Apo B secretion is regulated by hepatic triglyceride, and not insulin, in a model of increased hepatic insulin signaling

Abstract

Objective

States of insulin resistance, hyperinsulinemia, and hepatic steatosis are associated with increased secretion of triglycerides (TG) and apolipoprotein B (apoB), even though insulin targets apoB for degradation. We used hepatic-specific (h) Pten knockout (ko) mice, with increased hepatic insulin signaling, to determine the relative roles of insulin signaling and hepatic TG in regulating apoB secretion.

Results/Methods

TG and apoB secretion were elevated in hPten-ko mice. When hepatic TG was reduced by inhibition of DGAT1/DGAT2 or SREBP-1c, both TG and apoB secretion fell without changes in hepatic insulin signaling. Acute reconstitution of hPten reduced hepatic TG content and both TG and apoB secretion fell within 4 days despite decreased hepatic insulin signaling. Acute depletion of hepatic Pten by adenoviral introduction of Cre into Pten Floxed mice caused steatosis within 4 days, and secretion of TG and apoB both increased despite increased hepatic insulin signaling. Even when steatosis after acute Pten depletion was prevented by pre-treatment with SREBP-1c ASO, apoB secretion was not reduced after 4 days. Ex vivo results were in primary hepatocytes were similar.

Conclusion

Either hepatic TG is the dominant regulator of apoB secretion or any inhibitory effects of hepatic insulin signaling on apoB secretion is very short-lived.

Introduction

During the past 25 years, studies in cells, small animals, and humans have demonstrated a key role for hepatic insulin signaling in the regulation of very low density lipoprotein (VLDL) assembly and secretion ^1-3. In cultured hepatocytes, acute exposure of cells to insulin inhibits the secretion of both triglyceride (TG) and apolipoprotein B (apoB) despite concomitant stimulation of TG synthesis ^4-9. The acute inhibition of VLDL secretion by insulin results from direct targeting of apoB for degradation, at least in part, via PI3-kinase (PI3-K) mediated mechanisms ^10-12. In vivo, acute glucose-stimulated hyperinsulinemia suppressed secretion of VLDL in rats ¹³, while short-term hyperinsulinemia with euglycemia inhibited the secretion of VLDL in normal humans ^{14, 15}. On the other hand, chronic exposure of hepatocytes to insulin increased apoB secretion ^{6, 16, 17} suggesting the development of insulin resistance in the pathway for apoB degradation. Mouse models of hyperinsulinemia and insulin resistance have increased VLDL secretion ^{18, 19}. Similarly, VLDL secretion was not inhibited by short-term hyperinsulinemia in obese individuals ¹⁵ or people with type 2 diabetes mellitus ²⁰, two states characterized by insulin resistance.

Individuals with insulin resistance typically have increased secretion of both VLDL apoB and TG, despite ambient hyperinsulinemia ^21-23. However, insulin resistance is also associated with increased hepatic TG (steatosis) ^{24, 25}, and hepatic TG availability is a critical determinant of the assembly and secretion of VLDL ^26-28. This raises the question of whether increased apoB secretion present in people with insulin resistance results from the development of insulin resistance in the pathway of insulin-mediated targeting of apoB for degradation, or because increased availability of hepatic TG is dominant over increased hepatic insulin signaling ^{3, 29}. Relevant to this issue is our recent observation of dissociation between TG and apoB secretion in the complete absence of hepatic insulin signaling. Thus, in liver-specific insulin receptor knockout (LIRKO) mice, apoB secretion was significantly increased while TG secretion was markedly reduced ³⁰. These findings suggested that reduced TG-targeting of apoB for secretion was not as potent as the absence of insulin-targeting of apoB for degradation ¹. In the present studies, we took advantage of the availability of hepatic-specific (h) “phosphatase and tensin homologue deleted on chromosome 10” (Pten) knockout (ko) mice ³¹, which have markedly increased hepatic insulin signaling and steatosis, to further interrogate the relative roles of increased hepatic insulin signaling and increased hepatic TG availability in the assembly and secretion of VLDL. Of note, Stiles et al. showed previously that these hPten-ko mice have marked steatosis despite increased TG secretion, suggesting that insulin-targeting of apoB for degradation might be limiting the number of VLDL particles than can be secreted by these mice ³².

Materials and Methods

Animals

Liver-specific Pten knock out (hPten-ko) mice were generated as described ³². hPten-ko and Floxed littermate mice were maintained under a 12 hour light:dark cycle and had free access to water and standard rodent chow. All mice were studied at 3-4 months of age if not otherwise specified. C57BL/6J mice, used for primary hepatocyte studies were housed and fed in a similar manner. All procedures were approved by the Institutional Animal Care and Use Committee of Columbia University College of Physicians and Surgeons.

Antisense Treatments

Dgat1 (ISIS 191761), Dgat 2 (ISIS 217376), Srebp-1c( ISIS 219676) and control antisense oligonucleotide (ISIS 141923) were provided by ISIS Pharmaceuticals (Carlsbad, CA). ASO treatments were carried out as described ³³. Briefly, animals were injected intraperitoneally with ASO twice a week at a concentration of 25 ug per gram of body weight. Treatments were started at 12 weeks of age and were 4 weeks in duration.

Adenovirus treatments

A recombinant adenovirus containing a Pten cDNA (Pten) was kindly provided by Ramon Parsons (Columbia University, NY). A recombinant adenovirus containing a Cre-recombinase cDNA (Cre) under control of the albumin promoter was kindly provided by Derek Leroith (Mount Sinai School of Medicine, NY). A recombinant virus containing a GFP cDNA (GFP) was provided by Alan Tall (Columbia University, NY). Recombinant adenovirus was amplified in 293 cells and purified with a Vivapure Adenopack 100 column (Sartorius, Goettingen, Germany). Mice were injected with adenovirus through tail or femoral veins at the concentration of 4 × 10 ⁴ particles per gram of body weight. Mice were studied and sacrificed 4 days later.

Cell studies

Primary hepatocytes were isolated using a two-step collagenase perfusion procedure as described previously ³⁰. Insulin treatments were as previously reported ⁸. Primary hepatocytes were incubated in serum-free DMEM with or without 10 nM of insulin for 16 and 72 hours before assay. In the 72 hour incubations, we replaced the serum-free DMEM, with or without insulin, every 24 hours. For steady-state apoB labeling and immunoprecipitation, primary hepatocytes were labeled with [³⁵S]methionine in methionine-free DMEM for 2 hours as described ³⁴.

Blood Chemistry

Mice were fasted for 4 hours and blood samples were collected from retro-orbital veins. Triglyceride, glucose, cholesterol and free fatty acids (FFA) were measured using commercial assay kits (Wako Diagnostics, Richmond, VA). Insulin concentration was determined by a radioimmunoassay (Linco Research, St. Charles, MO).

Liver triglyceride

Liver triglyceride content was measured as described ¹⁸. Briefly, tissues were homogenized in phosphate buffered saline (PBS) and lipids were extracted twice with 6 volumes of chloroform and methanol (2:1). Dried lipids were resuspended in 15 % triton X-100 and triglyceride concentration was determined by a colorimetric method (Wako Diagnostics, Richmond, VA).

Assessment of TG and apoB Secretion Rates in Vivo

TG and apoB secretion rates were determined as described ¹⁸. Mice fasted for 4 hour were injected with both 15% Triton WR 1339, at the concentration of 5 ug /gm of body weight, and 200 uCi of [³⁵S] methionine per mouse through a femoral vein. Blood samples were collected just prior to injection and 30, 60, 90, and 120 min after injection by retro-orbital bleeding. For TG secretion rates, the concentration of TG was determined by a commercial assay kit (Wako Diagnostics, Richmond, VA). For apoB secretion rates, plasma from 60 and 120 min blood samples containing equal amounts of TCA-precipitable radioactivity (to correct for total protein synthesis) were separated on 4 % SDS-PAGE gels. The gels were dried under vacuum and heat (65°C) for 1 hour, and exposed on X-ray films to visualize apoB 100 and apoB 48 bands. Corresponding bands on the dried gel were cut out and counted to quantitate the rate of appearance of newly synthesized (³⁵S-labeled) apoB in plasma.

Dual-energy X-ray absorptiometry (DEXA)

Mice were anesthetized by i.p. injection of Ketamine /xylazine mixture (100 mg and 10 mg/kg body weight, respectively) and scanned using Lunar FIXImus2 (Lunar PIXImus Corp, Madison, WI). Body composition including body weight and fat was estimated by the system software.

Quantitative PCR

Total liver RNA was extracted with Trizol RNA extraction reagent, and cDNA was synthesized using SuperScript cDNA synthesis kits following the manufacturer's instructions (Invitrogen, Carlsbad, CA). Quantitative PCR (qPCR) was performed with SYBR Green in a MX 3005P (Stratagen, La Jolla, CA). Primers for qPCR reactions are described in Supplemental Table I.

Western blot analysis

Aliquots of liver were homogenized in a lysis buffer consisting of 25 mM Tris-HCL (pH7.4), 10 mM Na3PO4, 10 mM NaF, 10 mM Na4P2O7, 1 mM EGTA, 1 mM EDTA, 1% NP-40, 1 mM PMSF, 5 ug/ml leupeptin, and 5 ug/ml aprotinin. Proteins were separated on SDS-PAGE gels, transferred onto a nitrocellulose membrane, and immunoblotted with specific antibodies. AKT, p-AKT, and PTEN antibodies were purchased from Cell Signaling (Boston, ME). Anti β-actin antibody was obtained from Sigma (Saint Louis, MO).

Fast Performance Liquid Chromatography

FPLC was performed as described ¹⁸. Briefly, a 200 ul plasma sample, pooled from 4-5 animals bled after a 4-hr fast, was chromatographed through two Superose columns connected in series using a FPLC system (AKTA FPLC, UPC-900 with Unicorn software, GE healthcare, Pittsburgh, PA, USA). 54 samples of 0.5 ml fractions were collected and the concentrations of TG and cholesterol determined as described above.

Statistical Analysis

All data are presented as means ± S.E.M. Differences in means between two groups were evaluated for significance using a two-tailed Student's t test. A P value <0.05 was considered significant.

Results

Liver specific Pten knockout mice have increased hepatic insulin signaling, increased hepatic de novo lipogenesis, and hepatic steatosis

To evaluate the importance of increased hepatic insulin signaling relative to hepatic triglyceride content in the regulation of VLDL apoB secretion, we studied the hPten-ko mouse. The hPten-ko mouse was characterized in some detail previously by Stiles et al. ³², but we felt it was necessary to validate those original findings in our laboratory before extending studies in this model. Supplemental Figure IA confirms the obvious steatotic liver present in an hPten-ko mouse. The steatosis was confirmed quantitatively in Supplemental Figure IB, which shows that hepatic TG levels were increased almost 10-fold in hPten-ko mice. Hepatic cholesterol levels were not significantly different between Floxed and hPten-ko mice (data not shown). Increased hepatic insulin signaling in the absence of PTEN is demonstrated by significantly elevated levels of phospho-AKT (Suppl. Fig. IC). Total AKT was modestly reduced in the hPten-ko mice in this experiment, as well in others experiments where PTEN activity was absent or reduced and phospho-AKT was increased; a similar finding was suggested by Stiles et al. in their Supplemental Figure IC ³² . We also confirmed that male hPten-ko mice had lower plasma glucose and insulin levels, and similar TG levels compared to Floxed mice (Table 1); Stiles et al. only studied male mice ³¹. Fasting total cholesterol levels were significantly higher in both male and female hPten-ko mice, findings not noted by Stiles et al. ³². We did not, however, confirm the finding of Stiles et al. ³¹ of lower plasma levels of FFA in the hPten-ko mice: this difference may be due to the older age of our mice at time of studies. As described by Stiles et al. ³², we observed a small but significant reduction in body fat (1 gm or 3% of body fat; p = 0.004) in male hPten-ko mice (Suppl Fig. ID). We observed a smaller and non-significant reduction in body fat in female hPten-ko mice.

Table 1.

Characteristics of hPten-ko and Floxed mice

	Male Flox	Male hPten-ko	Female Flox	Female hPten-ko
	Mean ± SEM	Mean ± SEM	Mean ± SEM	Mean ± SEM
Body weight (g)	24.5 ± 0.9	22.6 ± 0.9	17.6 ± 0.5	18.7 ± 0.7
Fasting TC (mg/dl)	88.8 ± 21.8	122.0 ± 43.7**	76.2 ± 3.6	118.3 ± 2.8**
Fasting TG (mg/dl)	41.6 ± 2.3	43.9 ± 5.0	31.6 ± 0.5	48.9 ± 6.5*
Fasting FFA (mM)	0.70 ± 0.06	0.65 ± 0.08	0.62 ± 0.06	0.72 ± 0.05
Glucose (mg/dl)	139.2 ± 4.8	110.3 ± 7.9*	142.4 ± 5.2	137.2 ± 7.9
Insulin (ng/ml)	0.43 ± 0.04	0.09 ± 0.02**	0.55 ± 0.08	0.25 ± 0.03*

Body weight in gram, TG, Glucose in mg/dl; FFA in mMol; Insulin in ng/ml All measurements were from blood samples obtained after a 4 hour fast.

^*p<0.01

^**p <0.001.N=14-16, 10-12 week old mice

Stiles et al. demonstrated increased de novo lipogenesis in hPten-ko mice ³² and that finding was in accord with our demonstration of increased expression of Fas and Srebp-1c (Suppl. Fig. IE). Other genes that are associated with hepatic steatosis, SCD1, PPARγ2, and CD36, were also dramatically elevated (Supplemental Table II). Levels of LXRα, DGAT1, DGAT2, apoB, and microsomal triglyceride transfer protein (MTP) were not different between Floxed and hPten-ko mice.

TG and ApoB secretion rates are increased in hPten-ko mice

Increased TG secretion had previously been observed ³² and we confirmed that finding; TG secretion in hPten-ko mice was double that in Floxed mice (Fig. 1A and 1B). We hypothesized that despite steatosis and increased hepatic TG secretion, there would be increased insulin-mediated targeting of apoB into a degradation pathway, with decreased VLDL apoB secretion. We tested our hypothesis directly by injecting [³⁵S]-methionine intravenously, together with Triton WR1339, to quantitate the appearance of newly synthesized, radiolabeled apoB in plasma. In contrast to our expectations, rates of secretion of newly synthesized apoB100 and apoB48 were increased 50-100 % in hPten-ko mice (Fig. 1C and 1D). Furthermore ex vivo studies of primary hepatocytes from hPten-ko mice confirmed increases in both cellular and secreted apoB100 compared to cells from Floxed mice (Fig. 1E). We did not observe differences in apoB48 in cells or media between hPten-ko and Floxed mice. Thus, rather than finding evidence of increased insulin-mediated targeting of apoB for degradation in hPten-ko mice, we had in vivo data supporting increased secretion of both apoB100 and apoB48 as well as ex vivo data demonstrating increased apoB100 secretion in these mice.

Figure 1. hPten-ko mice have increased secretion of TG and apoB in vivo and apoB100 secretion ex vivo.

TG and apoB turnover studies were performed after a 4 hour fast. Mice were injected with Triton WR 1339 and [³⁵S]-methionine and blood samples obtained during the next 2 hrs. (A) TG secretion was higher over the 2 hr time course. (B) TG secretion rate, in mg/dl/hr, was greater in hPten-ko mice. (C) ApoB secretion rates were plotted as the percentage of the rates in Floxed mice 60 and 120 min after injection of [³⁵S]-methionine. ApoB 100 secretion rates were increased in hPten-ko mice. (D) ApoB 48 secretion rates were increased in hPten-ko mice. All data (A-D) are expressed as the mean ± SEM. N=7-8. * p < 0.05; ** P < 0.01. (E) ApoB 100 secretion was increased in primary hepatocytes isolated from hPten-ko mice. All data are normalized to cell apoB100 in cells. (N=4). ** p <0.01.

We could not rule out, however, the possibility that increased insulin signaling in the hPten knockout mouse actually limited what would have been an even greater response in apoB secretion to the prevailing level of steatosis. Therefore we designed experiments to reduce hepatic steatosis, with the idea that the targeting of apoB for degradation by increased hepatic insulin signaling in hPten-ko mice would be unmasked.

ASO knockown of Dgat1 and Dgat2 in hPten-ko mice decreased hepatic TG levels and TG secretion rate, but did not unmask insulin medicated targeting of apoB for degradation

To reduce hepatic steatosis without altering hepatic insulin signaling, we treated hPten-ko mice with ASO to both Dgat1 and Dgat2, twice weekly for 4 weeks. Administration of the two ASOs did not alter hepatic insulin signaling (data not shown), but resulted in marked reductions in Dgat1 and Dgat2 mRNA (Suppl. Fig. IIA). Hepatic TG levels and the rate of TG secretion were concomitantly and dramatically reduced by 50 % compared to hPten-ko mice receiving control ASO (Suppl. Fig. IIB and IIC). However, the secretion rates of apoB100 and apoB48 were not significantly decreased in Dgat ASO treated mice compared to control ASO treated groups (Suppl. Fig. IID and IIE).

Even though these treatments dramatically reduced hepatic TG levels, livers of hPten-ko mice treated with DGAT ASOs still contained very high TG levels (743 ug/mg protein); in our laboratory, C57Bl/6J mice on chow diets have liver TG concentrations between 50 and 150 ug/mg protein. We could not rule out, therefore, that this high liver TG content might have been adequate to protect apoB from insulin-mediated degradation in hPten-ko mice. Thus, we hypothesized that further reduction of hepatic TG might be needed to unmask the effects of increased insulin signaling on apoB secretion in hPten-ko mice.

Srebp-1c ASO treatment of hPTEN-ko mice decreased hepatic TG and TG secretion, but only minimally affected apoB secretion

To test this hypothesis, we targeted Srebp-1c, a global lipogenic transcription factor in the liver, for ASO treatment. Srebp-1c ASO treatment of hPten-ko mice twice weekly for 4 weeks did not alter hepatic insulin signaling (data not shown) but successfully decreased Srebp-1c mRNA more than 50% (Suppl. Fig. IIIA) and this was paralleled by a reduction in hepatic TG content of 70 %, to 338 ug/mg protein, compared to hPten-ko mice treated with control ASO (Suppl. Fig. IIIB). TG secretion was decreased concomitantly by approximately 40% (Suppl. Fig. IIIC). These Srebp-1c ASO-mediated changes did not, however, affect the secretion rate of apoB 100 (Suppl. Fig. IIID). Secretion of apoB 48 was reduced by 25%, and this was significant only at the 2 hr time-point (Suppl. Fig. IIIE).

Overall, the results of these two experiments with specific ASO-treatments that reduced hepatic TG levels dramatically, indicated that increased hepatic insulin signaling in hPten-ko mice did not appear to significantly target apoB for degradation even when hepatic TG was only modestly elevated; there was no change in apoB100 secretion, whereas apoB48 secretion may have tracked with TG content and secretion.

Primary hepatocytes from hPten-ko mice are not sensitive to the acute effects of insulin on apoB secretion or cellular apoB content

Our findings to this point were unexpected and indicated that we needed to demonstrate well-known effects of insulin to target apoB100 for degradation and thereby reduce apoB secretion ^4-9. We confirmed this well-established observation both with primary hepatocytes isolated from C57BL/6J wild type mice and with hepatocytes isolated from Floxed control mice (Fig. 2A and 2B). In both wildtype and Floxed hepatocytes, treatment with 10 nM insulin for 16 hrs reduced cellular and secreted apoB100 by 40-60%. As reported in prior literature, insulin treatment did not affect apoB48 cell or media content. By contrast, primary hepatocytes from hPten-ko mice did not respond to acute insulin treatment with a decrease in either apoB secretion or cellular content, (Fig. 2C). Interestingly, and as reported previously ^6,16,17, longer term insulin treatment (72 hrs) of primary hepatocytes from Floxed control mice was associated with a loss of insulin-mediated inhibition of apoB secretion (Fig. 2B) suggesting insulin resistance had developed during that period of time. Importantly, prolonged insulin treatment of primary hepatocytes isolated from hPten-ko mice did not alter either apoB secretion or cellular content; thus these cells were insulin resistant at both 16 and 72 hour (Fig. 2C). Overall, our results indicate that primary hepatocytes from hPten-ko mice, which had been exposed chronically in vivo to increased insulin signaling, did not respond to acute insulin treatment by reducing apoB100 secretion. Because prior studies had suggested that chronic hyperinsulinemia could reduce the ability of insulin to acutely target apoB for degradation ^{6, 16, 17, 35}, we next attempted to acutely alter hepatic insulin signaling in vivo.

Figure 2. ApoB 100 secretion is inhibited by insulin treatments in primary hepatocytes from C57BL/6J and Floxed control mice, but not from hPten-ko mice.

Primary hepatocytes from wildtype, Floxed, and hPten-ko mice were treated with or without 10 nM insulin for either 16 hours or 72 hours and then labeled with [³⁵S] methionine at steady state for 2 hours. Immunoprecipiated [³⁵S] methionine-labeled apoB100 and apoB48 were separated (left panels) and amounts of newly synthesized protein in cell or media was quantitated by scintillation counting of the bands (right panels). (A) Acute insulin treatment reduced apoB100 in cells and media of primary hepatocytes from C57BL/6J. (B) Acute insulin treatment reduced apoB100 in cells and media of primary hepatotocytes from Floxed control mice whereas chronic treatment did not reduce apoB in cells or media. (C) Neither acute nor chronic insulin treatment affected apoB100 in cells or media of hPten-ko primary hepatocytes. Insulin treatment did not affect apoB48 in cells or media of primary hepatocytes from any of the three groups. N=3 for each type of mice; * p < 0.05, **p=0.001

Acute restoration of PTEN in hPten-ko mice reduced insulin signaling, hepatic TG content, and both TG and apoB secretion rates

To acutely reduce hepatic insulin signaling, we introduced wild type Pten via adenoviral infection into hPten-ko mice. As expected administration of Pten adenovirus restored hepatic PTEN protein within 4 days and was associated with decreased phosphorylation of AKT compared to GFP adenovirus-treated mice (Fig. 3A). Decreased hepatic insulin signaling was associated with a significant reduction in the expression of Fas and a trend toward decreased expression of Srebp-1c (Fig. 3B). Liver TG levels and TG secretion rate also were decreased after restoration of PTEN protein (Fig. 3C and 3D). Surprisingly, despite the significant decrease in hepatic insulin signaling, secretion rates of apoB100 were decreased by 50% in hPten-ko mice after acute restoration of PTEN (Fig. 3E). ApoB48 secretion was similarly reduced after PTEN restoration (Fig. 3F). Acute reconstitution of PTEN in hPten-ko mice did not alter plasma levels of lipids, glucose, or insulin and did not affect body weight or body fat compared to control GFP virus-treated mice (Fig. 3G and 3H). These data indicate that despite reduced hepatic insulin signaling, which might have been expected to result in increased secretion of apoB, secretion of apoB fell in parallel with the reductions in hepatic TG content and secretion.

Figure 3. Injection of wild type of Pten adenovirus into hPten-ko mice reduced insulin signaling, expression of lipogenic genes, hepatic TG content, and secretion of both TG and apoB without changes in bodyweight, body fat, or plasma values.

hPten-ko null mice were injected with wild-type recombinant Pten adenovirus (Pten) or a control adenovirus (GFP) through a tail vein at the dose of 4 × 10⁹ particles per gram of body weight. Studies were performed 4 days after injection of virus. N=3 for each group. (A) Western blotting shows restoration of PTEN in livers of mice receiving Pten adenovirus; there was some variability in expression. Insulin signaling was reduced in mice receiving PTEN adenovirus, as assessed by Western blotting and by the ratio of p-AKT/AKT. (B) Hepatic expression of Fas and Srebp-1c genes was reduced after restoration of hepatic PTEN protein. (C) Hepatic TG content decreased in mice receiving Pten adenovirus. (D) TG secretion was lower in mice with restored hepatic PTEN protein. (E and F) Secretion rates of apoB100 and apoB48 were lower after restoration of hepatic PTEN. (G) Body weight and body fat content were determined by Dexa scan analysis 4 days after virus injection; there were no significant changes although there was a trend toward reduced body fat (%). (H) There were no significant changes in plasma levels of TG, cholesterol, glucose, FFA at days after virus injection. N=3 for each group.*p<0.05.

Acute liver-specific knockout of Pten in Floxed mice increased insulin signaling, hepatic TG content, and both TG and apoB secretion

Next, we acutely increased hepatic insulin signaling by deletion of the Pten gene in Floxed mice using a Cre recombinase adenovirus. Four days after infection of Floxed mice with adeno Cre under control of the albumin promoter, PTEN protein was dramatically reduced and insulin signaling, assessed by the level of p-AKT, was significantly increased compared to GFP adenovirus-treated mice (Fig. 4A). Increased hepatic insulin signaling was accompanied by trends toward increased expression of the lipogenic genes, Srebp-1c and Fas (Fig. 4B) and significantly increased hepatic TG mass (Fig. 4C). Concomitantly, TG secretion was 2 fold higher in the Floxed mice in whom hepatic Pten had been acutely knocked down compared to Floxed mice that had received GFP adenovirus (Fig. 4D). Although we expected that the acute increase in insulin signaling in mice with acute hPten knockout would target apoB for degradation, there was no change in the secretion rate of apoB100, and secretion of apoB48 actually increased by 50% compared to Floxed mice that had received GFP adenovirus (Fig. 4E and 4F).

Figure 4. Acute deletion of hepatic Pten in Floxed mice resulted in increased insulin signaling, expression of lipogenic genes, hepatic TG content and secretion of both TG and apoB48.

Floxed mice received injections of recombinant adenovirus containing either Cre (Cre) or Gfp (GFP). Studies were performed 4 days after injection of virus. (A) Western blotting shows marked reductions of Pten in livers of mice receiving Cre adenovirus. Insulin signaling was increased significantly as assessed by the p-AKT/AKT ratio. (B) Hepatic expression of Fas and Srebp-1c genes increased after acute PTEN knockout. (C) TG contents increased after acute hepatic PTEN knockout. (D) TG secretion increased after acute PTEN knockout. (E) ApoB100 secretion did not change after acute knockout of hepatic PTEN. (F) ApoB48 secretion increased after acute hepatic PTEN knockout. N=6-8. *p<0.05; **p<0.01

Acute liver-specific knock-out of Floxed Pten mice that were pre-treated with Srebp-1c ASO did not result in inhibition of apoB secretion despite increased insulin signaling without an increase in hepatic TG content

In the previous experiment, acute increases in hepatic insulin signaling resulted in significant increases in hepatic TG, confounding, somewhat, interpretation of the results. Therefore, we repeated the acute knockout experiment in Floxed mice that had been treated with Srebp-1c ASO for 4 weeks to block de novo lipogenesis and steatosis. Acute knockout of hepatic Pten resulted in a decrease in PTEN protein and increased insulin signaling confirmed by increased levels of p-AKT (Fig.5A). However, Srebp-1c expression did not increase at all while FAS expression trended to increase only slightly in the mice pre-treated with Srebp-1c ASO (Fig. 5B). TG content rose modestly (compare Fig. 5C with Fig. 4C) and was still within the normal range for our laboratory for chow-fed C57BL/6J mice. TG secretion also increased only slightly (compare Fig 5D with Fig 4D). Thus, we had created a model of increased insulin signaling without steatosis and expected, therefore, to demonstrate insulin-mediated targeting of apoB for degradation. However, 4 days of increased hepatic insulin signaling in the acute hPten knockout mice, in the absence of steatosis, had no effect on either apoB100 or apoB48 secretion (Fig. 5E and 5F). These data provide very strong evidence that increasing insulin signaling over a period as short of 4 days led to resistance in the pathway of insulin-mediated targeting of apoB for degradation.

Figure 5. Treatment of Floxed mice with Srebp-1c ASO prior to acute depletion of hepatic Pten blocks the increase in hepatic TG but does not result in decreased apoB secretion.

Floxed mice were pre-treated for 4 weeks with Srebp-1c antisense oligonucleotides (ASO) before being injected with recombinant adenovirus containing either Cre (Cre) or Gfp (GFP). Studies were performed 4 days after injection of virus. (A) Insulin signaling was increased significantly in mice receiving Cre adenovirus as assessed by the p-AKT/AKT ratio. (B) Srebp-1c and Fas gene expression were not increased in ASO-pretreated mice receiving Cre adenovirus. (C) Hepatic TG levels were not significantly increased and were in the normal range in ASO-pretreated mice receiving Cre adenovirus. (D) TG secretion was not significantly increased in ASO-pretreated mice receiving Cre adenovirus. (E and F) Secretion rates of apoB100 and apoB48 were not affected in ASO-pretreated mice receiving Cre adenovirus. N=3 for each group. *p<0.01

Discussion

The assembly and secretion of VLDL is one of the key pathways, together with fatty acid (FA) oxidation and de novo lipogenesis, by which the liver maintains control over TG content and attempts to avoid steatosis ^{26-28,36, 37}. It is not surprising, therefore, that although the regulation of the assembly of VLDL is complex ^{3, 26, 28, 29}, the availability of hepatic TG is a potent factor regulating TG secretion. Increased delivery of either albumin-bound FA or TGFA-enriched remnant lipoproteins to cultured hepatocytes ^{29, 38-40}, perfused livers ^{41, 42}, or livers in vivo ⁴³ can increase the secretion of both TG and apoB. Increased numbers of VLDL particles that are either the same size or larger are secreted. In contrast to increasing hepatic TG by delivery of FA or TGFA from plasma, increasing TG availability by increasing hepatic lipogenesis has been reported to increase secretion of TG but not apoB. For example, diets high in carbohydrates increased hepatic lipogenesis and TG secretion in VLDL without increasing apoB secretion, indicating assembly and secretion of larger VLDL ⁴⁴. Pharmacologic stimulation of the liver X-receptor (LXR) in mice increased lipogenesis and secretion of VLDL TG, but apoB secretion was unaffected ⁴⁵, Similarly, overexpression of SREBP-1c resulted in secretion of very large VLDL, suggesting more TG but not more apoB secretion ⁴⁶. On the other hand, a recent study in hamsters suggested increases in lipogenesis can stimulate both TG and apoB secretion ⁴⁷.

Published studies where TG synthesis is increased without directly increasing FA delivery or hepatic lipogenesis indicate that additional complexity is possible. Thus, two studies in which short-term overexpression of Dgat 1 or Dgat 2 was achieved using recombinant adenovirus demonstrated increased hepatic TG content, but had differing effects on TG and apoB. In one study, short term overexpression of Dgat1 increased both TG and apoB secretion while overexpression of Dgat2 had no effects ⁴⁸. In the second paper, there was no effect of overexpression of either of these enzymes on TG or apoB secretion ⁴⁹. Consistent with the latter report, mice transgenic for Dgat 2 developed fatty liver without increases in TG secretion ⁵⁰. In contrast, overexpression of Dgat1 in McAdle RH7777 cells increased cellular TG content and the secretion of both TG and apoB ⁵¹.

Some of the experiments noted above could have been confounded by alterations in hepatic insulin signaling. For example, high carbohydrate feeding, with concomitant hyperinsulinemia, could have targeted apoB for degradation, resulting in secretion of the same number of VLDL particles each containing more TG ⁴⁴. Similarly, increases in LXR signaling may increase hepatic insulin sensitivity ⁵², although a recent study indicates significant complexity in this regard ⁴⁷. It is also very likely that the time course of insulin-mediated targeting of apoB for degradation is complex; acute hyperinsulinemia targets apoB for degradation whereas longer exposure to insulin results in a loss of such targeting ^{1, 6, 16, 17}. Therefore, the background sensitivity of the liver to insulin will also be a critical determinant of the response of apoB to any alterations in ambient insulin levels in plasma or hepatic insulin signaling ^{15, 20}.

Our findings in mice specifically lacking hepatic PTEN provide significant, new insights regarding the relative importance of hepatic TG content and hepatic insulin-signaling for the secretion of both TG and apoB. First, our studies demonstrate clearly that, in the presence of longstanding increases in both hepatic insulin signaling and hepatic TG, the availability of TG for assembly into VLDL is the dominant stimulus for apoB secretion. Thus in the hPten-ko mice, where markedly increased insulin signaling should have reduced apoB secretion, the latter was actually increased in parallel with increased TG secretion. This was true both in vivo and ex vivo, the latter in primary hepatocytes isolated from hPten-ko and Floxed mice. In hPten-ko mice where hepatic TG levels were reduced by either inhibiting TG synthesis with Dgat ASOs or inhibiting de novo lipogenesis with Srebp-1c ASO, apoB secretion fell modestly and appropriately concomitant with reductions in hepatic TG content and TG secretion; we had expected more significant reductions in apoB secretion in the presence of high levels of insulin signaling but less hepatic TG-associated stimulation of VLDL assembly. More importantly, when hepatic PTEN activity was acutely restored in the hPten-ko mice, and insulin signaling was significantly reduced, apoB secretion did not increase, but actually fell in parallel with reductions in hepatic TG content and secretion. The strongest evidence for the importance of hepatic TG was apparent in studies where we acutely knocked out hepatic Pten and insulin signaling acutely increased; in studies performed only 4 days after adenoviral infection, apoB48 secretion increased in parallel with increased TG secretion while apoB100 secretion was unchanged. Even when we acutely knocked out hepatic Pten but prevented the subsequent acute increase in TG content by pre-treating the mice with Srebp-1c ASO, apoB secretion was appropriate for the level of TG secretion. The acute knockout studies indicated that either hepatic TG is always dominant over hepatic insulin signaling as a determinant of apoB secretion, or that 4 days of increased insulin signaling is long enough for development of resistance to insulin-mediated apoB degradation ^{6, 15-17, 20}. Our ex vivo studies showing loss of insulin-mediated inhibition of apoB secretion in primary hepatocytes from floxed mice after incubation with insulin for 72 hrs confirms those earlier results. Notably, Pten knockout hepatocytes had no response to incubation of insulin at either 16 or 72 hrs.

Our results must be compared with a recent report by Qiu et al. ⁵³ in which a different liver-specific Pten knockout mouse-model was found to have steatosis and increased expression of lipogenic genes, but decreased hepatic levels of apoB100 and apoB48. Furthermore, in that paper ⁵³, acute expression of a dominant negative Pten vector in HepG2 cells, McA-RH7777 cells, and primary hamster hepatocytes resulted in increased insulin signaling and reduced secretion of apoB100 and apoB48 due to increased proteasomal degradation. Differences in strains (we studied hPten-ko mice on a C57Bl/6 background ³¹ whereas the Pten knockout mouse in the paper by Qiu et al ⁵³ was on a mixed background ⁵⁴) could have resulted in different phenotypes, but we believe that differences in methodologic approaches may be the basis for the contrasting findings in this case. Qiu et al conducted only limited studies in their Pten knockout mouse, and did not perform any in vivo studies of TG and apoB secretion. Their finding of decreased hepatic levels of apoB does not rule out the possibility of increased secretion of apoB; more efficient secretion could result in reduced steady-state levels of hepatic apoB. The remainder of the studies by Qiu et al utilized cultured hepatocytes. They found that reduced PTEN activity secondary to over-expression of a dominant negative Pten resulted in decreased secretion of apoB; these observations are consistent with previously published work demonstrating the acute effects of increased insulin signaling on apoB degradation ¹. Indeed, we also demonstrated that insulin treatment of primary hepatocytes isolated from either wildtype C57BL/6J or Floxed mice resulted in reductions in the accumulation of newly synthesized apoB100 in cells and media. In our studies, apoB48 cell and media levels were not affected by acute insulin treatment, also concordant with published data ¹. As noted above, we demonstrated that after incubation of Floxed primary hepatocytes with insulin for 72 hrs, apoB100 secretion was no longer inhibited. On the other hand, were unable to demonstrate altered apoB100 secretion after acute insulin treatment (either 16 or 72 hrs) in hepatocytes isolated from hPten-ko mice. These results are in agreement with studies showing that “insulin resistance” develops in the pathway of insulin-mediated targeting of apoB100 for degradation after prolonged exposure of cultured liver cells to the hormone ^{6, 16, 17, 35}. Overall, however, we cannot explain all of the differences between our results and those of Qiu et al. ⁵³.

MTP plays a critical role in the assembly and secretion of VLDL, and insulin plays a key role in the regulation of MTP at the transcriptional level ^{55, 56}, working via the MAPK/ERK pathway ^{57, 58} and FoxO1 ⁵⁹. MTP protein has a long half-life, so insulin signaling must be increased chronically to result in reduced levels of MTP protein ^{55, 57}, and Qiu et al. did see reduced MTP protein and activity, albeit without altered mRNA levels in their hPten knockout mouse ⁵³. We did not see changes in the levels of either MTP mRNA (Supplemental Table II) or MTP protein (data not shown) in our hPten-ko mice. Studies from several laboratories have identified sites upstream of the apoB gene where transcriptional regulation may occur ^{60, 61}; however, there appears to be very little in vivo regulation of transcription of the apoB gene. Expression of apoB, as measured by its mRNA level, was the same in hPten-ko as in Floxed mice. Severe insulin deficiency can inhibit translation of apoB mRNA ⁶², as can inhibition of MTP ⁶³; there are no data suggesting increased translational efficiency in the presence of increased insulin signaling.

Of note, hPTEN-ko mice had significantly increased plasma levels of cholesterol, which were the result of increased HDL cholesterol concentrations (data not shown). This was an unexpected finding that may provide novel insights into the role of hepatic insulin signaling in the regulation of HDL metabolism, particularly in view of our prior observation that mice with severe hepatic insulin resistance, due to the absence of hepatic insulin receptors, had very low HDL cholesterol levels ³⁰. Further studies will be needed to better define the mechanisms underlying these observations.

In summary, despite evidence for markedly increased hepatic insulin signaling in hPten-ko mice, there was no evidence for specific targeting of apoB for degradation. In contrast, apoB secretion paralleled hepatic TG content. Our results indicate that although there is a compelling literature from in vitro, ex vivo, and in vivo studies demonstrating that acute increases in insulin signaling can reduce apoB secretion by targeting the protein for intracellular degradation, this effect appears to dissipate by 4 days in vivo, consistent with long-term in vitro studies. Our results are also supported by observations in mouse models of insulin resistance and increased VLDL secretion ^{18, 19} as well as demonstrations that VLDL secretion is not inhibited by insulin in humans with insulin resistance ^{14, 15} . The loss of insulin-mediated regulation of apoB secretion leaves hepatic TG as the major determinant of VLDL assembly and secretion ²³. Thus, in humans with insulin resistance-associated increases in VLDL secretion, the concomitant steatosis that is commonly present must result from other abnormalities that can limit apoB secretion ³⁴.