Schoenheimer effect explained – feedback regulation of cholesterol synthesis in mice mediated by Insig proteins (original) (raw)
Figure 1A shows the gene-targeting strategy for mouse Insig1. A neomycin resistance cassette (neo) flanked by loxP and frt sites was inserted into the 5′-flanking region of the Insig1 gene. An additional loxP site was inserted into the first intron. Cre-mediated recombination removes exon 1 of Insig1. Mice carrying the floxed Insig1 allele were bred to transgenic mice that express Cre recombinase driven from the interferon-responsive MX1 promoter (_MX1_-Cre) to derive mice homozygous for the floxed Insig1 allele and hemizygous for the _MX1_-Cre transgene (Insig1_f/f_MX1-Cre; f/f denotes flox/flox). When injected 4 times with a synthetic double-stranded ribonucleotide polyinosinic acid–polycytidylic acid (pIpC), these mice produced interferon, which activates the MX1 promoter, thereby producing the Cre recombinase in liver and other tissues (26). Figure 1B shows Southern blot analysis of _NcoI_-digested genomic DNA extracted from the livers of pIpC-treated mice of the indicated genotypes and probed with a 299-bp genomic fragment at the 5′ end of the Insig1 gene. In wild-type mice and mice hemizygous for the _MX1_-Cre transgene, a 9-kb band is observed that is unaffected by the expression of the Cre transgene. In _Insig1_f/f mice, a 3-kb band is observed which is markedly decreased in Insig1_f/f_MX1-Cre mice and is replaced by a 6-kb band that results from recombination between loxP sites. This recombination eliminates sequences encoding the first 119 of the 259 amino acids of Insig1.
Targeted disruption of Insig1 and Insig2 genes in mice. (A) Schematic of Insig1 gene-targeting strategy. Cre-mediated excision of the sequences between loxP sites deletes exon 1. The location of the probe used for Southern blot analysis is denoted by the horizontal filled rectangle labeled “probe.” (B) Representative Southern blot analysis of _Nco_I-digested DNA from livers of mice with the indicated genotypes that were treated with 4 intraperitoneal injections of pIpC (300 μg/injection). (C) Northern blot analysis of hepatic RNA of mice indicated in B. Total RNA from liver was pooled, and 20-μg aliquots were subjected to electrophoresis and blot hybridization with 32P-labeled cDNA probes for mouse Insig1 and mouse cyclophilin. (D) Immunoblot analysis of livers of mice indicated in B. Liver membrane fractions were prepared as described in Methods, and aliquots (45 μg) were subjected to SDS-PAGE and immunoblot analysis. (E) Schematic of Insig2 gene-targeting strategy. The Insig2 allele was disrupted by replacement of exons II and III of the Insig2 gene with a polIIsneopA expression cassette. The DNA probe used for Southern blot analysis is denoted by the horizontal filled rectangle labeled “probe.” (F) Representative Southern blot analysis of _NcoI_-digested tail DNA of the offspring from mating of Insig2+/− mice. (G) Northern blot analysis of hepatic RNA of mice described in F. Total RNA from livers of mice was subjected to electrophoresis and blot hybridization with 32P-labeled cDNA probes for mouse Insig2 and mouse cyclophilin. (H) Immunoblot analysis of liver membranes from mice with the indicated Insig2 genotype, as described above.
Figure 1, C and D, shows the expression of Insig1 by Northern blot analysis of total RNA and by immunoblotting of membrane proteins prepared from the same livers as those represented in Figure 1B. Levels of hepatic Insig1 mRNA and protein were identical in wild-type and _MX1_-Cre mice and were reduced by 20% in _Insig1_f/f mice, perhaps owing to transcriptional interference from the neo cassette. Insig1 mRNA and protein declined by more than 90% in Insig1_f/f_MX1-Cre mice injected with pIpC as compared with that in wild-type mice. In other studies using animals that were not injected with pIpC, the expression of Insig1 mRNA in Insig1_f/f_MX1-Cre was the same as in _Insig1_f/f mice (data not shown).
Figure 1E shows the strategy for germline targeting of mouse Insig2. Exons 2 and 3 of Insig2 were replaced by a neo cassette, which eliminated sequences encoding the first 123 of 225 amino acids of Insig2. Mating between Insig2+/– mice yielded Insig2+/+, Insig2+/–, and _Insig2_–/– mice at the expected 1:2:1 ratio. Figure 1F shows representative Southern blots of _NcoI_-digested genomic DNA extracted from the tails of Insig2+/+, Insig2+/–, and _Insig2_–/– littermates probed with a 505-bp genomic fragment containing exon 6 of the Insig2 gene. The expression of Insig2 mRNA and protein in these mice is shown in Figure 1, G and H, respectively. When the RNA was probed with a cDNA containing part of the 3′-untranslated region of Insig2, the result is an aberrant transcript in Insig2+/– and _Insig2_–/– mice, perhaps owing to a transcript initiating in exons 1a or 1b and splicing directly into exons 4, 5, or 6 (Figure 1G). In such a transcript, the first available inframe methionine is located in exon 5. If translated, this transcript would yield a peptide of only 23 amino acids. Levels of hepatic Insig2 protein, as detected by immunoblotting with antiserum raised against full-length mouse Insig2, were reduced by approximately 50% in Insig2+/– livers and were undetectable in _Insig2_–/– livers (Figure 1H).
Mice carrying the floxed Insig1 allele were also bred to transgenic mice that express Cre recombinase driven from the adenovirus EIIA promoter (EIIA-Cre). EIIA-Cre mice express Cre in germ cells, which results in germline deletion of the floxed Insig1 allele. Insig1+/–Insig2+/– mice were mated to each other in an attempt to produce Insig1–/–Insig2–/– mice. When the surviving adults were genotyped, 0 of 6 expected Insig1–/–Insig2–/– mice were observed, and only 3 of 12 expected Insig1–/–Insig2+/– mice were observed (Table 1).
Segregation of disrupted Insig1 and Insig2 alleles in mice
Further breeding experiments were conducted in which Insig1+/–Insig2–/– mice were interbred, and pregnant females were killed at 12.5 and 18.5 days post coitum (dpc). At 12.5 dpc, the observed ratio of Insig1+/+Insig2–/–, Insig1+/–Insig2–/–, and Insig1–/–Insig2–/– embryos was 46:81:41, which is consistent with the expected 1:2:1 ratio (Table 1). At 18.5 dpc, 82% of the expected Insig1–/–Insig2–/– embryos were observed. Further studies into the cause of the neonatal lethality in Insig1–/–Insig2–/– mice are ongoing and are beyond the scope of this work.
In order to study the metabolic effects of total Insig deficiency in adult livers, Insig1_f/f_MX1-Cre mice were bred with Insig2–/– mice. Mice of 4 genotypes were produced: _Insig1_f/f (designated control), Insig1_f/f_MX1-Cre (designated L-Insig1–/– to denote the conditional deficiency of Insig1 in liver when induced with pIpC), Insig1_f/f_Insig2–/– (hereafter designated Insig2–/–), and Insig1_f/f_Insig2–/–MX1-Cre (designated L-Insig1–/–Insig2–/–). For the studies described in Figures 2–6, all of the mice in each group were injected 4 times with pIpC and then studied between 7 and 14 days after the last injection.
Immunoblot (A) and lipid analysis (B) of livers from control and Insig-deficient mice. The mice used in this figure are the same as those compared in Table 2. (A) Immunoblot analysis from nuclear extract and membrane fractions obtained from mice with the following 4 genotypes: (a) _Insig1_f/f (designated control); (b) Insig1_f/f_Insig2_−/−_MX1-Cre (designated L-Insig1–/–Insig2–/–); (c) _Insig1_f/f_MX1_-Cre (designated L-Insig1–/–); and (d) Insig1_f/f_Insig2–/– (designated Insig2–/–). Each mouse was treated with 4 intraperitoneal injections of pIpC (300 μg/injection), and tissues were obtained 14 days after the final injection. Livers (n = 6) were separately pooled, and 45-μg aliquots of the pooled membrane and nuclear extract fractions were subjected to SDS-PAGE and immunoblot analysis. CREB protein and the transferrin receptor were used in the immunoblots as loading controls for the nuclear extract and membrane fractions, respectively. P, pSREBP; N, nSREBP. (B) Hepatic cholesterol and triglyceride content of control and Insig-deficient livers. Each bar represents the mean ± SEM of data from 6 mice.
In vivo synthesis rates of sterols (A) and fatty acids (B) in livers and brains from control and L-Insig1–/–Insig2–/– mice. Mice (20- to 24-week-old males; 5 or 6 per group) were treated with 4 intraperitoneal injections of pIpC (300 μg/injection). Five and a half days after the final injection, mice were fed ad libitum a chow diet containing 0.02% (low) or 1.5% (high) cholesterol for 2.5 days prior to sacrifice, at which time the mice were injected intraperitoneally with 3H-labeled water (50-mCi in 0.20 ml of isotonic saline). One hour later the tissues were removed for measurement of 3H-labeled fatty acids and digitonin-precipitable sterols. Each bar represents the mean ± SEM of the values from 5 or 6 mice.
When fed a chow diet ad libitum and sacrificed 14 days after the last pIpC injection, mice from all genotypes were indistinguishable by external appearance and body weight (Table 2). Plasma cholesterol was slightly elevated, and plasma triglycerides were slightly reduced in L-Insig1–/–Insig2–/– mice (P < 0.05, 2-tailed Student’s t test). Levels of plasma insulin, glucose, and free fatty acids were similar among the 4 groups.
Comparison of control, L-Insig1–/–Insig2–/–, L-Insig1–/–, and Insig2–/– mice
Figure 2A shows immunoblots of pooled hepatic membrane fractions and nuclear extracts from pIpC-treated mice of the 4 genotypes compared in Table 2. Mice with single deficiencies of hepatic Insig1 or Insig2 did not compensate by increasing the expression of Insig2 or Insig1, respectively. Levels of hepatic Insig1 protein were dramatically reduced in L-Insig1–/–Insig2–/– and L-Insig1–/– mice. Hepatic Insig2 protein was undetectable in L-Insig1–/–Insig2–/– and Insig2–/– mice. Levels of the membrane-bound precursors of SREBP-1 and SREBP-2 (pSREBP-1 and -2) were largely unaffected by deficiency of one or both Insig proteins. A small increase in Scap protein was consistently observed in the livers of mice deficient in Insig1. nSREBP-1 levels were increased by 2-fold in L-Insig1–/–Insig2–/– mice compared with those in control mice, while nSREBP-2 levels were unchanged. Hepatic HMG-CoA reductase protein levels were slightly elevated in Insig2–/– mice and were dramatically elevated in L-Insig1–/–Insig2–/– mice.
Figure 2B shows liver cholesterol and triglyceride content in the mice compared in Table 2. The hepatic content of total cholesterol and triglycerides was increased by 4- and 6-fold, respectively, in the L-Insig1–/–Insig2–/– mice compared with the hepatic content in control mice on the same diet. In the double-knockout livers, the free cholesterol content rose by 1.4-fold, and the cholesteryl ester content rose by 15-fold (data not shown). Hepatic lipid levels were normal in mice of the other genotypes. These data indicate that loss of both Insigs is necessary in order to perturb lipid metabolism grossly in mouse liver. For this reason, in all subsequent studies we used L-Insig1–/–Insig2–/– mice.
Table 3 shows the relative expression of various mRNAs in livers from L-Insig1–/–Insig2–/– mice studied in 3 different experiments. The values in control mice for each experiment were assigned as 1. The mean value for Insig1 mRNA was reduced by 95% compared with that of control mice, which reflects the direct action of the Cre recombinase. The Insig2 gene has 2 promoter/enhancer regions that give rise to 2 transcripts, denoted Insig2a and Insig2b (24). The expression of both transcripts was undetectable in L-Insig1–/–Insig2–/– livers. The mean mRNA values for SREBP-1a, SREBP-2, and Scap were unaffected by Insig deficiency whereas a small increase in SREBP-1c expression (mean of 1.8-fold) was observed. The mRNAs of 5 SREBP target genes involved in fatty acid synthesis were increased by 2.6- to 7.7-fold, consistent with the increase in nSREBP-1 in L-Insig1–/–Insig2–/– livers (see Figure 2A). mRNAs of 4 SREBP targets involved in cholesterol synthesis were generally the same as those in the control mice except for a 2-fold increase in HMG-CoA reductase mRNA. When considered in light of the large increase in the cholesterol content of the Insig-deficient livers, these 4 mRNAs were all inappropriately elevated (see below). mRNAs for ABCG5, ABCG8, and lipoprotein lipase — 3 genes known to be stimulated by LXRα and LXRβ (27) — were increased in L-Insig1–/–Insig2–/– livers by 3.0- to 5.3-fold. We attribute their increase to the activation of LXRs by the sterols (27) that accumulate in L-Insig1–/–Insig2–/– livers. The mRNA for LXRα itself was unaltered.
mRNAs in liver of L-Insig1–/–Insig2–/– mice as compared with values in control mice
Figure 3A shows the gross appearance of the livers of control and L-Insig1–/–Insig2–/– female mice fed a chow diet ad libitum 12 days after injection with pIpC. The liver of the L-Insig1–/–Insig2–/– mouse is pale, owing to the accumulation of lipids. The livers from mice of the other 3 genotypes were normal in appearance (data not shown). When liver sections were stained with oil red O, a fat-specific dye, the increase in lipids in the L-Insig1–/–Insig2–/– mice was dramatic (Figure 3B).
Lipid accumulation in livers of L-Insig1–/–Insig2–/– mice. Adult female mice were treated with 4 intraperitoneal injections of pIpC (300 μg/injection), and the livers were removed either 12 days (A and B) or 7 days (C and D) after the final injection. (A and B) Photographs of livers from chow-fed control and L-Insig1–/–Insig2–/– mice. (C and D) Oil red O–stained histologic sections of the livers from chow-fed control (left) and L-Insig1–/–Insig2–/– mice. Mice were perfused through the heart with HBSS and then with 10% (v/v) formalin in PBS; frozen sections of livers were stained with oil red O. Magnification, ×20.
In cultured cells, the reduction of Insig expression creates a relative resistance to the effects of sterols in blocking SREBP processing (14) and accelerating HMG-CoA reductase degradation (21). A similar resistance was apparent in the livers of L-Insig1–/–Insig2–/– mice. This resistance can be inferred from the observation that nSREBPs were still present and HMG-CoA reductase protein was increased in these livers despite the accumulation of cholesterol to levels that would have reduced these proteins in wild-type mice.
To enable direct comparison of cholesterol regulation in livers of control and L-Insig1–/–Insig2–/– mice, the mice were fed diets containing varying amounts of cholesterol for 2.5 days, after which pSREBPs, nSREBPs, and HMG-CoA reductase were measured by immunoblotting (Figure 4A). Metabolic measurements in these mice are provided in Table 4. When L-Insig1–/–Insig2–/– mice consumed the unsupplemented chow diet, levels of nSREBP-1 were elevated by 2-fold, and levels of nSREBP-2 were unaltered. In control mice fed a 2% cholesterol diet, nSREBP-2 was reduced by 93% (Figure 4, A and B). This decline was severely blunted in L-Insig1–/–Insig2–/– mice, in whom nSREBP-2 decreased only 17%. When mice were fed high-cholesterol diets, nSREBP-1 decreased slightly in controls and even less in L-Insig1–/–Insig2–/– mice. nSREBP-1 remained elevated in the knockout animals at all levels of dietary cholesterol.
Markedly elevated levels of nSREBPs and HMG-CoA reductase in the livers of Insig-deficient mice fed high-cholesterol diets. The mice used for this figure are the same as those compared in Table 4. Each mouse was treated with 4 intraperitoneal injections of pIpC (300 μg/injection); 8.5 day after the final injection, mice were fed ad libitum a chow diet containing the indicated amount of cholesterol for 2.5 days prior to study. (A) Immunoblot analysis of SREBP-1 and SREBP-2 from livers of control and L-Insig1–/–Insig2–/– mice fed with the indicated amount of cholesterol. Livers (4 or 6 per group) were separately pooled, and 45-μg aliquots of the membrane and nuclear extract fractions were subjected to SDS-PAGE and immunoblot analysis. Nonspecific bands are denoted by the asterisk. Arrows indicate the position of migration on SDS gels of monomeric HMG-CoA reductase (97 kDa). Immunoblots of CREB and transferrin receptor were used as loading controls for the nuclear extract and membrane fractions, respectively. (B) The gels of nuclear extract fractions (nuclear) shown in A were scanned and quantified by densitometry. The intensities of cleaved nSREBP-1 and nSREBP-2 in lane 1 (control mice fed with 0.02% cholesterol) were arbitrarily set at 100%. (C) Hepatic cholesterol content of control and Insig-deficient mice. Each value represents the mean ± SEM of data from 4 or 6 mice.
Effects of dietary cholesterol feeding in control and L-Insig1–/–Insig2–/– mice
Whereas HMG-CoA reductase protein levels were markedly reduced in control mice fed the 0.2% and 2.0% cholesterol diets, the elevated levels in L-Insig1–/–Insig2–/– mice were not affected (Figure 4A). Insig1 and Insig2 protein levels were unaffected by cholesterol feeding in control mice and were undetectable in the knockout animals (Figure 4A). Hepatic cholesterol content was markedly elevated in the knockout mice, and it did not increase further with cholesterol feeding (Figure 4C). Hepatic cholesterol increased in the control mice, but it did not reach the high levels seen in the knockout animals.
To confirm that the elevated levels of HMG-CoA reductase protein in the L-Insig1–/–Insig2–/– livers were associated with increases in HMG-CoA reductase activity, we assayed the enzyme activity in liver homogenates. Inasmuch as most HMG-CoA reductase molecules are phosphorylated and inactive (28, 29), we homogenized the livers in 2 different buffers. One of these contained sodium fluoride, which stops dephosphorylation during homogenization and is therefore believed to reveal true endogenous HMG-CoA reductase activity. The other contained sodium chloride, which permits dephosphorylation and therefore represents the maximal activity from the enzyme (28). In control mice, cholesterol feeding markedly reduced HMG-CoA reductase activity regardless of the homogenization buffer (Table 5). In the L-Insig1–/–Insig2–/– mice on a chow diet, reductase activity was markedly elevated, and it failed to decline after cholesterol feeding. The relative results were the same when the livers were homogenized in fluoride or chloride.
HMG-CoA reductase activity in livers of cholesterol-fed mice
Relevant hepatic mRNAs were quantified by real-time PCR in the livers of mice fed low-cholesterol chow (0.02%) or the high-cholesterol (2.0%) diet (Figure 5). Liver expresses 2 major isoforms of SREBPs: SREBP-1c and SREBP-2 (30). Although their functions overlap, SREBP-1c preferentially activates the genes involved in fatty acid and triglyceride synthesis while SREBP-2 preferentially activates the genes required for cholesterol synthesis (31). mRNAs for both SREBP-1c and -2 are enhanced by nSREBPs themselves in a feed-forward activation loop and, in addition, SREBP-1c mRNA is induced by insulin and by agonists of LXRs (32). In control mice, SREBP-1c mRNA was induced 2-fold by high-cholesterol feeding, which we attribute to the activation of LXRs by oxysterols generated from dietary sterols (33). In the sterol-overloaded livers of L-Insig1–/–Insig2–/– mice, SREBP-1c mRNA was elevated by 2-fold and did not rise further after cholesterol feeding. SREBP-2 mRNA was suppressed by 46% in control mice fed the high-cholesterol diet; this decline was blunted in L-Insig1–/–Insig2–/– mice. The decline of SREBP-2 mRNA in control mice is likely due to the loss of feed-forward activation as SREBP processing is reduced by the increase in hepatic cholesterol. In the double-knockout mice, the loss of Insigs prevents cholesterol from blocking SREBP processing, and thus the SREBP-2 mRNA does not decline. Consistent with this hypothesis, SREBP-2 target mRNAs for genes involved in cholesterol synthesis (HMG-CoA reductase, HMG-CoA synthase, farnesyl diphosphate [FDP] synthase, and squalene synthase) declined by more than 75% when control mice were fed the high-cholesterol diet. Levels of these mRNAs were elevated by 1.7- to 2.6-fold in L-Insig1–/–Insig2–/– mice. They declined partially when the animals were fed the high-cholesterol diet, but remained greater than or equal to levels in control mice fed the low-cholesterol diet.
Relative amounts of various mRNAs in livers from control and L-Insig1–/–Insig2–/– mice fed with diets containing a low (L) or high (H) amount of cholesterol (0.02% or 2.0% cholesterol, respectively). The mice used here are the same as those used for Figure 4 and Table 4. Total RNA from livers of mice was pooled and subjected to real-time PCR quantification as described in Methods. Each value represents the amount of mRNA relative to that in the control mice fed with a chow diet (0.02% cholesterol), which is arbitrarily defined as 1. LCE, long-chain fatty acyl–CoA elongase; SCD-1, stearoyl-CoA desaturase-1; GPAT, glycerol-3-phosphate acyltransferase; PEPCK, phosphoenolpyruvate carboxykinase.
Levels of mRNAs for genes involved in fatty acid and triglyceride synthesis were elevated 2.5- to 5.5-fold in the L-Insig1–/–Insig2–/– mice. These mRNA levels declined only slightly with cholesterol feeding in both the control and the knockout animals (Figure 5). mRNAs for the LXR-responsive genes ABCG5 and ABCG8 were induced by high-cholesterol feeding in control mice. These mRNAs were elevated in L-Insig1–/–Insig2–/– mice, and they did not increase with cholesterol feeding, which is consistent with the hypothesis that hepatic LXRs were already activated by sterols in the livers of the L-Insig1–/–Insig2–/– mice.
To test the effect of Insig deficiency on hepatic lipid synthesis, male control and L-Insig1–/–Insig2–/– mice were fed diets containing 0.02% (low) or 1.5% (high) cholesterol for 2.5 days, after which in vivo lipid synthesis was determined by measurement of the incorporation of intraperitoneally injected 3H-labeled water into digitonin-precipitable sterols and fatty acids (Figure 6). High-cholesterol feeding reduced hepatic sterol synthesis by 93% in control mice. Hepatic sterol synthesis in L-Insig1–/–Insig2–/– mice was elevated by 5-fold, and it did not decline with the high-cholesterol diet. Hepatic fatty acid synthesis was 3-fold higher in L-Insig1–/–Insig2–/– mice fed the low-cholesterol diet compared with that in control mice. The synthesis of sterols and fatty acids in brain, a tissue with a very low level of interferon-induced _MX1_-Cre recombination (26), was normal in L-Insig1–/–Insig2–/– mice.