Inflammatory NF-κB activation promotes hepatic apolipoprotein B100 secretion: evidence for a link between hepatic inflammation and lipoprotein production

Julie Tsai, Rianna Zhang, Wei Qiu, Qiaozhu Su, Mark Naples, Khosrow Adeli


Insulin-resistant states are commonly associated with chronic inflammation and hepatic overproduction of apolipoprotein B100 (apoB100), leading to hypertriglyceridemia and a metabolic dyslipidemic profile. Molecular mechanisms linking hepatic inflammatory cascades and the pathways of apoB100-lipoprotein production are, however, unknown. In the present study, we employed a diet-induced, insulin-resistant hamster model, as well as cell culture studies, to investigate the potential link between activation of hepatic inflammatory nuclear factor-κB (NF-κB) signaling cascade and the synthesis and secretion of apoB100-containing lipoproteins. Using an established insulin-resistant animal model, the fructose-fed hamster, we found that feeding fructose (previously shown to induce hepatic inflammation) for as little as 4 days reduced hepatic IκB (inhibitor of NF-κB) level, indicating activation of the inflammatory NF-κB cascade. Importantly, IKK (IκB kinase) inhibition was found to suppress apoB100 overproduction in fructose-fed hamster hepatocytes. As IKK, the upstream activator of NF-κB has been shown to inhibit insulin signaling, and insulin is a major regulator of apoB100, we modulated IKK activity in primary hamster hepatocytes and HepG2 cells and assessed the effects on hepatic apoB100 biosynthesis. Inhibition of the IKK-NF-κB pathway by BMS345541 and activation of the pathway by adenoviral-mediated IKK overexpression decreased and increased newly synthesized apoB100 levels, respectively. Pulse-chase and metabolic labeling experiments revealed that IKK activation regulates apoB100 levels at the levels of apoB100 biosynthesis and protein stability. Inhibition of the IKK-NF-κB pathway significantly enhanced proteasomal degradation of hepatic apoB100, while direct IKK activation led to reduced degradation and increased apoB100 mRNA translation. Together, our results reveal important links between modulation of the inflammatory IKK-NF-κB signaling cascade and hepatic synthesis and secretion of apoB100-containing lipoproteins. Hepatic inflammation may be an important underlying factor in hepatic apoB100 overproduction observed in insulin resistance.

  • fructose
  • insulin resistance
  • hamster
  • proteasomal degradation
  • triglycerides
  • cholesterol
  • translation

a major metabolic complication of insulin-resistant states (such as obesity and type 2 diabetes) is a profound dyslipidemia, which contributes to increased cardiovascular disease risk (1). One key mechanism underlying diabetic or metabolic dyslipidemia is hepatic overproduction of apolipoprotein B100 (apoB100), the major protein component of triglyceride-rich very-low-density lipoproteins (VLDL) (1). apoB100 is a 550-kDa amphipathic glycoprotein that is secreted from the liver, acting as a transport protein to shuttle hepatic lipid store to peripheral tissues (6). Under most metabolic stimuli, apoB secretion is regulated posttranscriptionally (8). Final apoB100 output is largely determined by availability and recruitment of neutral lipids, which govern co- and posttranslational stability of newly synthesized apoB100 and help prevent its degradation by the proteasome (5). The initial lipidation is catalyzed by microsomal triglyceride transfer protein (MTP) (28). Further lipidation at later presecretory stages is also essential for maturation of apoB100-containing lipoproteins. Several enzymes, including acyl-coenzyme A-diacylglycerol acyltransferase (DGAT) 1 and 2, which catalyze the final step in triglyceride synthesis, have been suggested to participate in this maturation process (17).

Insulin is known as an important physiological regulator of hepatic lipid metabolism and acutely inhibits hepatic apoB secretion (9, 32). Although the inhibitory effect of insulin on apoB100 levels appears to involve the phosphatidylinositol 3-kinase signaling cascade, Akt1, a classical downstream effector of phosphatidylinositol 3-kinase, has been shown to have no effect on apoB100 secretion (4). Parallel signaling cascades, such as the mitogen-activated protein (MAP) kinases ERK, p38, and JNK, have been demonstrated to affect apoB secretion (2, 3, 21, 30, 37), but knowledge regarding their downstream mechanisms remains limited. Since MAP kinases can be stimulated by both insulin and inflammatory cytokines, and considerable cross talk exists between the insulin signaling cascades and inflammatory pathways, inflammation could play an important role in the overproduction of apoB100-containing lipoproteins observed in insulin-resistant states.

Insulin resistance is associated with chronic low-grade inflammation (14). It has been suggested that inflammation may precede and induce insulin insensitivity in glucose-metabolizing tissues, because inflammatory nuclear factor-κB (NF-κB) and JNK signaling pathways can phosphorylate insulin receptor substrates (IRS) 1/2 at serine residues, consequently blocking the insulin signal (16, 33). In addition, in diet-induced insulin resistance, there is an increased flux of free fatty acids, as well as adipose and macrophage-derived cytokines, into the bloodstream (13). These all contribute to the chronic activation of inflammatory pathways and can lead to cellular perturbations and apoptosis in various tissues.

Inflammatory cytokines and fatty acids are known to activate NF-κB. There are five members in this family: NF-κB1 (p50), NF-κB2 (p52), RelA (p65), RelB, and c-Rel. The most common form of NF-κB is the NF-κB1/RelA heterodimer (24). Under basal conditions, NF-κB exists in a complex with inhibitor of NF-κB (IκB) in the cytoplasm. Upon activation of upstream inflammatory pathways, IκB is phosphorylated by the IκB kinase (IKK) complex, ubiquitinated, and degraded, allowing NF-κB to translocate into the nucleus and bind its target genes (24). There are three subunits in the IKK complex: IKK-α, IKK-β, and IKK-γ. IKK-α and -β are both catalytic, but IKK-β is the main subunit in the classical IKK-NF-κB cascade (18). There is evidence that this pathway can also regulate fatty acid oxidation (29). In addition, NF-κB is activated by protein kinase C (11), a signaling molecule that has been shown to regulate apoB100 translation (32) and is upregulated in insulin resistance (31).

To elucidate the potential link between activation of hepatic inflammation and regulation of apoB100 production, we first investigated the activation state of inflammatory signaling molecules in the livers of hamsters fed a high-fructose diet. High-fructose feeding has been shown previously to induce hepatic inflammation (22), hepatic insulin resistance (35), and apoB100 overproduction (34, 35). We then further studied the effects of modulating IKK activity on apoB100 secretion in primary hamster hepatocytes and a human hepatoma cell line, HepG2. Our results suggest that apoB100 production can be regulated by the inflammatory IKK-NF-κB pathway, providing novel insights to how insulin resistance, inflammation, and dyslipidemia may be linked.


Animal protocols.

HepG2 cells were from American Type Culture Collection (ATCC, Manassas, VA). Male Syrian golden hamsters (Mesocriceus auratus) were purchased from Charles River Canada (Montreal, QC, Canada). Hamsters were housed individually with 12-h light and dark cycles, as well as free access to food and water. All animal protocols followed national guidelines and were approved by the animal ethics committee of The Hospital for Sick Children (Toronto, ON, Canada). After acclimatization, animals were placed on either a normal hamster diet (chow) or high-fructose diet (60% fructose, Dyets, Bethlehem, PA) for 0–16 days. At the end of the feeding period, hamsters were anesthetized by isoflurane, and their livers were infused for 2 min with 5 U/kg insulin (Eli Lilly, Toronto, ON, Canada) or saline via the vena cava. Livers were then homogenized with solubilization buffer [150 mM NaCl, 10 mM tris(hydroxymethyl)aminomethane (pH 7.4), 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 1% Nonidet P-40, 100 mM sodium fluoride, 10 mM sodium pyrophosphate, 2 mM sodium orthovanadate, and cocktail protease inhibitor (Roche Applied Science, Indianapolis, IN)]. Alternatively, after a 2-wk feeding period, primary hamster hepatocytes were isolated.

Liver perfusion and isolation of primary hamster hepatocytes.

Hamsters were anesthetized with isoflurane, their livers perfused with collagenase, and hepatocytes were isolated as previously described (34). Isolated hepatocytes were incubated in 5% fetal bovine serum (FBS; Hyclone, Logan, UT), 1% l-glutamine (Wisent, St. Bruno, QC, Canada), 0.5% penicillin/streptomycin (Wisent)-containing William's E medium (Gibco, Invitrogen, Burlington, ON, Canada), and 10 μU/mL insulin at a density of 1.2 × 106 cells/well in six-well plates. After 4 h, medium was changed to 20% FBS-containing Dulbecco's modification of Eagle's medium (DMEM, Wisent).

Cell culture.

Primary hepatocytes and HepG2 cells were maintained in 20% or 5% FBS-containing DMEM, respectively, at 37°C, 5% CO2. For adenovirus-mediated β-galactosidase (β-gal) (control) or wild-type IKK-β (Vector BioLabs, Philadelphia, PA) overexpression experiments, HepG2 cells were seeded on collagen-coated plates at 1 × 106 cells/35-mm2 plate the day before infection.

Treatments and adenoviral infections.

Primary hamster hepatocytes were treated overnight with a dimethylsulfoxide (DMSO) vehicle control or a final concentration of 10 μM IKK inhibitor BMS345541 (Sigma-Aldrich, Oakville, ON, Canada). For IKK overexpression, hepatocytes were infected with either β-gal or wild-type IKK-β adenovirus at a multiplicity of infection (MOI) of 0–20 for 1 day. Primary hepatocytes subjected to metabolic radiolabeling were either treated with DMSO or 10 μM BMS345541 or infected with β-gal or IKK adenovirus at MOI 5. HepG2 cells were more sensitive to chemical inhibitors and were, therefore, treated with BMS345541 at 0.5 μM for a total of 2–4 h, depending on the following cotreatments. For IKK overexpression in HepG2 cells, cells were infected with β-gal or IKK adenovirus at MOI 10 for 1 day. To investigate the involvement of proteasomal degradation, primary hamster hepatocytes were incubated with 25 μM MG132, and HepG2 cells were incubated with 20 μg/ml proteasomal inhibitor N-acetyl-leucyl-leucyl-norleucinal (ALLN; Sigma-Aldrich) from the 1 h methionine and cysteine-free prepulse period before metabolic radiolabeling. For intracellular lipid determinations, HepG2 cells were incubated with media supplemented with 360 μM oleic acid (OA) for 4 h.

Luciferase assay.

HepG2 cells were cotransfected with NF-κB-pGL2 firefly luciferase construct that was a generous gift from Dr. G. Lukacs (Department of Physiology, McGill University, Montreal, QC) and Renilla luciferase construct (Promega). Cells were transfected using Lipofectamine (Invitrogen) for 2 days before luciferase activities were measured using the Dual Luciferase-Reporter Assay kit (Promega). Firefly luciferase activities were normalized to Renilla luciferase activities to account for transfection efficiencies.

Metabolic labeling of primary hepatocytes and HepG2 cells.

To study newly synthesized apoB100, cells were incubated in methionine and cysteine-free DMEM (Wisent), supplemented with 5% FBS for HepG2 or 20% FBS for primary hepatocytes for 1 h (prepulse), followed by 100 or 50 μCi/ml [35S]methionine pulse labeling for primary hamster hepatocytes or HepG2 cells, respectively. For the detailed pulse chase in primary hepatocytes, a 30-min pulse was performed, followed by a 0- to 120-min chase with 20% FBS containing DMEM. For ALLN treatment in HepG2 cells, total pulse was 1.5 h. For VLDL-apoB100 isolation, primary hepatocytes were pulse labeled for 2 h. Total protein synthesis was assessed by trichloroacetic acid (TCA) precipitation and scintillation counting.

Cell-free in vitro translation.

Cell-free in vitro translation was performed as previously described (36). In short, HepG2 cells treated with BMS345541 or infected with IKK adenovirus were incubated in 5% FBS-containing methionine and cysteine-free DMEM for 1 h. The cells were then washed twice with ice-cold buffer A [150 mM RNase-free sucrose, 33 mM NH4Cl, 7 mM KCl, 1.5 mM Mg(OAc)2, 30 mM HEPES, 150 mM, pH 7.4], and perforated using 150 μg/ml lysolecithin (Sigma-Aldrich). Permeabilized cells were scraped into in vitro translation buffer [100 mM HEPES, 200 mM KCl, 7 mM NH4Cl, 0.5 mM Mg(OAc)2, 1 mM DTT, 1 mM spermidine trihydrochloride, 10 mM creatine phosphate, 1 mM ATP, 1 mM GTP, 40 μM amino acid mix (−methionine), 0.1 mM S-adenosylmethionine, 40 U/ml creatine phosphokinase, pH 7.4] and centrifuged at 12,000 rpm for 1 min. The supernatants were collected, and in vitro translation was performed by incubating the supernatants with 400 μCi/ml [35S]methionine for 1 h at 30°C. Total protein synthesis was assessed by TCA precipitation and scintillation counting. apoB100 and control protein albumin were immunoprecipitated as described below.

VLDL-apoB100 isolation.

Chow or fructose-fed hamster hepatocytes were treated with BMS345541 overnight and pulse labeled for 2 h. Cells were lysed, and the media were subjected to ultracentrifugation for 18 h at 35,000 rpm. The top 10% of each sample was collected as VLDL and immunoprecipitated for apoB100. As a control, transferrin was immunoprecipitated from the cell lysates and whole media.

Immunoprecipitation and chemiluminescent immunoblot analysis.

Cell lysates and media were immunoprecipitated for apoB100, or albumin, or transferrin with rabbit anti-hamster apoB, or goat anti-human apoB, or goat anti-human albumin, or goat anti-human transferrin antibody (Midland Bioproducts, Boone, IA), followed by zysorbin addition (Zymed, South San Francisco, CA) for immunocomplex pull-down. Immunoprecipitates, or cell lysates, or plasma samples were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotted for the following proteins at indicated primary antibody dilutions: IκBα (1:1,000; Cell Signaling), phospho-ERK (1:1,000; Cell Signaling), ERK2 (1:1,000; Santa Cruz), phospho-p38 (1:1,000; Cell Signaling), p38 (1:10,000; Stressgen), phospho-JNK (1:1,000; Cell Signaling), JNK (1:1,000; Cell Signaling), ubiquitin (1:500; Stressgen), β-actin (1:1,000; Sigma-Aldrich), and TNF-α (1:1,000; Cell Signaling). Bands visualized with enhanced chemiluminescence (PerkinElmer, Shelton, CT) were quantified by densitometry.


Total RNA was extracted using QIAGEN RNeasy mini kit (Mississauga, ON). cDNA was synthesized using Applied Biosystems TaqMan Reverse Transcription Reagents (Foster City, CA). RT-PCR was performed using the following sets of primers to measure the mRNA levels of apoB, MTP, DGAT1, and DGAT2: apoB 5′AGACAGCATCTTCGTGTTTCAA, 3′ATCATTTAGTTTCAGCCCAGGA; MTP 5′TAATCCTGGGAGGACTTGAAAA, 3′CCTTCTCCTGCTTCTGCATACT; DGAT1 5′TCCACTCCTGCCTGAATGC, 3′AAGTGTCTGATGCACCACTTG; DGAT2 5′CAGCTACAGGTCATCTCAGTGC, 3′CAAACACCAGCCAAGTGAAGTA; and 18S 5′TAAGTCCCTGCCCTTTGTACACA; 3′GATCCGAGGGCCTCACTAAAC.

Triglyceride and cholesterol assays.

BMS345541-treated or IKK-overexpressing HepG2 cells were treated with OA for 3.5 h. Intracellular lipids were then extracted by hexane-isopropanol (3:2), dried, and resuspended in ethanol. Triglyceride and cholesterol levels were determined by triglyceride and cholesterol assay kits (Randox, Mississauga, ON, Canada).

Fluorography by phosphorimaging.

Newly synthesized apoB100 molecules were visualized by phosphorimaging and quantified with ImageQuant software (Molecular Dynamics, GE Healthcare Life Sciences, Piscataway, NJ).

Statistical analysis.

Data obtained by densitometry or fluorography were analyzed using t-tests or ANOVAs, depending on the experimental conditions. All results are presented as means ± SE. Asterisks (*) indicate statistically significant differences (P < 0.05) compared with respective controls.


Fructose feeding increases hepatic NF-κB activity.

To investigate the relationship between fructose-induced insulin resistance and inflammation, hamsters were fed a high-fructose diet for 0–16 days. The fructose-fed hamster model has been extensively characterized (34, 35) and exhibits hypertriglyceridemia, hepatic insulin resistance, and VLDL overproduction. Following a 2-min insulin infusion, hamster livers were harvested, solubilized, and immunoblotted for various signaling molecules. The insulin-induced phosphorylation of the insulin receptors, IRS-1, IRS-2, and Akt, have been previously published (34, 35) and are thus not presented here. While insulin infusion resulted in Akt phosphorylation in chow-fed hamster livers at all time points (data not shown), there were significant decreases in total Akt level in as little as 4 days and Akt phosphorylation in 8 days with fructose feeding (data not shown), demonstrating that induction of hepatic insulin resistance could be achieved within a short period of fructose feeding. Interestingly, fructose feeding also resulted in a significant reduction in IκB levels at 8 and 16 days (Fig. 1C, left), suggesting activation of NF-κB signaling in the livers of fructose-fed hamsters. Insulin appeared to further enhance this reduction in IκB level in fructose-fed hamsters (Fig. 1C, right). Insulin is known to also stimulate the mitogenic ERK pathway (15). However, ERK activity upon insulin treatment was significantly lower with 4- and 8-day fructose feeding, indicating insensitivity in both canonical and mitogenic insulin cascades (Fig. 1). The basal activity of the stress MAP kinase p38 was low in chow-fed hamster livers (Fig. 1A). In contrast, the livers from hamsters fed fructose for 8 days had a significant increase in basal p38 activity (Fig. 1C, left), but insulin treatment appeared to normalize this activation (Fig. 1C, right). Surprisingly, JNK activity, which has been shown to be activated in sucrose-fed rats and fructose-treated rat hepatocytes (39, 40), did not change significantly with fructose feeding up to 16 days, either with or without insulin stimulation. It appears that short-term, high-fructose feeding activates the inflammatory NF-κB cascade and the stress p38 cascade, while downregulating both canonical and mitogenic arms of the insulin cascades.

Fig. 1.

Fructose feeding induces hepatic inhibitor of nuclear factor-κB (NF-κB) (IκB) kinase (IKK)-NF-κB activation. Syrian golden hamsters were fed a control chow or high-fructose diet for 0–16 days. Their livers were then harvested after a 2-min insulin or saline infusion in vivo. Liver homogenates were subjected to SDS-PAGE and immunoblotting to determine the activities of indicated signaling molecules. A: representative immunoblots of signaling molecules in chow-fed hamster livers, with or without insulin treatment. B: representative immunoblots of signaling molecules in fructose-fed hamster livers, with or without insulin treatment. C: quantifications of immunoblots in fructose-fed hamster livers relative to 0-day control, without (left) and with insulin stimulation (right). IκB expression was normalized to β-actin. Phospho (P) proteins were expressed as a ratio to their total protein levels. A total of three animals were used for each time point. Results are presented as means ± SE. *Statistically significant differences compared with 0-day feeding (P < 0.05) by one-way ANOVA followed by Tukey's multiple comparison test. D: plasma TNF-α level. 20 μl of 4-day chow or fructose-fed hamster plasma samples were subjected to immunoblot analysis. 100 ng of recombinant mouse TNF-α were used as a positive control. Results are presented as means ± SE from 3 animals. *Statistically significant difference compared with chow control (P < 0.05) by t-test.

To further confirm the inflammatory status of these hamster livers, we tried detecting the mRNA levels of several NF-κB target genes (IκBα, growth arrest and DNA damage-45β, IL-1, IL-6, TNF-α, transforming growth factor-α, X-linked inhibitor of apoptosis protein, and Bcl-xL) by RT-PCR. Unfortunately, our primers, which had to be designed from human, rat, and mouse sequence alignments (due to the lack of hamster gene sequence information), did not always yield PCR products. Nevertheless, among the two genes that we were able to assess, IκBα mRNA level showed a nonsignificant increase at 4- and 8-day fructose feeding, and Bcl-xL mRNA level was not significantly different between the diet groups at all time points (data not shown). We also immunoblotted for the proinflammatory cytokine TNF-α in our hamster plasma samples (Fig. 1D). The trimeric form of TNF-α was significantly elevated at 4-day fructose feeding (Fig. 1D, bottom), although there was no difference in plasma TNF-α level between the diet groups at 8 and 16 days (data not shown). Our laboratory has previously shown that the plasma free fatty acid level, another stimulus of the NF-κB pathway (13, 16), is increased in our 2-wk fructose-fed hamster model (34), suggesting that the fructose-enriched diet most likely induced systemic inflammation in our hamsters. The proinflammatory stimuli, however, may change during the course of the feeding.

Modulation of the inflammatory IKK-NF-κB pathway affects apoB100 production.

Since our previous studies have reported increased hepatic apoB100 production in fructose-fed hamsters, and the present study revealed increased hepatic NF-κB activation with fructose feeding, we postulated that NF-κB activation, an indicator of hepatic inflammation, may contribute to the increased hepatic apoB100 production. To examine the relationship between NF-κB activity and apoB100 production, primary hamster hepatocytes and HepG2 cells were treated with a specific chemical inhibitor of IKK, BMS345541 (Fig. 2, AC), subjected to metabolic radiolabeling, and newly synthesized apoB100 were detected from the cell lysates and culture media. BMS345541 decreased the level of newly synthesized apoB100 in primary hepatocytes (Fig. 2A) and the secretion of newly synthesized apoB100 by HepG2 cells (Fig. 2B). As free IκB (not bound to NF-κB) can be rapidly degraded, even in the absence of IKK activation (26), immunoblotting for IκB following BMS345541 treatment in the absence of a stimulus might not directly reflect NF-κB activity. Therefore, to confirm NF-κB inhibition by BMS345541, luciferase activity was measured in HepG2 cells transfected with a plasmid encoding luciferase under the control of an NF-κB promoter. A 2-h treatment with BMS345541 was sufficient to decrease luciferase activity by ∼60%, confirming inhibition of NF-κB activity by the chemical inhibitor (Fig. 2C). Immunoblot analysis suggested that IKK overexpression might have increased total cellular apoB100 level in primary hamster hepatocytes (Fig. 2D, right). To confirm this observation, metabolic pulse labeling was performed for the IKK-transduced cells. IKK overexpression increased the level of newly synthesized apoB100 level in both primary hepatocytes (Fig. 2D, bottom) and HepG2 cells (Fig. 2E, bottom), and increased apoB100 secretion by HepG2 cells (Fig. 2E). Interestingly, primary hepatocytes and HepG2 cells seemed to respond to IKK modulations differently. In primary hepatocytes, the IKK modulations seem to affect cellular apoB100 level more than secreted apoB100, while, in HepG2 cells, the effects appeared to be greater on secreted apoB100 than cellular apoB100. The reasons for this discrepancy in response are unclear, but may relate to potential differences in inflammatory signaling cascade activation.

Fig. 2.

Effect of modulating IKK-NF-κB on apolipoprotein B100 (apoB100) production in primary hamster hepatocytes and HepG2 cells. Primary hamster hepatocytes were incubated overnight with IKK inhibitor BMS345541 (BMS) at 10 μM, or infected with β-galactosidase (β-gal) or IKK adenovirus at indicated multiplicity of infection (MOI) for 1 day before immunoblotting or metabolic pulse labeling. HepG2 cells were incubated with BMS345541 from the prepulse period, or infected with β-gal or IKK adenovirus at indicated MOI for 1 day before metabolic pulse labeling. NF-κB activity was assessed by IκB immunoblot in primary hepatocytes and HepG2 cells or by NF-κB-luciferase activity. A: IKK inhibition by BMS345541 in primary hamster hepatocytes (3 animals; a total of 9 plates). B: IKK inhibition by BMS345541 in HepG2 cells (3 experiments; a total of 9 plates). C: NF-κB-luciferase activity in HepG2 cells treated with 0.5 μM BM345541 for 2 h. Firefly luciferase activities were normalized against renilla luciferase activities and presented in relation to DMSO control treatment (a total of 3 plates). D: IKK overexpression in primary hamster hepatocytes (4 animals; a total of 11 plates). E: IKK overexpression in HepG2 cells (2 experiments; a total of 5 plates). Results are presented as means ± SE. *P < 0.05 compared with DMSO or β-gal control by t-test.

IKK inhibition suppressed apoB100 overproduction in fructose-fed primary hamster hepatocytes.

We also investigated the effect of IKK inhibition on apoB100 synthesis and secretion in chow- or fructose-fed hamster hepatocytes. To account for animal-to-animal variability and possible global effects that IKK inhibition induced, apoB100 levels were normalized against TCA counts (TCA = TCA-precipitable counts; a measure of total labeled protein synthesized) or transferrin (a control protein). When apoB100 was normalized against TCA counts, both cellular and secreted VLDL-apoB100 levels were significantly higher in fructose-fed hamster hepatocytes (Fig. 3A, top). When apoB100 was normalized against transferrin, only secreted VLDL-apoB100 level was significantly higher from the fructose-fed hamster hepatocytes (Fig. 3, A and B, bottom). IKK inhibition by BMS345541 decreased cellular apoB100 level, normalized either way, in both chow- and fructose-fed hamster hepatocytes (Fig. 3, A and B, top). Interestingly, when normalized to TCA counts, BMS345541 was found to suppress the elevated cellular apoB100 level in fructose-fed hamster hepatocytes to a level comparable to that of DMSO-treated, chow-fed hamster hepatocytes (Fig. 3A, top). When VLDL-apoB100 secretion was normalized to transferrin, IKK inhibition reduced apoB100 secretion from the fructose-fed hamster hepatocytes to a level similar to DMSO-treated control hepatocytes (Fig. 3B, bottom). These results suggest that IKK inhibition may have the potential to reverse apoB100 overproduction observed in insulin resistance.

Fig. 3.

IKK inhibition suppressed the overproduction of apoB100 in fructose-fed primary hamster hepatocytes. Hepatocytes from hamsters fed with control chow or fructose diet for 2 wk were treated overnight with DMSO or BMS345541, followed by 2-h [35S]methionine pulse labeling. A: total protein synthetic rate was assessed by trichloroacetic acid (TCA) precipitation. B: very-low-density lipoprotein (VLDL)-apoB100 was obtained by collecting the top 10% of media samples ultracentrifuged at 35,000 rpm overnight. apoB100 and the control protein transferrin were immunoprecipitated from cell lysates and media (VLDL fractions and whole media, respectively) and visualized by fluorography. To account for animal-to-animal variation and any global effect of IKK inhibition, apoB100 levels were normalized against TCA counts (A) or transferrin (B). Results are presented as means ± SE (3 animals; a total of 9 plates). *P < 0.05 between BMS345541 treatment and DMSO control. #P < 0.05 between the fructose and chow control diets by two-way ANOVA followed by Bonferroni posttests.

IKK regulates apoB100 synthesis at an early stage.

To examine mechanisms by which the IKK-NF-κB cascade regulates apoB100 levels, we performed a detailed pulse-chase in primary hepatocytes treated with BMS345541 or overexpressing IKK to determine the effects on apoB100 synthesis and cellular accumulation. ApoB100 accumulation was lower overall in cells treated with the IKK inhibitor (Fig. 4A, inset). Conversely, cells overexpressing IKK had higher apoB100 accumulation compared with cells transduced with a β-gal control adenovirus (Fig. 4B, inset). Interestingly, adenoviral infection itself appeared to affect apoB100 synthesis as both β-gal and IKK infection shifted the highest peak on the cellular apoB100 graph from ∼45 min (Fig. 4A) to 20 min (Fig. 4B). This shift appeared to be partially contributed by the rapid degradation of cellular apoB100 as proteasomal inhibition by MG132 blunted the peak (Fig. 4D). Nevertheless, IKK-NF-κB activation appeared to stimulate apoB100 accumulation at an early stage, as IKK activation had statistically significant effects on cellular apoB100 levels, whether or not the proteasomal inhibitor was present (Fig. 4B, inset, Fig. 4D, inset). Moreover, IKK inhibition also appeared to affect apoB100 synthesis and degradation, as MG132 was able to rescue the BMS345541-induced apoB100 reduction to a nonstatistically significant level (Fig. 4C, inset). These results suggest that IKK affected apoB100 biogenesis at an early step of protein synthesis, possibly by influencing protein synthetic rate and/or protein stability/degradation.

Fig. 4.

Modulation of IKK-NF-κB pathway influences biosynthesis of newly synthesized apoB100. Primary hamster hepatocytes were preincubated with BMS345541 overnight or infected with IKK adenovirus (MOI 5) for 1 day. Detailed pulse chase was then performed, as described in materials and methods. A: newly synthesized apoB100 accumulation in cells and media treated with DMSO or BMS345541. Inset: total apoB100 accumulation, expressed as area under the curve (AUC). B: newly synthesized apoB100 accumulation in cells overexpressing β-gal or IKK and their culture media. Inset: total apoB100 accumulation, expressed as AUC. C: newly synthesized apoB100 accumulation in cells and media treated with DMSO or BMS345541 in the presence of MG132. Inset: total apoB100 accumulation, expressed as AUC. D: newly synthesized apoB100 accumulation in cells and media overexpressing β-gal or IKK in the presence of MG132. Inset: total apoB100 accumulation, expressed as AUC. Results are presented as means ± SE (3 animals; a total of 6 plates). *P < 0.05 against DMSO or β-gal control using two-way ANOVA followed by Bonferroni posttests for the line graphs and t-test for the AUC graphs.

IKK modulates apoB100 levels partially through inhibition of proteasomal degradation.

To determine whether the IKK-NF-κB pathway affected apoB100 levels by affecting proteasomal degradation, newly synthesized apoB100 was monitored in HepG2 cells in the presence or absence of the proteasomal inhibitor ALLN. HepG2 cells were used because they are known to actively degrade newly synthesized apoB100. HepG2 cells were either pretreated with BMS345541 or transduced with IKK and then treated with the proteasomal inhibitor ALLN. IKK inhibition by BMS345541 caused a ∼30% decrease in the secretion of newly synthesized apoB100, which was reversed upon proteasomal inhibition (Fig. 5A), suggesting that proteasomal degradation may be responsible for the BMS345541-induced apoB100 reduction. However, IκB degradation also relies on the proteasome (18), and, therefore, in the presence of ALLN, NF-κB activation may be impaired, which would result in reduced apoB100 levels. The fact that no significant difference was observed between the apoB100 levels in DMSO- and BMS345541-treated cells upon addition of ALLN could be the result of the dual effects of ALLN, protecting both apoB100 and IκB from proteasomal degradation. In contrast to BMS345541, IKK overexpression increased apoB100 secretion by ∼2.5-fold, an increase that could not be blocked by ALLN, as expected (Fig. 5A). Since proteasomal degradation accounts for the majority of apoB100 degradation in HepG2 cells (41), ALLN treatment would result in retention of most of the total amount of apoB100 synthesized during the pulse period. The increase in apoB100 secretion by IKK-overexpressing cells treated with ALLN compared with control cells treated with ALLN indicates that IKK increased the total amount of apoB100 synthesized.

Fig. 5.

Proteasomal inhibition blocks the suppressive effects of the IKK inhibitor BMS345541 on apoB100 secretion. A: newly synthesized apoB100 in IKK inhibited or activated cells in the presence or absence of N-acetyl-leucyl-leucyl-norleucinal (ALLN). HepG2 cells treated with BMS345541 or infected with IKK adenovirus were subjected to 1.5 h pulse in the presence or absence of proteasomal inhibitor ALLN. B: apoB100 ubiquitination in IKK inhibited or activated cells in the presence or absence of ALLN. The ratio of ubiquitinated apoB100 to total apoB100 was determined by immunoblotting for ubiquitin and apoB100 on samples immunoprecipitated for apoB100. Results represent the means ± SE of apoB100 or ubiquitinated apoB100 level (2 experiments; a total of 6 plates). *P < 0.05 against DMSO or β-gal control by two-way ANOVA followed by Bonferroni posttests. NT, no treatment; OA, oleic acid.

The activity of the ubiquitin-proteasome system was confirmed by immunoblotting for ubiquitinated apoB100. Surprisingly, there appeared to be less ubiquitination of apoB100 in BMS345541-treated cells, as indicated by the ubiquitinated apoB100-to-total apoB100 ratio (Fig. 5B, bottom). It is possible that IKK inhibition simply promoted apoB100 turnover to the extent that a lower accumulation of ubiquitinated apoB100 was observed. Cells overexpressing IKK also showed a trend toward a decrease in apoB100 ubiquitination, which was expected (Fig. 5B). ALLN treatment resulted in the accumulation of ubiquitinated apoB100 (Fig. 5B), confirming the efficacy of the proteasomal inhibitor. The fact that the differences in ubiquitinated apoB100 in cells with modulations of the IKK pathway compared with their respective controls were eliminated by ALLN treatment suggests that the IKK-NF-κB pathway was influencing apoB100 turnover. The unexpected decrease in ubiquinated apoB100-to-apoB100 ratio with IKK inhibition was likely due to increased rate of apoB100 degradation. Since there was a lack of difference in the ubiquitinated apoB100-to-total apoB100 ratios in the presence of ALLN, the rate of apoB100 ubiquitination was likely unaffected. Therefore, any change in the ratio in the absence of ALLN may likely indicate changes in the rate of apoB100 degradation. Overall, these results suggest that modulation of proteasomal degradation may partially explain the regulation of apoB100 production by the IKK-NF-κB pathway.

Intracellular lipid levels were unaffected by modulations of the IKK-NF-κB pathway.

Since the IKK-NF-κB pathway has been shown to affect fatty acid oxidation in cardiomyocytes (29), it is possible that IKK modulations in this study also affected lipid metabolism and the intracellular availability of triglyceride or cholesterol for lipoprotein assembly, which would, in turn, affect apoB100 stability and secretion. First, RT-PCR analysis of mRNA levels for apoB, MTP, DGAT1, and DGAT2 indicated that expression of their genes was not affected by IKK modulation (Fig. 6A). Fructose-fed hamster livers and hepatocytes also showed no change in the mRNA levels of these genes (except DGAT1, which our primers could not recognize; data not shown). Consistent with the lack of change in DGAT1 and DGAT2 mRNA levels, intracellular triglyceride level was unaffected by modulations of the IKK-NF-κB pathway, either in the presence or absence of OA (Fig. 6B, left). Cholesterol level also remained unchanged (Fig. 6B, right), suggesting that intracellular lipid availability was not involved in IKK-NF-κB-mediated modulation of apoB100.

Fig. 6.

Effect of modulating IKK on intracellular lipids and mRNA levels. HepG2 cells were treated with BMS345541 or infected with IKK adenovirus. A: mRNA levels of apoB, microsomal triglyceride transfer protein (MTP), diacylglycerol acyltransferase (DGAT) 1, and DGAT2. Total RNA was extracted after BMS345541 treatment or IKK adenoviral infection. mRNA levels were detected by RT-PCR using primers described in materials and methods. All expressions were normalized to 18S RNA and presented as percentage of DMSO or β-gal control. Results are presented as means ± SE (2 experiments; a total of 6 plates). B: intracellular triglyceride and cholesterol levels. HepG2 cells treated with BMS345541 or infected with IKK adenovirus were incubated with or without 360 μM OA for 4 h. Lipids were then extracted from the cells, and total intracellular triglyceride and cholesterol levels were determined. Results are presented as means ± SE (3 experiments; a total of 9 plates).

IKK affected apoB100 translation.

Since the IKK-NF-κB pathway did not affect intracellular lipid levels, and earlier experiments suggested a potential effect of IKK overexpression on apoB synthesis, we investigated whether IKK signaling influenced apoB100 mRNA translation by performing a cell-free in vitro translation assay. Since IKK modulations affected total protein synthesis as indicated by TCA (Fig. 7A), the amount of newly synthesized apoB100 was assessed by normalizing against levels of newly synthesized albumin (a control protein) or total TCA counts (reflecting total newly synthesized labeled protein) (Fig. 7B). Normalized to albumin translation, apoB100 translation was ∼20% lower with BMS345541 treatment compared with the DMSO vehicle control and ∼25% higher with IKK overexpression compared with β-gal control (Fig. 7B, top). If apoB100 levels were normalized to TCA counts, IKK inhibition by BMS345541 reduced apoB100 translation by ∼25%, and IKK overexpression increased apoB100 level by ∼45% (Fig. 7B, bottom). This indicates that, although total protein synthesis was affected by the IKK-NF-κB pathway, apoB100 synthesis was influenced to a larger extent than synthesis of other proteins. In contrast to the in vitro cell-free system, total newly synthesized proteins, as estimated by TCA counts, were not affected by IKK modulations in intact cells (Fig. 7C, top). Interestingly, BMS345541 treatment actually led to significantly higher transferrin level in primary hamster hepatocytes (Fig. 7C, left middle), suggesting that the IKK-NF-κB cascade may play a unique role in hepatic apoB100 synthesis and secretion.

Fig. 7.

Effect of IKK modulation on apoB100 protein synthesis. apoB100 protein synthesis was investigated by cell-free in vitro translation. Cell-free lysates were prepared from methionine-deprived HepG2 cells treated with BMS345541 or infected with IKK adenovirus. In vitro translation was performed by incubating the lysates with 400 μCi/ml [35S]methionine for 1 h at 30°C. A: total protein synthesis, determined by TCA precipitation and scintillation counting. B: in vitro translated apoB100 levels, normalized against control protein albumin levels (top) or TCA counts (bottom). Results are presented as means ± SE (2 experiments; a total of 4 plates). *P < 0.05 against DMSO or β-gal control by t-test. C: effect of IKK modulations on general protein synthesis in intact cells. Primary hamster hepatocytes (left) and HepG2 cells (right) were treated with DMSO or BMS345541, or infected with control β-gal or IKK adenovirus, followed by [35S]methionine metabolic labeling. Total protein synthesis was assessed by TCA protein precipitation (top), and the control protein transferrin was immunoprecipitated from cell lysates (middle) and media (bottom), subjected to SDS-PAGE, and visualized by fluorography. Results are presented as means ± SE (2 experiments; a total of 5 plates). *P < 0.05 compared with DMSO control by t-test.


Although hepatic apoB100 metabolism is thought to be perturbed in insulin resistance, resulting in hyperlipidemia and an increased risk of cardiovascular disease, the mechanisms linking insulin resistance to apoB100 overproduction are largely unknown. Since insulin resistance is associated with chronic inflammation, we postulated, in the present study, that dysregulation of inflammatory cascades could contribute to hepatic apoB100 overproduction. Evidence obtained using a previously established fructose-induced, insulin-resistant hamster model, as well as cultured primary hamster hepatocytes and HepG2 cells, appears to suggest that development of hepatic insulin resistance is associated with inflammatory NF-κB activation, and that apoB100 production is influenced by this pathway at the levels of translation and proteasomal degradation. We found that short-term fructose feeding is sufficient to induce systemic inflammation, as indicated by the early elevation in plasma TNF-α level that likely contributed to the activation of NF-κB in our hamster livers, as indicated by the decrease in IκB level (Fig. 1). Consistent with results of other insulin resistance or type 2 diabetes studies (10, 23), the livers of fructose-fed hamsters also exhibited an increase in basal p38 activation. However, the activation of ERK, another MAP kinase, was impaired with fructose feeding in the presence of insulin stimulation (Fig. 1). Since it has been shown that ERK inhibition upregulates hepatic apoB100-VLDL production (37), this decrease in ERK activity corresponds with the apoB100 overproduction previously reported in this model (34). One unexpected observation in this study was the lack of a significant change in JNK activation, since it has been previously demonstrated that fructose or sucrose feeding activates this signaling cascade in rats (22, 39, 40). It is possible that fructose feeding of hamsters could take longer to achieve the same degree of activation. The method we used to detect JNK activity might have also contributed to the difference between our observations, since Wei and Pagliassotti assessed the activity by an in vitro kinase assay (39), and we performed immunoblotting to detect phosphorylated JNK. Finally, the anti-JNK antibodies used in our study were raised against human JNK and may not adequately cross-react with hamster JNK (no hamster-specific JNK antibody is available).

The activation of the IKK-NF-κB pathway was not only associated with the development of insulin resistance, its inhibition by a chemical inhibitor and its activation by adenovirus-mediated IKK overexpression also, respectively, decreased and increased apoB100 production in our cultured cell models. Moreover, IKK inhibition was able to suppress the overproduction of apoB100 in fructose-fed hamster hepatocytes to a level comparable to the vehicle-treated control chow-fed hamster hepatocytes, indicating the importance of this signaling cascade in the induction of apoB100 overproduction in insulin resistance. The effects of IKK modulations on apoB100 level were not due to changes in apoB100 gene expression or intracellular lipid levels (Fig. 6), but were due to synergistic modulations of proteasomal degradation of apoB100 (Fig. 5) and apoB100 mRNA translation (Fig. 7). Although our results suggested that IKK activation protected apoB100 from proteasomal degradation, and IKK inhibition had the opposite effect, it was unexpected that both conditions resulted in reduced ubiquitinated apoB100 level. However, it is possible that IKK inhibition led to an increase in the delivery of ubiquitinated apoB100 to the proteasome, resulting in an apparent decrease in ubiquitinated apoB100.

ApoB100 is not the only protein that is negatively regulated by inhibition of the IKK-NF-κB pathway. Zangar et al. (42) have seen similar effects of the NF-κB pathway on CYP3A4 stability. NF-κB inhibition suppresses CYP3A4 production without decreasing CYP3A4 mRNA level. Unlike apoB100, however, proteasomal inhibition actually results in lower CYP3A4 level. The authors suggested that this is because the proteasome is essential in NF-κB activation, but is not a major proteolytic pathway for CYP3A4 (42). In contrast, the proteasome is a key catabolic regulator for apoB100. Proteasomal inhibition could thus reverse the loss of apoB100 induced by NF-κB inhibition in our study, but not CYP3A4.

IKK overexpression increased apoB100 synthesis at the level of mRNA translation without influencing transcription. Although IKK overexpression also increased overall protein synthesis, as indicated by TCA counts, it had a greater effect on apoB100 mRNA translation, demonstrating the susceptibility of apoB100 to overproduce in response to inflammation. The effect of IKK on global protein synthesis is not surprising, as the catalytic IKK subunits (both α and β) have been demonstrated to activate mammalian target of rapamycin (mTOR) by phosphorylating tuberous sclerosis complex (TSC)-1/2, and TSC-1/2 have been shown to suppress mTOR activity (12). Since 4E binding protein (4EBP) 1 and S6 kinase, key proteins in mRNA translation, are downstream effectors of mTOR (20), IKK overexpression would lead to increased protein synthesis. Interestingly, IKK modulation did not affect total protein synthesis in intact primary hamster hepatocytes or HepG2 cells. Moreover, IKK inhibition actually increased the control protein transferrin in primary hamster hepatocytes, indicating that apoB100 was regulated by the IKK-NF-κB cascade in a manner very different from other proteins.

One interesting observation we made in this study was that adenovirus infection itself seemed to affect apoB100 production. Adenoviral infection, using the control β-gal adenovirus, appeared to increase the rates of apoB100 synthesis and degradation, as indicated by a shift in accumulation of newly synthesized apoB100 to an earlier time point (Fig. 4B). Akt activation induced by the adenoviral backbone could have contributed to this effect (27). Since Akt is a known activator of mTOR, adenoviral infection on its own could affect apoB100 production at the level of synthesis. However, immunoblot analysis of cells transfected with the control β-gal adenovirus for a number of inflammatory pathway components did not show any evidence of adenoviral-induced inflammation (data not shown).

Although our results demonstrate regulation of apoB100 by the IKK-NF-κB pathway, many questions remain to be answered. First, modulations of this signaling cascade were performed at the level of IKK, not NF-κB. There is, therefore, a possibility that apoB100 levels were controlled directly by IKK, and that the downstream NF-κB activity was not required. There are several proteins that are influenced by IKK itself. For example, IRS1 is a well-known substrate of the IKK complex (16). IKK phosphorylates IRS1 at serine residues, thereby inhibiting its activity. Bcl10 and FOXO3a are two other such substrates for IKK. Both proteins have been shown to be phosphorylated by IKK, consequently leading to their proteolysis (19, 25). Furthermore, due to the extensive cross talk between the IKK-NF-κB cascade and other signaling pathways, such as those involving JNK and ERK (7, 38), it is possible that these other pathways could play a role in regulation of apoB100 levels.

In summary, our data suggest that hepatic apoB100 production is under the control of the inflammatory IKK-NF-κB signaling pathway. We have also demonstrated for the first time that proteasomal degradation and mRNA translation of apoB100 can be influenced by this signaling cascade. Our results contribute to the present understanding of the relationship between insulin resistance, inflammation, and lipid dysregulation. Nevertheless, more studies will be required to conclusively define the underlying mechanisms by which this pathway regulates apoB100 metabolism.


This work was supported by an operating grant (T-6658) from the Heart and Stroke Foundation of Ontario and NSERC to K. Adeli. J. Tsai is the recipient of the PGSD graduate scholarship from the Natural Sciences and Engineering Research Council of Canada and Peterborough K. M. Hunter graduate studentship of the University of Toronto.


View Abstract