Inhibition of Microsomal Triglyceride Transfer Protein Expression and Apolipoprotein B100 Secretion by the Citrus Flavonoid Naringenin and by Insulin Involves Activation of the Mitogen-Activated Protein Kinase Pathway in Hepatocytes

HepG2 cells were obtained from the American Type Culture Collection (Rockville, MD) and grown as described previously (21). For experiments, HepG2 cells were plated in either 100-mm or six-well (35-mm) culture plates from Falcon Scientific (VWR, Mississauga, ON) and maintained in minimal essential medium containing 5% human lipoprotein-deficient serum (LPDS). Naringenin and wortmannin (obtained from Sigma, St. Louis, MO) were solubilized in DMSO (concentration in cell cultures did not exceed 0.5% for any of the treatments). Bovine pancreatic insulin (Sigma) was solubilized in 0.01 N HCl. MEK1/2 and MAPK^p38 inhibitors and their nonactive isoforms UO126, UO124, SB203580, and SB202474 were purchased from Calbiochem (San Diego, CA) and solubilized in DMSO.

HepG2 cells grown in 100-mm culture dishes were preincubated for 30 min with UO126 (10 μmol/l), UO124 (10 μmol/l), SB203580 (10 μmol/l) or SB202474 (10 μmol/l), or wortmannin (1 μmol/l) followed by a further 6 h with either insulin (100 nmol/l) or naringenin (100 μmol/l). Total RNA was isolated using Trizol reagent (Invitrogen, Mississauga, ON). Custom oligonucleotide primers were generated corresponding to human MTP, human apoB, and human GAPDH as previously reported (22). mRNA was quantitated using an RNAse protection assay described previously (22). Data are expressed as band intensity for MTP or apoB compared with that obtained for GAPDH and converted to percent of control.

ApoB secretion into the media was measured by immunoblot analysis as previously described (9). HepG2 cells grown in six-well (35-mm) culture dishes were preincubated for 30 min with UO126 (10 μmol/l), UO124 (10 μmol/l), SB203580 (10 μmol/l), or SB202474 (10 μmol/l) followed by a further 23.5 h with either insulin (100 nmol/l) or naringenin (100 μmol/l). Media apoB100 was determined after 4.5% SDS-PAGE, transfer to polyvinylidene difluoride membranes, and immunoblot analyses (9).

ERK1/2 or MAPK^p38 phosphorylation and protein levels were measured by immunoblot analyses using a phospho-specific ERK1/2 or MAPK^p38 antibody and an ERK1/2 or MAPK^p38 antibody (all from Cell Signaling Technology, Beverly, MA). HepG2 cells were grown in six-well (35-mm) culture dishes and were incubated overnight in serum-free minimal essential medium containing 0.5% wt/vol insulin-free fatty acid–free BSA (Sigma) to induce quiescence. Initially, the effect of insulin (100 nmol/l) or naringenin (100 μmol/l) on ERK1/2 or MAPK^p38 phosphorylation was measured at 15, 30, and 60 min. In subsequent ERK1/2 experiments, cells were preincubated for 30 min with UO126 (10 μmol/l), UO124 (10 μmol/l), SB203580 (10 μmol/l), or SB202474 (10 μmol/l) followed by 30 min with or without insulin (100 nmol/l) or naringenin (100 μmol/l). Cells were lysed in 500 μl lysis buffer (20 mmol/l Tris-HCl [pH = 7.4], 50 mmol/l NaCl, 1 mmol/l EDTA, 1 mmol/l EGTA, 1% Triton X-100, 25 mmol/l NaF, 1 mmol/l Na₃VO₄, 1 mmol/l PMSF, 0.1 mmol/l leupeptin, 100 units/ml aprotinin) and incubated on ice for 10 min. Lysates were homogenized by 15 passes through a 25-gauge needle followed by centrifugation (10 min, 10,000g, 4°C) to pellet cell debris. Aliquots of the supernatant, normalized to cell protein, were resolved by 12% SDS-PAGE, and proteins were electrophoretically transferred onto polyvinylidene difluoride membranes. Immunoblot analyses were then completed according to the manufacturer’s instructions (Cell Signaling Technology). Bands were visualized as described for apoB (10). ERK1/2 or MAPK^p38 phosphorylation was expressed as a ratio of the phosphorylated band to the protein band and converted to a percentage of control.

All values are presented as means ± SE of at least three experiments done in duplicate. Means were compared by t tests to determine statistical significance compared with control unless otherwise stated. P < 0.05 was considered significant.

Insulin decreases MTP expression via activation of the MAPK^erk pathway in HepG2 cells (16). Because we previously showed that naringenin induces a 32% decrease in MTP expression within 24 h (11), we determined whether naringenin mediates this effect via MAPK^erk activation. Incubation of HepG2 cells with naringenin (100 μmol/l) or insulin (100 nmol/l) for 6 h significantly decreased the expression of MTP to 67 and 45% of control, respectively (P < 0.01) (Fig. 1A). Preincubation with the selective MEK1/2 inhibitor, UO126 (10 μmol/l), completely blocked the naringenin-induced inhibition of MTP (P < 0.01 compared with naringenin alone) (Fig. 1A). In contrast, UO124 (10 μmol/l), the inactive derivative of UO126, did not significantly alter the level of MTP expression in the presence of naringenin (Fig. 1A). In addition, UO126 (10 μmol/l) partially restored the insulin-induced decrease in MTP expression up to 77% of control (P < 0.01 compared with insulin alone), whereas the null isoform (UO124) did not alter the insulin effect.

Activation of the MAPK^p38 side-pathway normally decreases the activation of the MAPK^erk pathway and therefore acts as a dampener or fine-tunes the signal. When MAPK^p38 was inhibited with the selective inhibitor SB203580 (10 μmol/l), the basal level of MTP expression was significantly decreased to 52% of control (Fig. 1B). However, the addition of naringenin or insulin in the presence of the MAPK^p38 inhibitor did not potentiate the decrease in MTP expression after 6 h (Fig. 1B). Also, SB202474 (10 μmol/l), the inactive derivative of SB203580, did not alter MTP expression on its own or in the presence of naringenin or insulin. Preincubation of cells with the PI 3-kinase inhibitor wortmannin did not affect MTP mRNA levels in the presence of naringenin or insulin (data not shown).

To confirm that naringenin and insulin were activating the MAPK^erk/MAPK^p38 pathway, we examined the time course of ERK1/2 or MAPK^p38 phosphorylation by naringenin and insulin. HepG2 cells were preincubated with naringenin (100 μmol/l) or insulin (100 nmol/l) for 15, 30, or 60 min, and the phosphorylation of ERK1/2 or MAPK^p38 relative to total ERK1/2 or MAPK^p38 protein levels, respectively, were compared with control values. At all time points, there was basal phosphorylation of ERK1/2 or MAPK^p38 in control cells (Fig. 2A and B). Naringenin had no effect at 15 min, whereas at 30 min, phosphorylation of ERK1/2 increased by 1.8-fold (P < 0.05). In contrast, the phosphorylation of ERK1/2 by insulin peaked earlier, reaching 2.9-fold over control by 15 min (P < 0.05), and levels were still significantly elevated 2.1-fold at 30 min (P < 0.05). Values for both compounds returned to baseline by 60 min (Fig. 2A). Naringenin significantly increased the phosphorylation of MAPK^p38 2.3-fold at 15 min (P < 0.05) and stayed at similar levels up to 60 min. Similarly, the phosphorylation of MAPK^p38 by insulin was enhanced 3.4-fold by 15 min (P < 0.05) and also remained elevated up to 60 min (Fig. 2B).

ERK1/2 phosphorylation was determined in HepG2 cells preincubated for 30 min in the absence or presence of the MEK1/2 inhibitor UO126 (10 μmol/l) or the nonactive isoform UO124 (10 μmol/l), followed by a further 30-min incubation with naringenin (100 μmol/l) or insulin (100 nmol/l) (Fig. 3A). ERK1/2 phosphorylation were abolished in the presence of UO126 confirming that this compound inhibits MEK1/2, whereas the inactive UO124 had no significant effect. Both naringenin (P < 0.05) and insulin (P < 0.01) significantly increased the phosphorylation of ERK1/2, which was completely inhibited in cells preincubated with UO126 (P < 0.01 for both). The nonactive isoform had no effect on the naringenin-induced increase in ERK1/2 phosphorylation; however, it reduced ERK1/2 phosphorylation relative to levels observed with insulin alone (P < 0.01).

HepG2 cells were preincubated for 30 min in the absence or presence of the MAPK^p38 inhibitor SB203580 (10 μmol/l) and the nonactive isoform SB202474 (10 μmol/l), followed by further 30-min incubation with naringenin (100 μmol/l) or insulin (100 nmol/l) (Fig. 3B). ERK1/2 phosphorylation was significantly increased in the presence of SB203580, whereas this increase was not observed with SB202474 (P < 0.01). Naringenin and insulin significantly increased the phosphorylation of ERK1/2 compared with control (P < 0.05). Preincubation with the MAPK^p38 inhibitor caused a significant increase in the naringenin-induced ERK1/2 phosphorylation; however, this was not greater than the inhibitor alone. In addition, SB203580 caused a significant 14-fold induction in insulin-induced ERK1/2 phosphorylation, compared with insulin or the inhibitor alone (P < 0.01). In contrast, the addition of SB202474 to naringenin- or insulin-treated cells had no significant effect on ERK1/2 phosphorylation.

As previously reported (9), incubation of HepG2 cells with naringenin (100 μmol/l) or insulin (100 nmol/l) for 24 h resulted in a significant decrease in the secretion of apoB100 to 31 and 56% of control, respectively (P < 0.01) (Figs. 4A and B). Preincubation of cells with the MEK1/2 inhibitor (UO126 [10 μmol/l]) alone increased basal apoB100 secretion to 145% of control (Fig. 4A), consistent with inhibition of basal ERK1/2 phosphorylation (Fig. 3A). The UO126 compound blocked ∼50% of the naringenin-induced inhibition of apoB100 secretion (P < 0.05 compared with naringenin alone), whereas UO126 plus insulin completely blocked the insulin-induced decrease in apoB100 secretion such that secretion reached 121% of control (P < 0.05 compared with insulin alone). Conversely, the inactive isoform, UO124 (10 μmol/l), had no significant effect under any experimental condition.

Incubation of HepG2 cells with the MAPK^p38 inhibitor, SB203580 (10 μmol/l), resulted in a significant decrease in secreted apoB100 to 59% of control (P < 0.01), and the inactive inhibitor slightly reduced apoB100 to 87% of control (not significant) (Fig. 4B). Preincubation with SB203580 caused apoB100 secretion to decrease a further 24% in the presence of naringenin and similarly enhanced the insulin-induced decrease by 25% (P < 0.05 compared with naringenin or insulin treatment alone, respectively). The inactive SB202474 did not significantly alter the naringenin- or insulin-mediated decrease in apoB100 secretion (Fig. 4B).

To determine if any of the treatments influenced apoB100 secretion through effects on apoB expression, we measured apoB mRNA abundance. HepG2 cells were treated with naringenin (100 μmol/l), insulin (100 nmol/l), UO126 (10 μmol/l), UO124 (10 μmol/l), SB203580 (10 μmol/l), SB202474 (10 μmol/l), or wortmannin (1 μmol/l) for 6 h. UO124 and SB202474 modestly increased apoB mRNA to 109 and 129% of control, respectively (P < 0.05), whereas other values were unchanged from control (data not shown).

Previously, we demonstrated that incubation of HepG2 cells with the PI 3-kinase inhibitor wortmannin prevented most of the insulin-mediated decrease in apoB100 secretion (9). However, the naringenin-induced reduction in apoB100 secretion was only partially blocked. Therefore, we investigated the effect of combined inhibition of MAPK^erk and PI 3-kinase on apoB100 secretion over a 24-h period. Preincubation of HepG2 cells with wortmannin (1 μmol/l) resulted in a significant 30% increase in the basal levels of apoB100 secretion (P < 0.01) (Fig. 5). Preincubation with wortmannin followed by incubation with naringenin or insulin blunted the naringenin-induced inhibition of apoB100 secretion by 31%, whereas the insulin effect was completely blocked such that apoB100 levels reached 126% of control (P < 0.05 compared with insulin alone, but similar to wortmannin alone). Preincubation of HepG2 cells with both wortmannin (1 μmol/l) and UO126 (10 μmol/l) before the addition of naringenin completely attenuated the naringenin effect relative to control; however, levels remained significantly below (50%) those observed for co-incubation with the two inhibitors alone. In comparison, the combined inhibition of PI 3-kinase and MEK1/2 in the presence of insulin resulted in complete ablation of the insulin effect and stimulated apoB100 secretion to 219% of control levels (P < 0.05), a value similar to that observed for co-incubation with the two inhibitors alone.

The prevalence of insulin resistance and obesity is increasing rapidly worldwide and represents an increased risk for the development of dyslipidemia and ultimately atherosclerosis (23). Often the insulin resistance is due to defective intracellular insulin signaling; therefore, treatments that can potentially overcome this defect may prove very useful (24–27). Previously, we have shown that the grapefruit flavonoid naringenin activates the insulin-stimulated PI 3-kinase pathway within hepatocytes and does so independent of IRS-1/2 phosphorylation (9). We now extend those findings and show for the first time that naringenin decreases the expression of MTP via activation of the insulin-sensitive MAPK^erk pathway and specifically by phosphorylation of ERK1/2 within this pathway. Importantly, a significant proportion of the naringenin-mediated decrease in hepatic apoB100 secretion also depends on activation of the MAPK^erk pathway and is modulated by MAPK^p38 activation. Furthermore, simultaneous inhibition of PI 3-kinase and MEK1/2 demonstrated that both pathways contribute significantly to the naringenin-stimulated decrease in apoB100 secretion. Insulin was previously shown to decrease MTP expression via the MAPK^erk pathway (16); however, our results clearly identify that part of the insulin-mediated decrease in apoB100 secretion is also via activation of the MAPK^erk pathway, an effect modulated by MAPK^p38 activation. These results are summarized in Fig. 6.

We had previously reported a significant 30% decrease in MTP mRNA expression after 24-h exposure of HepG2 cells to 200 μmol/l naringenin (11). In the present set of experiments, 100 μmol/l naringenin decreased MTP expression by 33% after 6 h, suggesting a rapid response to this flavonoid. Furthermore, insulin rapidly decreased MTP mRNA expression by 55% after 6 h. The naringenin- and insulin-induced decreases in MTP expression both require activation of MAPK^erk. Our observations for insulin are consistent with a recent report demonstrating that insulin inhibits an MTP-luciferase reporter construct in HepG2 cells within 6 h (16). Furthermore, naringenin and insulin decreased apoB100 secretion by 69 and 44%, respectively, after 24 h, which also required MAPK^erk signaling. Whether the rapid and significant changes we observe in MTP mRNA translate into a significant decrease in MTP protein activity and confer the decrease in apoB100 secretion after 24 h is not entirely clear. The half-life of the MTP protein in HepG2 cells is reported to be 4.4 days (6). Previously, we demonstrated that naringenin is a direct inhibitor of MTP activity, which we concluded contributed to the rapid decrease in apoB100 secretion observed with this flavonoid (8). Indeed, pulse-chase experiments revealed similar kinetics for apoB100 secretion in HepG2 cells exposed to naringenin or an MTP inhibitor (10). In contrast, insulin does not inhibit MTP activity (6). This difference provides an explanation for why insulin induces greater ERK1/2 activation but decreases apoB100 secretion less than naringenin. Collectively, these results indicate that decreased MTP expression by naringenin and insulin may not mediate acute changes in apoB100 secretion. Nevertheless, our novel observation that MAPK^erk and MAPK^p38 mediate a significant proportion of the acute decrease in media apoB100 directly links signaling through these pathways to hepatic apoB100 secretion.

The identity of the MAPK^erk-sensitive pathway, responsible for the acute regulation of apoB100 secretion, remains to be elucidated. Activation of MAPK^erk and/or MAPK^p38 may directly affect apoB stability and/or degradation. It has previously been reported that insulin and naringenin do not alter the levels of apoB mRNA, which was confirmed in this study (11,28). However, pulse-chase experiments revealed that naringenin increases intracellular degradation of newly synthesized apoB100 in HepG2 cells (8). Early experiments using primary rat hepatocytes suggested that phosphorylation of apoB100 could be directly increased by insulin stimulation resulting in decreased apoB100 secretion (29–32). It is possible that MAPK^erk activation directly alters the phosphorylation of apoB100 and both naringenin and insulin decrease its secretion, in part through this mechanism. Whether apoB100 contains the docking domains potentially required for direct MAPK^erk-mediated apoB100 phosphorylation (33) remains to be elucidated.

The downregulation of MTP expression via the MAPK^erk pathway we identified with naringenin and insulin likely results in longer-term adaptive decreases in MTP protein and activity leading to reduced apoB100 secretion. In support of this concept, we observed a significant decrease in MTP protein after a 5-day exposure of HepG2 cells to naringenin (11). Furthermore, the fructose-fed insulin-resistant Syrian Golden hamster displays significantly higher MTP expression and protein levels in the liver and intestine and overproduction of apoB-containing lipoproteins (4,34). Correction of the insulin resistance in this model results in normalization of MTP expression and apoB secretion (5). The mechanism(s) for naringenin- or insulin-induced decreases in MTP expression through MAPK^erk activation are not completely understood. SREBP-2 is known to negatively regulate MTP expression (35). Recently, it was reported that insulin activates the MAPK^erk cascade resulting in the phosphorylation of SREBP-2, thereby increasing its activity (36). Whether the naringenin-induced activation of MAPK^erk acts through a similar mechanism remains to be determined.

There is evidence that SREBP-1a and -2 are downstream effectors of the MAPK^erk cascade leading to transactivation of the LDL-r gene (36,37). Furthermore, increased LDL-r expression decreases hepatocyte apoB100 secretion (38). In contrast, we previously demonstrated in HepG2 cells that the naringenin- and insulin-induced stimulation of LDL-r expression was completely blocked by inhibitors of PI 3-kinase, demonstrating that LDL-r expression was entirely upregulated by this pathway (9). However, it is unlikely that inhibition of apoB secretion in vivo by insulin or naringenin occurs via LDL-r upregulation, since apoB secretion was inhibited similarly by both compounds in wild-type and LDL-r (−/−) mouse hepatocytes (9). Furthermore, inhibition of PI 3-kinase only partially recovered the naringenin-induced decrease in apoB100 secretion from HepG2 cells (9). Importantly, in the present study, we discovered that both PI 3-kinase and MEK1/2 pathways contribute significantly to the decreased apoB100 secretion induced by naringenin or insulin. Our results suggest that regulation of apoB100 secretion by naringenin is partly mediated (50%) through activation of both PI 3-kinase and MAPK^erk, whereas the insulin effect is entirely through activation of these two pathways. Furthermore, we show that even in the absence of naringenin or insulin, MAPK^erk and MAPK^p38 regulate hepatocyte apoB100 secretion. Basal phosphorylation of ERK1/2 inhibits apoB100 secretion, which is attenuated by basal activity of MAPK^p38 (Figs. 2 and 4).

Emerging evidence suggests that aberrant MAPK^p38 signaling occurs in insulin-resistant or type 2 diabetic skeletal myocytes, adipocytes, and hepatocytes (16,26,27,39–41). Stimulation of MAPK^p38 normally inhibits MEK1/2, thereby decreasing signaling through this pathway, and we show here that both naringenin and insulin increase activation of MAPK^p38 and MAPK^erk. This suggests that acute regulation of apoB100 secretion reflects a balance between activation of MAPK^erk and MAPK^p38 signaling pathways. Our results indicate that activation of MAPK^erk by either naringenin or insulin predominates under normal conditions. However, in insulin-resistant hepatocytes, MAPK^p38 is chronically activated (26) and thus the normal insulin-induced MEK1/2-mediated decrease in MTP expression and apoB100 secretion may not occur. This concept is consistent with elevated MTP expression, apoB100 overproduction, and dyslipidemia in animal models of insulin resistance (4,42). Therefore, naringenin, which appears to bypass the insulin receptor (9), may restore normal activation of the MAPK^erk pathway.

Naringenin stimulated the phosphorylation of ERK1/2 but not as potently as insulin. In addition, the time to reach peak activation of ERK1/2 phosphorylation was longer for naringenin (30 min) compared with insulin (15 min). The variation in the time course of ERK1/2 activation suggests that the mechanism whereby naringenin activates the MAPK^erk pathway differs from that of insulin. This concept is consistent with our previous observation that naringenin activates PI 3-kinase independent of IRS-1/2 phosphorylation, suggesting activation of insulin signaling independent of the insulin receptor (9). The small membrane-bound GTPase Ras is the classic upstream activator of the Raf-1/MEK/ERK MAPK^erk cascade (33) and can be activated either dependent on or independent of insulin receptor phosphorylation (43) (Fig. 6). A recent report showed that insulin inhibits MTP expression in HepG2 cells via Raf-1 but does so independent of Ras (16). Therefore, it is possible that naringenin activates the MAPK^erk pathway via Ras, which may account for the longer time required for ERK1/2 phosphorylation compared with insulin.

Collectively, we demonstrate here that naringenin, like insulin, inhibits the expression of MTP via MEK1/2 activation and ERK1/2 phosphorylation. Importantly, naringenin and insulin cause an acute decrease in the secretion of apoB100 from HepG2 cells mediated by MAPK^erk. Furthermore, MAPK^p38 is directly linked to apoB100 secretion such that blocking its activity enhances the naringenin- and insulin-induced activation of MAPK^erk and decreases apoB100 secretion (Fig. 6). The activation of these insulin-signaling pathways in hepatocytes indicates that naringenin and related compounds may provide insights into novel treatments for insulin resistance in humans, especially if activation of peripheral kinases in these signaling cascades occurs independently of the insulin receptor and IRS-1/2.

This work was supported by an operating (T-4386) and program (PG-4854) grant from the Heart and Stroke Foundation of Ontario (to M.W.H.). E.M.A. is the recipient of a joint Astra Zeneca/Heart and Stroke Foundation of Canada research fellowship.