Sodium-glucose cotransporter 2 inhibitors such as canagliflozin lower blood glucose and reduce cardiovascular events in people with type 2 diabetes through mechanisms that are not fully understood. Canagliflozin has been shown to increase the activity of the AMP-activated protein kinase (AMPK), a metabolic energy sensor important for increasing fatty acid oxidation and energy expenditure and suppressing lipogenesis and inflammation, but whether AMPK activation is important for mediating some of the beneficial metabolic effects of canagliflozin has not been determined. We, therefore, evaluated the effects of canagliflozin in female ApoE−/− and ApoE−/−AMPK β1−/− mice fed a western diet. Canagliflozin increased fatty acid oxidation and energy expenditure and lowered adiposity, blood glucose and the respiratory exchange ratio independently of AMPK β1. Canagliflozin also suppressed liver lipid synthesis and the expression of ATP-citrate lyase, acetyl-CoA carboxylase and sterol response element-binding protein 1c independently of AMPK β1. Canagliflozin lowered circulating IL-1β and studies in bone marrow-derived macrophages indicated that in contrast with the metabolic adaptations, this effect required AMPK β1. Canagliflozin had no effect on the size of atherosclerotic plaques in either ApoE−/− and ApoE−/−AMPK β1−/− mice. Future studies investigating whether reductions in liver lipid synthesis and macrophage IL-1β are important for the cardioprotective effects of canagliflozin warrant further investigation.
Canagliflozin is a sodium-glucose transporter 2 inhibitor (SGLT2i), that inhibits the reabsorption of glucose in the renal proximal convoluted tubules to promote urinary glucose excretion and thereby reduce blood glucose . In addition to reducing blood glucose, canagliflozin treatment lowers BMI and triglycerides in patients with type 2 diabetes , effects which have been associated with increases in energy expenditure and the suppression of the RER indicative of increases in fatty acid oxidation [3,4]. Importantly, these changes in metabolism are associated with a reduction in cardiovascular events compared with placebo  and additional analysis has revealed that these effects are unlikely explained by improved glycemic control [5–10]. These data suggest that in addition to urinary glucose excretion, SGLT2i's may activate additional pathways that exert beneficial effects.
Atherosclerotic cardiovascular disease (CVD) is a result of dysregulation of lipid metabolism and a maladaptive immune response that leads to a build-up of lipid-laden macrophages (foam cells) within the intima of large blood vessels . Lowering LDL cholesterol by inhibiting liver HMG-CoA reductase (HMGR) or ATP-citrate lyase (ACLY) is associated with reduced atherosclerosis and death from cardiovascular events [12–16]. In addition to LDL-lowering, a recent clinical trial has shown that an IL-1β neutralizing antibody reduces cardiovascular events independently of changes in lipid levels . Interestingly, these beneficial effects on cardiovascular events appear to be specific to IL-1β neutralization since the anti-inflammatory agent methotrexate had no discernable benefit . Collectively, these data suggest that therapies that suppress LDL cholesterol and/or inhibit IL-1β may be effective for reducing CVD, potentially through distinct mechanisms.
SGLT2i's increase the activity of AMP-activated protein kinase (AMPK) [19–22]; a ubiquitously expressed heterotrimer (consisting of α, β, γ subunits), that is a central regulator of lipid metabolism and inflammation (reviewed here [23,24]). Mouse hepatocytes, macrophages and white adipose tissue are composed of primarily an AMPK α1, β1, γ1 heterotrimers. In hepatocytes pharmacological activation of AMPK β1 complexes increases phosphorylation and inhibition of downstream substrates acetyl-coA carboxylase (ACC) and HMGR resulting in reductions in fatty acid and cholesterol synthesis, which lowers liver lipids and LDL cholesterol [25–27]. In white adipose tissue, AMPK is important for increasing mitochondrial biogenesis, the browning of white fat and enhancing energy expenditure [28–30]. In macrophages, where inflammatory status and metabolic activity are directly linked [31,32], the activation of AMPKβ1 switches macrophages from an M1, pro-inflammatory phenotype, to an M2, anti-inflammatory phenotype associated with reductions in pro-inflammatory signaling (NFκB and JNK) and cytokine production (Il-1β, TNFα) [33–37]. These data suggest that pharmacologically targeting AMPK in the liver, white adipose tissue and macrophages may improve risk factors underpinning CVD by suppressing fatty acid and cholesterol synthesis and reducing inflammation.
Given that canagliflozin activates AMPK in hepatocytes  and macrophages , we hypothesized that some of the effects of canagliflozin may be mediated through AMPK activation in these tissues. To test this hypothesis we examined the effects of a clinically relevant dose (30 mg/kg) of canagliflozin [39,40] delivered daily for 6 weeks to female ApoE−/− mice (with normal AMPK expression) or female ApoE−/−AMPK β1−/− mice (low levels of AMPK activity in liver, white adipose tissue and macrophages ). We also examined the in vitro effects of canagliflozin on macrophage inflammation and hepatocyte lipid metabolism.
All animal experiments were performed in accordance with McMaster Animal Care Committee guidelines (AUP #: 16-12-41, Hamilton, ON). The generation and characterization of AMPKβ1−/−, ApoE−/− AMPKβ1−/− and NLRP3−/− mice have been described previously [12,41,42]. Specific Pathogen Free female mice were group housed in a temperature-controlled facility (22–23°C) on a 12 h light/dark cycle where food and water were provided ad libitum. ApoE−/− and ApoE−/−AMPKβ1−/− were fed standard rodent chow diet (Envigo 8640) until 6–8 weeks of age, then switched to a high-fat, high-cholesterol (0.2%) ‘western diet’ (TD.09821, Envigo Diet) and simultaneously treated daily by oral gavage for 6 weeks with vehicle or a canagliflozin (30 mg/kg). For tissue collection, a subset of mice were fasted for 6 h then administered insulin (0.4 U/kg) for 15 min before tissue harvesting from anesthetized mice. Additional tissue collection was performed following in vivo lipogenesis described below from a separate cohort of mice.
For glucose tolerance test (GTT), all mice were fasted for 6 h beginning at 7 am. Blood glucose levels were assessed by a small tail vein nick using a handheld Aviva glucometer (Roche). Intraperitoneal injections of glucose (2.25 g/kg body mass) were administered to initiate the GTT, and blood glucose measures obtained at the indicated time points. Body composition measures were obtained by TD-NMR (Bruker Corporation). Measurements of RER, energy expenditure, ambient activity, food and water intake were assessed using a Comprehensive Laboratory Animal Monitoring System (Columbus Instruments) as described . Commercially available kits were used to measure Insulin (EMD Millipore). homeostatic model assessment of insulin resistance (HOMA-IR) was calculated using the following formula: (fasting insulin (µU/ml) × fasting glucose (mM)/22.5).
Hepatic lipogenesis was performed as described [44,45] with the following modifications; mice treated with canagliflozin for 6 weeks (as described above) were fasted for 12 h overnight, followed by a 2 h refeed to enhance lipogenesis. Mice were then treated with vehicle or canagliflozin (30 mg/kg) by oral gavage for 1 h before intraperitoneal injection with [3H]-acetate (12μCi/mouse). Tissues were collected 1 h after injection of acetate, and frozen in liquid nitrogen. Fatty acid and sterols fractions were extracted as previously described [25,29].
Cytokine analysis in plasma was done using the Bio-Plex Pro Mouse Cytokine 23-Plex Immunoassay (Bio-Rad) performed according to the manufacturer's instructions, as described . IL-1β and TNFα were measured using Mouse DuoSet ELISAs (R&D Systems) from cell culture media and in the liver (following homogenization in 1% BSA (reagent diluent)). For cell culture measurements values were corrected to total cellular protein per well.
Immunoblotting and histological analyses
For protein analyses, tissue or cell lysates were diluted with Western sample buffer and loaded in SDS–PAGE gels as described previously . Proteins were resolved by molecular mass and transferred to polyvinylidene difluoride membranes prior to blocking in 5% BSA. All primary antibodies were obtained from Cell Signaling Technologies (P-Ser79/212-ACC Cat# 3661, ACC Cat# 3662, P-Thr172-AMPK Cat # 2531, AMPK panα Cat # 2532, β-actin Cat # 5125), except for UCP1 (Alpha Diagnostic International, Cat# UCP11-A) and SREBP1 (Santa Cruz Biotechnology Inc Cat # sc-13551). Antibodies were used at concentrations of 1 : 1000 except β-actin and UCP-1 which were used at 1 : 5000 and SREBP1 which was used at 1 : 250. For lesion histology, hearts with aortic roots intact were fixed in 10% neutral buffered formalin, cut transversely, and embedded in paraffin. Serial sections 4 µM thick were cut beginning at the aortic root origin and mounted on slides for hematoxylin and eosin staining (sections were stained every 80 µM to assess lesion size). Images were captured using Nikon 90 Eclipse microscope (Nikon) and lesions and necrotic areas were traced manually and measured using ImageJ. Lesion sizes and necrotic areas were calculated from five sections of the aortic root at 80 μM intervals from each animal.
Real-time quantitative PCR (RT-qPCR)
RNA was isolated from cells using the High Pure RNA Isolation Kit (Roche), and from tissues using TRIzol reagent (Invitrogen) and purified using the RNeasy kit (Qiagen). All Taqman primers were purchased from Invitrogen, and relative gene expression was calculated using (2–ΔCT) method. Values were normalized to housekeeping gene β-actin and expressed relative to wildtype with LPS (cells) or wild-type vehicle (tissues).
Cell culture experiments
Primary hepatocytes were isolated following EGTA and collagenase perfusion. Cells were allowed to adhere overnight in Williams Media E (10% FBS, 1% antibiotic-antimycotic, 2 mM l-glutamine). The following day media was replaced with serum-free media with [3H]-acetate with the addition of 30 μM Canagliflozin or vehicle (DMSO) for 4 h. Cells were washed 3× with ice-cold PBS and scraped in KOH. Sterol isolation was performed as previously described . A portion of the sterol fraction was counted for radioactivity. Bone marrow-derived macrophages (BMDMs) were isolated and differentiated for 7 days with conditioned media (from L929 cells, a source of macrophage colony-stimulating factor; MCSF) . Macrophages were then reseeded in DMEM (10% FBS, 1% antibiotic-antimycotic) at 2 × 106 cells/ml for 6 and 12 well tissue-culture plates and allowed to adhere overnight. The following day, media was replaced with serum-free media and cells were treated with the indicated drugs dissolved in DMSO (final concentration 0.1%). For protein signaling by immunoblotting, cells were treated with drugs for 90 min in serum-free media then rapidly lysed in cell lysis buffer. For mRNA analysis and cytokine release experiments, cells were treated with the drug for 90 min followed by the addition of 10 ng/ml LPS for 24 h.
When comparing two factors, data were analyzed using two-way ANOVAs to determine interaction and main effects, and Sidak's post-hoc tests used for comparisons between groups. One-way ANOVA with Sidak's post-hoc test was used when one variable was being compared. All data are represented as mean ± s.e.m., and p < 0.05 is considered statistical significance.
Canagliflozin reduces adiposity and blood glucose independently of AMPKβ1
Dyslipidemia and inflammation are significant contributors to atherosclerosis, diabetes and insulin resistance. To investigate the role of canagliflozin induced AMPK activation on atherosclerosis and metabolic parameters female ApoE−/− and ApoE−/− AMPK β1−/− were fed a proatherogenic diet for 6 weeks (starting at 6 weeks of age) while receiving daily gavage of either vehicle or 30 mg/kg canagliflozin; a dose that has been shown to result in serum concentrations of 10–20 µM, similar to humans taking 180–300 mg/day [39,40], and activate AMPK [19,39,40]. Daily canagliflozin treatment did not affect body mass but did reduce adiposity and percent fluid mass (Figure 1A–C). Consistent with previous findings , ApoE−/−AMPKβ1−/− mice had a lower body mass and adiposity than control ApoE−/− mice (Figure 1A,B). As expected, canagliflozin reduced 12 h fasted blood glucose and improved glucose tolerance in both ApoE−/− and ApoE−/−AMPKβ1−/− mice (Figure 1D–F); an effect which occurred independently of reductions in fasting insulin levels (Figure 1G) and changes to the HOMA-IR (Figure 1H).
Canagliflozin reduces adiposity and blood glucose in ApoE null mice independently of AMPK β1
Canagliflozin increases energy expenditure and suppresses liver fatty acid synthesis, ACC and ACLY expression independently of AMPK β1
To examine mechanisms which might contribute to the lower adiposity with canagliflozin treatment mice were housed in metabolic cages to assess energy expenditure, locomotor activity as well as food and water intake. Canagliflozin treatment increased energy expenditure (heat) as well as food and water intake, without affecting ambient locomotor activity in both ApoE−/− and ApoE−/−AMPKβ1−/− mice (Figure 2A–D). Canagliflozin treatment reduced RER in both genotypes (Figure 2E). Calculations based on the RER suggested canagliflozin increased rates of fatty oxidation (Figure 2F) while tending to reduce carbohydrate oxidation (Figure 2G). However, in contrast with other SGLT2i's , canagliflozin did not increase the expression of Ucp1 in white or brown adipose tissue, or alter the expression of other markers of adipose tissue browning, or reduce BAT triglycerides (Figure 3A–G). These data indicate that increases in fatty acid oxidation with canagliflozin occur independently of AMPK β1.
Canagliflozin increases energy expenditure, food and fluid intake and decreases RER in ApoE−/− and ApoE−/− AMPKβ1−/− mice.
Canagliflozin reduces UCP1 expression but does not change other markers of browning in iWAT or BAT.
In lipogenic tissues such as the liver, blood glucose is also used for the synthesis of lipids through de novo lipogenesis (DNL). In addition to increases in fatty acid oxidation, a reduction in DNL can also contribute to a reduction in RER . We subsequently examined liver DNL and found that consistent with previous studies [41,47], AMPKβ1−/− mice had higher liver DNL compared with controls, but surprisingly, canagliflozin suppressed DNL independently of AMPKβ1 (Figure 4A). AMPK primarily regulates DNL through phosphorylation of ACC [27,47], and consistent with the activation of AMPK, canagliflozin tended to increase the ratio of phosphorylated ACC (pACC) over total ACC in the liver of ApoE−/− mice but not ApoE−/− AMPKβ1−/− mice (Figure 4B,C). However, total ACC protein expression was significantly reduced in the absence of AMPK β1 and by canagliflozin (Figure 4B,D). Consistent with lower ACC there were also significant reductions in Acc1 and Acc2 mRNA (Figure 4E,F). In addition to ACC, the enzyme ACLY, which is upstream of ACC and generates acetyl-CoA from citrate, is also critical for regulating DNL. We found that consistent with reductions in ACC expression, there was a strong trend for a reduction in Acly expression with canagliflozin treatment (p = 0.058) and that Acly expression was also reduced basally in ApoE−/− AMPKβ1−/− mice (Figure 4G). Consistent with reductions in Acly mRNA we found that ACLY protein expression was also reduced following treatment with canagliflozin in both genotypes (Figure 4H,I). Both ACLY and ACC expression are regulated by sterol response element-binding protein 1c (SREBP1c). Therefore, we examined protein levels of this transcription factor, and found that canagliflozin treatment reduced SREBP1c protein in the liver of both ApoE−/− and ApoE−/− AMPKβ1−/− mice (Figure 4J,K). To directly evaluate the effects of canagliflozin independently of changes in glucose and other circulating factors we treated primary mouse hepatocytes generated from WT and AMPK β1−/− mice and found canagliflozin suppressed DNL in hepatocytes from wild-type mice but that these effects were blunted in cells lacking AMPK β1 (Figure 4L). These data indicate that in isolated hepatocytes AMPK β1 is important for suppressing DNL, however, chronic treatment of mice with canagliflozin leads to the inhibition of DNL independently of AMPK β1. These data suggest that the in vivo effects of canagliflozin to suppress DNL may be secondary to reductions in blood glucose and the subsequent inhibition of SREBP1c and transcriptional down-regulation of key lipogenic enzymes ACLY and ACC.
Canagliflozin reduces hepatic lipogenesis independently of AMPK by suppressing lipogenic gene expression.
Canagliflozin reduces liver sterol synthesis
in vivo and in vitro
Circulating lipids are central to the development of atherosclerosis. Consistent with reductions in fatty acid synthesis, canagliflozin lowered plasma ApoB in ApoE−/− mice. However, surprisingly, ApoE−/− AMPKβ1−/− mice, had reduced levels of ApoB that did not decline further with canagliflozin (Figure 5A). The low level of ApoB in ApoE−/− AMPKβ1−/− mice may have been secondary to the reductions in blood glucose and liver ACLY which would be expected to lower acetyl-CoA availability for HMGR-mediated sterol synthesis. Consistent with reduced ApoB levels, canagliflozin treatment reduced hepatic cholesterol synthesis in ApoE−/− but not ApoE−/− AMPKβ1−/− (Figure 5B). To directly investigate the effects of canagliflozin on sterol synthesis independent from changes in blood glucose we isolated hepatocytes from wild-type mice, AMPK β1−/− mice and mice lacking the key inhibitory phosphorylation site on HMG-CoA reductase (HMGR KI). Consistent with the effects in vivo we found that canagliflozin suppressed sterol synthesis through a pathway requiring AMPK and the phosphorylation of HMGR (Figure 5C). However, there were no changes in total cholesterol or triglycerides in plasma (Figure 5D,E). These data suggest that in vivo a complex feedback loop regulating sterol synthesis and circulating lipids may impact circulating lipids. Consistent with this idea, LDLr which is also regulated by SREBP1c, was down-regulated by canagliflozin in both genotypes (Figure 5F), an effect which would be expected to increase circulating cholesterol levels and may explain increases in LDLc observed clinically in some studies .
Canagliflozin reduces sterol synthesis in an ApoE−/− but not ApoE−/− AMPKβ1−/− mice.
Canagliflozin reduces IL-1β but not atherosclerosis
In addition to lipids, inflammation is critical for the development of atherosclerosis. Canagliflozin reduced IL-1β in the plasma of ApoE−/− mice with the same trend present in the livers (p = 0.069); an effect which was blunted in ApoE−/−AMPKβ1−/− mice (Figure 6A,B). Surprisingly, other markers of inflammation in the plasma remained unchanged (Supplementary Table S1). To examine the potential mechanisms by which canagliflozin might suppress IL-1β we conducted studies in primary BMDMs. Previously we have shown that canagliflozin, but not other SGLT2i's such as dapagliflozin and empagliflozin, can increase AMPK activity in primary hepatocytes and cancer cells, resulting in reduced lipogenesis and proliferation, respectively [19,20] but whether this effect is conserved in primary BMDMs had not been tested. We found that 10 μM canagliflozin increased the phosphorylation of AMPK (Thr172) and ACC (Ser79) in wildtype but not AMPK β1−/− macrophages (Figure 6C–E). The activation of AMPK has been shown to reduce inflammation [34,50,51], but we observed no change in the phosphorylation of NFκB and IKKβ with canagliflozin treatment (Figure 6F,G), however, a marker of macrophage polarization, P-Stat-1 was reduced with canagliflozin independently of AMPK (Figure 6H). Consistent with changes in STAT-1, TNFα secretion was reduced in both wildtype and AMPK β1 KO macrophages, however, IL-1β secretion was reduced with canagliflozin in an AMPK β1 dependent manner (Figure 6I,J). The secretion of IL-1β by macrophages often requires cleavage of pro-IL-1β by the NLRP3 inflammasome, an effect which is inhibited by AMPK activation [52,53]. Therefore, we hypothesized that AMPK inhibition of the NLRP3 inflammasome may be important for the effects of canagliflozin to lower circulating IL-1β, however, canagliflozin reduced IL1β secretion in both wildtype and NLRP3−/− macrophages (Figure 6K). Given that the effect on IL-1β appears to be upstream of the NLRP3 inflammasome, we next investigated the regulation of IL-1β mRNA, and found that canagliflozin reduced IL-1β mRNA in a dose-dependent manner in wild-type macrophages, but this was attenuated in AMPKβ1−/− macrophages (Figure 6L). Collectively, these data indicate that canagliflozin potently activates macrophage AMPK, and results in both AMPK dependent and independent anti-inflammatory effects. Furthermore, the effects of canagliflozin on IL-1β secretion appear to occur through a mechanism that is upstream of the NLRP3 inflammasome.
Canagliflozin decreases IL-1β.
Surprisingly, despite the effects of canagliflozin to lower ApoB and IL-1β atherosclerotic plaque size and the necrotic area was unalteredin in ApoE−/− mice over the 6 weeks treatment period (Figure 7A–C). There was also no change in these parameters in ApoE−/− AMPKβ1−/− mice.
Canagliflozin treatment does not alter atherosclerotic development.
CVD is the leading cause of morbidity and mortality worldwide  and individuals with type 2 diabetes are more likely to die of CVD than individuals without type 2 diabetes . In people with type 2 diabetes, the SGLT2 inhibitor canagliflozin reduces cardiovascular events [5,56,57] but the mechanisms mediating these beneficial effects are currently unclear. In the current study, we find that canagliflozin exerts beneficial effects on metabolism through both AMPK dependent and independent pathways. Specifically, we find that canagliflozin lowers blood glucose, liver fatty acid synthesis and RER and increases energy expenditure independently of AMPK while reductions in hepatic cholesterol synthesis and IL-1β requires AMPK.
The defining feature of SGLT2is is that they lower blood glucose due to glycosuria. A reduction in blood glucose enhances rates of fatty acid oxidation due to a reduction in malonyl-CoA [58,59]. Malonyl-CoA is an allosteric inhibitor of CPT-1 and also the first committed step in DNL [58,59]. Consistent with previous studies [4,19], we observed that canagliflozin induced a drop in RER indicative of increased fatty acid oxidation and/or inhibition of DNL and led to an increase in energy expenditure. This increase in energy expenditure was not associated with increased markers of adipose tissue browning and importantly also occurred independently of AMPK. Given the drop in RER, our data suggest that an increase in fatty acid oxidation may have been important for enhancing energy expenditure. Mechanistically this increase in fatty acid supply, potentially driven by enhanced adipose tissue lipolysis, promotes fatty acid oxidation which results in a lower NADH/FADH2 ratio and enhances oxygen use per ATP produced because NADH provides three electrons compared with two for FADH2. Therefore, oxygen utilization and energy expenditure are elevated . Collectively, these data suggest that the increase in energy expenditure may be due to the increased reliance on lipids as a fuel source which occurs independently of AMPK β1.
Consistent with previous findings we found increased DNL in mice lacking AMPKβ1−/− [41,47]. Surprisingly, this increase in DNL occurred despite a reduction in ACLY which would be expected to starve ACC of its substrate acetyl-CoA. Future studies investigating ACLY activity and acetyl-CoA levels in the ApoE−/−AMPKβ1−/− mice are warranted. In contrast with other AMPK agonists such as metformin  or direct AMPK activators [25,61], the effects of canagliflozin to lower fatty acid synthesis and/or increase fatty acid oxidation in vivo were independent of AMPK. To examine the potential mechanisms mediating these effects we measured the expression of ACLY and ACC and found they were markedly inhibited independently of AMPK and this was associated with a reduction in the master regulator of lipogenesis SREBP1c, which is consistent with previous findings demonstrating canagliflozin induced a fasting-like transcriptional state . The inhibition of ACC is currently under development for the treatment of NAFLD [63,64] and hepatocellular carcinoma (HCC) , however, increases in hypertriglyceridemia have been a potential complication of these therapies . ACLY inhibitors also lower liver fat in preclinical animal models  and are associated with weight loss . Our current findings cannot determine whether the effects seen here on ACC and ACLY expression are directly mediated by reduced blood glucose but it is clear that the suppression of flux through this pathway occurs independently of AMPK.
In addition to suppressing liver fatty acid synthesis, canagliflozin also lowered sterol synthesis. Canagliflozin suppressed ACLY, which reduces the conversion of citrate to acetyl-CoA which is required for both fatty acid synthesis as described above as well as sterol synthesis. Interestingly, liver ACLY expression was reduced in ApoE−/−AMPKβ1−/− compared with ApoE−/− controls and was not reduced further with canagliflozin treatment. These findings are consistent with previous studies which have established an important reciprocal link between AMPK and ACLY activity [12,66,67]. These data indicate that acutely canagliflozin-induced activation of AMPK suppresses hepatic sterol synthesis through phosphorylation of HMGR but chronically reductions in sterol synthesis are likely mediated through lower blood glucose and transcriptional inhibition of ACLY. Plasma cholesterol content is a balance between cholesterol synthesis and clearance, with the liver being a major contributor to both components. Given the reduced expression of LDLr (responsible for cholesterol clearance) and reduced cholesterol synthesis it is not surprising that there were no changes in plasma cholesterol, however, interestingly we did see reductions in circulating ApoB levels (a major component of LDL particles), suggesting that there may be alterations in types of circulating cholesterol in the model, however, given that mice largely carry cholesterol in VLDL particles these findings are difficult to translate to human studies. Taken together, these data suggest potential feedback mechanisms in the cholesterol pathway which may help elucidate the unexpected increase in LDLc seen in some clinical studies .
Anti-IL-1β therapies have been associated with reduced CV events, independent of changes in serum lipid levels . Therefore, we sought to examine if canagliflozin reduced IL-1β, which could partially explain the outcomes seen clinically. The relationship between canagliflozin and IL-1β is especially interesting because IL-1β has been directly implicated with heart failure [68,69], which seems to be the indication that canagliflozin is most protective against CVD mortality [56,57]. In the current study, we found that canagliflozin treatment reduced circulating IL-1β. Importantly, this effect appears to be specific to IL-1β rather than a broad anti-inflammatory effect since none of the 17 other inflammatory cytokines measured in the plasma of mice were reduced with canagliflozin treatment. Studies in BMDMs demonstrated that canagliflozin activated AMPK and reduced IL-1β mRNA and secretion, an effect we hypothesized was potentially mediated by AMPK inhibition of the NLRP3 inflammasome (recently reviewed here ). However, canagliflozin dose-dependently reduced IL-1β in both WT and NLRP3−/− macrophages suggesting, that these effects are independent of the activation of the canonical NLRP3 inflammasome. Importantly, we found direct effects on macrophage IL-1β mRNA suggesting that these effects are occurring at a transcriptional level. However, other inflammasomes, or other regulatory pathways of caspase-1 could be contributing to reductions in IL-1β seen here. Interestingly, other SGTL2i's have been shown to reduce IL-1β in diabetic mice, however, this was determined to not be a direct effect on the macrophage, but rather a result of glucose lowering . How AMPK suppresses IL-1β expression is not understood but is an area of active investigation. Additionally, AMPK independent effects of canagliflozin on macrophage polarization (measured by Stat1 phosphorylation), as well as reductions in TNFα secretion, warrant further investigation for potential anti-inflammatory effects of canagliflozin.
Despite reductions in both circulating ApoB and IL-1β we did not see any changes in atherosclerosis with canagliflozin treatment (30 mg/kg) in female ApoE−/− mice. This is in contrast with two recent studies, which showed that canagliflozin reduces atherosclerosis in hyperglycemic male mice [72,73]. While there are many potential differences between studies one possibility for why we did not observe an effect on atherosclerosis compared with previous studies is because we utilized female mice that did not have pronounced hyperglycemia. As such we only observed modest reductions in fasting blood glucose compared with the ∼50% reduction observed in the previous reports [72–74]. Alternatively, there may be a yet to be identified sex-specific hormonal response to canagliflozin. Future studies investigating whether canagliflozin improves atherosclerosis in male mice independently of reductions in blood glucose, potentially by treating ApoE−/− SGLT2−/− mice, are warranted.
In conclusion, our data indicate that in addition to reducing blood glucose, canagliflozin inhibits liver fatty and sterol synthesis and circulating IL-1β, effects which may be important for reducing cardiovascular events in clinical populations. In addition, paradoxical effects on cholesterol synthesis and clearance pathways may provide important mechanistic insights to increases in LDLc seen clinically with canagliflozin. Whether reductions in liver lipid synthesis and macrophage IL-1β are important for the cardioprotective effects of canagliflozin warrant further investigation.
The authors declare that there are no competing interests associated with the manuscript.
These studies were supported by research grants from the Canadian Institutes of Health Research (201709FDN-CEBA-116200 to GS) and Diabetes Canada (DI-5-17-5302-GS).
E.A.D., R.J.F., J.S.V.L, J.D.S. and G.R.S. designed the experiments. E.A.D., R.J.F., J.H.L., R.L., L.L., E.M.D. and A.E.G. performed the experiments and testing. E.A.D., R.J.F. and G.R.S. wrote the manuscript. All authors edited the manuscript and provided comments. G.R.S. is the guarantor of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
We thank Dr. Thomas J. Hawke for allowing access to the microscope, and Kevin P. Foley for technical assistance. This study was supported by a grant from the Diabetes Canada (G.R.S.). E.A.D was a recipient of an Ontario Graduate Scholarship (Queen Elizabeth II Graduate Scholarship in Science and Technology) and a Douglas C. Russell Memorial Scholarship, J.D.S is supported by a Canada Research Chair in Metabolic Inflammation. G.R.S. is supported by a Tier 1 Canada Research Chair, the J. Bruce Duncan Chair in Metabolic Diseases and a Diabetes Canada Investigator Award (DI-5-17-5302-GS). These studies were supported by research grants from the Canadian Institutes of Health Research (201709FDN-CEBA-116200 to GS) and Diabetes Canada (DI-5-17-5302-GS).
AMP-activated protein kinase
bone marrow-derived macrophages
de novo lipogenesis
glucose tolerance test
homeostatic model assessment of insulin resistance
respiratory exchange ratio
sodium-glucose cotransporter 2 inhibitors
sterol response element-binding protein1c