Plasma cholesterol-lowering and transient liver dysfunction in mice lacking squalene synthase in the liver.

Squalene synthase (SS) catalyzes the biosynthesis of squalene, the first specific intermediate in the cholesterol biosynthetic pathway. To test the feasibility of lowering plasma cholesterol by inhibiting hepatic SS, we generated mice in which SS is specifically knocked out in the liver (L-SSKO) using Cre-loxP technology. Hepatic SS activity of L-SSKO mice was reduced by >90%. In addition, cholesterol biosynthesis in the liver slices was almost eliminated. Although the hepatic squalene contents were markedly reduced in L-SSKO mice, the hepatic contents of cholesterol and its precursors distal to squalene were indistinguishable from those of control mice, indicating the presence of sufficient centripetal flow of cholesterol and/or its precursors from the extrahepatic tissues. L-SSKO mice showed a transient liver dysfunction with moderate hepatomegaly presumably secondary to increased farnesol production. In a fed state, the plasma total cholesterol and triglyceride were significantly reduced in L-SSKO mice, primarily owing to reduced hepatic VLDL secretion. In a fasted state, the hypolipidemic effect was lost. mRNA expression of liver X receptor α target genes was reduced, while that of sterol-regulatory element binding protein 2 target genes was increased. In conclusion, liver-specific ablation of SS inhibits hepatic cholesterol biosynthesis and induces hypolipidemia without increasing significant mortality.


MATERIALS AND METHODS
The detailed procedures for the generation of liver-specifi c SS knockout (L-SSKO) mice and other assays are available in the supplementary Materials and Methods.

RESULTS
The heterozygous fl oxed SS (SS +/f ; f denotes fl anked by loxP) carrying one copy of the Cre recombinase gene under the control of the albumin gene promoter (Alb-Cre) ( 10 ) were interbred with SS +/f littermates lacking Alb-Cre to generate L-SSKO mice. L-SSKO mice and littermate controls [SS +/+ , SS +/+ Alb-Cre, and SS f/f (fSS)] were generated. Because there were no differences in growth curves or metabolic parameters such as plasma lipid and glucose levels between the SS +/+ , SS +/+ Alb-Cre, and SS f/f (fSS) mice, we used fSS mice as a control.
To determine whether SS expression was ablated in the liver, we performed a Southern blot analysis (supplementary Fig. 2A). At 4 weeks of age, the livers of L-SSKO mice contained both disrupted and fl oxed alleles with the former being more predominant (89% based on relative band density). Northern blot analysis (supplementary Fig.  2B) and real-time PCR ( Fig. 1A ) showed that the mRNA (HMGCR). However, development of SS inhibitors such as zaragozic acid and lapaquistat acetate (TAK-475) has been halted because of safety concerns ( 6,7 ).
To investigate the consequences of the systemic ablation of SS, we generated mice lacking SS in the whole body ( 8 ). The SS Ϫ / Ϫ mice exhibited developmental defects and did not survive beyond E12.5. Supplementation of the dams' diet with squalene or cholesterol did not allow survival of SS Ϫ / Ϫ fetuses to term. In contrast, SS +/ Ϫ mice were apparently normal and their plasma lipoprotein profi les were indistinguishable from those of wild-type mice, even though hepatic SS activity was reduced by 50%. These results suggest that fetal demands for cholesterol were not met by maternal supplies and/or that the accumulation of precursors of squalene such as FPP was toxic. In this context, it is noteworthy that naturally occurring inborn errors of cholesterol metabolism are frequently associated with severe developmental abnormalities particularly in the central nervous system ( 9 ).
Because deletion of SS theoretically does not block the nonsterol pathway, mice in which SS is specifi cally knocked out in the liver can be a viable model in which hepatic sterol pathway is selectively abrogated. To test this hypothesis and further examine the effi cacy and safety of inhibition of SS, we used tissue-specifi c gene targeting with the Cre-loxP system to generate mice lacking SS in a liver-specifi c manner. Conditional deletion of SS in the livers. A: Quantitative real-time PCR of SS mRNA levels from the livers, intestines, adrenal glands, and testes of control and L-SSKO male mice at the age of 12 weeks (n = 5 ‫ف‬ 6 in each group). The internal standards were GAPDH mRNA for livers and ␤ actin mRNA for intestines, adrenal glands, and testes. B: Immunoblot analysis for SS protein of the liver of control and L-SSKO male mice at the age of 12 weeks (n = 3 in each group). Protein disulfi de isomerase (PDI) was used as a loading control for SS protein levels. C: SS activity in the hepatic microsomal fractions of control (n = 4) and L-SSKO (n = 6) male mice at the age of 12 weeks. In vitro synthesis rate of cholesterol (D) and fatty acid (E) in whole liver slices from male mice at the age of 12 weeks (n = 5 in each group). Whole liver slices were incubated with [ 14 C]acetate, and 14 C-labeled cholesterol was determined. Two portions of liver from each mouse were used for the assay of liver slices. F: In vivo synthesis rate of cholesterol in liver from male mice at the age of 12 weeks (n = 5 in fSS and n = 6 in L-SSKO mice). Mice were injected with [ as lanosterol, zymosterol, lathosterol, desmosterol, and cholesterol were not altered ( Next, we compared liver weight, liver functions, and plasma lipid levels between the control and L-SSKO mice ( Fig. 3 , Table 1 , and supplementary Table 2). The livers of L-SSKO mice were larger than those of the control mice ( Fig. 3A ). The ratio of liver weight to body weight was 43% larger in L-SSKO mice at 12 weeks of age and 27% larger at 24 weeks of age ( Fig. 3B ). In proportion to enlargement of the liver, plasma levels of alanine aminotransferase (ALT) in the L-SSKO were signifi cantly elevated compared with those in the control mice at 12 weeks of age ( Fig. 3C ). To determine the causes of the liver dysfunction associated with hepatomegaly, we performed a microscopic analysis of the liver. Hematoxylin and eosin (H and E) staining revealed cellular swelling and spotty necrosis of parenchymal cells ( Fig. 3D ). Oil Red O staining showed no evidence for accumulation of neutral lipids ( Fig. 3E ). The livers of the L-SSKO mice contained an increased number of deoxynucleotidyltransferase-mediated dUTP nick end labeling (TUNEL)-positive parenchymal cells (lower panels), indicating apoptotic cell death ( Fig. 3F ). Consistently, liver caspase-3 activity in the L-SSKO mice was 50% higher than that in the fSS mice ( Fig. 3G ). In contrast, the number of Ki-67-positive hepatocytes was signifi cantly increased in the L-SSKO mice ( Fig. 3H, I ), indicating that the hepatocytes regenerated at an increased rate.
The accumulation of FPP may be causally related to the increased apoptosis and proliferation of hepatocytes expression of SS in the livers of L-SSKO mice was reduced to 5% of that in the fSS mice. In contrast, the mRNA expressions of SS in small intestines, adrenal glands, and testes were not different between fSS and L-SSKO mice ( Fig. 1A ).
In an immunoblot, SS protein was barely detectable in the livers of L-SSKO mice ( Fig. 1B ). Consistent with the profound reduction in SS at both the mRNA and protein levels, the SS activity in the livers of L-SSKO mice was <10% of that in fSS mice ( Fig. 1C ).
L-SSKO mice were born at a rate in accordance with the rule of Mendelian inheritance and survived at a rate that was not signifi cantly different from that of the control mice up to 24 weeks of age.
Cholesterol synthesis was reduced by 71% in liver slices ( Fig. 1D ), while fatty acid synthesis was increased by 58% ( Fig. 1E ). In parallel, though less signifi cantly, cholesterol synthesis from [ 3 H]water injected intraperitoneally in vivo was decreased by 49% ( Fig. 1F ). Nevertheless, the hepatic levels of cholesterol were not decreased ( Fig. 2H ). A compensatory increase in the intestinal cholesterol absorption was not observed (supplementary Fig. 2C).
L-SSKO mice, indicating that hepatic VLDL production was decreased ( Fig. 4D ). ␤ -VLDL clearance was severely delayed in LDL receptor (LDLR) KO mice compared with fSS mice, while it was not signifi cantly different between the fSS and L-SSKO mice ( Fig. 4E ). Together, these results indicate that the hypolipidemia of L-SSKO mice primarily resulted from impaired secretion of VLDL by the liver.
The changes in protein expression of LDLR, SR-BI, and HMGCR were largely in parallel with the changes in mRNA expression: no change and drastic decrease and increase, respectively ( Fig. 5B ). Liver HMGCR activity, expressed as the rate of biosynthesis of mevalonate from HMG-CoA, was 8.9-fold higher in the L-SSKO mice than in to make fatty liver. No difference was observed in the plasma ALT levels between fSS and L-SSKO mice (supplementary Fig. 3C).
In a fed state, plasma total cholesterol levels were decreased by ‫ف‬ 40% at 4 and 12 weeks of age ( Table 1 ). Plasma triglyceride levels were decreased by 48% at 12 weeks of age. Plasma free fatty acid levels were not different between fSS and L-SSKO mice. After a 16 h fast, however, plasma levels of both total cholesterol and triglycerides were increased, and their differences between the control and L-SSKO mice disappeared.
To determine which lipoproteins were affected by the abrogation of SS, we performed HPLC analyses of plasma lipoproteins. All of the lipoprotein fractions were affected by the elimination of SS synthesis: cholesterol contents were signifi cantly decreased in the LDL and HDL fractions ( Fig. 4A ), and triglyceride contents were decreased in all the lipoprotein fractions ( Fig. 4B ). Consistent with the decreases in apoB-containing lipoproteins including chylomicron, VLDL, and LDL, plasma levels of apoB-100 and apoB-48 were decreased ( Fig. 4C ).
The mice were intravenously injected with Triton WR1339 to inhibit the plasma clearance of triglyceriderich lipoprotein by interfering with lipoprotein lipase and the receptor-mediated lipoprotein uptake by the liver. The rate of increase in plasma triglyceride levels after the injection of Triton WR1339 was decreased by 38% in the male mice at 12 weeks of age. C: Immunoblot analysis of plasma apoB for fSS and L-SSKO male mice at the age of 12 weeks. Two microliters of plasma taken from mice in the regular chow-fed state was subjected to SDS-PAGE. Each lane represents an individual mouse (n = 3, in each group). D: Hepatic VLDL production rate after injection of Triton WR1339 in fSS (n = 4) and L-SSKO (n = 6) male mice at the age of 12 weeks. This rate is expressed as the increase in plasma triglyceride levels. E: Decay of 125 I-labeled rabbit ␤ -VLDL from the plasma in fSS, L-SSKO, and LDLR KO male mice at the age of 12 weeks (n = 5 in each groups). Each value represents the mean ± SD. * P < 0.01 by two-way repeated-measures ANOVA in E. liver. Why was the reduction of cholesterol synthesis in vivo less prominent compared with that in the liver-specifi c SREBP-cleavage activating protein knockout (L-SCAPKO) mice where 82% of cholesterol synthesis was eliminated ( 13 )? In contrast to L-SSKO mice, L-SCAPKO mice showed a signifi cant decrease in the expression of LDLR. Conceivably, lower LDLR-mediated centripetal fl ux of cholesterol in L-SCAPKO mice might mitigate the contribution of cholesterol synthesis in the extrahepatic tissues, particularly in the intestine, to the overall cholesterol synthesis in vivo, thereby increasing the apparent contribution of the liver.
It is also surprising to note that hepatic contents of cholesterol and its precursors distal to squalene were not altered in L-SSKO mice ( Fig. 2 ). As discussed above, they may be centripetally derived from the peripheral tissues such as the intestine, another potent cholesterogenic organ. According to Strandberg et al. ( 14 ), methyl sterols, most of which are cholesterol precursors, are preferentially released out of the intestinal cells compared with squalene. ABCA1 and HDL may not be involved in this process, because the centripetal movement of cholesterol was not altered in ABCA1 KO mice despite the virtual absence of HDL ( 15 ). The changes in the gene expression shown in Fig. 5A (decreases in the gene expressions of Cyp7A1, ABCG5, and ABCG8 and increases in the gene expressions of HMGCR, FPPS, and SQLE) also appear to favor the preservation of cholesterol in the liver.
Many of these changes may result from activation of the SREBP-2 pathway ( Fig. 5A ) probably owing to the reduced functional cholesterol or oxysterol pool in the livers of the control mice ( Fig. 5C ). Consistent with the increased HMGCR activity, hepatic content of mevalonate was increased by 7.9-fold in the L-SSKO as compared with the control mice ( Fig. 2A ).

DISCUSSION
In contrast to the severe liver dysfunction and death resulting from the defi ciency of HMGCR in the liver ( 11 ), the liver dysfunction of L-SSKO mice was mild and transient and did not compromise survival. More surprisingly, despite the almost complete absence of cholesterol synthesis ex vivo, hepatic cholesterol synthesis in vivo was suppressed only partially (49%), and hepatic contents of cholesterol and its precursors distal to squalene in L-SSKO mice were indistinguishable from those in fSS mice. Despite the apparent maintenance of cholesterol homeostasis in normally feeding L-SSKO mice, VLDL production was reduced, which thereby lowered the plasma concentration of apoB-containing lipoproteins.
Why was the hepatic cholesterol synthesis in vivo suppressed only partially? According to Turley et al. ( 12 ), 24% of cholesterol synthesis occurs in small intestine of rats, and a signifi cant fraction of it is transported to liver (4.6 mol/h) where nearly 50% of total cholesterol synthesis occurs (17.0 mol/h). Probably a fraction of intraperitoneally injected [ 3 H]water is used for cholesterol synthesis in the intestine and ultimately recovered in the liver, thereby diluting the effects of the elimination of cholesterol synthesis in the Hepatic expression of mRNA and proteins involved in lipid metabolism and hepatic HMGCR activity. A: Total RNA from the liver of male mice (n = 5 in each group) at the age of 12 weeks was subjected to quantitative real-time PCR. Relative mRNA levels were normalized to GAPDH. B: Immunoblot analysis of LDLR, SR-BI, and HMGCR of fSS and L-SSKO male mice at the age of 12 weeks. PDI was served as loading controls. Each lane represents an individual mouse. C: HMGCR activity (the rate of biosynthesis of mevalonate from HMG-CoA) in the liver microsomal fractions of fSS and L-SSKO male mice at 12 weeks of age (n = 4 in each group). Each value represents the mean ± SD. * P < 0.05, ** P < 0.01, and *** P < 0.001 by Student's t -test. LRP1, LDL receptor-related protein 1; MTTP, microsomal triglyceride transfer protein.
SS somehow suppresses LXR-␣ activity, thereby decreasing VLDL production. Because we failed to detect changes in hepatic contents of the oxysterols that can transactivate LXR-␣ , other pathways may underlie the suppression of LXR-␣ activity. In this context, it is interesting to note that the upstream nonsterol isoprenoid derivative of mevalonate, geranylgeranyl diphosphate (GGPP), and its alcohol derivative, geranylgeranyol, inhibit LXR-␣ signaling ( 36 ). This mechanism may play a dominant role because FPP, a precursor of GGPP, was robustly increased in the livers of L-SSKO mice ( Fig. 2B ).
The hypolipidemia observed in the L-SSKO mice in a fed state disappeared in a fasted state ( Table 1 ). Fasting may mask the phenotypic manifestations caused by genetic inhibition of cholesterol biosynthesis, by suppressing the overall biosynthesis of cholesterol in the liver ( 37 ).
Finally, the transient liver dysfunction deserves discussion. Farnesol has been reported to cause cell-cycle arrest and apoptosis in some transformed cells ( 38 ). We confi rmed a similar cytotoxicity on hepatocytes (supplementary Fig. 3B). Furthermore, liver dysfunction and increased synthesis of farnesol had similar time courses ( Fig. 3C ; supplementary Fig. 3A). Therefore, it is highly probable that the transient buildup of farnesol induced death of hepatocytes in L-SSKO mice. The increased cell proliferation might have occurred as a compensatory regeneration ( Fig.  3H, I ). The hepatotoxicity of lapaquistat acetate, or TAK-475, was the major reason for the suspension of its clinical development ( 7 ). If a buildup of farnesol is a causative factor, coadministration of statins can alleviate the liver toxicity by reducing the synthesis of FPP. A similar strategy has been proposed to avoid nonsterol isoprenoid depletion associated with treatment with statins ( 39 ). Judging from our results, the liver toxicity of SS inhibition is modest and self-limited even though the inhibition is complete. Therefore, it might be premature to give up the development of SS inhibitors as a novel strategy to treat hyperlipidemia and atherosclerosis.
In conclusion, L-SSKO mice are hypolipidemic due to a decrease in VLDL production. They develop mild and transient liver dysfunction. This animal model can be used to further defi ne the role of hepatic cholesterol synthesis in the regulation of plasma lipoproteins and to understand the mechanism of toxicity of SS inhibition.