Statins reduce lipid levels and are widely prescribed. Statins have been associated with an increased incidence of type 2 diabetes, but the mechanisms are unclear. Activation of the NOD-like receptor family, pyrin domain containing 3 (NLRP3)/caspase-1 inflammasome, promotes insulin resistance, a precursor of type 2 diabetes. We showed that four different statins increased interleukin-1β (IL-1β) secretion from macrophages, which is characteristic of NLRP3 inflammasome activation. This effect was dose dependent, absent in NLRP3−/− mice, and prevented by caspase-1 inhibition or the diabetes drug glyburide. Long-term fluvastatin treatment of obese mice impaired insulin-stimulated glucose uptake in adipose tissue. Fluvastatin-induced activation of the NLRP3/caspase-1 pathway was required for the development of insulin resistance in adipose tissue explants, an effect also prevented by glyburide. Fluvastatin impaired insulin signaling in lipopolysaccharide-primed 3T3-L1 adipocytes, an effect associated with increased caspase-1 activity, but not IL-1β secretion. Our results define an NLRP3/caspase-1–mediated mechanism of statin-induced insulin resistance in adipose tissue and adipocytes, which may be a contributing factor to statin-induced development of type 2 diabetes. These results warrant scrutiny of insulin sensitivity during statin use and suggest that combination therapies with glyburide, or other inhibitors of the NLRP3 inflammasome, may be effective in preventing the adverse effects of statins.
Introduction
Therapy with statins inhibits hydroxymethylglutaryl-CoA reductase, a rate-limiting enzyme in cholesterol biosynthesis, and reduce LDL cholesterol levels. Statins have actions beyond lipid-lowering effects, which include modulation of immune function (1). Statins decrease intermediates in the mevalonate pathway that lie upstream of cholesterol formation, reducing protein prenylation, a post-translational lipid modification that occurs on many proteins, including those involved in immune responses (1). Immune proteins such as pattern recognition receptors (PRRs) have emerged as integrators of nutrient- and pathogen-sensing systems, and the inflammation that occurs during obesity (i.e., metaflammation) has been characterized in terms of excess nutrients and energy (2). Drug-mediated changes in inflammation via engagement of PRRs should also be considered, particularly for therapeutic agents such as statins that are used to treat aspects of metabolic disease.
Statin-mediated decreases in protein prenylation are generally associated with anti-inflammatory responses and can reduce levels of tumor necrosis factor and interleukin (IL)-6 in lipopolysaccharide (LPS)-treated peripheral blood (3). In contrast, statins have been associated with increased secretion of the proinflammatory cytokine IL-1β; an effect that requires caspase-1 activity and priming with another immunogenic agent such as LPS (4). These features are indicative of regulation by the inflammasome containing the PRR, NOD-like receptor family, pyrin domain containing 3 (NLRP3; also referred to as NACHT, leucine-rich repeat, and pyrin domains-containing protein 3 or cryopyrin) (5,6). The NLRP3 inflammasome is causally linked to the development of insulin resistance in rodents (7) and has recently been shown to be activated in macrophages of patients with newly diagnosed insulin-resistant type 2 diabetes (8). Statin therapy has been associated with increased incidence of type 2 diabetes, as high as 48% in certain populations (9,10). Glyburide, a drug that is widely prescribed for the treatment of diabetes, inhibits the NLRP3 inflammasome independently of cyclohexylurea-mediated insulin secretion (11). We hypothesized that statin-mediated activation of the NLRP3 inflammasome promotes insulin resistance, which could be attenuated by glyburide.
Research Design and Methods
Mice and Materials
The McMaster University Animal Ethics Review Board approved all procedures. Male wild-type (WT) C57BL/6 (catalog #000664) and leptin-deficient ob/ob (catalog #000632) JAX mice were from The Jackson Laboratory. NLRP3−/− mice (>10 generations backcrossed to C57BL/6) were from Professor Nicolas Fasel (Université de Lausanne, Lausanne, Switzerland) and were provided by Dr. Dana Philpott (University of Toronto, Toronto, ON, Canada). To determine the effect of long-term statin treatment on insulin-stimulated tissue glucose uptake, ob/ob mice were orally administered 40–50 mg/kg fluvastatin or vehicle 5 days a week for 6 weeks, a dose of fluvastatin that has been used in other mouse models (12). Twenty-four hours after the last dose, mice were injected with 2 μCi of 3H-2-deoxy-d-glucose (2DG) via tail vein, immediately followed by the administration of insulin (4 units/kg i.p.). Blood samples were taken at baseline, 5, 10, 15, and 20 min, and were analyzed for 2DG radioactivity. Mice were killed by cervical dislocation, and tissues were snap frozen in liquid nitrogen. Brown adipose tissue (BAT) and gonadal white adipose tissue (WAT) were analyzed for 2DG radioactivity with and without deproteinization (0.3 mol/L BaOH and 0.3 mol/L ZnSO4) to calculate the rates of tissue-specific glucose uptake. Statins were from Cayman Chemical (Ann Arbor, MI). InvivoGen (San Diego, CA) supplied ultra-pure LPS (Escherichia coli 0111:B4). z-WHED-FMK and caspase-1/3 kits were from R&D Systems (Denver, CO). The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) kit and all other chemicals were from Sigma-Aldrich (St. Louis, MO).
Macrophages
Bone marrow–derived macrophages (BMDMs) were cultured for 7–10 days in DMEM containing 10% FBS and 15% L929 conditioned media. BMDMs were washed in serum-free media and exposed to statin (1 μmol/L fluvastatin, unless otherwise stated) for 18 h in serum-free DMEM and LPS (200 ng/mL) was added during the final 4 h. GGPP (10 μmol/L), z-WHED-FMK (10 μmol/L), and glyburide (200 μmol/L) were used during the statin treatment period. IL-1β and IL-6 were quantified by ELISA. Transcript levels were analyzed by quantitative PCR, as described previously (13,14).
Adipose Explants and Adipocytes
Mice were killed by cervical dislocation, and PBS-rinsed gonadal adipose tissue was minced into ∼5-mg pieces in DMEM containing 10% FBS. After 2 h of equilibration, 25 mg explants were placed in serum-free DMEM and exposed to 10 μmol/L fluvastatin (18 h) and 2 μg/mL LPS (4 h), and were stimulated with 0.3 nmol/L insulin for 10 min. Adipose tissue lysates were used for determination of caspase-1/3 activity (over 4 h), immunoblotting, or ELISA determination of cytokines, as described previously (14). 3T3-L1 preadipocytes (ATCC, Rockville, MD) were differentiated (14), and fluvastatin/LPS treatment was similar to explants. 3T3-L1 media were used for ELISAs, and lysates were used to measure caspase-1 enzymatic activity fluorometrically or for immunoblotting after insulin stimulation at 0.3 or 100 nmol/L for 10 min.
Statistical Analysis
Significance was determined by unpaired, two-tailed t tests or ANOVA, as appropriate. A Bonferroni or Tukey post hoc test was used when appropriate (Prism 4–6; GraphPad Software).
Results
Statins Activate the NLRP3 Inflammasome
All statins (10 μmol/L, 18 h) increased the secretion of IL-1β from WT BMDMs compared with LPS alone (Fig. 1A). Fluvastatin increased IL-1β secretion in a dose-dependent manner only with LPS priming (Fig. 1B), but LPS alone increased IL-6 secretion in BMDMs (Fig. 1C). Fluvastatin up to 100 μmol/L did not lower BMDM viability detected using the MTT assay (data not shown). The isoprenyl intermediate GGPP prevented fluvastatin-induced IL-1β secretion in LPS-primed BMDMs (Fig. 1D), suggesting that decreased prenylation drives statin-mediated inflammasome activation. Inhibition with z-WHED or glyburide prevented statin-induced IL-1β secretion in LPS-primed BMDMs (Fig. 1E). LPS treatment, but not fluvastatin treatment alone, increased transcript levels of NLRP3, IL-1β, and IL-6 (Fig. 1F and G). Therefore, statins alone did not alter inflammasome-priming events such as increased NLRP3 transcript levels (15). The combination of fluvastatin and LPS synergistically increased both IL-1β and IL-6 transcript levels (Fig. 1F). Fluvastatin did not increase IL-1β secretion in LPS-primed BMDMs from NLRP3−/− mice (Fig. 1H). LPS increased IL-6 secretion BMDMs from NLRP3−/− mice (Fig. 1I).
Fluvastatin Impairs Adipose Tissue Insulin Signaling Via the NLRP3 Inflammasome
We first established that long-term oral administration of fluvastatin impaired insulin-simulated glucose disposal into adipose tissue using an in vivo mouse model of obesity. 2DG uptake was >50% lower in WAT, but not BAT, of fluvastatin-treated ob/ob mice (Fig. 2A). We then used WAT explants to determine the mechanisms of statin-induced insulin resistance. Fluvastatin increased caspase-1 activity in LPS-primed adipose tissue from WT mice, but not from NLRP3−/− mice (Fig. 2B). Glyburide prevented this increased caspase-1 activity (Fig. 2B). Fluvastatin increased caspase-3 activity in LPS-primed adipose tissue from WT and NLRP3−/− mice and independently of glyburide treatment (Fig. 2C). Therefore, fluvastatin activated an NLRP3-dependent, glyburide-sensitive, caspase-1 inflammasome in adipose tissue.
Surprisingly, LPS alone increased IL-1β in adipose explants from both WT and NLRP3−/− mice (Fig. 2D). Fluvastatin plus LPS further increased IL-1β levels compared with LPS in adipose explants from WT mice, but not NLRP3−/− mice (Fig. 2D). LPS alone did not change the ability of insulin to phosphorylate Akt at serine 473 in adipose tissue explants (Fig. 2E). Fluvastatin alone impaired insulin-mediated phosphorylated Akt (pAkt) in adipose tissue from WT mice, but not NLRP3−/− mice (Fig. 2E). The combination of LPS and fluvastatin prevented the ability of insulin to increase the levels of pAkt in adipose tissue explants from WT mice, but not NLRP3−/− mice (Fig. 2E). Glyburide reversed fluvastatin-induced suppression of insulin-mediated pAkt in LPS-primed adipose explants, but glyburide did not increase pAkt levels on its own (Fig. 2F).
Interestingly, changes in caspase-1 activity, but not Il-1β secretion mirrored statin-induced insulin action in adipose explants. There are many nonadipocyte cell types and potential sources of IL-1β processing in adipose tissue (16), so we next tested the adipocyte cell-autonomous response. Treatment with fluvastatin plus LPS increased caspase-1 activity in 3T3-L1 adipocytes, but did not increase IL-1β or IL-6 secretion (Fig. 3A–D). However, treatment with fluvastatin plus LPS significantly lowered insulin-stimulated pAkt in 3T3-L1 adipocytes (Fig. 3E).
Discussion
Treatment with statins lowers blood lipid levels and reduces the number of cardiovascular disease–related events (17). Paradoxically, statins have been associated with an increased incidence of diabetes. This has sparked debate over reassessing the benefits and risks of statin use (18). Understanding how statins promote adverse effects such as the progression to diabetes may promote improvements in this drug class. We show that statins activate the NLRP3 inflammasome in various immune and metabolic cells of adipose tissue. Fluvastatin-induced impairments in insulin signaling were dependent upon the NLRP3 inflammasome. The commonly used diabetes drug glyburide inhibited statin-induced inflammasome activation and prevented defects in adipose tissue insulin action.
Endogenous and exogenous stimuli activate the NLRP3 inflammasome, which prompted the theory that this PRR is a metabolic danger sensor (19). Production of bioactive IL-1β (or IL-18) by the NLRP3 inflammasome requires priming and stimuli-promoting assembly of a caspase-1 protein complex. We confirm that statins increased IL-1β secretion in adequately primed macrophages (5,6,20), and we demonstrated the requirement of the NLRP3 inflammasome. Glyburide, an existing diabetes drug, inhibited statin-induced increases in IL-1β in macrophages, which is consistent with its inhibitory effect on other inflammasome activators (11). All statins tested activated NLRP3-mediated increases in IL-1β, which is similar to the class effect of these hydroxymethylglutaryl-CoA reductase inhibitors increasing the risk of diabetes, independently of potency or lipophilic properties (18). A standard dose of fluvastatin can equate to micromolar serum levels in humans (21), and other statins can reach serum levels >10 μmol/L (22), which corresponds with the effective dose range of our in vitro models. The dose response of fluvastatin-induced IL-1β secretion that we report is consistent with higher doses of statins increasing the risk of diabetes to a greater extent (18). This is important because of the diminishing returns of lipid lowering as the dose of statins is increased, the high dose of statins required to achieve adequate lowering in many patients, and the incidence of statin intolerance in clinical practice (23).
Adipose tissue is a key site of inflammation during insulin resistance, and the NLRP3/caspase-1 inflammasome regulates adipose tissue inflammation and function (24). We first showed that 6 weeks of oral fluvastatin treatment in ob/ob mice impaired insulin-stimulated glucose uptake in WAT, a depot where insulin normally increases glucose disposal from the blood. Statin feeding in these mice had no effect on glucose uptake in BAT, highlighting the specificity of this statin-mediated effect. We then provided genetic evidence that statins impaired adipose tissue insulin action via the NLRP3 inflammasome, which was also prevented with glyburide ex vivo. The combination of LPS and fluvastatin was most effective in preventing insulin-mediated signals in adipose tissue explants. However, fluvastatin did not require LPS to cause impaired insulin action, suggesting that adipose tissue contains endogenous NLRP3 inflammasome-priming signals. This is consistent with the NLRP3 inflammasome mediating obesity-associated insulin resistance in response to saturated lipids (7). Intriguingly, NLRP3 was not necessarily required for IL-1β secretion from adipose tissue explants. Since the regulation of IL-1β did not mirror changes in insulin action, our results suggest that caspase-1 rather than IL-1β provides the link to adipose tissue insulin resistance. Further, a cell-autonomous program that increases caspase-1 activity can be engaged by fluvastatin in LPS-primed clonal adipocytes. This response in 3T3-L1 adipocytes did not increase IL-1β secretion, but impaired insulin action. Therefore, our results suggest that fluvastatin acts through the NLRP3/caspase-1 inflammasome in multiple cells within adipose tissue and culminates in insulin resistance that does not necessarily require Il-1β. Our results concerning insulin resistance have focused on fluvastatin, but the type of statin and pleiotropic actions on inflammation (which are often conflicting) in immune cells, liver, muscle, adipose tissue, and pancreas should be considered in obese and prediabetic mice and patients (25,26). Little is known about the contributing factors to the statin-diabetes relationship. We propose a role for inflammation and that inflammasome-mediated insulin resistance is positioned as a contributor to the development of diabetes. It is enticing to speculate that metabolic endotoxemia or other priming agents for the inflammasome may play a role (27).
We propose that the inhibition of NLRP3/caspase-1 inflammasome may attenuate statin-induced insulin resistance. This is particularly relevant in mitigating any contribution of statins to insulin resistance leading to diabetes in obese hyperlipidemic patients who commonly use this class of drugs for lipid lowering, but are often at risk for the development of diabetes. The next generation of statins may be driven by combination therapy or statin derivatives that maintain or enhance lipid-lowering properties, but allay adverse effects by evading the NLRP3 inflammasome.
See accompanying article, p. 3569.
Article Information
Acknowledgments. The authors thank the Université de Lausanne (Lausanne, Switzerland) and the Institute of Arthritis Research for the NLRP3−/− mice.
Funding. This work was supported by grants to J.D.S. from the Canadian Institutes of Health Research (CIHR) (grants PNI123793 and MOP130432), the Canadian Diabetes Association (CDA), Natural Sciences and Engineering Research Council (grant 435474-2013), and a CIHR grant to G.R.S. J.F.C. was supported by a Canada Graduate Scholarship (CIHR). J.D.C. was supported by MAC-Obesity research funds (McMaster University). M.D.F. was supported by Banting and CIHR postdoctoral fellowships. G.R.S. holds a Tier II Canada Research Chair. J.D.S. is supported by a CDA Scholar award.
Duality of Interest. G.R.S. is on the scientific advisory board and received honoraria from Esperion Therapeutics. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. B.D.H. researched the data, contributed to the discussion, and edited the manuscript. T.C.L., J.F.C., E.D., W.C., J.S.L., J.D.C., K.P.F., B.M.D., and M.D.F. researched the data. M.A.T. and G.R.S. contributed to the discussion and edited the manuscript. J.D.S. researched the data, derived the hypothesis, and wrote the manuscript. J.D.S. 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.