Deletion of Gdf15 Reduces ER Stress-induced Beta-cell Apoptosis and Diabetes

Abstract

Endoplasmic reticulum (ER) stress contributes to pancreatic beta-cell apoptosis in diabetes, but the factors involved are still not fully elucidated. Growth differentiation factor 15 (GDF15) is a stress response gene and has been reported to be increased and play an important role in various diseases. However, the role of GDF15 in beta cells in the context of ER stress and diabetes is still unclear. In this study, we have discovered that GDF15 promotes ER stress-induced beta-cell apoptosis and that downregulation of GDF15 has beneficial effects on beta-cell survival in diabetes. Specifically, we found that GDF15 is induced by ER stress in beta cells and human islets, and that the transcription factor C/EBPβ is involved in this process. Interestingly, ER stress-induced apoptosis was significantly reduced in INS-1 cells with Gdf15 knockdown and in isolated Gdf15 knockout mouse islets. In vivo, we found that Gdf15 deletion attenuates streptozotocin-induced diabetes by preserving beta cells and insulin levels. Moreover, deletion of Gdf15 significantly delayed diabetes development in spontaneous ER stress-prone Akita mice. Thus, our findings suggest that GDF15 contributes to ER stress-induced beta-cell apoptosis and that inhibition of GDF15 may represent a novel strategy to promote beta-cell survival and treat diabetes.

In pancreatic beta cells, insulin is folded and matured in the endoplasmic reticulum (ER). Insulin mutations or need for excessive insulin production such as in response to hyperglycemia lead to accumulation of misfolded and unfolded insulin in the ER, resulting in ER stress (1-3). This triggers the unfolded protein response (UPR), which involves activation of 3 transmembrane proteins (IRE1, PERK, and ATF6), to alleviate ER stress (2, 3). However, prolonged ER stress converts this homeostatic response to terminal UPR, leading to beta-cell apoptosis and diabetes development (2, 3). Some processes involved in beta-cell ER stress/UPR have been extensively studied, but others are still not fully understood. To develop efficient strategies to halt beta-cell death and treat diabetes, a better understanding of the factors involved in beta-cell ER stress/UPR is needed.

Growth differentiation factor 15 (GDF15), a distant member of the TGFβ family (4), is also known as macrophage inhibitory cytokine-1 (4), nonsteroidal anti-inflammatory drug-activated gene (5), and prostate-derived factor (6). GDF15 is a stress response gene, and its increase has been associated with cancers (7, 8), cardiovascular disease (9,10), insulin resistance (11, 12), diabetes, and diabetic complications (13-18). Until now, the function of GDF15 has mainly been studied in the context of its role as a circulating factor in regulating food intake and body weight through binding and activation of its unique brain-restricted receptor, GDNF Family Receptor Alpha Like (19-22). In terms of intracellular GDF15, several studies have shown that various antitumor factors induce the expression of GDF15, which in turn promotes cancer cell apoptosis (23-29). In this context, GDF15 has been shown to translocate into mitochondria and induce reactive oxygen species production (30). In addition, GDF15 has also been shown to accumulate in the nucleus and modulate gene expression (31).

Surprisingly, a recent study reported that treatment with exogenous GDF15 could decrease cytokine-induced beta-cell death and that administration of GDF15 could reduce the incidence of diabetes in nonobese diabetic (NOD) mice (32). This is in contrast to the pro-apoptotic effects of GDF15 observed in tumors (25-29) and the elevated circulating GDF15 levels found in subjects with diabetes (16-18). These seemingly contrasting effects raise the question as to what role endogenous, and especially intracellular, GDF15 plays in beta-cell death and diabetes development. The present study was therefore aimed at elucidating the role of intracellular GDF15 and its downregulation in the context of diabetes. Using human and mouse islets, INS-1 cells, Akita cells, Gdf15 knockout mice, and 2 diabetes mouse models, we found that intracellular GDF15 contributes to ER stress-mediated beta-cell apoptosis and that downregulation of intracellular GDF15 has beneficial effects on beta-cell survival and delays diabetes development.

Materials and Methods

Tissue Culture

INS-1 cells were grown in RPMI 1640 (Thermo Fisher Scientific) with 11.1 mmol/L glucose, 10% fetal bovine serum, 1% penicillin/streptomycin, 1 mmol/L sodium pyruvate, 2 mmol/L L-glutamine, 10 mmol/L HEPES, and 0.05 mmol/L 2-mercaptoethanol, as previously described (33, 34). To induce ER stress, INS-1 cells were treated with a commonly used ER stress inducer, thapsigargin (TG), which inhibits the sarco/ER Ca²⁺-ATPase and induces cell death via ER Ca²⁺ depletion and the UPR (35). To study the effect of pro-inflammatory cytokines on Gdf15 expression, INS-1 cells were treated with cytokine cocktail (0.5 ng/mL TNFα, 0.5 ng/mL interferon γ, and 0.1 ng/mL IL-1β) or PBS for 16 hours. To knockdown gene expression, INS-1 cells were transfected with siRNA (Horizon Discovery) using DharmaFECT 1 transfection reagent (Horizon Discovery). The Akita and wild-type (WT) cells were grown in DMEM (Thermo Fisher Scientific) supplemented with 15% fetal bovine serum and 1% penicillin/streptomycin as previously described (36, 37). Human islets were obtained from the Integrated Islet Distribution Program (Table 1); islets from the same donor were treated with vehicle or drug. A minimum of 3 donor islet preparations were used per experiment. To induce ER stress, human islets were treated with 1 µM TG for 24 hours. Mouse pancreatic islets were isolated by collagenase digestion (33) and used for apoptosis assays.

Table 1.

Human islet preparations used in the study

Islet prep	Origin	Islet isolation center	Unique identifier	Age (y)	Sex	BMI (kg/ m2)	HbA1c (%)/blood glucose (mg/dL)	Hx of DM	Cause of death	Cold ischemia (h)	Purity (%)	Viability (%)	Handpicked
1	IIDP	Southern California	ADGU064	30	M	35.9	NA/176.4	N	Stroke	12.5	90	95	Y
2	IIDP	Southern California	SAMN11791244	47	M	36.1	NA/166.2	N	Stroke	6.5	95	95	Y
3	IIDP	University of Miami	SAMN12274306	37	M	25.3	NA/192.8	N	Anoxia	152	80	80	Y
4	IIDP	University of Wisconsin	SAMN12670838	40	M	30.7	NA/152.6	N	Head trauma	6.2	90	98	Y
5	IIDP	Southern California	SAMN12713942	41	M	23.5	NA/144.0	N	Stroke	10.7	85	95	Y

Abbreviations: BMI, body mass index; DM, diabetes mellitus; HbA1c, glycated hemoglobin; Hx, history; IIDP, Integrated Islet Distribution Program; M, male; N, no; NA, not available; Y, yes.

Animal Studies

All mouse studies were approved by the University of Alabama at Birmingham Animal Care and Use Committee. NOD mice and nonobese diabetes-resistant mice were purchased from the Jackson Laboratory. To follow the development of diabetes in mice, blood glucose was measured using a glucometer (Life Scan), and mice were considered diabetic if their blood glucose was >250 mg/dL on 2 consecutive days. Gdf15 knockout mice on a C57BL/6N background were generated as previously described (38). To induce diabetes, 13-week-old male Gdf15 knockout mice and WT C57BL/6N control mice (Charles River Laboratories) received multiple low-dose injections of streptozotocin (STZ: 40 mg/kg body weight per day for 5 days, prepared in 0.1 mmol/L sodium citrate at pH 4.5). To generate Gdf15^-/-Ins2^+/Akita (double mutant) mice, we crossed Gdf15 knockout mice with Ins2^+/Akita (Akita) mice (The Jackson Laboratory).

Quantitative Real-time PCR

Total RNA was extracted using the miRNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. RNA was reverse transcribed to cDNA using the first-strand cDNA synthesis kit (Roche). Quantitative real-time PCR was performed on a LightCycler 480 System (Roche) using SYBR Green (Thermo Fisher Scientific). All primers used are listed in Table 2. All samples were corrected for the 18S ribosomal subunit run as an internal standard.

Table 2.

Primers used in the study

Description	Sequences (5′-3′)
1.18s qPCR 5′ primer	AGTCCCTGCCCTTTGTACACA
2.18s qPCR 3′ primer	GATCCGAGGGCCTCACTAAAC
3.Human GDF15 qPCR 5′ primer	TCCCGGGACCCTCAGAGT
4.Human GDF15 qPCR 3′ primer	TCGTAGCGTTTCCGCAACT
5.Mouse Gdf15 qPCR 5′ primer	ACCCCGGTGGTTCTTATGC
6.Mouse Gdf15 qPCR 3′ primer	CAGGTCATCATAAGTCTGCAGTGA
7.Rat Gdf15 qPCR 5′ primer	GGAGGACTCGAACTCAGAACCA
8.Rat Gdf15 qPCR 3′ primer	TCGCACCTCTGGACTGAGTATC
9.Human GDF15 promoter 5′ primer	CACACGCGTCAGGCATGTAATCCCACCACCAA
10.Human GDF15 promoter 3′ primer-1	CACAGATCTGCGGCAGCCACGAGAGCA
11.Human GDF15 promoter 3′ primer-2	CACAGATCTGGCTGTGCAGGTTGCGGCTC
12.Human C/EBPβ qPCR 5′ primer	CAAGAAGACCGTGGACAAGCA
13.Human C/EBPβ qPCR 3′ primer	CGCACGGCGATGTTGTT
14.Mouse C/ebpβ qPCR 5′ primer	AGCGGCTGCAGAAGAAGGT
15.Mouse C/ebpβ qPCR 3′ primer	GGCAGCTGCTTGAACAAGTTC
16.Rat C/ebpβ qPCR 5′ primer	ACGAGCGGCTGCAGAAGA
17.Rat C/ebpβ qPCR 3′ primer	GGCAGCTGCTTGAACAAGTTC
18.Rat Gadd34 qPCR 5′ primer	GCTGGCCTGCCTACACTACAG
19.Rat Gadd34 qPCR 3′ primer	CCACCTCCCCAACTTTCTTATCT

Western Blotting

Protein extracts from WT and Akita cells were prepared using lysis buffer containing HEPES (50 mM), Nonidet P-40 (10%), sodium fluoride (100 mM), sodium pyrophosphate (10 mM), EDTA (4 mM), phenylmethanesulphonyl fluoride (1 mM), leupeptin (2 μM), activated sodium orthovanadate (2 mM), and okadaic acid (100 nM). Protein concentrations were measured by Pierce BCA protein assay (Thermo Fisher Scientific, Hudson, NH) and equal amounts of protein were loaded. Bands were visualized by ECL plus (GE Healthcare). The following antibodies were used: rabbit anti-GDF15 IgG (Bioss Antibodies, catalog no. bs-3818R-TR, RRID: AB_10857699) (39), mouse anti-beta-actin IgG (Cell Signaling Technology, catalog no. 3700S, RRID: AB_2242334) (40), horse anti-mouse IgG (Cell Signaling Technology, catalog no. 7076S, RRID: AB_330924) (41), and goat anti-rabbit IgG (Cell Signaling Technology, catalog no. 7074S, RRID: AB_2099233) (42).

Plasmid Construction, Transfection, and Luciferase Assay

The human GDF15 proximal promoter region was amplified by 2-round PCR using the primers listed in Table 2. The PCR product was subcloned into the Mlu І and Bgl II sites of the pGL3-Enhancer luciferase reporter plasmid (Promega). The reporter plasmid containing the human GDF15 promoter region was confirmed by sequencing. For luciferase assays, HEK293 cells were plated in 12-well plates and grown overnight to ~60% confluence. Cells were transfected with the reporter plasmid and the internal control plasmid pRL-TK (Promega) using DharmaFECT Duo transfection reagent (Horizon Discovery). Eight hours after transfection, the cells were changed to new medium containing 0.5 µM of TG or DMSO for another 16 hours. At 24 hours of transfection, firefly as well as Renilla luciferase activity were determined using the Dual Luciferase Assay Kit (Promega).

Apoptosis Assay

Apoptosis was analyzed by using the Cell Death Detection ELISAPLUS kit (Roche, catalog no. 11774425001, RRID: AB_2909412) (43) according to the manufacturer’s instruction and as previously described (44). In brief, INS-1 cells were seeded in 6-well plates and transfected with rat Gdf15 siRNA or scrambled control (Horizon Discovery) for 72 hours using DharmaFECT 1 transfection reagent (Horizon Discovery). For the last 16 hours of the transfection, the cells were treated with 0.1 µM of TG or DMSO. The cells were then lysed in 1 mL of lysis buffer and 20 µL of 5 time-diluted cell lysate was used for the apoptosis assay. The results were normalized to DNA content by using the Quant-iT PicoGreen dsDNA Reagent and Kit (Thermo Fisher Scientific). To detect apoptosis in mouse islets, 50 islets were treated with 1 µM of TG or DMSO for 24 hours. The islets were lysed in 100 µL of lysis buffer and 20 µL of cell lysate were used for the apoptosis assay.

ELISA

Serum GDF15 in mice and GDF15 protein levels in cells were measured by using 50 µL of diluted serum or cell lysate and the Mouse/Rat GDF15 Quantikine ELISA kit (R&D Systems, catalog no. MGD150, RRID: AB_2909411) (45) according to the manufacturer’s instructions. Mouse serum insulin levels were assessed by using the Ultra Sensitive Mouse Insulin ELISA Kit (Crystal Chem, catalog no. 90080, RRID: AB_2783626) (46) as previously described (47, 48).

Glucose Tolerance Test and Insulin Tolerance Test

Glucose tolerance tests and insulin tolerance tests were performed in Gdf15 knockout (KO) mice and control mice as previously described (33, 49). Briefly, age- and sex-matched mice were fasted for 4 hours and then given glucose (3 g/kg) or regular insulin (1 U/kg) by IP injection. Blood glucose was measured at 0, 15, 30, 60, 90, and 120 minutes.

Immunohistochemistry

Insulin staining was performed as previously described (33). The guinea pig anti-insulin antibody (Abcam, catalog no. ab7842, RRID: AB_306130) (50) and the Alexa Fluor 594 AffiniPure donkey anti-guinea pig IgG (Jackson ImmunoResearch, catalog no. 706-065-148, RRID: AB_2340451) (51) were used. After staining, the Vectashield Mounting Solution with 4′,6-diamidino-2-phenylindole (Vector Labs) was used for visualization of nuclei.

Statistical Analysis

The Student t test were used to calculate the significance of a difference between 2 groups. One-way ANOVA was used for temporal analyses of datasets containing 2 or more groups. Log-rank test was used to compare disease-free survival in 2 groups. A P value < 0.05 was used to indicate significant difference between groups.

RESULTS

Beta-cell GDF15 Is Induced by ER Stress

To test whether GDF15 expression responds to ER stress in beta cells, we treated INS-1 cells with TG. We found that both Gdf15 mRNA (Fig. 1A) and GDF15 protein (Fig. 1B) were significantly induced by TG. Of note, a similar induction of GDF15 expression was recapitulated in human islets (Fig. 1C). Next, we assessed GDF15 expression in Akita cells. These cells harbor a mutation in the Ins2 gene, which causes misfolding and accumulation of insulin in the ER, leading to spontaneous ER stress and beta-cell death (36). Gdf15 mRNA (Fig. 1D) and GDF15 protein (Fig. 1E and 1F) were also significantly increased in the Akita cells compared with WT cells. Pro-inflammatory cytokines have been shown to induce ER stress and contribute to beta-cell death in type 1 diabetes (52, 53). To test whether the expression of GDF15 is regulated by type 1 diabetes-associated cytokines, we treated rat INS-1 cells with a cytokine cocktail (TNFα, interferon γ, and IL-1β). We found that the levels of GDF15 mRNA and protein were also significantly induced by cytokine treatment (Fig. 1G and 1H). In addition, to investigate the mechanism of GDF15 regulation by ER stress, we cloned a fragment of the proximal promoter of human GDF15 (-1082 to -1) into a luciferase reporter plasmid (pGL3-Enhancer Vector). We found that TG treatment significantly induced luciferase activity driven by the human GDF15 promoter (Fig. 1I), suggesting that ER stress induces GDF15 at the transcriptional level. Furthermore, to test the effect in vivo, we measured the expression of Gdf15 in islets of NOD mice, whose beta cells have been shown to be exposed to ER stress during diabetes progression (1, 54). The results showed that Gdf15 mRNA levels are significantly increased in islets of diabetic NOD mice compared with nondiabetic controls (Fig. 1J). Together, these results suggest that the expression of beta-cell GDF15 is induced by ER stress.

Figure 1.

Effects of ER stress on GDF15 expression. (A) Gdf15 mRNA and (B) GDF15 protein expression by ELISA were assessed in INS-1 cells treated with 0.5 µM TG or DMSO for 16 hours. (C) GDF15 mRNA expression was assessed in nondiabetic isolated human islets treated with 1 µM TG or DMSO for 24 hours. (D) Gdf15 mRNA and (E) GDF15 protein expression by ELISA and (F) by Western blot were assessed in wild-type (WT) and Akita beta cells. (G) Gdf15 mRNA and (H) GDF15 protein expression by ELISA were measured in INS-1 cells treated with cytokine cocktail (0.5 ng/mL TNFα, 0.5 ng/mL IFNγ, and 0.1 ng/mL IL-1β) for 16 hours. (I) Luciferase activity driven by human GDF15 promoter was analyzed in HEK293 cells treated with 0.5 µM TG or DMSO for 16 hours. (J) Gdf15 mRNA expression levels were measured in islets of 10-week female nondiabetic NOD mice, 20-week female NOR mice, 20-week male non-diabetic NOD mice and 20-week female diabetic NOD mice. All data are shown as the mean ± standard error of the mean of at least 3 independent experiments or 3 mice; n = 5 human islet donors. *P < 0.05, **P < 0.01, and ***P < 0.001.

C/EBPβ Is Involved in the Induction of GDF15 Expression

Previous studies have shown that GDF15 can be regulated by several transcription factors including nuclear factor-κB (55), CHOP (56), P53 (57), and C/EBPβ (26) in non-beta cells. We therefore tested these factors in beta cells and found that only C/EBPβ was able to regulate beta-cell GDF15 in the context of ER stress. Specifically, although siRNA transfection led to significant knockdown of P53, Nfκb (P65 subunit), Chop, and C/ebpβ, even in the presence of TG-induced ER stress (Fig. 2A), only knockdown of C/ebpβ could blunt TG-induced Gdf15 in INS-1 cells (Fig. 2B). In addition, C/EBPβ expression was increased in human islets treated with TG (Fig. 2C) and Akita cells (Fig. 2D), which was also associated with elevated GDF15 expression (Fig. 1C and 1D). These findings suggest that the transcription factor C/EBPβ is involved in the induction of GDF15 in beta cells.

Figure 2.

Role of C/EBPβ in the regulation of GDF15 expression. INS-1 cells were transfected with siP53, siP65, siChop, siCebp/β, or scrambled control, followed by treatment with TG (0.5 µM for 5 hours) or DMSO and the expression of these transcription factors (A) and of Gdf15 (B) was measured. (C) Nondiabetic isolated human islets treated with TG (1 µM for 24 hours) or DMSO, and (D) wild-type (WT) and Akita beta cells were analyzed for C/EBPβ mRNA expression. All data are shown as the mean ± standard error of the mean of 3 independent experiments; n = 5 islet donors. *P < 0.05 and ***P < 0.001.

GDF15 Promotes ER Stress-induced Beta-cell Apoptosis

Next, we tested the possibility that GDF15 is not only induced by ER stress, but also plays a role in ER stress-mediated beta-cell apoptosis. Indeed, although TG treatment of INS-1 cells promoted apoptosis as expected, knockdown of Gdf15 significantly reduced ER stress-induced cell apoptosis (Fig. 3A). The activation of UPR is thought to be a cellular defense response to ER stress. However, prolonged activation of UPR also leads to cell apoptosis. To test whether GDF15 exerts its effects on beta-cell apoptosis via regulation of UPR, we analyzed the expression of CHOP, which is a common downstream factor of UPR (58, 59). Indeed, knockdown of Gdf15 effectively reduced TG-induced Gdf15 expression (Fig. 3B) and significantly decreased the TG-induced increase in Chop (Fig. 3C). Similarly, we discovered that knockdown of Gdf15 also significantly decreased TG-induced expression of Gadd34, which is a target of CHOP (60) (Fig. 3D). These results suggest that inhibition of GDF15 has beneficial effect on ER stress-induced beta-cell apoptosis, which is at least in part mediated by the regulation of CHOP signaling.

Figure 3.

Role of GDF15 in beta cell apoptosis and UPR. (A) INS-1 cells with knockdown of Gdf15 (siGdf15) or scrambled control followed by treatment with TG (0.1 µM for 16 hours) or DMSO were analyzed for apoptosis by Cell Death Detection ELISAPLUS. (B) Gdf15, (C) Chop, and (D) Gadd34 mRNA expression was measured in INS-1 cells with knockdown of Gdf15 followed by TG treatment (0.1 µM for 16 hours). All data are shown as the mean ± standard error of the mean of at least 3 independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001. N.S., not significant.

Gdf15 Deletion Has No Effect on Glucose Homeostasis Under Normal Conditions

To study the in vivo role of GDF15 in glucose homeostasis, we obtained Gdf15KO (38). Gdf15KO mice are healthy and fertile as previously described (38). In addition, Gdf15KO mice showed no difference in glucose (Fig. 4A), insulin (Fig. 4B), glucose tolerance (Fig. 4C and 4D), and insulin sensitivity (Fig. 4E and 4F) compared with control mice, all while having no detectable GDF15 in their serum (Fig. 4G). These findings are consistent with previous reports (56, 61). The results further suggest that deletion of Gdf15 has no effect on glucose homeostasis under normal conditions.

Figure 4.

In vivo effects of GDF15 deletion on glucose homeostasis under normal conditions. (A) Blood glucose, (B) serum insulin, (C-D) glucose tolerance test (IP: 3 g/kg), (E-F) insulin tolerance test (IP: 1 U/kg), and (G) serum GDF15 levels were measured in Gdf15 KO mice and control mice. All data are shown as the mean ± standard error of the mean of at least 6 mice. ***P < 0.001. N.S., not significant.

Gdf15 Deletion Protects Mice From STZ-induced Diabetes and Islets From ER Stress-induced Apoptosis

To test the in vivo role of Gdf15 in the context of diabetes, 13-week-old male Gdf15KO mice and WT controls were injected with multiple low-dose STZ. Although WT control mice developed overt diabetes (blood glucose >250 mg/dL), Gdf15KO mice maintained significantly lower blood glucose levels (Fig. 5A). Interestingly, although most beta cells were destroyed in the control mice, Gdf15KO mice still maintained healthy insulin-producing beta cells (Fig. 5B). In addition, Gdf15KO mice showed significantly higher serum insulin levels compared with controls (Fig. 5C). To test the effect of Gdf15 deletion on ER stress-induced apoptosis in islets, we treated islets isolated from Gdf15KO mice and WT control mice with TG. Although TG treatment significantly induced apoptosis in islets of control mice, this effect was completely abolished in islets of Gdf15KO mice (Fig. 5D). These results suggest that deletion of Gdf15 helps improve STZ-induced diabetes by maintaining functional beta cells and circulating insulin levels and helps prevent islets from ER stress-induced apoptosis.

Figure 5.

In vivo effects of Gdf15 deletion on STZ-induced diabetes and Akita diabetes. (A) Blood glucose, (B) islet immunohistochemistry, and (C) serum insulin in male Gdf15 knockout mice (n = 6) and control mice (n = 5) treated with multiple low-dose STZ injections (40 mg/kg/d) for 5 days. (D) Apoptosis analyzed by Cell Death Detection ELISAPLUS in primary mouse islets of Gdf15 KO mice and controls treated with TG (1 µM for 24 hours) or DMSO. (E) Kaplan-Meier survival curve of the diabetes incidence of female Akita mice (n = 6) and double mutant mice (n = 9) (diabetes is defined as blood glucose >250 mg/dL for 2 consecutive measurements). All data are shown as the mean ± standard error of the mean. Representative islet pictures are shown. *P < 0.05 and **P < 0.01.

Gdf15 Deletion Delays the Development of Diabetes in Akita Mice

To further test the in vivo role of Gdf15 in the context of an additional diabetes model, we crossed Gdf15KO mice with Ins2^+/Akita (Akita) mice, whose Ins2 gene harbors a mutation leading to severe beta-cell ER stress and apoptosis, to create Gdf15^-/-Ins2^+/Akita (double mutant) mice. Consistent with the early onset and very severe diabetes found in male Akita mice reported previously (58), Gdf15 deletion was not able to change the disease course in males. However, interestingly, although half of the female Akita mice developed overt diabetes at 3 weeks, all of the same-aged female double mutant mice remained nondiabetic (Fig. 5F). In addition, although all the female Akita developed severe diabetes at 7 weeks, more than one-half of the same-aged double mutant mice still remained nondiabetic (Fig. 5F). These results suggest that deletion of Gdf15 significantly delays the development of diabetes in female Akita mice.

Discussion

In summary, our results demonstrate for the first time that GDF15 plays an important role in ER stress-induced beta-cell death, reveal a novel C/EBPβ-GDF15-CHOP signaling cascade, and indicate that deletion of Gdf15 has beneficial effect on beta-cell survival and diabetes.

ER stress/UPR has been shown to play an important role in beta-cell loss in diabetes. However, many of the factors involved in this process are still unknown. Here, we discovered that ER stress induced a dramatic increase in GDF15 expression in human islets and beta cells. Of note, this GDF15 elevation was not restricted to 1 condition, but was found in response to different stimuli (TG, cytokines) and in different diabetes models (Akita, NOD). We also elucidated the molecular mechanisms involved and found that the regulation of GDF15 in human isles and beta cells occurs at the transcriptional level and involves C/EBPβ. Furthermore, downregulation of Gdf15 decreased Chop and ER stress-induced beta-cell apoptosis and helped improve or delay diabetes in 2 different models of diabetes.

Interestingly, a previous study showed that ablation of C/ebpβ could also alleviate ER stress-induced beta-cell death and diabetes (62), which is in alignment with our current findings. In addition, our results suggest that GDF15-mediated ER stress-induced beta-cell apoptosis is at least in part mediated via regulation of the proapoptotic factor CHOP. Interestingly, deletion of Chop has also previously been shown to decrease ER stress-induced beta-cell death and delay the development of diabetes in Akita mice (58). In fact, we observed that Gdf15 deletion also has protective effects in Akita mice. Moreover, given the identified role of CHOP-mediated ER stress in this process, it is tempting to speculate that Gdf15 deletion would also have beneficial effects in the context of lipid overload, which previously has been shown to induce beta-cell apoptosis via the same pathway of CHOP-mediated ER stress (63-65). Taken together with our current findings, this suggests the existence of a previously unappreciated signaling cascade (C/EBPβ–GDF15-CHOP) that contributes to beta-cell death in the context of ER stress and diabetes.

Although we cannot rule out the possibility that deletion of Gdf15 has additional beneficial effect on whole body glucose homeostasis in peripheral tissues such as liver and adipose tissue, it is important to note that the glucose tolerance tests and insulin tolerance tests remained unaltered in Gdf15 knockout mice and the phenotype only became apparent in the context of beta-cell stress such as STZ injections and Akita mutation. In addition, isolated islets of Gdf15KO were also protected ex vivo from ER stress-induced apoptosis, supporting the notion of an islet effect. This suggests that inhibition of beta-cell GDF15 may represent a novel strategy to help protect against beta-cell death and diabetes.

So far, the role of GDF15 in diabetes has remained controversial. Although a recent report suggested that administration of exogenous GDF15 could delay the development of diabetes in NOD mice (32) and decrease cytokine-induced beta-cell death, previous work has found that diabetes is actually associated with increased GDF15 levels (13-18) and that elevation of GDF15 promotes cell apoptosis (23-29). Our studies now demonstrate that downregulation of endogenous, intracellular GDF15 has beneficial effects in the context of diabetes and protects beta cells from ER stress-induced apoptosis. Of note, these beneficial effects were observed even in the face of unmeasurably low circulating GDF15 levels in the Gdf15 knockout mice, suggesting that low serum levels are not necessarily detrimental. This is consistent with the notion that the expression of the only known GDF15 receptor, GDNF Family Receptor Alpha Like, has been reported to be restricted to the brain (19-22) and therefore would not be anticipated to contribute to any peripheral systemic functions caused by circulating GDF15.

Taken together, we found that intracellular GDF15 contributes to ER stress-induced beta-cell death and that deletion of GDF15 can prevent or delay diabetes development in different mouse models. Hence, inhibition of intracellular GDF15 may represent a novel strategy to promote beta-cell survival and treat diabetes. In fact, we recently demonstrated that the principle of targeting beta-cell expression of a detrimental factor can be very promising, both by pursuing novel drug discovery using small molecule screening (47) and by repurposing existing drugs that then can be used more readily in clinical trials (66).

Obviously, there are also some limitations to our study. Although we were able to demonstrate that like GDF15, the transcription factor C/EBPβ is increased in Akita beta cells and in response to TG in human islets and INS-1 and that C/EBPβ knockdown completely blunted the GDF15 increase, we could not show direct C/EBPβ binding to the GDF15 promoter by chromatin immunoprecipitation because of the unavailability of a functioning C/EBPβ chromatin immunoprecipitation antibody. In addition, although our results clearly demonstrate that GDF15 deletion affects beta-cell survival and glucose homeostasis in the context of ER stress, we cannot rule out any contributions from GDF15 deficiency in other metabolically active tissues because we used a whole-body Gdf15 knockout mouse. Future studies using a beta-cell-specific Gdf15 knockout will therefore address the tissue specific effects of GDF15.

Glossary

Abbreviations

ER

endoplasmic reticulum

GDF15

growth differentiation factor 15

NOD

nonobese diabetic

STZ

streptozotocin

TG

thapsigargin

UPR

unfolded protein response

WT

wild-type

Funding

This work was supported by R01DK078752 and Human Islet Research Network U01DK120379 granted to A.S. and UAB Diabetes Research Center Pilot & Feasibility Award (NIH P30 DK079626) granted to G.X. Human pancreatic islets and/or other resources were provided by the NIDDK-funded Integrated Islet Distribution Program (IIDP) at City of Hope, NIH grant #2UC4DK098085.

Author Contributions

G.X. conceived the project, designed, performed, and analyzed the experiments, wrote the manuscript, and is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. J.C. isolated the mouse islets and helped with the mouse experiments. S.J. and T.B.G. helped with the mouse experiments. S.R. and A.K. provided the Akita and wild-type cells and edited the manuscript. E.L.G.-L. and S.J.L. provided the Gdf15KO mice. A.S. supported the work and revised the manuscript.

Conflict of Interest

No potential conflicts of interest relevant to this article were reported.

Data Availability

The data sets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.