Interleukin 6 protects pancreatic β cells from apoptosis by stimulation of autophagy

Abstract

IL-6 is a pleiotropic cytokine with complex roles in inflammation and metabolic disease. The role of IL-6 as a pro- or anti-inflammatory cytokine is still unclear. Within the pancreatic islet, IL-6 stimulates secretion of the prosurvival incretin hormone glucagon-like peptide 1 (GLP-1) by α cells and acts directly on β cells to stimulate insulin secretion in vitro. Uncovering physiologic mechanisms promoting β-cell survival under conditions of inflammation and stress can identify important pathways for diabetes prevention and treatment. Given the established role of GLP-1 in promoting β-cell survival, we hypothesized that IL-6 may also directly protect β cells from apoptosis. Herein, we show that IL-6 robustly activates signal transducer and activator of transcription 3 (STAT3), a transcription factor that is involved in autophagy. IL-6 stimulates LC3 conversion and autophagosome formation in cultured β cells. In vivo IL-6 infusion stimulates a robust increase in lysosomes in the pancreas that is restricted to the islet. Autophagy is critical for β-cell homeostasis, particularly under conditions of stress and increased insulin demand. The stimulation of autophagy by IL-6 is regulated via multiple complementary mechanisms including inhibition of mammalian target of rapamycin complex 1 (mTORC1) and activation of Akt, ultimately leading to increases in autophagy enzyme production. Pretreatment with IL-6 renders β cells resistant to apoptosis induced by proinflammatory cytokines, and inhibition of autophagy with chloroquine prevents the ability of IL-6 to protect from apoptosis. Importantly, we find that IL-6 can activate STAT3 and the autophagy enzyme GABARAPL1 in human islets. We also see evidence of decreased IL-6 pathway signaling in islets from donors with type 2 diabetes. On the basis of our results, we propose direct stimulation of autophagy as a novel mechanism for IL-6-mediated protection of β cells from stress-induced apoptosis.—Linnemann, A. K., Blumer, J., Marasco, M. R., Battiola, T. J., Umhoefer, H. M., Han, J. Y., Lamming, D. W., Davis, D. B. Interleukin 6 protects pancreatic β cells from apoptosis by stimulation of autophagy.

Diabetes mellitus is characterized by elevated blood glucose due to insufficient insulin production. The pathophysiology of diabetes hinges on failure of the insulin-producing β cells, which is at least partially due to increased β-cell apoptosis (1). However, the processes that lead to apoptosis in the development of diabetes are not yet fully understood. Identifying therapeutic targets that could prevent β-cell apoptosis would therefore allow prevention of diabetes and delay the progression to insulin dependence.

One potential player in this pathway is the cytokine IL-6. IL-6 can be produced by and act on multiple tissues throughout the body (2). However, during rest, up to 35% of IL-6 is produced by adipose tissue (3). Muscle-produced IL-6 is naturally increased during exercise (4), where it is linked to improved insulin sensitivity and nutrient availability (5, 6). IL-6 may also be elevated in obesity and type 2 diabetes (7, 8), likely originating from adipose tissue. Under these conditions, its role in islet function and blood glucose homeostasis is still unclear, though recent work indicates that IL-6 may function similar to leptin in the CNS to suppress feeding and improve glucose tolerance (9). Clues to the role of IL-6 in obesity and diabetes can be derived from knockout studies. Genetic deletion of IL-6 in the mouse leads to obesity and glucose intolerance after 6 to 9 mo of age (10). When these mice are fed a high-fat diet they have impaired α-cell expansion and insulin secretion, leading to glucose intolerance (11). Collectively, these results suggest that IL-6 is important for islet adaptation and survival in obesity.

The IL-6 signaling cascade is activated by the binding of IL-6 to the IL-6 specific receptor, IL-6 receptor α (IL-6Rα), which causes recruitment of gp130 (12). Dimerization of the IL-6/IL-6Rα/gp130 complex enables activation of the receptor and stimulates second messenger signaling through both the Janus kinase/signal transducer and activator of transcription (JAK/STAT) and MAPK/ERKs pathways (2). In the JAK/STAT signaling cascade, IL-6 robustly activates the transcription factor signal transducer and activator of transcription 3 (STAT3) to control diverse functions such as cell cycle progression and apoptosis (13). For this reason, STAT3 phosphorylation is widely used as a marker of IL-6 activity. It was recently discovered that STAT3 is also able to regulate the process of autophagy through both its localization in the cell (14) and regulation of transcription (15). Interestingly, like mice lacking Il6, Stat3 β-cell-specific knockout mice also become obese and exhibit impaired islet architecture and insulin secretory defects (16). This raises the question of whether IL-6 plays a role in islet autophagy through STAT3 signaling.

Autophagy is a regulated process where specific cellular contents or proteins are segregated within the cell, then degraded. At least 3 distinct autophagy pathways have been described: microautophagy, macroautophagy, and chaperone mediated autophagy (17). Herein, the term “autophagy” will be used to reference the process of macroautophagy, which involves the sequestering of proteins in membrane bound vesicles called autophagosomes that fuse with acidic lysosomes to degrade the autophagosome contents. Autophagy is normally stimulated in nutrient limiting conditions to liberate amino acids for new protein synthesis as well as generate energy (18, 19). However, in the β cell, excess nutrients from high-fat diet feeding actually stimulate autophagy (20). Further, loss of β-cell autophagy in mouse models through tissue- specific deletion of the autophagy enzyme, Atg7, leads to reduced β-cell mass in part due to increased apoptosis (20, 21). When fed a high-fat diet, these mice do not have adequate β-cell mass compensation and develop diabetes (20).

A critical role for autophagy in the adaptation to obesity is further supported by the observation that stimulation of autophagy can reduce the incidence of endoplasmic reticulum stress-induced diabetes in mice (22). Considering that IL-6 can directly activate downstream signaling in the β cell to regulate insulin secretion and prevent cell death (23–25), we hypothesized that IL-6 acts directly on the pancreatic β cell to promote autophagy. Thus, we set out to determine the role of the IL-6 activated signaling cascade in β-cell survival with respect to autophagy.

MATERIALS AND METHODS

Antibodies and reagents

Phospho-Stat3 Y705 (9145), p-Stat3 S727 (9134), total Stat3 (4904), β-actin (3700), LC3A/B (12741), pAMPK Thr172 (4188), AMPK (5831), pS6K Thr309 (9234), pS6 S240/244 (2215), total S6 (2217), pAkt S473 (4060), total Akt (2965 or 2920), Atg7 (8558), and Atg4c (5262), and cleaved caspase 3 (9661), as well as horseradish peroxidase–conjugated secondary antibodies, were purchased from Cell Signaling Technology (Danvers, MA, USA) and IRDye secondary antibodies were from Li-Cor Biosciences (Lincoln, NE, USA). LC3A-AF488 conjugate (ab185036) and Lamp1-Cy3 conjugate (ab67283) antibodies were purchased from Abcam (Cambridge, MA, USA). Recombinant mouse or human IL-6 for use in cell/islet culture was from Miltenyi Biotec (San Diego, CA, USA), and carrier-free recombinant mouse IL-6 for use in the osmotic pumps was from R&D Systems (Minneapolis, MN, USA). Triciribine (Akt inhibitor V) was purchased from EMD Millipore (Billerica, MA, USA), rapamycin was from LC Labs (Woburn, MA, USA), chloroquine and active human glucagon-like peptide 1 (GLP-1) were from Tocris Bioscience (Bristol, United Kingdom), and bafilomycin A1 was from Cayman Chemicals (Ann Arbor, MI, USA). Mouse TNF-α and mouse IL-1β were from Miltenyi Biotec and IFN-γ was from Thermo Fisher Scientific (Waltham, MA, USA).

Cell/islet culture and treatments

INS-1 cells were grown in RPMI 1640 containing 11.1 mM glucose, 10% fetal bovine serum, and 1% antibiotic/antimycotic. Human islets were obtained through the Integrated Islet Distribution Program [National Institutes of Health (NIH), National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD, USA] and were cultured overnight to confirm viability and sterility before any treatments. Islets were cultured in uncoated Petri dishes with RPMI 1640 containing 8 mM glucose, 10% heat inactivated fetal bovine serum, and 1% penicillin/streptomycin. For IL-6 treatments, cells were treated with 200 ng/ml recombinant mouse IL-6 for 24 h and islets with 200 ng/ml recombinant human IL-6 for 96 h. For rapamycin treatment, cells were treated with 10 to 100 nM rapamycin for 24 h, and in certain experiments pretreated for 1 h with rapamycin before IL-6 treatment. To inhibit Akt, we used 1 μM triciribine for 24 h. In IL-6 + triciribine treatment conditions, cells were pretreated for 1 h with triciribine then IL-6 was added for the remaining 23 h.

Analysis of protein and RNA

For protein analysis, whole-cell lysate was prepared using a 50 mM HEPES buffer containing 1% Nonidet P-40 as well as protease and phosphatase inhibitors, then run on 4–20 or 10% SDS-PAGE gels (Bio-Rad, Hercules, CA, USA). Protein was transferred to polyvinylidene fluoride membranes, and nonspecific binding was blocked by incubation with either Tris-buffered saline with 0.1% Tween 20/5% milk or Li-Cor blocking buffer. All primary antibodies were used according to the manufacturers’ recommendations. ECL (GE Healthcare Life Sciences, Little Chalfont, United Kingdom) was used to develop membranes after incubation with horseradish peroxidase–conjugated secondary antibodies, and blots were imaged using a GE ImageQuant LAS 4000 imaging station. Membranes incubated with IRDye secondary antibodies were briefly dried and imaged on the Li-Cor Odyssey CLx. Protein band intensities were quantified by ImageJ software (Image Processing and Analysis in Java; NIH; http://imagej.nih.gov/), and unpaired Student’s t tests were used to determine significance at a level of P < 0.05 in GraphPad Prism (GraphPad Software, La Jolla, CA, USA).

Total RNA was isolated from cells using the Qiagen RNeasy kit. RNA quantity and quality was measured using a Nanodrop 2000 spectrophotometer and 100 to 500 ng of RNA per sample was used to make cDNA with an Applied Biosystems High Capacity cDNA synthesis kit. Quantitative PCRs using Power SYBR Master Mix (Thermo Fisher Scientific) were run on a StepOne Plus System (Thermo Fisher Scientific). All transcripts were normalized to β actin, and Student’s t tests were used to determine significance at a level of P < 0.05 in GraphPad Prism; these values are reported in the text and figures. Significance values of human autophagy genes were further subjected to a multitest correction using a 2-stage linear step-up procedure of Benjamini, Krieger, and Yekutielli with Q = 5%. This resulted in an adjusted Q value of 0.002 for GABARAPL1 and 0.052 for LAMP1. Primer sequences are as follows: rat Atg7 (forward) 5′-GGGCTATTACTACAATGGTGACT-3′, rat Atg7 (reverse) 3′-CTCAAGTGTGTTGGTGTTGTG-5′; rat Atg4c (forward) 5′-CTGCTTGGGACAACATGAATTATAG-3′, rat Atg4c (reverse) 3′-GGCAACACCTTGCTTTCATC-5′; human IL-6RA (forward) 5′-GTGCTCTTGGTGAGGAAGTT-3′, reverse 3′-TTCTGGGACTCCTGGGAATA-5′; human GABARAPL1 (forward) 5′-CTGACCTTACTGTTGGCCAGT-3′, (reverse) 3′-TGCAACCAGAACCATTACCTCA-5′; human LAMP1 (forward) 5′-GTGTCACGAAGGCGTTTTCA-3′, (reverse) 3′-AGCAGACACTCCTCCACAGA-5′; human IL-6 (forward) 5′-ACAACCTGAACCTTCCAAAGA-3′, (reverse) 3′-TCAGCAGGCTGGCATTT-5′.

Osmotic minipump implants

Mice were housed in facilities with a standard light–dark cycle and provided with free access to water. All protocols were approved by the University of Wisconsin and William S. Middleton Memorial Veterans Affairs Institutional Animal Care and Use Committees to meet acceptable standards of humane animal care. Alzet osmotic minipumps (Model 1007D; Durect Corp, Cupertino, CA, USA) were preloaded with either 16 μg/ml carrier free recombinant mouse IL-6 (R&D Systems, 406-ML/CF) diluted in sterile saline, or sterile saline. These osmotic pumps continuously infused IL-6 into the mice for 1 wk at a rate of 0.5 μl/h. Prefilled pumps were placed in sterile saline for ∼1 to 2 h before surgeries. The pumps were implanted subcutaneously in the subscapular region of 12- to 15-wk-old male C57BL/6J mice. Blood glucose and weight from 4 to 6 h unfed mice were measured before surgery and again at 1, 4, and 7 d after surgery. Before surgery, blood was collected by retro-orbital bleed in unanesthetized mice to also allow for measurement of insulin. At 1 and 4 d postsurgery, blood was collected by tail nick. Intraperitoneal glucose tolerance tests (2 mg/kg glucose) were performed on d 7 after 4 h unfed and immediately before euthanasia by tribromoethanol injection followed by terminal bleed via cardiac puncture. As circulating IL-6 has a circadian rhythm (26), all animals were humanely killed between 2 and 4 pm for blood collection and tissue harvesting to ensure comparable results. Blood was collected from unanesthetized mice at indicated time points via retroorbital bleeds for glucose and insulin measurements. IL-6 measurements were taken from terminal bleed samples. Five pairs of mice were implanted with pumps and analyzed. Mice that were not implanted with osmotic pumps were used for comparison of autophagy. These mice were euthanized at 12 to 15 wk of age and cryopreserved pancreas samples were processed in the same manner as those for the pump implant mice.

Immunofluorescent analysis of cells and tissues

For immunofluorescence, INS-1 cells were grown on coverslips and fixed with ice-cold 100% methanol before staining according to the antibody manufacturer’s protocol. Experiments were repeated in triplicate and representative images are shown. For the mouse tissue analysis, after euthanizing each mouse, the pancreas was immediately removed and fixed in 10% formalin for 24 h. Each pancreas was then dehydrated in 30% sucrose buffer for 24 h and cryopreserved in OTC until sections were cut on a Leica cryostat (Leica, Wetzlar, Germany). Background was reduced using Dako Protein Block (Agilent, Santa Clara, CA, USA), and sections were incubated with antibodies diluted according to manufacturer’s instructions in Dako Antibody Diluent (Agilent). All slides (pancreas and INS-1 cell staining) were mounted with ProLong Diamond-containing DAPI (Thermo Fisher Scientific), and images were obtained with a Nikon A1Rs confocal microscope (Nikon, Tokyo, Japan) with a ×60 oil objective. For pancreas sections, autophagosomes and lysosomes were counted from 3 to 5 islets in each of the 5 biologic replicate pairs with ImageJ software using maximum intensity projections from z-stack images. Before counting autophagosomes/lysosomes, we applied a background reduction threshold that was standardized across all samples analyzed. Counting of punctate foci greater than 10 pixels in size with fluorescence intensity above a standardized minimum signal intensity threshold was automated in ImageJ. The total number of autophagosomes or lysosomes in each islet were then divided by the total number of cells in each islet (identified by counting nuclei in each islet with ImageJ) to normalize for differences in islet size. Data were plotted and analyzed by GraphPad Prism.

Measurement of serum glucose, insulin, and IL-6

Glucose was measured using a Bayer Contour Next (Bayer, Pittsburgh, PA, USA) glucose monitor with the appropriate test strips recommended by the manufacturer. Insulin was measured using an Ultra-Sensitive Mouse Insulin ELISA kit according to the manufacturer’s instructions (Crystal Chem, Downers Grove, IL, USA). IL-6 was measured with a Quantikine Mouse IL-6 ELISA according to the manufacturer’s instructions (R&D Systems). Data were plotted by GraphPad Prism, and unpaired 2-tailed t tests were used to determine differences between vehicle and IL-6-treated mice.

Apoptosis analysis

For apoptosis analysis, INS-1 cells were plated at a density of 20,000 cells/well in a 96-well tissue culture dish with an optically clear surface. Two days later, cells were pretreated for 6 h with GLP-1 (50 ng/ml) or 4 h with IL-6 (200 ng/ml) alone or in the presence of either chloroquine (20 μM) or bafilomycin A1 (200 nM). A proinflammatory cytokine cocktail consisting of TNF-α (50 ng/ml), IL-1β (10 ng/ml), and IFN-γ (50 ng/ml) was then added and cells were cultured for an additional 24 h. Cells were then analyzed for apoptosis with a Caspase-Glo 3/7 Assay (Promega, Madison, WI, USA) according to the manufacturer’s instructions, and luminescence was measured on a Tecan M1000 Pro plate reader (Tecan, Zurich, Switzerland). Data were plotted by GraphPad Prism, and unpaired Student’s t tests were used to determine significance.

RESULTS

IL-6 stimulates β-cell autophagy in vitro

IL-6 activates a JAK/STAT signaling cascade to modulate intracellular signaling. Activation of STAT3 by phosphorylation at tyrosine 705 (Y705) enables dimerization and entry into the nucleus where STAT3 can then regulate gene transcription. We find that IL-6 is able to stimulate phosphorylation of STAT3 Y705 in INS-1 β cells (Fig. 1A; P < 0.0001). STAT3 can also be phosphorylated at serine 727 (S727), and this phosphorylation may be cell or context dependent to confer activation of specific promoters (27–29). We find that in the β cell, IL-6 is also able to stimulate robust phosphorylation of Stat3 S727 (Fig. 1A; P = 0.0006).

Figure 1.

IL-6 activates autophagy in β cells. A) INS-1 cells were treated with 200 ng/ml recombinant mouse IL-6 or were serum starved for 24 h; then Stat3 activation by phosphorylation was measured by Western blot analysis. B) INS-1 cells were treated with 200 ng/ml recombinant mouse IL-6 over time course from 0 to 24 h (0, 1, 2, 4, 6, 24 h); then Stat3 activation by phosphorylation, LC3, and p62 were measured by Western blot analysis. C) INS-1 cells were grown on coverslips, serum starved or treated for 24 h with IL-6, then analyzed by immunofluorescence for autophagosome formation by LC3A/B staining (red). Nuclei are stained with DAPI (blue). Insets show zoomed-in view of autophagosomes in IL-6 and serum-starved conditions. Quantifications in A and B are result of at least 3 separate experiments. *P < 0.05, ***P < 0.001.

Cytosolic STAT3 inhibits autophagy by sequestering specific proteins important for this process (14), and we hypothesized that IL-6 might promote autophagy through its actions on STAT3. Therefore, Y705 phosphorylation of STAT3 by IL-6 in the β cell should activate STAT3 and thereby promote autophagy. We analyzed this by looking at the autophagy mediator LC3, a protein that is critical for autophagosome formation. Conversion of LC3 from its long form (ProLC3) to a shorter form (LC3-I) and subsequent fatty acid conjugation (LC3-II) is essential for its entry into the autophagosome membrane (30). STAT3 is rapidly activated within 1 h of IL-6 treatment and this activation is sustained for at least 24 h (Fig. 1B). Coincident with STAT3 phosphorylation, we observed that IL-6 treatment of β cells stimulates a rapid 2.5-fold increase in LC3-II relative to β actin control after 1 h (Fig. 1B; P = 0.01) This is followed by a less robust induction in both LC3-I and -II at 2, 4, and 6 h of treatment (Fig. 1B). By 24 h of IL-6 treatment, there is again a 2.5-fold increase in LC3-II protein overall (P = 0.02) as well as relative to LC3-I, reflecting fatty acid conjugation of LC3-I (Fig. 1B; P = 0.03). The observed cycling of LC3-II levels with a peak at 1 h and again at 24 h of sustained IL-6 exposure suggests that autophagic flux is intact, and this is supported by our observation that p62 levels track with LC3-II levels over time (Fig. 1B). We also observed robust incorporation of LC3 into autophagosome vesicles following IL-6 treatment to a similar degree as serum starvation, a known stimulator of autophagy (Fig. 1C). Taken together, we show that IL-6 stimulates autophagy in the β cell.

Autophagy is dependent on a series of enzymes that are critical for various steps of the process. The autophagy enzyme, ATG4, cleaves ProLC3 to LC3-I. We find that IL-6 is able to stimulate transcription of Atg4c (P = 0.01; Fig. 2A). However, ATG4C protein production is not consistently stimulated by IL-6 (Fig. 2B). It is unclear why protein levels do not appear to be robustly affected by IL-6, though this may be due to assessment at a single timepoint (24 h) and/or protein turnover rates since induction of a dynamic process such as autophagy can contribute significantly to flux in global protein levels over time. We also analyzed the effects of IL-6 on a second autophagy enzyme, ATG7. ATG7 is responsible for the fatty acid conjugation that converts LC3-I to LC3-II. This activity is required for LC3 entry into the autophagosome membrane and subsequent closure of the vesicle. Tissue specific loss of Atg7 in the β cell leads to reduced β-cell mass in part due to increased apoptosis (21). We find that IL-6 stimulates both mRNA (P = 0.03) and protein expression (P = 0.03) of ATG7 (Fig. 2C, D).

Figure 2.

IL-6 stimulates autophagy enzyme production. INS-1 cells were treated with IL-6 for 24 h; then Atg4c transcript (A), ATG4C protein (B), Atg7 transcript (C), and ATG7 protein (D) were measured. *P < 0.05 (n ≥ 3).

IL-6 stimulates β-cell autophagy in vivo

We therefore hypothesized that IL-6 could stimulate β-cell autophagy in vivo. To determine this, we implanted lean C57BL/6J mice with osmotic pumps that continuously infused IL-6 for a period of 7 d. We chose infusion over periodic injections to more closely mimic the chronic IL-6 elevation observed in obesity as opposed to acute elevations during exercise. Mice implanted with pumps containing recombinant mouse IL-6 had significantly higher circulating IL-6 than the mice implanted with saline-containing pumps (Fig. 3A). The increase in circulating IL-6 had no effect on weight or fasting blood glucose over the course of the 7 d (Fig. 3B, C). There was also no effect on glucose tolerance or glucose-stimulated insulin release after 7 d of IL-6 infusion (Fig. 3D, E).

Figure 3.

IL-6 activates autophagy in vivo. A) IL-6 levels in plasma from mice continuously infused with vehicle or IL-6 for 7 d. B, C) Body weight (B) and fasting blood glucose (C) measurements at d 0, 1, 4, and 7 relative to pump implant surgery. D, E) Glucose (D) and insulin (E) measurements after 2 mg/kg glucose delivered intraperitoneally in mice. F) Representative images showing nuclei (DAPI, blue), LC3A (autophagosome, green) and Lamp1 (lysosome, red) staining in cryopreserved pancreas sections from mice with no pump (a), saline pump (39 pg/ml IL-6) (b), and IL-6 pump (102 pg/ml IL-6) (c). Insets are representative of white outlined box on LC3 and overlaid images to show higher magnification of punctate staining indicative of autophagosomes and increased lysosomes in islet β cells. G, H) Autophagosomes (G) and lysosomes (H) from all mice implanted with pumps were quantified, and linear regression relative to IL-6 levels was analyzed. *P < 0.05 (n = 5 mouse pairs).

Because of the mildly elevated IL-6 levels in the mice with saline pumps, we included another control group of mice that did not have any pump implants. Pancreata from all animals were fixed, cryopreserved, and analyzed for autophagosome and lysosome formation to measure autophagy and autophagic flux. IL-6 stimulated only a modest qualitative increase in autophagosome formation in response to elevated levels of IL-6. This is visualized in representative images in Fig. 3F as an increase in LC3 incorporation into punctate autophagosomes, in contrast to the more diffuse staining in animals with no implanted pump (Fig. 3Fa, LC3A staining; inset, higher-magnification view). However, the autophagosomes were qualitatively less brightly stained in sections from animals implanted with IL-6 pumps (cf. Fig. 3Fb, c). Lamp1 was used as a marker for lysosomes, and we found that IL-6 stimulated a robust increase in lysosomes within the islet (Fig. 3F, Lamp1 staining and overlay inset). Interestingly, IL-6 also stimulated a dramatic decline in lysosomes in the exocrine pancreatic tissue.

Because we observed measurable IL-6 levels even with saline administration, we were unable to directly determine the impact of IL-6 on β-cell autophagy by comparing saline and IL-6-treated mice. Therefore, we quantified islet autophagy using autophagosome and lysosome numbers as a function of measured IL-6 levels across both treatment groups. There was no significant correlation between IL-6 levels and autophagosome number (Fig. 3G). However, we observed a significant positive correlation between the average number of lysosomes per cell and circulating IL-6 levels (Fig. 3H; R² = 0.4247, P = 0.0412). Collectively, these results are consistent with our hypothesis that circulating IL-6 may stimulate autophagy in the β cell.

IL-6 regulates autophagy through multiple complementary mechanisms

After establishing that IL-6 stimulates autophagy in INS-1 cells and in islets in vivo, we wanted to determine the mechanism. Autophagy is stimulated by the energy sensor AMPK. We find that in INS-1 β cells, IL-6 stimulates AMPK by inducing a 1.5-fold increase in phosphorylation of the α-catalytic subunit (Fig. 4A; P = 0.0075). AMPK promotes autophagy in part through the inhibition of the mammalian target of rapamycin (mTOR) complex 1 (mTORC1), a well-characterized inhibitor of autophagy (31–33). As expected, IL-6 treatment of INS-1 cells also reduces mTORC1 activity as measured by reduced phosphorylation of S6 (Fig. 4B; P = 0.017). A second mTOR protein complex, mTORC2, has recently been reported to stimulate autophagy and is known to play a role in pancreatic β-cell homeostasis (34, 35). mTORC2 is the principal protein kinase for Akt S473, which is a readout for its activity (36). Akt S473 phosphorylation is robustly stimulated ∼2.5 fold after IL-6 treatment in β cells (Fig. 4C; P = 0.02). Therefore, IL-6 activates AMPK, inhibits mTORC1, and activates Akt in β cells; together, these effects individually and in combination promote autophagy.

Figure 4.

IL-6 regulates autophagy through multiple mechanisms. A–C) Analysis of AMPK activation (A), S6 phosphorylation (B), and Akt phosphorylation (C) by Western blot analysis in INS-1 cells treated with 200 ng/ml IL-6 for 24 h. Representative blots are shown along with quantification of at least 3 biologic replicates. *P < 0.05, **P < 0.01. D) Analysis of S6 phosphorylation, STAT3 activation, LC3, and p62 in presence of rapamycin and/or IL-6. Rapamycin concentration was 10 nM (lanes 2 and 5) and 100 nM (lanes 3 and 6). E) Akt phosphorylation, STAT3 activation, LC3, and p62 in presence of IL-6 and/or 1 μM Akt inhibitor triciribine. Representative blots are shown from at least 3 replicates.

It has been reported that phosphorylation of STAT3 S727 is mTORC1 dependent in neuroblastoma cells (37), a result that would be incongruent with our findings that IL-6 both inhibits mTORC1 (Fig. 4B) and phosphorylates STAT3 S727 (Fig. 1A) in the β cell. To determine the dependence of IL-6-mediated STAT3 S727 phosphorylation on mTORC1 in the β cell, INS-1 cells were treated with rapamycin to repress mTORC1 activity. This treatment resulted in a complete loss of mTORC1 activity, as shown by loss of S6 phosphorylation (Fig. 4D). Despite loss of mTORC1 activity, IL-6 is still able to stimulate phosphorylation of STAT3 at both Y705 and S727 in the β cell (Fig. 4D). Rapamycin and IL-6 together stimulate a decrease in both LC3-I and -II relative to either rapamycin or IL-6 alone after 24 h of treatment (Fig. 4D). We also observe that p62 protein levels are further decreased in the presence of both IL-6 and rapamycin, suggesting that the decrease in LC3 is due to additive stimulation of autophagy.

As STAT3 S727 phosphorylation did not appear to be mediated by mTORC1 under our conditions, we instead hypothesized that in the β cell, IL-6-mediated STAT3 S727 phosphorylation instead relies on Akt signaling. To test this hypothesis, we treated INS-1 cells with the Akt inhibitor triciribine and analyzed STAT3 activation in response to IL-6. As shown in Fig. 4E, the Akt inhibitor reduced Akt S473 phosphorylation, and this inhibition of Akt significantly impaired the ability of IL-6 to stimulate phosphorylation of STAT3 S727. Inhibition of Akt activity does not impair the IL-6-mediated stimulation of STAT3 Y705 (data not shown). Akt inhibition does, however, correlate with reduced LC3-I, LC3-II, and p62 levels after 24 h of treatment (Fig. 4E). This may be due to increased autophagic flux, but could also reflect a more general requirement of Akt activity in cell survival and function (38). Alternatively, the decrease in both LC3-I and -II levels with Akt inhibition relative to IL-6 treatment alone could be due to reduced activity of ATG4C in the cleavage of ProLC3 to LC3-I. In conclusion, we find that IL-6-mediated phosphorylation of Stat3 S727 is dependent on its activation of Akt activity in β cells, and is not dependent on mTORC1. The Stat3 S727 phosphorylation and/or Akt signaling is also likely important for the full effect of IL-6 on LC3-II formation.

We also wanted to determine how IL-6 might regulate the autophagy enzymes ATG4C and ATG7 in the β cell to promote autophagy, as demonstrated in Fig. 2. To determine whether mTORC1 inhibition was involved, we treated β cells with rapamycin to inhibit mTORC1 activity as shown in Fig. 4D. Although rapamycin is unable to independently activate STAT3 (Fig. 4D), it independently stimulates transcription of Atg7 (Fig. 5A; P = 0.0054) but does not significantly alter ATG7 protein levels (Fig. 5B).

Figure 5.

IL-6 selectively uses mTOR/Akt signaling to modulate specific autophagy enzymes. A, B) Analysis of Atg7 and Atg4c transcription (A) and ATG4C and ATG7 protein levels (B) after rapamycin treatment of INS-1 cells. Rapamycin concentration was 10 nM (lanes 2 and 5) and 100 nM (lanes 3 and 6). C, D) Atg4c transcription (C) and ATG4C and ATG7 protein levels (D) after 1 μM Akt inhibitor (triciribine) and 200 ng/ml IL-6 treatments. *P < 0.05, **P < 0.01 (n ≥ 3). E) Model for stimulation of autophagy by IL-6.

Atg4c transcription is not affected by rapamycin treatment, suggesting that its transcriptional regulation by IL-6 is independent of mTORC1 inhibition (Fig. 5A). The ATG4C promoter also contains a STAT3 binding site in its promoter [site identified using MotifMap (39)], whereas Atg7 does not, suggesting that transcription of these genes may be differentially regulated by IL-6. Since Akt is important for full activation of STAT3 in the β cell (Fig. 4E), we analyzed Atg4c activation in the presence of Akt inhibitor to determine whether its transcriptional regulation is dependent on Akt-driven STAT3 activation. We find that inhibition of Akt activity rendered IL-6 incapable of stimulating Atg4c expression at the transcript level (Fig. 5C), and protein levels are somewhat reduced when Akt is inhibited (Fig. 5D), suggesting that Akt is required for ATG4C protein production. In combination, these data demonstrate that IL-6 stimulates Atg4c transcription through Akt, likely mediated by STAT3 S727 phosphorylation, while stimulating Atg7 transcription through inhibition of mTORC1. Collectively, this suggests that IL-6 stimulates autophagy through multiple parallel mechanisms (Fig. 5E).

Autophagy is required for IL-6-mediated protection of β cells from apoptosis

To determine whether IL-6 is able to directly protect β cells from apoptosis, we induced INS-1 cell death with a cocktail of proinflammatory cytokines—TNF-α, IL-1β, and IFN-γ—for 24 h. These proinflammatory cytokines did not activate STAT3 signaling to the same degree as IL-6, and IL-6 was still able to stimulate STAT3 phosphorylation in the presence of this cytokine cocktail (Fig. 6A). The slight decrease in STAT3 phosphorylation in the presence of cytokines relative to IL-6 alone may be due to activation of STAT1 by IFN-γ, which has previously been shown to reduce, but not completely abolish, IL-6 stimulated phosphorylation of STAT3 (40). Apoptosis was then measured by cleaved caspase 3 levels and caspase 3/7 activity. Cytokine treatment stimulated an increase in cleaved caspase 3 protein levels (Fig. 6B) and a 2.6-fold increase in caspase activity (Fig. 6C; P = 0.0001). Pretreatment with IL-6 reduced the cytokine-stimulated increase in cleaved caspase 3. Furthermore, treatment with either GLP-1 (6 h pretreatment; P = 0.01) or IL-6 (4 h pretreatment; P = 0.02) was able to similarly reduce caspase activity (Fig. 6C). To determine whether autophagy stimulation by IL-6 was required for apoptosis protection, we used chloroquine to inhibit autophagy. Chloroquine itself does not stimulate apoptosis, as measured by cleaved caspase 3 protein levels or caspase 3/7 activity (Fig. 6B, C). Treatment of INS-1 cells with chloroquine results in an accumulation of LC3-II due to a block in autophagy downstream of autophagosome formation at the stage of fusion with lysosomes (Fig. 6B). When autophagy was blocked at the time of IL-6 pretreatment, IL-6 was subsequently unable to protect the β cells from apoptosis as seen by increases in both cleaved caspase 3 levels (Fig. 6B) and caspase 3/7 activity (Fig. 6C). This indicates that autophagy is required for IL-6-mediated protection of β cells from proinflammatory cytokine-induced apoptosis.

Figure 6.

IL-6 requires autophagy to protect β cells from proinflammatory cytokine-induced apoptosis. A) Western blot analysis of STAT3 activation in INS-1 in presence of IL-6 and proinflammatory cytokine cocktail. Representative blot is shown (n = 3). B) Western blot analysis of LC3 and levels of cleaved caspase 3 in INS-1 cells treated with cytokine cocktail, 200 ng/ml IL-6, and/or chloroquine. C) Analysis of caspase 3/7 activity in INS-1 treated with cytokine cocktail in presence of GLP-1, IL-6, or chloroquine. *P < 0.05, **P < 0.01, ***P < 0.001 (n ≥ 3).

IL-6 regulates autophagosome components in human islets and signaling is perturbed in type 2 diabetics

We next looked at whether IL-6 could similarly stimulate STAT3 activation in isolated human islets. In nondiabetic human islets we see a robust stimulation of Y705 phosphorylation of STAT3 by IL-6 (Fig. 7A). Autophagy is increased in mice fed a high-fat diet, and loss of β-cell autophagy impairs both baseline cell survival as well as compensation for obesity (20). Therefore, to determine whether changes in IL-6 signaling and autophagy might occur in humans with diabetes, we analyzed transcripts from human islets of nondiabetic donors and donors with type 2 diabetes (defined here as either hemoglobin A1C > 6.0% or confirmed history of diabetes). We initially examined transcription of the IL-6 receptor components IL-6RA and GP130, and we found that IL-6RA transcript was decreased 3-fold (P = 0.02) and GP130 was decreased 2.1-fold (P = 0.015) in islets from donors with diabetes (Fig. 7B, C). IL-6 expression was also decreased in islets from donors with diabetes compared to nondiabetic donor islets (P = 0.017; Fig. 7D). This suggested that autophagy gene transcripts regulated downstream of IL-6 signaling might also be down-regulated in islets from donors with diabetes.

Figure 7.

IL-6 signaling is perturbed in human islets in type 2 diabetes. A) Human islets were cultured for 96 h in absence or presence of 200 ng/ml recombinant human IL-6; then Stat3 activation was analyzed by Western blot. Representative blot is shown (n = 3). B–D) Islets from nondiabetic donors and donors with type 2 diabetes were analyzed for transcription of IL-6RA (B), GP130 (C), IL-6 (D), GABARAPL1 (E), and LAMP1 (F). G) Nondiabetic human islets were treated with 200 ng/ml recombinant human IL-6 or vehicle for 4 d, and transcription of GABARAPL1 was measured. *P < 0.05, ****P < 0.0001 (n ≥ 4).

We did not observe significant differences in the transcript levels of ATG4C or ATG7 in the diabetic vs. nondiabetic islet. We then looked at the other autophagy enzymes critical in autophagosome formation. There are 6 mammalian homologs of the autophagosome component Atg8: GABARAP, GABARAPL1, GATE16, MAP1LC3A, MAP1LC3B, and MAP1LC3C. Analysis of the human ATG8 homologs revealed a striking 57-fold down-regulation of GABARAPL1 transcription (P < 0.0001) (Fig. 7E). All other homologs were not significantly different (data not shown). Additionally, we detected a modest 1.6-fold increase in transcription of the lysosome component LAMP1 (P = 0.0128; Fig. 7F). We hypothesize that the lysosome transcript LAMP1 is regulated independently of IL-6 signaling in the diabetic islet.

While our sample size of islets from diabetic donors was small and our overall analysis of transcripts was likely underpowered to detect small changes in autophagy transcripts, the dramatic decrease in GABARAPL1 led us to hypothesize a specific role for GABARAPL1 in human islet autophagy. To explore a more direct role for IL-6 regulation of autophagy in human islets, we therefore sought to determine whether GABARAPL1 was regulated by IL-6. We treated nondiabetic human islets in culture with IL-6 and found that GABARAPL1 was stimulated by nearly 2-fold after IL-6 treatment (Fig. 7G; P = 0.03), suggesting that its down-regulation in diabetic islets is due to decreased IL-6 signaling through the membrane bound receptor. These findings suggest that IL-6 mediated islet signaling is likely altered in type 2 diabetes. IL-6-mediated transcription of GABARAPL1 may be of particular importance in the adaptive stimulation of autophagy in the human islet.

DISCUSSION

We have shown that IL-6 stimulates autophagy in the pancreatic β cell in vitro (Fig. 1B, C) and in the islet in vivo (Fig. 3F–H). IL-6 mediated regulation of autophagy occurs through multiple complementary mechanisms involving both repression of mTORC1 and stimulation of mTORC2. IL-6 treatment leads to the coordinated stimulation of autophagy through activation of AMPK (Fig. 4A) and inhibition of mTORC1 (Fig. 4B). IL-6 also stimulates transcription of the autophagy enzyme Atg7 (Fig. 2C, D), an effect likely mediated by the repression of mTORC1 activity (Fig. 5A), independent of STAT3 (Fig. 4D). IL-6 also stimulates Akt activation (Fig. 4C) and Atg4c transcription (Fig. 2A, B), which appear to be linked through Akt-mediated Stat3 phosphorylation at S727 (Figs. 4E and 5C). This is in contrast to previous reports linking phosphorylation at this site to mTORC1 activity (37), and may represent a β-cell-specific response. This is of particular interest given the conflicting observations of IL-6 or STAT3-mediated inhibition of autophagy throughout the literature (41) and general down-regulation of autophagy in other tissues of obese type 2 diabetics (42). In light of our in vivo data suggesting IL-6 mediated inhibition of lysosome formation in the exocrine pancreas alongside a clear stimulation of lysosomes by IL-6 in the islets, autophagy regulatory pathways may indeed be cell-type specific. Therefore, we hypothesize that cell-type specific IL-6 signaling plays a role in many of the divergent effects of IL-6 on autophagy and on signaling in general. It is quite clear that the β-cell autophagy response is different from many other cell types where autophagy is stimulated by nutrient deprivation, as β cells have increased autophagy in response to high nutrient load and do not have an increase in autophagy after mice were unfed overnight (20). Accordingly, we show that the regulation of autophagy by IL-6 and its downstream signaling molecules is also different than in other tissues. It will be interesting to determine whether this is a characteristic of the IL-6 response in other highly secretory cell types or is truly specific to the β cell.

The question remains as to the overall purpose of IL-6 mediated autophagy stimulation and how that might spare a β cell from apoptosis in acute vs. long-term conditions. We attempted to address that question by using a model of continuous IL-6 infusion for 7 d, as opposed to multiple daily IL-6 injections. Unexpectedly, the mice with saline pumps still had higher circulating IL-6 levels in a terminal bleed sample than expected for lean mice. In previous studies where IL-6 was continuously infused into mice using the same methods, IL-6 levels in saline pump mice were not reported (43). We hypothesize that the mild elevation in IL-6 in the mice with a saline pump was due to one of several possibilities: 1) an inflammatory response to the subcutaneous infusion pump, 2) a stress response to the glucose tolerance test performed just before death, or 3) a rise due to the effects of tribromoethanol used for euthanasia (44). Therefore, the saline pump mice may have had mildly elevated IL-6 throughout the study or only in response to terminal procedures. We were unable to obtain adequate sample to measure circulating IL-6 in mice with implanted pumps before they were humanely killed. In case the effect was due to inflammatory responses from the pump implant, we added a separate control group for visualization of autophagy where no pumps were implanted. Notably, we saw no stimulation of autophagy or lysosome formation in the animals without pumps (Fig. 3F), suggesting that an acute effect of IL-6 from terminal procedures was unlikely to explain our findings. In our study, despite elevated basal IL-6 in saline controls, we did observe an approximately 2-fold increase in circulating IL-6 levels when IL-6 containing pumps were used (Fig. 3A). For these reasons, we think that that this is a valid experimental model to examine the impact of chronic IL-6 elevations on islet autophagy. We find that chronic IL-6 elevation increased the number of lysosomes, which may be indicative of increased autophagic flux in the β cell. However, future experiments will be required to confirm this and further analyze the effects of chronic IL-6 signaling on β-cell autophagy.

Importantly, we find that IL-6 mediated induction of autophagy protects β cells from apoptosis (Fig. 6B). This is consistent with previous observations that exercise (which stimulates IL-6) activates autophagy in muscle cells and is critical for cellular homeostasis (45), and we propose that IL-6 similarly mediates an important adaptive response to obesity in the islet. Given our observation that the IL-6 receptor and its downstream signaling target GABARAPL1 are both down-regulated in human islets in type 2 diabetes (Fig. 6), IL-6-mediated autophagy stimulation appears to be impaired in type 2 diabetes. We hypothesize that inability of the diabetic islet to respond to increased circulating IL-6 in obesity may contribute to β-cell dysfunction and death. Surprisingly, LAMP1 expression is increased in islets from type 2 diabetics, contrary to what one might expect if there is a down-regulation of autophagy. Differential regulation of this autophagy component in human islets suggests that there may be an uncoupling of autophagosome and lysosome biogenesis. This may represent compensatory up-regulation to account for decreased flux, rampant untargeted degradation of cellular components, or may perhaps involve independent regulation of additional transcription factors such as the transcription factor EB (46). Future work will be necessary to determine whether GABARAPL1 is a key player in human islet autophagy and how both LAMP1 and lysosome biogenesis are regulated in the islet. Furthermore, it will be interesting to determine whether loss of IL-6 signaling plays a role in the increased mTORC1 and decreased mTORC2 activity that is observed in islets from humans with type 2 diabetes (47). Intriguingly, the incidence of diabetes increases dramatically with age (48), and it is well-documented that autophagy in tissues including the pancreatic islet are impaired by aging (49, 50); future work will be necessary to determine whether these 2 phenomena are causally linked, and whether restoration of autophagy in aged islets can protect against diabetes.

We know that autophagy is critical for cellular homeostasis, but it is also important for energy generation from protein breakdown and is a natural response to nutrient deprivation. The increase in circulating IL-6 during exercise suggests an induction of IL-6 activity under conditions of energy demand or nutrient deprivation. Similarly, prolonged nutrient deprivation in mice can stimulate IL-6 receptor transcription in the islet (11). Obesity is clearly not a condition of nutrient deprivation. However, significantly increased demand for insulin under conditions of obesity may uniquely stimulate autophagy in the β cell to promote energy generation and mobilization of amino acids for insulin production. Autophagy under these conditions could also aid in alleviating mitochondrial stress through the clearance of dysfunctional organelles. This could, in turn, serve as a mechanism of quality control in the β cell that when perturbed leads to apoptosis and diabetes development.

Collectively, we have demonstrated a protective role for IL-6 through the direct stimulation of β-cell autophagy. Our work addresses a key aspect of the fundamental process of islet adaptation to obesity and suggests that increased IL-6 in obesity may serve a positive role in this process. Considering IL-6 in this light therefore opens up new avenues of study with the perspective that endogenous inflammatory response can stimulate adaptation and β-cell survival to prevent diabetes.

Glossary

GLP-1

glucagon-like peptide 1

IL-6Rα

IL-6 receptor α

JAK/STAT

Janus kinase/signal transducer and activator of transcription

mTOR

mammalian target of rapamycin

mTORC

mammalian target of rapamycin complex

STAT3

signal transducer and activator of transcription 3

AUTHOR CONTRIBUTIONS

A. K. Linnemann, D. B. Davis, and D. W. Lamming designed the study; A. K. Linnemann, J. Blumer, M. R. Marasco, T. J. Battiola, H. M. Umhoefer, and J. Y. Han completed the experiments and A. K. Linnemann compiled the data; A. K. Linnemann, D. W. Lamming, and D. B. Davis analyzed the data; A. K. Linnemann, D. B. Davis, and D. W. Lamming wrote the paper; and all authors reviewed and edited the article.