The Role of Insulin in the Regulation of PCSK9

Abstract

Objective

Proprotein convertase subtilisin/kexin type 9 (PCSK9), which binds the low density lipoprotein (LDL) receptor and targets it for degradation, has emerged as an important regulator of serum cholesterol levels and cardiovascular disease risk. Although much work is currently focused on developing therapies for inhibiting PCSK9, the endogenous regulation of PCSK9, particularly by insulin, remains unclear. The objective of these studies was to determine the effects of insulin on PCSK9 in vitro and in vivo.

Approach and Results

Using rat hepatoma cells and primary rat hepatocytes, we found that insulin increased PCSK9 expression and increased LDL receptor degradation in a PCSK9-dependent manner. In parallel, hepatic Pcsk9 mRNA and plasma PCSK9 protein levels were reduced by 55-75% in mice with liver-specific knockout of the insulin receptor; 75-88% in mice made insulin deficient with streptozotocin; and 65% in ob/ob mice treated with antisense oligonucleotides against the insulin receptor. However, antisense olignonucleotide mediated knockdown of insulin receptor in lean, wildtype mice had little effect. In addition, we found that fasting was able to reduce PCSK9 expression by 80% even in mice that lack hepatic insulin signaling.

Conclusions

Taken together, these data indicate that though insulin induces PCSK9 expression, it is not the sole or even dominant regulator of PCSK9 under all conditions.

Introduction

LDL receptors in the liver mediate clearance of more than 70% of LDL from the serum ¹. Consequently, the LDL receptor is a key determinant of serum cholesterol levels and cardiovascular disease risk. Proprotein convertase subtilisin/kexin type 9 (PCSK9) has emerged as an important regulator of the LDL receptor. PCSK9 is a secreted protein which binds to the extracellular domain of the LDL receptor and targets it for degradation ^{2, 3}. Consequently, individuals with gain of function mutations in PCSK9 show increased levels of LDL cholesterol and increased risk of cardiovascular disease ⁴. Conversely, individuals with loss of function mutations in PCSK9 manifest decreased levels of LDL cholesterol and reduced risk of cardiovascular disease ⁵.

Given the importance of PCSK9 in regulating the LDL receptor, a great deal of effort has recently been placed on producing inhibitors of PCSK9 for therapeutic use. Thus, antisense oligonucleotides and antibodies against PCSK9 have been developed, and preclinical studies have been promising^{6, 7}. However, much remains to be learned about the biology of PCSK9. In particular, identifying the endogenous regulators of PCSK9 is important as they could potentially be harnessed for therapeutic intervention. In addition, an understanding of the endogenous regulation of PCSK9 could guide the use anti-PCSK9 therapies in different patient populations.

PCSK9 levels vary over 100-fold between individuals⁸. Though some of this variability is due to genetic differences, and the use of cholesterol lowering drugs, it is also likely that changes in the hormonal milieu contribute. For example, thyroxine, estrogen and glucagon have all been implicated in the regulation of PCSK9 ⁹. The effects of insulin, however, have been controversial ^{10, 11,12}.

To better understand the role of insulin in the regulation of PCSK9, we studied the effects of insulin in vitro and in vivo. We find that insulin, in hepatoma cells and primary rat hepatocytes, increases PCSK9 mRNA and protein, and decreases the half-life of the LDL receptor protein in a PCSK9-dependent manner. In parallel, the ablation of insulin signaling in the hepatocyte either by genetic knockout using albumin-Cre driven recombination of LoxP sites in the insulin receptor, destruction of the β-cells of the pancreas, or antisense oligonucleotide-mediated knockdown against the insulin receptor in ob/ob mice, decreases PCSK9.

Materials and Methods

Material and Methods are available in the online-only Data Supplement.

Results

As expected, insulin increased mRNA levels of LDL receptor (Ldlr) in rat hepatoma cells ¹³. It also induced the lipogenic enzyme fatty acid synthase (Fasn), and suppressed the gluconeogenic enzyme phosphoenolpyruvate carboxykinase (Pck1, Fig. 1A). However, the effects on LDLR protein were more complex. Insulin treatment for four to twelve hours increased LDLR protein, in parallel with Ldlr mRNA. At 24-30 hours, however, insulin suppressed LDLR protein, and the amount of LDLR protein in cells treated with insulin for 24 hours fell below those observed in untreated cells (Fig. 1B). In contrast, protein levels of fatty acid synthase, as well as transferrin receptor (TFRC), were increased by insulin treatment at 24-30 hours (Fig. 1B). Thus, insulin has a biphasic effect on LDLR protein, inducing it at early time points, and suppressing it at later time points.

Figure 1. Biphasic regulation of LDLR by insulin.

Rat hepatoma cells (A to C) or primary rat hepatocytes (D, E) were treated with insulin. Gene expression was measured by real time PCR (A, D) and protein levels were determined by immunoblotting whole cell lysates (B, C, E). (A, B) Rat hepatoma cells were placed in serum free medium 30 hours prior to harvest, and insulin was added for the indicated amount of time prior to harvest. (C, E) Cells were treated with cycloheximide (CHX) for the indicated amount of time and insulin for a total of twelve hours prior to harvest. Primary rat hepatocytes were serum starved overnight and treated with insulin for six hours prior to harvest (D) or treated with cycloheximide for the indicated amount of time and insulin for a total of twelve hours prior to harvest (E). In B, C and E, representative gel images are shown on the left and protein quantifications on the right. Data represent mean and s.e.m. of three to five independent experiments. In A and D, *p < 0.05 compared to non-stimulated controls; in B, *, #, & p < 0.05 versus non-stimulated controls, for LDLR, FASN and TFRC, respectively; in C and E, *p < 0.05 for insulin treatment and #p < 0.05 for cycloheximide treatment.

The fact that insulin reduced LDLR protein but not Ldlr mRNA at later time points suggested that insulin might increase LDLR degradation. To test this, we measured the stability of LDLR protein by treating cells with cycloheximide, which blocks protein synthesis. LDLR protein levels fell more rapidly in the presence of insulin than in the absence of insulin (Fig. 1C). Quantification of the immunoblots of five independent experiments revealed that insulin reduced the half-life of the LDLR by almost 33% (Figure IA, B in the online-only Data Supplement). In parallel, pulse-chase studies also showed that insulin increased the degradation of the LDLR (Figure IC, D in the online-only Data Supplement).

Primary rat hepatocytes are the most robust model system currently available for studying insulin action in vitro, particularly on lipid-related genes^{14, 15}. In these cells, insulin increased mRNA levels of Ldlr by greater than two-fold and fatty acid synthase by seven-fold; in parallel, insulin suppressed Pck1 by thirty-fold (Fig. 1D). At the same time, insulin increased LDLR degradation: in the absence of insulin, 44% of LDLR protein remained after cycloheximide treatment whereas in the presence of insulin, only 16% of LDLR protein remained after cycloheximide treatment (Fig. 1E).

PCSK9 ^{2, 3} and inducible degrader of the LDLR (IDOL) ¹⁶ promote the degradation of the LDLR. In rat hepatoma cells, Idol mRNA was not detectable in either the absence or presence of insulin (data not shown). However, insulin did induce Pcsk9 mRNA, cellular PCSK9, and secreted PCSK9 (Fig. 2A). To determine whether PCSK9 is necessary for the decrease in LDLR protein observed with prolonged insulin treatment, cells were treated with either a control adenovirus or an adenovirus encoding a shRNA against PCSK9, and then incubated in the presence or absence of insulin for 24 hours (Fig. 2B). The shPCSK9 adenovirus effectively decreased PCSK9 protein in the cells and the media, both in the presence and absence of insulin (Fig. 2B). In the absence of insulin, knockdown of PCSK9 increased LDLR as expected. Importantly, the ability of insulin to decrease LDLR protein was blunted by knockdown of PCSK9 (Fig. 2B). Knockdown of PCSK9 did not, however, alter expression of fatty acid synthase or transferrin receptor (Fig. 2B).

Figure 2. PCSK9 mediates insulin induced degradation of the LDLR.

Rat hepatoma cells (A, B) or primary rat hepatocytes (C, D) were treated with insulin. Gene expression was measured by real time PCR (A, C) and protein levels were determined by immunoblotting whole cell lysates or medium (A, B, D). (A) Rat hepatoma cells were placed in serum free medium 30 hours prior to harvest, and insulin was added for the indicated amount of time prior to harvest. (B) Rat hepatoma cells were infected with adenovirus expressing a control shRNA (shControl) or an shRNA against PCSK9 (shPCSK9) and then treated with insulin for 24 hours. (C, D) Primary rat hepatocytes were serum starved overnight and then stimulated with insulin for six hours (C) or 24 hours (D). (D) PCSK9 protein in the medium was measured by immunoblotting; silver staining of the upper portion of the gel shows that similar amounts of protein were loaded. In A, B and D, representative gel images and protein quantifications are shown. Data represent mean and s.e.m. of three to six independent experiments. In A, *, #, & p < 0.05 versus non-stimulated controls for Pcsk9 mRNA levels, PCSK9 levels in cell and medium, respectively; in B, *p < 0.05 for the effects of insulin and # p < 0.05 for the effects of Ad-shPCSK9; in C and D, *p < 0.05 versus non-stimulated controls.

In primary rat hepatocytes, insulin produced an even more robust increase in PCSK9. Pcsk9 mRNA levels were increased three-fold, and secreted PCSK9 protein was increased three-fold (Fig. 2C, 2D). The effects of insulin were mediated at the transcriptional level, as insulin induced the activity of a PCSK9 promoter luciferase construct by two-fold (Figure IIB in the online-only Data Supplement) and the ability of insulin to induce Pcsk9 mRNA in primary rat hepatocytes was entirely abolished by treatment with actinomycin D, which interferes with transcription (Figure IIA in the online-only Data Supplement). Moreover, insulin induction of the PCSK9 promoter was blocked by inhibition of the phosphoinositide 3-kinase signaling pathway, which mediates many of insulin’s metabolic effects (Figure IIC in the online-only Data Supplement); it was also blocked by mutation of the sterol response element (SRE) and hepatocyte nuclear factor 1 (HNF1) binding sites, but not the Sp1 site (Figure IID in the online-only Data Supplement). We therefore examined expression of sterol regulatory element binding protein-1a (Srebp-1a), Srebp-2 and Srebp-1c as well as HNF-1α and HNF-1β in response to insulin. In rat hepatoma cells, only Srebp-2 was increased at the mRNA level (Figure IIE in the online-only Data Supplement). In primary rat hepatocytes, Srebp-1c was increased twelve-fold and Srebp-1a was increased two-fold, whereas Srebp-2 and the HNFs were not changed (Figure IIF in the online-only Data Supplement). Collectively, these data support the notion that insulin induces Pcsk9 transcription via the SREBPs, but do not rule out a role for other factors.

To determine the effects of insulin on PCSK9 in vivo, we used Liver Insulin Receptor Knockout (LIRKO) mice and their controls. LIRKO hepatocytes lack insulin receptors making them incapable of insulin signaling¹⁷. Consequently, LIRKO mice are hyperglycemic and hyperinsulinemic¹⁷. We have previously shown that LIRKO mice show decreased levels of SREBP-1, SREBP-2, and the SREBP target genes^{18, 19}. Moreover, on an atherogenic Paigen diet (15% dairy fat, 1% cholesterol, 0.5% cholic acid)²⁰, LDLR protein levels are markedly decreased¹⁸.

We therefore measured PCSK9 and LDLR in LIRKO and control mice fed either a chow or atherogenic Paigen diet for four weeks (Fig. 3). On the chow diet, LIRKO mice showed decreased levels of plasma PCSK9, decreased Pcsk9 and Ldlr mRNA levels, and normal LDLR protein levels (Fig. 3A-C, 3E). These data suggest that insulin increases both LDLR synthesis and degradation, and is consistent with our in vitro data showing increased Pcsk9 and Ldlr mRNA in the presence of insulin.

Figure 3. LIRKO mice show decreased PCSK9 expression.

Eight week old male LIRKO (L) mice and their littermate controls (C) were fed either a chow or Paigen diet for four weeks and sacrificed in the non-fasted state. Plasma PCSK9 was measured by ELISA (A, n=5-9 mice per group). Gene expression was measured by real time PCR (B-D, n=4-8 mice per group). (E) Hepatic LDLR was measured by immunoblotting (left) and quantified (right). (F) Plasma cholesterol was measured using a colorimetric kit. *P < 0.05 versus control mice on the same diet; ^# p < 0.05 versus chow-fed mice of the same genotype.

The Paigen diet reduced Pcsk9 and Ldlr mRNA in both control and LIRKO mice. However, control mice maintained normal levels of LDLR protein on the Paigen diet (Fig. 3E). LIRKO mice, on the other hand, showed a marked reduction in LDLR protein and developed severe hypercholesterolemia on the Paigen diet (Fig. 3E, F). Why LDLR protein was decreased in the Paigen-fed LIRKO mice is still under investigation, but may be due to decreased Ldlr mRNA and increased Idol expression (Fig. 3C, D).

Pcsk9 has also been previously shown to be decreased by fasting^{21, 22}. To determine whether the effects of fasting on Pcsk9 were mediated by insulin, we subjected control and LIRKO mice to a 24-hour fast. Interestingly, fasting reduced Pcsk9 mRNA by 95% in control mice, and 80-90% in LIRKO mice (Fig. 4A). Similar effects were observed on Ldlr mRNA, but the effects were more modest (Fig. 4A). Conversely, re-feeding (i.e., feeding mice a high carbohydrate diet after a 24-hour fast) increased Pcsk9 mRNA by almost 60-fold in control mice, but only 30-fold in LIRKO mice (Fig. 4B). In parallel, re-feeding induced Ldlr four-fold in control mice, but only two-fold in LIRKO mice (Fig. 4B). Idol mRNA, on the other hand, was again slightly higher in LIRKO mice, but unchanged by fasting or re-feeding (Fig. 4A, 4B). Interestingly, LDLR protein was similar in control and LIRKO mice in the fasted state (0 hours of re-feeding) and after 24 hours of re-feeding, just as it was in chow-fed mice (Fig. 4C). However, after six or twelve hours of re-feeding, LIRKO mice showed lower levels of LDLR protein than controls, despite the fact that Pcsk9 levels were lower.

Figure 4. Fasting and Feeding Effects on PCSK9.

Eight to twelve week old chow fed male LIRKO (L) mice and their littermate controls (C) were sacrificed in the non-fasted state (A, white bars), after a 24 hour fast (A, black bars) or after a 24 hour fast followed by re-feeding a high carbohydrate diet for 0-24 hours (B, C). Gene expression was measured by real time PCR (A, B n=4-8 mice per group). (C) Hepatic LDLR was measured by immunoblotting (left), and quantified (right). In A, *p< 0.05 versus similarly treated control mice; ^# p < 0.05 versus non-fasted mice of the same genotype. In B and C, * p < 0.05 versus control mice; ^#, & p < 0.05 versus 0 h refeeding in control or LIRKO mice, respectively.

The fact that Pcsk9 is still regulated by fasting in LIRKO mice indicates that other factors besides hepatic insulin action are important in the regulation of Pcsk9. One such factor could potentially be glucagon, which has previously been shown to suppress hepatic Pcsk9 mRNA in vivo in rats⁹. Indeed, glucagon suppressed PCSK9 mRNA and protein levels by 50% in primary rat hepatocytes (Figure IIIA, IIIC in the online-only Data Supplement). Glucagon also suppressed mRNA levels of Srebp-1c and Srebp-2 by 20-50%, while it increased expression of Pck1 nine-fold (Figure IIIA, IIIB in the online-only Data Supplement). More importantly, though glucagon decreased Ldlr mRNA by 20%, it increased LDLR protein by two-fold (Figure IIIA, IIID in the online-only Data Supplement).

We also examined the effects of insulin in the context of insulin deficiency and selective insulin resistance. Selective insulin resistance is a key feature of Type 2 diabetes, in which not all signaling pathways appear to become resistant to insulin – some appear to remain sensitive to insulin and are driven to excess by the hyperinsulinemia which co-evolves with insulin resistance¹⁵. As a model of insulin deficiency, we chose streptozotocin treated mice, as streptozotocin destroys the β-cells of the pancreas. As a model of selective insulin resistance, we chose leptin deficient, obese ob/ob mice. To compare directly the effects of insulin deficiency and selective insulin resistance, we studied in parallel lean, wildtype mice treated with vehicle (WT); lean, wildtype mice treated with streptozotocin (STZ); and ob/ob mice treated with vehicle (ob/ob). Prior studies have reported both STZ and ob/ob mice to be hyperglycemic and hyperglucagonemic, though STZ mice are hypoinsulinemic and lean or underweight and ob/ob mice are hyperinsulinemic and obese²³ (see also Supplemental Table II in the online-only Data Supplement).

Liver gene expression analysis by real time PCR revealed that Pcsk9 was markedly decreased in streptozotocin treated mice but not ob/ob mice (Table 1). In parallel, the SREBP transcription factors which drive Pcsk9, Srebp-1c¹⁰ and Srebp-2²⁴, and their other targets, fatty acid synthase, stearoyl CoA desaturase 1 (Scd1), 3-hydroxy-3-methylglutaryl-CoA reductase (Hmgcr) and farnesyl diphosphate synthase (Fdps), were decreased by 44-98% in streptozotocin treated mice. In ob/ob mice, on the other hand, these genes were generally unchanged, or in the case of Srebp-1c, Fasn and Scd1, increased, consistent with prior studies^{12, 25}. The changes in FASN and SCD1 were also evident at the protein level, as they were markedly induced in ob/ob livers (Fig. 5A). In contrast, both streptozotocin-treated mice and ob/ob mice showed increased expression of the gluconeogenic genes, glucose-6-phosphatase (G6pc), Pck1 and Ppargc1a (Table 1), and were hyperglycemic (Supplemental Table II in the online-only Data Supplement). Idol mRNA was slightly elevated in ob/ob livers (Table 1).

Table 1.

Effects of diabetes on hepatic gene expression.

	WT	STZ	ob/ob
Pcsk9	1.00 ± 0.07	0.12 ± 0.03 *	0.91 ± 0.12 &
Ldlr	1.00 ± 0.11	0.49 ± 0.06 *	1.36 ± 0.16 *&
Srebp-1c	1.00 ± 0.10	0.12 ± 0.03 *	1.61 ± 0.14 *&
Fasn	1.00 ± 0.20	0.21 ± 0.03 *	7.90 ± 2.15 *&
Scd1	1.00 ± 0.15	0.02 ± 0.01 *	17.56 ± 5.38 *&
Srebp-2	1.00 ± 0.05	0.56 ± 0.08 *	0.89 ± 0.05 &
Hmgcr	1.00 ± 0.11	0.30 ± 0.08 *	1.47 ± 0.18 &
Fdps	1.00 ± 0.09	0.18 ± 0.06 *	0.74 ± 0.15 &
Idol	1.00 ± 0.09	0.74 ± 0.05	1.27 ± 0.17 &
G6pc	1.00 ± 0.21	2.31 ± 0.24 *	4.47 ± 0.35 *&
Pck1	1.00 ± 0.26	2.76 ± 0.24 *	1.61 ± 0.14
Ppargc1α	1.00 ± 0.08	6.60 ± 0.34 *	2.15 ± 0.37 *&

Vehicle treated ob/ob mice (ob/ob) and their lean, wildtype controls treated with either vehicle (WT) or streptozotocin (STZ) were sacrificed in the non-fasted state. Hepatic gene expression was measured by real time PCR (n= 5 per group).

^*p < 0.05 compared to WT mice;

^&p < 0.05 compared to STZ mice.

Figure 5. PCSK9 expression is decreased in mice with insulin deficiency, but not selective insulin resistance.

Eight to twelve week old male mice were sacrificed in the non-fasted state (n = 4 to 5 mice per group). (A to C) Vehicle treated ob/ob mice (ob/ob) and lean, wildtype controls treated with either vehicle (WT) or streptozotocin (STZ) were studied in parallel. Protein levels were measured by immunoblotting (A), plasma PCSK9 levels were measured by ELISA (B), and LDL-cholesterol levels were measured by colorimetric assays (C). *p < 0.05 compared to lean mice treated with vehicle and ^&p < 0.05 compared to STZ treated group. (D to H) ob/ob mice (ob/ob) and their lean, wildtype controls (WT) were treated with control ASO or ASO against the insulin receptor (INSR) for four weeks. Protein levels were measured by immunoblotting liver extracts (D), hepatic gene expression was measured by real time PCR (E, G and H), and plasma PCSK9 levels were measured by ELISA (F). *p < 0.05 compared to control ASO treated mice and ^#P < 0.05 compared to WT treated with the same ASO. In A and D, representative gel images (left) and protein quantifications (right) are shown. Data present the mean and s.e.m.; control mice treated with vehicle or control mice treated with control ASO were set to 1.

Plasma PCSK9 levels mirrored liver Pcsk9 mRNA levels (Fig. 5B), with STZ mice showing a 70% reduction in plasma PCSK9 levels. However, both STZ and ob/ob mice showed increased LDL cholesterol, despite a tendency towards elevated LDLR protein (Fig. 5A, 5C), likely because of increased secretion of VLDL, the precursor of LDL²⁶.

To directly test the role of insulin signaling in the regulation of PCSK9 in ob/ob mice, we knocked down the insulin receptor using antisense oligonucleotides (ASO). Thus, ob/ob mice and their wildtype controls were injected weekly for four weeks with either a control ASO or an ASO against the insulin receptor. Such treatment would abolish the ability of insulin to act on the hepatocyte.

We have previously shown that antisense-mediated knockdown of the insulin receptor worsens hyperglycemia, but decreases levels of Srebp-1c, its lipogenic targets, and hepatic steatosis in ob/ob mice²⁷. Consistent with this, treatment with ASO against the insulin receptor markedly decreased insulin receptor protein levels in the livers of both wildtype and ob/ob mice (Fig. 5D). Similarly, fatty acid synthase protein, which was increased in the livers of ob/ob mice, was decreased by knockdown of the insulin receptor in mice of both genotypes (Fig. 5D).

Knockdown of the insulin receptor decreased both Pcsk9 mRNA (Fig. 5E) and plasma PCSK9 in ob/ob mice (Fig. 5F), indicating that insulin promotes the expression of Pcsk9 in ob/ob mice. However, knockdown of the insulin receptor did not significantly alter Pcsk9 in lean mice. Knockdown of the insulin receptor also decreased Ldlr and Idol mRNA levels in ob/ob but not lean mice (Fig. 5G, 5H); nonetheless, LDLR protein levels were not markedly different between groups (Fig. 5D).

Discussion

PCSK9 has emerged as an important regulator of the LDL receptor and a novel therapeutic target. Here, we show that insulin promotes the degradation of the LDL receptor in vitro in a PCSK9-dependent manner. However, in vivo, decreased levels of PCSK9 in insulin deficient states are generally not associated with an increase in LDLR protein; indeed, in LIRKO mice re-fed a carbohydrate diet for 6-12 hours, or LIRKO mice fed a Paigen diet, LDLR protein is decreased. These data point to the fact that insulin regulation of LDLR is complex, and suggest that in vivo, insulin may act through PCSK9-independent mechanisms to increase LDLR protein expression.

Our data in vivo support the notion that insulin can directly induce PCSK9 expression. Thus, LIRKO mice, ob/ob mice treated with IR ASO, mice treated with streptozotocin, and fasted mice all show decreased levels of Pcsk9. However, our data also point to the fact that insulin is not always a major regulator of PCSK9 in vivo. First, LIRKO mice, despite their inability to respond to insulin, still show an 80% reduction of PCSK9 upon fasting. Consequently, in the fasted state, the effects of insulin receptor knockout on Pcsk9 expression are abolished. Second, in ob/ob mice, insulin clearly promotes the expression of PCSK9, as knockdown of the insulin receptor in these mice decreases PCSK9 expression. However, given that ob/ob mice are markedly hyperinsulinemic relative to their lean controls, it might be expected that PCSK9 levels would be supranormal. That they are not suggests that the effects of hyperinsulinemia are balanced by some other factor in the diabetic state which suppresses PCSK9. Finally, and perhaps most importantly, the acute knockdown of the insulin receptor in lean, wildtype mice by antisense oligonucleotides does not significantly decrease PCSK9 levels. In addition, adenoviral-mediated delivery of Cre recombinase into mice carrying a floxed allele of the insulin receptor significantly decreased PCSK9 expression in some cohorts, but not others (data not shown). Why knockdown of the insulin receptor in wildtype mice has little effect on Pcsk9 while LIRKO mice show markedly reduced Pcsk9 is not clear. However, LIRKO mice have chronic insulin resistance, starting in the perinatal period, which leads to peripheral insulin resistance²⁸. This peripheral insulin resistance may be necessary to unmask the effects of hepatic insulin signaling on the liver, perhaps by altering glucagon signaling or whole body cholesterol homeostasis.

Our data showing that insulin promotes PCSK9 expression is also consistent with prior reports showing decreased Pcsk9 by fasting^{21, 22} and increased Pcsk9 in mice subjected to a hyperinsulinemic euglycemic clamp^{10, 12}. However, studies by Ai and colleagues reported a two-fold increase in Pcsk9 mRNA upon knockdown of the insulin receptor in mice using an adenovirus encoding a shRNA ¹¹. One important difference in the experimental design of this study is that Ai and colleagues studied mice that had been fasted for five hours prior to sacrifice. As shown in Fig. 4B, a 24-hour fast abolishes the effects of insulin receptor knockout on Pcsk9. Indeed, even a fast of only five hours is sufficient to reduce Pcsk9 levels by 60% in wildtype mice (Figure IV in the online-only Data Supplement). This is not surprising given the facts that nuclear levels of SREBP-1 and SREBP-2 are markedly decreased by six hours of fasting ²⁹ and that mice lacking both SREBP-1 and SREBP-2, due to knockout of SREBP cleavage-activating protein (Scap), show markedly reduced Pcsk9 levels³⁰. Thus, it is likely that in the studies by Ai and colleagues, fasting abrogated the effects of insulin, unmasking the effects of other factors on Pcsk9.

One such factor could be glucagon. Despite its hyperglycemic effects, glucagon has beneficial effects on the LDL receptor. Interestingly, it was shown 20 years ago that glucagon increases LDLR protein, but not mRNA levels in rats³¹. Our data show that glucagon suppresses Pcsk9 expression in hepatocytes (Figure IIIA, IIIC in the online-only Data Supplement). Thus, fasting suppresses PCSK9 in two ways: it lowers insulin and raises glucagon. The increase in glucagon could account for the ability of fasting to decrease Pcsk9 mRNA even in the livers of LIRKO mice, which are unable to respond to insulin. Similarly, increased glucagon levels in ob/ob mice³² could potentially mitigate the effects of hyperinsulinemia on Pcsk9 in the liver.

Studies in humans show that PCSK9 levels are normal or increased in obesity/Type 2 diabetes^{8, 33}. As diabetic patients are at high risk for cardiovascular disease, they are important candidates for PCSK9 therapy. Our data indicate that though insulin promotes the expression of PCSK9, other factors may play a dominant role. Defining these other factors and understanding how they interact with insulin in the control of PCSK9, the LDL receptor and cardiovascular disease risk will be important to developing better treatments for diabetic patients.

Significance.

PCSK9 is a promising therapeutic target. Here, we show that insulin induces PCSK9, though other hormones and factors are clearly involved. Dissecting the endogenous regulators of PCSK9 is an important step towards the rational utilization of PCSK9 inhibitors in patients with Type 1 diabetes, Type 2 diabetes, and obesity.

Nonstandard Abbreviations

ASO

antisense oligonucleotides

FASN

fatty acid synthase

HNF1

hepatocyte nuclear factor 1

IDOL

inducible degrader of the LDLR

LDL

low density lipoprotein

LDLR

low density lipoprotein receptor

LIRKO

liver insulin receptor knockout

PCK1

phosphoenolpyruvate carboxykinase

PCSK9

proprotein convertase subtilisin/kexin type 9

SCD1

stearoyl CoA desaturase 1

SREBP

sterol regulatory element binding protein

STZ

streptozotocin