Potential conflict of interest: Nothing to report.
Accumulation of unfolded and misfolded proteins in the endoplasmic reticulum (ER) results in ER stress and lipid overload-induced ER stress has been implicated in the development of insulin resistance. Here, evidence is provided for a molecular link between hepatic apolipoprotein B100 (apoB100), induction of ER stress, and attenuated insulin signaling. First, in vivo upregulation of hepatic apoB100 by a lipogenic diet was found to be closely associated with ER stress and attenuated insulin signaling in the liver. Direct in vivo overexpression of human apoB100 in a mouse transgenic model further supported the link between excessive apoB100 expression and hepatic ER stress. Human apoB100 transgenic mice exhibited hypertriglyceridemia and hyperglycemia. In vitro, accumulation of cellular apoB100 by free fatty acid (oleate) stimulation or constant expression of wild-type or N-glycosylation mutant apoB50 in hepatic cells induced ER stress. This led to perturbed activation of glycogen synthase kinase 3 and glycogen synthase by way of the activation of c-Jun N-terminal kinase and suppression of insulin signaling cascade, suggesting that dysregulation of apoB was sufficient to disturb ER homeostasis and induce hepatic insulin resistance. Small interfering (si)RNA-mediated attenuation of elevated apoB level in the apoB50-expressing cells rescued cells from lipid-induced ER stress and reversed insulin insensitivity. Conclusion: These findings implicate apoB100 as a molecular link between lipid-induced ER stress and hepatic insulin resistance. (HEPATOLOGY 2009.)
Perturbations in lipid metabolism and lipid signaling underlie the pathogenesis of a cluster of chronic metabolic diseases, including insulin resistance, type 2 diabetes, fatty liver disease, and atherosclerosis. The atherogenic dyslipidemia commonly associated with insulin-resistant states consists of hypertriglyceridemia, a high level of very low-density lipoprotein (VLDL), a low level of high-density lipoprotein (HDL) cholesterol,1 and elevated small, low-density lipoprotein (LDL). High dietary fat intake has been shown to induce insulin resistance (IR) and the lipid synthetic rate by way of increased free fatty acid (FFA) flux, as well as assembly and secretion of both VLDL-apolipoprotein B (apoB) and triglyceride (TG) in animal and human models.2, 3 The association of IR and increased VLDL secretion is thought to be derived from increased FFA delivery to the liver resulting from increased lipolysis in adipose tissue, a phenomenon accompanied by increased hepatic lipogenesis, increased hepatic microsomal triglyceride transfer protein (MTP) level and activity, and loss of apoB regulation by insulin.
Recent studies implicate hepatic ER stress as a central abnormality linking obesity, hepatic IR, and hepatic steatosis.4, 5 Ozcan et al.4 have demonstrated that obesity-induced ER stress leads to hepatic IR by activating c-Jun N-terminal kinase (JNK) through inositol-requiring enzyme-1 (IRE-1), with subsequent inhibition of insulin receptor signaling. ER stress has also been linked to increased hepatic lipogenesis.6, 7 However, loss of MTP activity in mouse hepatocytes by either gene disruption or chemical inactivation exhibits marked accumulation of hepatic TGs without inducing typical ER stress markers.8 Even with the knock-down of acyl-coenzyme A:diacylglycerol acyltransferase 2 (DGAT2), an enzyme that catalyzes the last step of mammalian TG synthesis, oleate (OA) induces the same degree of Grp78 expression and eIF2α phosphorylation.9 This suggests that TG itself does not cause ER stress.
ApoB100 is essential for the hepatic assembly and secretion of TG-rich lipoproteins. It is a major protein component of plasma VLDL, IDL, and LDL. Numerous reports have demonstrated increased secretion of apoB-lipoproteins in response to increased hepatic FAs and/or TG.9, 10 A recent study that focused on the association of secreted apoB and FAs/TG demonstrated a parabolic relationship between lipid-induced ER stress and apoB secretion.9 However, how lipids induce ER stress and its effect on the metabolic processes remains an open question.
Because hepatic apoB protein synthesis is thought to far exceed its secretion,8, 10 conditions associated with excessive synthesis and accumulation of hepatic apoB in the secretory pathway, such as the addition of exogenous FA or TG, may directly induce ER stress. Hence, we postulated that excessive apoB production may act as a molecular link between high fat diet (lipid overload) and hepatic ER stress in diet-induced IR. To test this hypothesis, we investigated the role of apoB in the development of ER stress and IR using a diet-induced insulin-resistant hamster model. We further investigated the mechanisms linking apoB overproduction, lipid-induced ER stress, and perturbations in insulin signaling cascades in human apoB100 transgenic mouse model and apoB overexpressing cell models.
apoB, apolipoprotein B100; ATF4, activating transcription factor 4; ER, endoplasmic reticulum; FFA, free fatty acids; FFC, fructose, fat, and cholesterol; GS, glycogen synthase; GSK-3, glycogen synthase kinase 3; HDL, high-density lipoprotein; IR, insulin resistance; IRE-1, inositol-requiring enzyme-1; IRS, insulin receptor substrate; JNK, c-Jun N-terminal kinase; LDL, low-density lipoprotein; MTP, microsomal triglyceride transfer protein; OA, oleate; PERK, double stranded RNA-activated protein kinase-like ER kinase; PKB, protein kinase B/Akt; UTR, untranslated region; VLDL, very low-density lipoprotein.
Materials and Methods
ApoB Expression Plasmids.
Human apoB48, apoB50WT, and apoB50N158-1496 were generous gifts from Dr. Zemin Yao (University of Ottawa Heart Institute).11
Immunoprecipitation and Immunoblot Analyses.
In general, all immunoprecipitations and immunoblotting were performed as described.12 The antibodies used in this study are listed in the Supporting Information.
Luciferase Reporter Assays.
Luciferase reporter assays using the full-length mouse activating transcription factor 4 (ATF4) RNA leader wild-type (WT) or the ATF4 uORF1(Δ1) mutant fused to luciferase reporter gene13 were performed as described.14
Perturbation in Hepatic ER Function, Insulin Action, and ApoB Production In Vivo Induced by a Lipogenic Diet.
To examine the potential associations between hepatic apoB, insulin signaling, and ER stress, hamsters were fed a diet rich in fructose, fat, and cholesterol (FFC diet) for 4, 8, and 16 days. FFC-fed hamsters had significantly higher plasma TG and cholesterol levels (Supporting Fig. 1A,B) at all timepoints. The plasma apoB100 level was significantly increased by about 50% (P < 0.05) in the FFC diet over chow-fed controls (Supporting Fig. 1C). Histological analysis showed visible enlargement of the liver, an increase in cell size, and lipid accumulation in the liver of hamsters fed with the FFC diet when compared to the control chow diet at 2 weeks (Supporting Fig. 1D). Further examination of the liver extracts of FFC-fed hamsters showed significantly elevated apoB100 level throughout the 16-day time course (Fig. 1A). In contrast, the protein level of another VLDL component, apolipoprotein E (apoE), was not induced by the FFC diet, nor was the HDL protein component apolipoprotein AI (apoAI) (Fig. 1A), suggesting that apoB was the major apolipoprotein affected by this lipogenic diet. Together, these results demonstrated that increased apoB100 level was one of the essential hepatic responses to lipid loading.
Biosynthesis of apoB100 and its assembly and secretion as a lipoprotein particle (VLDL) is a highly regulated process that occurs in the ER and Golgi. Accumulation of apoB100 in the ER may thus exert an impact on the ER. Activations of the ER resident eIF2α kinase PERK, its substrate eIF2α, and JNK have been widely used as indicators of ER stress.15, 16 Phosphospecific antibodies against eIF2α and JNK revealed increased phosphorylations of eIF2α and JNK in the liver extracts of the FFC-fed hamsters (Fig. 1B), suggesting induction of hepatic ER stress by the FFC diet.
As described in the Supporting Information, our hamsters exhibited symptoms of IR after 2 weeks of FFC feeding. We then decided to investigate whether the development of IR is correlated with the elevation of apoB100 and ER stress. Compared to the chow group, basal phosphorylation of Akt on Ser473 was slightly (but not significantly) reduced in the FFC-fed hamster livers (Supporting Fig. 2). With insulin stimulation, there was a greater and statistically significant reduction in the phosphorylation of Akt on Ser473 and Thr308 as well as a dramatic decrease in total Akt level in the FFC-fed hamster livers (Fig. 1C). At first glance, the ratio of phosphorylated Akt to its total Akt level appeared to be higher in the FFC-fed hamster livers than the chow control (1.2 versus 1.0). However, the overall amount of phosphorylated Akt was markedly lower in the FFC-fed animals. In addition, our total Akt immunoblot indicated multiple hyperphosphorylated Akt isoforms in the chow-fed hamster livers, whereas only one Akt band was observed in the FFC group (Fig. 1C, middle panel), suggesting that Akt might not be fully activated and may therefore be functionally defective in the livers of FFC-fed hamsters.
Glycogen synthase kinase 3 (GSK-3) is one of the Akt substrates and a primary regulator of glycogen synthase (GS).17 It exists as two ubiquitously expressed isoforms, GSK-3α (51 kDa) and GSK-3β (46 kDa), and is constitutively active in unstimulated cells. Activation of Akt by insulin phosphorylates GSK-3α at Ser21 and GSK-3β at Ser9, resulting in the inactivation of GSK-3.18 Corresponding to our previous results on Akt activity, phosphorylation of GSK-3α/β at Ser21/9 was significantly reduced in the FFC-fed hamsters compared to the chow-fed controls (Fig. 1D), indicating a hyperactivated GSK-3α/β activity in the FFC-fed hamster liver.
Accumulation of ApoB in Hepatic ER-Induced ER Stress.
Lipid overload has been shown to induce ER stress and regulate apoB100 secretion.9 However, how lipids relay their effects to the ER remains to be determined. To investigate the potential link between lipid overload, cellular apoB protein production, and ER stress, we treated rat hepatoma McA-RH7777 (McA) cells with OA for 16 hours. Incubation of McA cells with OA induced a bulk accumulation of apoB100 without affecting apoB48 or another secretory protein albumin (Fig. 2A). Simultaneously, increased expression of an ER stress marker, Grp78, and enhanced phosphorylation of PERK and eIF2α were observed in the OA-stimulated cells but not the nontreated cells (Fig. 2A). To further confirm the role of apoB100, a small interfering (si)RNA against rat apoB100 was employed to silence the cellular apoB100 in McA cells. A 90% knock-down of apoB100 prevented the OA-induced eIF2α phosphorylation in McA cells (Fig. 2B). These results suggested that, in response to increased lipid delivery to hepatic cells, a significant amount of cellular apoB100 accumulated in the ER and induced ER stress.
A human apoB100 transgenic mice model19 was employed to further determine the effect of apoB100 overexpression on hepatic ER function in vivo. The transgenic mice overexpressing human apoB100 exhibited hypertriglyceridemia and hyperglycemia (Supporting Fig. 3A,C), but hypoinsulinemia on normal chow diet. Defective insulin action was also observed in the transgenic mice as indicated by the reduced basal phosphorylation of Akt (Supporting Fig. 3E). The presence of exogenous apoB100 in these mouse hepatocytes was associated with an increased hepatic Grp78 level and elevated eIF2α and JNK phosphorylations (Fig. 2C). On the other hand, The expression of apoAI, a control lipoprotein protein, was unaffected. In vitro constitutive expression of either C-terminal truncated apoB constructs, the apoB50WT, or the apoB50N158-1496 mutant, which is more prone to misfolding and ER retention,11 in McA cells induced Grp78 expression and eIF2α phosphorylation (Fig. 2D). Transient transfection of a shorter apoB isoform, apoB48 (48% of apoB100), in McA cells also induced ER stress (Supporting Fig. 4). In contrast, overexpression of another ER-resident protein, MTP, was unable to induce ER stress as indicated by the lack of effect on eIF2α phosphorylation at day 2 and day 8 posttransfection (Supporting Fig. 4). To verify the impact of apoB in the induction of ER stress, an siRNA against human apoB was employed to target the exogenous human apoB50 in McA cells. ApoB siRNA significantly decreased human apoB50 without affecting the control protein albumin (Fig. 2E). Reduction of apoB50 rescued cells from ER stress, as indicated by reduced eIF2α phosphorylation in the targeted cells compared to the control cells (Fig. 2E).
ATF4 is a fundamental transcription factor whose translation is ER stress-mediated.20, 21 To further characterize the status of ER stress in the apoB50-expressing cells, we examined the translation efficiency of ATF4 messenger (m)RNA, a process solely dependent on eIF2α phosphorylation.13 ApoB50WT or N158-1496-expressing cells and their control cells were transfected with a fusion luciferase reporter construct containing either ATF4 5′-untranslated region (UTR) or an ATF4 5′-UTR mutant lacking ORF1 function (Δ1) as negative control.13 eIF2α phosphorylation induced by the ER accumulation of apoB50WT or N158-1496 efficiently stimulated the activity of the reporter gene (Fig. 2F), but no induction of luciferase activity was observed in cells transfected with the ATF4 mutant ORF1 (Δ1)/luciferase (Fig. 2F), supporting that accumulation of apoB in the ER was sufficient to perturb ER homeostasis.
ER Stress Induced by ApoB Overload Impeded Insulin Action Through JNK-Mediated Phosphorylation of IRS-1.
The insulin receptor substrate (IRS-1) is a substrate of insulin receptor tyrosine kinase. In vitro study with the Fao liver cells has shown that ER stress induced by tunicamycin or thapsigargin inhibits insulin signaling through JNK.4 Consistent with our observation that JNK was activated in the FFC-fed hamster (Fig. 1B), JNK phosphorylation was significantly induced in McA cells incubated with OA (Fig. 3A). A similar observation was made in the apoB100 transgenic mouse hepatocytes and in apoB50WT and apoB50N158-1496-expressing cells (Figs. 2C, 3B). Phosphorylation of the JNK substrate, c-Jun at Ser73, was also enhanced in the apoB50-expressing cells (Fig. 3B).
JNK has been shown to phosphorylate IRS-1 at Ser307. Indeed, phosphorylation at Ser307 was increased in apoB50-expressing cells (Fig. 3C), which was accompanied by a reduction in the insulin-induced IRS-1 tyrosine phosphorylation (Fig. 3D). Insulin stimulation of apoB50-expressing cells induced much lower phosphorylation of Akt on Ser473 compared to the control cells (Fig. 3E). These results were in line with the in vivo data (Fig. 1B-D), and indicated that lipid- or apoB-induced ER stress inhibited insulin receptor signaling by way of JNK.
ApoB Overload Disturbed Hepatic Glycogenesis.
GSK-3 has been demonstrated to phosphorylate four of nine GS regulatory serine residues (Ser641, Ser645, Ser649, and Ser653) in response to insulin stimulation. This phosphorylation event plays a critical role in inhibiting GS activity and hence glycogen synthesis.22 Determination of GSK-3 activity showed that phosphorylation of GSK-3α/β on Ser21/9 was much lower in the apoB50-expressing cells (Fig. 4A) compared to the mock-transfected cells. Interestingly, phosphorylation at GSK-3α Ser21 appeared to be more sensitive to the presence of apoB than GSK-3β (Fig. 4A, top panel). Corresponding to the hyperactivation of GSK activity, the phosphorylation of GS on Ser641/645 was increased, hence less active, in apoB-expressing cells upon insulin stimulation (Fig. 4B). These results supported the notion that lipid-induced apoB accumulation in liver cells perturbed glycogenesis and contributed to the development of hepatic insulin resistance.
Inhibition of ApoB Overproduction Reversed Perturbations in Insulin Signaling.
Next, we were interested in knowing whether reducing apoB from an abnormally high level back to a physiological level would recover insulin action in our apoB-overexpressing models. Using the siRNA approach in Fig. 2E, we found that, consistent with its effect on releasing ER stress, knocking down human apoB50 in McA cells restored insulin signal transduction. The reduction of apoB50 protein level increased insulin-stimulated phosphorylations of Akt and GSK3α/β by 2.5-fold and 2.0/1.4-fold, respectively (Fig. 5A,B). These results further supported a potential link between apoB accumulation in the ER and hepatic insulin sensitivity.
Hepatic VLDL production is primarily substrate-driven, the most important regulatory substrate being FFA.23In vivo studies in humans support an important role of plasma FA delivery to the liver as a stimulus for apoB and TG secretion.24In vitro, OA has been shown to stimulate the assembly and secretion of VLDL even though TG synthesis and MTP activity were inhibited.25 One of the potential mechanisms for this increased VLDL secretion is increased availability of newly synthesized apoB due to enhanced co- and posttranslational stability, without changes in the apoB mRNA level.26 Thus, under constant lipid stimulation (e.g., high fat diet), enhanced apoB stability may eventually lead to the accumulation of apoB in the ER, perturbing ER homeostasis.
In the present study we examined the specific impact of apoB overproduction in lipid-induced ER stress and hepatic insulin resistance by employing a hamster model and cell culture systems. In vivo, feeding hamsters with the FFC diet induced dyslipidemia. The significantly elevated plasma TG and cholesterol levels in FFC-fed hamsters led to increased FFA flux into the liver and hepatic VLDL-apoB overproduction (data not shown). At the early stage of the FFC diet, from 4 to 8 days, the apoB protein level was increased despite the induction of eIF2α phosphorylation by ER stress (Fig. 1A, top panel). At day 16, although we observed a smaller increase in the apoB level compared to those at days 4 and 8, it was still significantly higher than that in the control group (Fig. 1A, top panel). The potential mechanisms for apoB elevation may include enhanced intracellular stability of newly synthesized apoB and/or reduced turnover of nascent apoB in hepatocytes.27 Development of fatty liver induced by the FFC diet at 2 weeks may contribute to the reduction in hepatic apoB100 at day 16 (Supporting Fig. 1D; Fig. 1A, top panel). Phosphorylation of eIF2α, which inhibits translational initiation, may also be involved in inhibiting apoB protein synthesis at this stage (day 16). We also noticed a slight increase that was not statistically significant in apoB100 at day 4 in the control chow-fed hamster livers and plasma compared to day 0.
Physiologically, constitutive apoB synthesis in the liver is essential for rapid assembly of lipids into lipoprotein particles for export. However, apoB synthesis in hepatocytes has been demonstrated to be much higher than what is secreted.28 Therefore, prolonged exposure of hepatocytes to lipids can result in a large pool of nascent apoB molecules retained in the ER, leading to ER stress. Indeed, we found that activation of ER stress markers by lipid overload was concomitant with in vitro or in vivo up-regulation of apoB. In vivo, eIF2α phosphorylation and JNK activation were coupled to the accumulation of cellular apoB100 (Fig. 1A,B). This association was further strengthened by an in vitro experiment demonstrating the inability of OA to induce ER stress when endogenous apoB was knocked down by RNA interference (Fig. 2B). Overproduction of apoB100 in vivo or apoB50 in vitro effectively activated the ER stress markers (Fig. 2C,D). Consistent with its liability to misfolding and ER retention, expression of apoB N158-1496 mutant showed a stronger impact on the ER than the WT apoB, suggesting that perturbations in ER function by apoB parallel its accumulation (Fig. 2C).
Studies using mouse models have yielded data linking ER stress and systemic insulin action.4 Here, we showed that lipid-induced apoB overload exerted its effect on the insulin signaling pathway by inducing ER stress and activating JNK (Fig. 3A,B). The FFC diet or apoB overexpression reduced insulin signaling as indicated by decreased Akt activation (Figs. 1C, 3E). The reduced Akt protein level in vivo may further compromise insulin action. Potential mechanisms for this reduction could be suppression of Akt gene expression or enhanced protein degradation in the development of insulin resistance in the FFC model.
Lipid and glucose metabolisms are closely associated in vivo. In insulin resistance, skeletal muscle exhibits defects in insulin-stimulated suppression of glycogenesis, diverting energy from ingested carbohydrate into increased hepatic de novo lipogenesis.29 GSK-3 is one of the key kinases in regulating glucose metabolism. Expression and activity of GSK-3 are elevated in diabetic rodents and humans, with consequential suppression of GS activity.30 Genetic targeting studies have demonstrated that GSK-3α and β display tissue-specific physiological functions. GSK-3α exerts a unique and essential role in hepatic glucose homeostasis,31 whereas GSK-3β is the primary muscle GSK.32 Although there was a consistent increase in the kinase activity of GSK-3α as a consequence of declined Akt activity in the livers of FFC-fed hamsters and cells constitutively expressing apoB, GSK-3β appeared less affected by exogenous apoB (Fig. 4A). These findings implicated GSK-3α as a point of convergence between lipid and glucose metabolism in FFC-induced insulin resistance. Hyperactivation of GSK-3 inhibited GS activity and reduced glycogen synthesis, thus perturbing glucose metabolism (Fig. 4B).
GSK-3 has been shown to catalyze phosphorylation of IRS-1 on Ser332, thus interfering with receptor-mediated tyrosine phosphorylation by the insulin receptor and attenuating insulin receptor signaling through negative feedback modulation.33 Therefore, in addition to the negative regulatory effect of JNK on IRS1 Ser307 (Fig. 3C), the hyperactive GSK-3 (Figs. 1D, 4A) may also negatively regulate the insulin signaling cascade. Our findings pointed to a fundamental mechanism underlying the molecular sensing of lipid overload-induced hepatic insulin resistance. Overproduction of apoB100 may play an important role in promoting lipid-induced ER stress that in turn impacts insulin signaling.
We thank Dr. R. Wek for the ATF4 (WT) and ATF4 uORF1(Δ1) luciferase cDNA constructs, and Dr. Z. Yao for the human apoB48 cDNA construct and the apoB50WT and apoB50N158-1350-expressing McA cells.