diabetic, lean donors (age: 57 ± 11 years; HbA1c: 5.7 ± 1.4 BMI range 19–39) and 4 T2D donors (age: 63 ± 5 years; HbA1c: 6.6 ± 0.9; BMI range- 29–37).

Capacitance measurement. Islets were dispersed in calcium-free dissociation buffer in 35 mm dishes and incubated overnight in RPMI containing 11mM (mouse) or 5.5 mM (human) glucose. GIP (10 nM), Ex4 (1 nM) or vehicle (H2O) was added to each dish 1 h before patch clamping, using the standard whole-cell technique with the sine + DC locking function of an EPC10 amplifier and Patchmaster software (HEKA Electronics). Experiments were performed as described previously7 at 32–35 °C using an extracellular bath solution (118 mM NaCl, 20 mM TEA, 5.6 mM KCl, 1.2 mM MgCl2, 2.6 mM CaCl2, 5 mM glucose, 5 mM Hepes, pH – 7.4) and pipette solution (125 mM CsGlutamate, 10 mM CsCl, 10 mM NaCl, 1 mM MgCl2, 0.05 mM EGTA, 5 mM Hepes, 3 mM MgATP, 0.1 mM cAMP, pH – 7.15). Capacitance responses to a train of 10 depolarizations from –70 to 0 mV at 1 Hz were normalized to initial cell size and expressed as femtofarad per picofarad (fF/pF).  cells were identified using positive insulin immunostaining.

RNA-seq. Total RNA from isolated islets was extracted using TRI Reagent. The yield and quality of total RNA was assessed using Nanodrop (Thermo Fisher) and BioAnalyzer (Agilent), respectively. Ribosomal RNA was removed using a bead-based hybridization kit (RiboZero, Epicentre) and cDNA libraries were prepared using the Illumina TruSeq RNA sample preparation kit. The quality and concentration of libraries were assessed using BioAnalyzer and qPCR, respectively. The libraries were loaded as two indexed samples per lane on an Illumina HiSeq 2000. Raw sequenced reads were obtained in fastq format, and mapped onto the mouse genome (mm9) using Tophate1.4.1, and then analyzed using a custom R-based pipeline to calculate gene-expression profiles using ENSEMBL annotation for coding genes. The number of reads mapped onto the gene was counted

Nature Medicine: doi:10.1038/nm.3997

regardless of transcription isoform and normalized to total mapped reads to obtain transcript union Read Per Million total reads (truRPMs). The data have been deposited in NCBI’s Gene Expression Omnibus and are accessible through GEO Series accession number GSE65361.

Cell-line culture. INS1823/32 cells were a generous gift from C. Newgard, Duke University. Cells were grown in RPMI (10% FBS, 1% P/S). For GIP experiments, cells were starved in RPMI (1% FBS, 1% P/S) for 3 h. MIN6 cells (from ATCC) were grown in DMEM (20% FBS, 1% P/S). siRNA knockdown experiments were performed by following the manufacturer’s instructions (Origene). All cell lines were previously tested for mycoplasma contamination.

Apoptosis assays. MIN6 beta cells. After 24 h exposure to thapsigargin (5 uM, Sigma) in culture media (DMEM, 20% FBS, 1% P/S), apoptosis was analyzed using a MitoPT assay (ImmunoChemistry Technologies) by following the kit’s instructions. Briefly, cells were exposed to 100 nM MitoPT for 10 min, washed 1 × with PBS and then detached from the plate with trypsin. An aliquot of cells was visualized on a microscope slide. Total cell number was counted using bright field and tetramethylrhodamine methyl ester (TMRM) uptake was visualized with a 545 nM filter. Apoptotic cells were calculated as (total cells – TMRM positive cells)/Total cells × 100%. Mouse islets. Dispersed islets were transduced with Ad-GFP or Ad-Tcf7-GFP at an MOI of 10 to achieve a >90% induction rate. After transduction, cells were exposed to control or thapsigargin (100 uM) for 72 h. Apoptosis was assessed using a MitoPT assay. Human islets. Dispersed human cells were transfected with human siTCF7 or siScram control duplexes (OriGene, Rockville, MD) and an Alexa488 labeled negative control siRNA (Qiagen, Toronto, ON), using Dharmafect (Thermo Scientific, Ottawa, ON, Canada). 24 h post-transfection culture medium was changed to fresh medium containing glucose and/or human recombinant IL1- (Sigma, Oakville, ON, Canada), as indicated. Cell-death assays were performed on dispersed human islet cells by use of the In situ Cell Death Detection Kit TMR Red (Roche, Mannheim, Germany), using TUNEL technology, according to the manufacturer’s directions. Images were obtained using a Zeiss AxioObserver Z1 with a Zeiss-Colibri light source at 488 nm and 594 nm, a × 40/1.3 NA lens, and an AxioCam HRm camera. Images were acquired in Axiovision 4.8 software (Carl Zeiss MicroImaging, Göttingen, Germany) and analyzed using ImageJ software (National Institutes of Health). Cell death was determined as ((#TUNEL+/Alexa488+) / (#Alexa488+)) and expressed as a fold increase over control, unstimulated conditions (5.5mM glucose, scrambled siRNA).

Statistical analysis. All values are presented as mean ± s.e.m. Statistical analysis was performed using GraphPad Prism 5.0. The appropriate t-test, one-way analysis of variance (ANOVA), or two-way ANOVA was completed using P < 0.05 to signify significant differences. Bonferroni post-hoc analysis was performed where appropriate. All data was assessed to

Nature Medicine: doi:10.1038/nm.3997

ensure normal distribution and equal variance between groups, using GraphPad Prism 5.0. Prior to the experiment, it was determined that individual data points would be excluded if their value was greater than 2 × SD from the mean, in an experiment with a sample size greater than seven.

41. Wicksteed, B. et al. Conditional gene targeting in mouse pancreatic ss-cells: analysis of ectopic Cre transgene expression in the brain. Diabetes 59, 3090– 3098 (2010). 42. Scrocchi, L.A. et al. Glucose intolerance but normal satiety in mice with a null mutation in the glucagon-like peptide receptor gene. Nat. Med. 2, 1254–1258 (1996). 43. Verbeek, S. et al. An HMG-box-containing T-cell factor required for thymocyte differentiation. Nature 374, 70–74 (1995). 44. Patterson, J.T. et al. A novel human-based receptor antagonist of sustained action reveals body weight control by endogenous GLP-1. ACS Chem. Biol. 6, 135–145 (2011). 45. Lamont, B.J. et al. Pancreatic GLP-1 receptor activation is sufficient for incretin control of glucose metabolism in mice. J. Clin. Invest. 122, 388–402 (2012).

Nature Medicine: doi:10.1038/nm.3997