Proprotein convertase subtilisin/kexin type 9 (PCSK9): Hepatocyte-specific low-density lipoprotein receptor degradation and critical role in mouse liver regeneration


  • Ahmed Zaid,

    1. Laboratorie of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, affiliated to the University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Anna Roubtsova,

    1. Laboratorie of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, affiliated to the University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Rachid Essalmani,

    1. Laboratorie of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, affiliated to the University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Jadwiga Marcinkiewicz,

    1. Laboratorie of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, affiliated to the University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Ann Chamberland,

    1. Laboratorie of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, affiliated to the University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Josée Hamelin,

    1. Laboratorie of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, affiliated to the University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Michel Tremblay,

    1. Laboratorie of Hyperlipidemia and Atherosclerosis, Clinical Research Institute of Montreal, affiliated to the University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Hélène Jacques,

    1. Laboratorie of Hyperlipidemia and Atherosclerosis, Clinical Research Institute of Montreal, affiliated to the University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Weijun Jin,

    1. Department of Anatomy and Cell Biology, SUNY Downstate Medical Center, Brooklyn, NY
    Search for more papers by this author
  • Jean Davignon,

    1. Laboratorie of Hyperlipidemia and Atherosclerosis, Clinical Research Institute of Montreal, affiliated to the University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Nabil G. Seidah,

    1. Laboratorie of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, affiliated to the University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Annik Prat

    Corresponding author
    1. Laboratorie of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, affiliated to the University of Montreal, Montreal, Quebec, Canada
    • Laboratory of Biochemical Neuroendocrinology, IRCM, 110 Pine Avenue West, Montreal, Quebec H2W 1R7, Canada
    Search for more papers by this author
    • fax: 514-987-5542

  • Potential conflict of interest: Nothing to report.


The gene encoding the proprotein convertase subtilisin/kexin type 9 (PCSK9) is linked to familial hypercholesterolemia, as are those of the low-density lipoprotein receptor (LDLR) and apolipoprotein B. PCSK9 enhances LDLR degradation, resulting in low-density lipoprotein accumulation in plasma. To analyze the role of hepatic PCSK9, total and hepatocyte-specific knockout mice were generated. They exhibit 42% and 27% less circulating cholesterol, respectively, showing that liver PCSK9 was responsible for two thirds of the phenotype. We also demonstrated that, in liver, PCSK9 is exclusively expressed in hepatocytes, representing the main source of circulating PCSK9. The data suggest that local but not circulating PCSK9 regulates cholesterol levels. Although transgenic mice overexpressing high levels of liver and circulating PCSK9 led to the almost complete disappearance of the hepatic LDLR, they did not recapitulate the plasma cholesterol levels observed in LDLR-deficient mice. Single LDLR or double LDLR/PCSK9 knockout mice exhibited similar cholesterol profiles, indicating that PCSK9 regulates cholesterol homeostasis exclusively through the LDLR. Finally, the regenerating liver of PCSK9-deficient mice exhibited necrotic lesions, which were prevented by a high-cholesterol diet. However, lipid accumulation in hepatocytes of these mice was markedly reduced under both chow and high-cholesterol diets, revealing that PCSK9 deficiency confers resistance to liver steatosis. Conclusion: Although PCSK9 is a target for controlling hypercholesterolemia, our data indicate that upon hepatic damage, patients lacking PCSK9 could be at risk. (HEPATOLOGY 2008;48:646–554.)

Proprotein convertase subtilisin/kexin type 9 (PCSK9)1 is the ninth member of the proprotein convertase family.2 The first seven members, including furin, cleave protein precursors of hormones, growth factors, receptors, or surface glycoproteins at basic sites (after Arg or Lys residues). The eighth member, SKI-13 or S1P,4 is known to cleave membrane-bound transcription factors such as the SREBPs5 in their luminal domains, resulting in the release of their DNA-binding domain. Proprotein convertases can also inactivate secreted substrates, such as endothelial lipase6 and PCSK9.7

PCSK9 is synthesized as a precursor that undergoes autocatalytic cleavage of its N-terminal prosegment in the ER,1 a step required for its exit from this compartment and its efficient secretion. Secreted PCSK9 remains associated with its prosegment.1 Different from the other proprotein convertases, this serine protease has no known substrate other than itself. In addition, the tight association of the prosegment with the active site8 raises the question of the existence of an in trans PCSK9 protease activity. PCSK9 binds the EGF-A domain of the low-density lipoprotein receptor (LDLR) through its catalytic domain9, 10 and favors the targeting of the LDLR to endosomes/lysosomes and its degradation.11, 12 In addition, binding of PCSK9 to the cell surface seems to implicate its C-terminal Cys/His-rich domain.13 The liver and small intestine are the richest source of PCSK9, in which it colocalizes with LDLR.1, 14

It was shown that human mutations affecting the level of PCSK9 and/or its activity toward the LDLR resulted in either hypercholesterolemia15, 16 or hypocholesterolemia.17 Accordingly, Pcsk9 knockout led to hypocholesterolemia in mice, with high levels of hepatic LDLR protein and low levels of circulating low-density lipoprotein (LDL) cholesterol.18 These data and the identification of two healthy women lacking functional PCSK919, 20 indicate that inhibitors of the PCSK9-mediated degradation of LDLR would be valuable tools to control plasma cholesterol levels. In this study, we aimed at dissecting the liver-specific role of PCSK9 and its contribution to liver regeneration. Accordingly, we generated and analyzed mice completely deficient in PCSK9 or lacking it specifically in the liver, as well as liver-specific transgenic mice. Our data showed that hepatic PCSK9 contributes to two thirds of the increase in circulating cholesterol resulting from total mouse PCSK9 degrading activity on LDLR. Because PCSK9 was highly expressed in the fetal liver and up-regulated during liver regeneration,1 we also examined the impact of its absence in this process. Our data suggest that the major role of PCSK9 during liver regeneration resides in its ability to regulate cholesterol levels.


HDL, high-density lipoprotein; LDL, low-density lipoprotein; LDLR, low-density lipoprotein receptor; %Li/bw, percentage of liver to body weight; mRNA, messenger RNA; PCNA, proliferating cell nuclear antigen; PCSK9, proprotein convertase subtilisin/kexin type 9; PHx, partial hepatectomy; QPCR, quantitative polymerase chain reaction; TC, total plasma cholesterol.

Material and Methods

For a description of the experimental procedures, see the Supplementary Material.


Tissue Expression of PCSK9.

Quantitative polymerase chain reaction (QPCR) analysis of PCSK9 distribution in adult mouse tissues (Fig. 1) showed that liver, the major site of expression, was followed by the colon, ileum, and duodenum expressing approximately 7-, 9-, and 20-fold less messenger RNA (mRNA), respectively. The kidney, reproductive organs (epididymis, uterus, ovary, and testis), other digestive tissues (jejunum and stomach), the cerebellum, and adrenal gland express much lower levels. PCSK9 transcripts were not detectable in the pancreas, olfactory bulb, muscle, or lung. The remarkably high levels of PCSK9 expression in liver prompted us to better define the in vivo role of PCSK9 in this tissue.

Figure 1.

QPCR analysis of PCSK9 expression in mouse tissues normalized to S16 mRNA level.

Complete and Hepatocyte-Specific Knockout of PCSK9.

To inactivate Pcsk9 specifically in the liver, we targeted its proximal promoter and exon 1 by framing them with loxP sites, efficiently recombined in vivo by the Cre recombinase (Fig. 2A). Complete knockouts (−/−) were obtained using CMV-cre transgenic mice that express Cre in germ cells. Northern blot analysis (Fig. 2C) did not reveal any residual RNA expression in the liver (Fig. 2B), nor did QPCR analyses in the liver and other tissues (data not shown). For hepatocyte-specific knockout, f/f mice carrying the Alb-cre transgene that expresses Cre under the control of the albumin promoter were obtained. In the absence of Cre, only floxed alleles were detected. In its presence, Δ1 were the major alleles in liver, but was not detected in other tissues (Fig. 2D). Furthermore, no PCSK9 transcripts were detected in the liver of f/f Alb-cre mice via northern blotting (Fig. 2E), QPCR and in situ hybridization (data not shown). Thus, flox allele recombination in hepatocytes was very efficient and abolished PCSK9 expression in the liver, demonstrating that PCSK9 is exclusively expressed in hepatocytes. The remaining flox alleles detected via PCR on genomic DNA originated from liver cell types other than hepatocytes, in which Pcsk9 is not transcribed.

Figure 2.

Conditional knockout of Pcsk9. (A) Pcsk9 proximal promoter and exon 1 were framed with loxP sites. The targeting vector also exhibited a neomycin resistance cassette framed with frt sites excisable by flipase. Embryonic stem cells were transfected with the purified ClaI insert, which generated a Pcsk9neo allele by homologous recombination, and used to produce Pcsk9neo/+ mice. Removal of the neo cassette resulted in Pcsk9flox/+ mice. Cre-mediated recombination led to Pcsk9Δ1/+ mice harboring a null allele. Black bars symbolize 5′ and 3′ probes, and half arrows indicate the P1 to P5 primers used for genotyping. (B) NcoI-digested tail DNA was analyzed via Southern blotting with an exon 2 probe. (C,E) PCSK9 transcripts were detected using a probe covering exons 4 and 5. (D) The liver-specific Cre recombination of flox into Δ1 alleles was evidenced by PCR genotyping of Pcsk9flox/floxmice missing (f/f) or exhibiting (f/f Alb-cre) the Alb-cre transgene.

Overexpression of PCSK9 in the Liver.

We also generated transgenic mice overexpressing wild-type (TgWT) or catalytically inactive (H229A; TgHA) mouse PCSK9 in the liver carrying a C-terminal V5 tag. Selected low (TgWTlow) and high (TgWThigh; TgHA) expressing lines showed an approximately 2.5- or 40-fold increase in mRNA, respectively, compared with endogenous levels (QPCR data not shown).

Analysis of Liver LDLR and Plasma PCSK9.

Whereas +/+ and f/f mice exhibited similar liver LDLR levels, −/− and liver-specific knockout (f/f Alb-cre) mice showed 2- to 3-fold higher LDLR levels (Fig. 3A). In contrast, overexpression of PCSK9 in the liver (TgWThigh) led to the quasi-disappearance of the LDLR signal, close to the Ldlr−/− background. The H229A mutant, expected to be retained in the ER,1 had no significant effect (TgHA), and only proPCSK9 could be detected. For TgWThigh, the dominant form was mature PCSK9. Also detected was its furin-cleaved (ΔN218) form.7

Figure 3.

Analysis of the liver and plasma of knockout and transgenic mice. (A) Western blot analysis of liver extracts (4 mice per genotype) using LDLR (top panels), V5 (middle right panel) and α-actin (lower panels) antibodies. (B) Immunoprecipitation and western blot analysis of plasma PCSK9. Endogenous (arrowheads) and V5-tagged transgenic (arrows) forms of PCSK9 are indicated. (C) LDLR immunohistochemistry (green). Nuclei were stained with TO-PRO-3 (blue).

Endogenous mature PCSK9 and its furin-cleaved form7, 21 were detectable in +/+ and f/f plasma, whereas neither form was observed in −/− or f/f Alb-cre plasma (Fig. 3B). In the mouse, the liver is thus the major source of circulating PCSK9. The plasma of TgWTlow mice exhibited two additional bands corresponding to the V5-tagged forms. Immunohistochemistry confirmed the above data and demonstrated that PCSK9 deficiency led to higher levels of cell surface LDLR in hepatocytes (Fig. 3C). Note that LDLR changes were more dramatic than in Fig. 3A, likely reflecting cell surface versus total cellular receptor levels. Finally, overexpression of PCSK9 (TgWThigh), but not of PCSK9-HA, led to an Ldlr−/−-like phenotype.

Cholesterol Analyses.

PCSK9 total and liver-specific knockouts resulted in a 42% and 27% drop in total plasma cholesterol (TC), respectively, suggesting that liver PCSK9 contributes to approximately two thirds of the observed hypocholesterolemia phenotype (Table 1). In contrast, TgWThigh mice exhibited a 57% increase in TC, whereas minor or no significant changes were detected in TgWTlow and TgHA mice. Pcsk9+/+ or Pcsk9−/− mice on an Ldlr−/− background were also obtained. The TC of Ldlr−/− mice was approximately 2.5-fold higher than that of wild-type mice (8.48 versus 3.39 mM) but was not different from that of the double Ldlr−/−Pcsk9−/− knockout mice, indicating that PCSK9 deficiency had no additive effect. Fast protein liquid chromatography analysis of plasma pools revealed that LDL and high-density lipoprotein (HDL) cholesterol in Pcsk9−/− mice were approximately 20% and 62% that of wild-type mice, respectively (Fig. 4A). In f/f Alb-cre mice, they were approximately 40% and 70% that of control f/f mice (Fig. 4B). The difference between the −/− and f/f Alb-cre profiles (Fig. 4A,B) is a measure of the impact of the PCSK9 deficiency in extrahepatic tissues on LDLR. Because LDLR binds the apolipoproteins B and E, it is not surprising that PCSK9 affected both LDL and HDL cholesterol. PCSK9 overexpression resulted in an approximately 5.2- and 1.3-fold increase in LDL and HDL cholesterol, respectively (Fig. 4C), generating a profile different from the Ldlr−/− one characterized by a higher LDL cholesterol (≈14.8-fold) but no change in HDL cholesterol. Finally, no difference was observed in the cholesterol profiles of single Ldlr−/− and double Ldlr−/−Pcsk9−/− mice (Fig. 4D), confirming that PCSK9 does not regulate plasma cholesterol independently from the LDLR.

Table 1. Total Plasma Cholesterol of the Various Mouse Strains Used in This Study
Mouse StrainTC (mM)Percent ChangeStudent t Test
  • *

    Significant ratios to control values.

Pcsk9+/+3.39 ± 0.62  
Pcsk9−/−1.97 ± 0.47−42*P = 0.0003*
Pcsk9f/f3.44 ± 0.22  
Pcsk9f/fAlb-cre2.51 ± 0.59−27*P = 0.0005*
No Tg3.55 ± 0.62  
TgHA4.12 ± 0.62+16P = 0.05
TgWTlow4.09 ± 0.88+15P = 0.4
TgWThigh5.57 ± 0.98+57*P = 0.00003*
Ldlr−/− (C57BL/6)8.48 ± 0.79  
Ldlr−/−Pcsk9+/+8.42 ± 0.92−1P = 0.4
Ldlr−/−Pcsk9−/−8.33 ± 1.40−2P = 0.4
Figure 4.

Plasma cholesterol fast protein liquid chromatography profiles. Plasma pools (500 μL) from 3 to 6 mice: (A) Pcsk9+/+ and Pcsk9−/−, (B) f/f and f/f Alb-cre, (C) control (No Tg), TgWThigh and Ldlr−/−, and (D) Pcsk9+/+ and Pcsk9−/− in a Ldlr−/− background. Cholesterol levels associated with LDL (fractions 11-24 or 15-24) or HDL (fractions 26-38) were compared in control (black symbols) and mutant (open symbols) mice.

Partial Hepatectomy.

Because PCSK9 favors neuronal differentiation in telencephalon primary cultures1 and is highly expressed in the developing liver and up-regulated in the regenerating rat liver,1 we examined whether PCSK9 was needed for proper liver regeneration, independently or not of its effect on cholesterol availability. This was especially of interest in view of its proposed inhibition in the future treatment of hypercholesterolemia.2 We thus performed partial hepatectomy (PHx) or sham operations on Pcsk9−/− and Pcsk9+/+ mice. In the latter, PCSK9 mRNA peaked at day 3 (Fig. 5A), instead of day 2 in the rat,1 but did not vary in sham operated mice (data not shown). The impact of the presence or absence of PCSK9 on liver regeneration was thus analyzed at 40 hours and 72 hours post-PHx.

Figure 5.

PHx in mice fed a chow or high-cholesterol diet. (A,C) PCSK9 and HMG-CoA reductase mRNA levels were measured by QPCR at various times (A) or 72 hours (C) postsurgery. (B) TC, %Liv/bw, and occurrence of necrosis are given for sham-operated or PHx mice, fed a chow or a high-cholesterol diet, and analyzed at 40 hours, 72 hours, and/or 2 weeks post-surgery. *Significantly different averages between Pcsk9−/− and corresponding Pcsk9+/+ mice (Student t test, P < 0.05).

As in untreated mice, TC was approximately 50% lower in Pcsk9−/− than in Pcsk9+/+ mice for sham-operated and PHx mice (Table 1), indicating that PHx per se had no effect on TC (Fig. 5B1,B2). QPCR analysis of HMG-CoA reductase mRNA levels, indicative of cellular cholesterol synthesis, revealed that at 72 hours hepatectomized Pcsk9−/− mice exhibited an approximately 2.5-fold up-regulation, but not the Pcsk9−/− sham-operated littermates (Fig. 5C, chow), which paradoxically have comparable TC levels (1.60 versus 1.66 mM; Fig. 5B2). This suggests that PHx profoundly depleted intracellular cholesterol, likely due to a higher cholesterol requirement in the regenerating liver.

One of 6 and 2 of 9 Pcsk9−/− mice analyzed 40 hours and 72 hours post-PHx, respectively, were very weak and sacrificed prematurely. The 5 and 7 remaining Pcsk9−/− mice looked weaker during the first 24 hours compared with sham-operated and operated Pcsk9+/+ mice. Sham operation did not significantly affect the percentage of liver to body weight (%Li/bw), which was significantly reduced 40 hours post-PHx (1.5- to 2-fold). At 72 hours, while Pcsk9+/+ mice recovered their original %Li/bw, Pcsk9−/− mice exhibited a significantly lower value (3.7% versus 4.8%; P = 0.012) (Fig. 5B2), indicating that regeneration was compromised. Although sham-operated and hepatectomized Pcsk9+/+ mice exhibited a normal liver morphology (Fig. 6A,C), 6 of 7 Pcsk9−/− mice showed necrotic foci located mostly at the periphery of the overgrown 4 remaining lobes (Fig. 6D,E). Indeed, after resection of the three larger lobes corresponding to two thirds of the liver mass, the 4 remaining lobes increased in volume (≈2- to 3-fold). In these foci, the liver architecture was disrupted with swollen hepatocytes undergoing ballooning degeneration. Infiltration of red blood cells and leukocytes, which are indicative of inflammation, was also observed at the border of the necrotic areas. The surrounding hepatocytes appeared hypertrophic. The lesion numbers and sizes varied from mouse to mouse, never exceeding 10% of the section. The observed lesions were due to the absence of PCSK9 and not to viral or bacterial infections, because PCR tests were negative for 6 species of Helicobacter, including Helicobacter hepaticus, the only pathogen detected in our mouse hepatitis virus-free SPF facility. At 72 hours post-PHx, 7 heterozygotes with intermediate TC values did not show any lesions (data not shown).

Figure 6.

PCSK9 deficiency leads to lesions in the regenerating liver. Liver sections from wild-type and knockout mice were stained with hematoxylin-eosin. (A,B) Sham-operated or (C-H) PHx livers were analyzed (C-E) 72 hours or (F-H) 2 weeks postsurgery. The framed areas (D,G) were enlarged 9 times (E,H). (I,J) Liver sections of mice fed a high-cholesterol diet.

Proliferating cell nuclear antigen (PCNA), whose accumulation correlates with DNA replication, was analyzed via immunohistochemistry (Fig. 7C-F). In Pcsk9+/+ mice, PHx induced PCNA expression in 70% and 59% of hepatocytes at 40 hours and 72 hours, respectively, whereas sham surgery did not (Fig. 7M). In Pcsk9−/− mice, PCNA induction was delayed and only apparent (77%) at 72 hours post-PHx. Evidence for delayed proliferation was also provided on western blot analysis of PCNA, cyclin E, and Cdk2 (Fig. 7N). Immunodetection of bromodeoxyuridine, injected 2 hours before sacrifice at 40 hours, confirmed that DNA replication was not significantly initiated in Pcsk9−/− mice at this time point (0.2% of labeled hepatocytes versus 18.8% for Pcsk9+/+ mice (Fig. 7M). Furthermore, at 72 hours post-PHx. Apoptosis (terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling) (Fig. 7G,H) was only observed in Pcsk9−/− livers.

Figure 7.

Analysis of Pcsk9+/+ and Pcsk9−/− livers 40 hours and 72 hours postsurgery. (A-H) Bromodeoxyuridine, PCNA analysis, and terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling at 40 hours and/or 72 hours post-PHx. (I-L) Oil red O-stained liver sections of mice fed chow and a high-cholesterol diet. (M) Percentage of bromodeoxyuridine or PCNA-positive cells (2282 to 3603 hepatocytes were counted from 4 to 5 mice per condition, with 3 random fields per mouse). Error bars (standard deviation) illustrate the interanimal variability (averages of 3 fields per mouse). Asterisks indicate significantly different averages between *Pcsk9−/− and **Pcsk9+/+ mice (Student t test, P < 0.0001). (N) Representative western blotting of PCNA, cyclin E, and Cdk2 in corresponding pooled liver extracts (4 to 5 mice per pool) and normalization to actin levels, with the 40 hours sham +/+ value fixed to 1.

To assess the evolution of the necrotic lesions with time, we analyzed the morphology of Pcsk9−/− livers 2 weeks post-PHx (Fig. 6F,G). Although 1 of 7 hepatectomized Pcsk9−/− mice died 48 hours later, the %Li/bw of Pcsk9+/+ and Pcsk9−/− mice was similar (5.4% and 4.6%) (Fig. 5B3), indicating that after 14 days, there was a full mass recovery in Pcsk9−/− mice. However, Pcsk9−/− livers still exhibited cavities. The latter were devoid of red blood cells and infiltrating leukocytes (Fig. 6G,H), suggesting a healing process. Whether these cavities can be further resorbed remains to be elucidated. Thus, PCSK9 plays a critical role in the first days of liver regeneration.

Because hypocholesterolemia is the only known phenotype associated with PCSK9 deficiency, we tested whether increasing circulating cholesterol levels affected the development of necrotic lesions. Pcsk9+/+ and Pcsk9−/− mice were fed a high-cholesterol diet (0.5% cholesterol) for 1 week before and for 72 hours after PHx (Fig. 5B4). Whereas the TC of sham-operated Pcsk9+/+ mice was not significantly changed, it increased 2.6-fold (4.20 versus 1.60 mM; P = 0.02) in sham-operated Pcsk9−/− mice (Fig. 5B2,B4). Following PHx, TC increased in both Pcsk9+/+ and Pcsk9−/− mice by 2.5-fold (7.28 versus 2.90 mM; P = 0.000005) and 3-fold (4.90 versus 1.66 mM; P = 0.0001), respectively, while their HMG-CoA reductase mRNA levels dropped by approximately 13-fold (Fig. 5C). All five Pcsk9−/− mice survived following PHx and behaved as Pcsk9+/+ mice in the first 24 hours. The high-cholesterol diet significantly compromised liver regeneration in Pcsk9+/+ mice (3.6 ± 0.4 versus 4.8 ± 0.7; P = 0.01). In contrast, the %Li/bw of Pcsk9−/− mice fed a high-cholesterol diet was slightly, though not significantly, higher than that obtained on chow diet (4.1 ± 0.4 versus 3.7 ± 0.5). Importantly, Pcsk9−/− liver sections no longer revealed any necrotic lesions (Fig. 5B4). Histological analyses revealed an accumulation of fat droplets, appearing as vacuoles, especially in Pcsk9+/+ livers (Fig. 6I,J). Staining with oil red O confirmed the above data and emphasized the higher density and larger size of fat droplets in Pcsk9+/+ than in Pcsk9−/− livers (Fig. 7I-L). Increased steatosis, which was reported to hamper liver regeneration,22 is the most likely explanation for the compromised regeneration observed in Pcsk9+/+ livers that exhibited a 3.6- and 5.4-fold increase in liver cholesterol and TG levels, respectively (data not shown).


Analysis of conditional knockout mice lacking PCSK9 specifically in hepatocytes revealed that: (1) hepatocytes are the only site of PCSK9 synthesis in the liver and are the source of circulating PCSK9, suggesting that the intestine, the main other site of PCSK9 synthesis (Fig. 1), does not significantly contribute to plasma PCSK9; and (2) the absence of hepatic PCSK9 contributes to two thirds of the hypocholesterolemia phenotype observed in Pcsk9−/− mice. According to reported cholesterol clearance activities of mouse tissues,23 the liver and the small intestine account for approximately 70% and 25% of clearance activity, respectively. Together with our data, these observations suggest that the absence of PCSK9 in the liver, and thereby in the plasma, had a minor or no impact on extrahepatic LDLR activity. Rather, the latter would be regulated by local PCSK9.

Mice expressing approximately 2.5-fold the endogenous level of PCSK9 protein (TgWTlow) (Fig. 3B) exhibited only a minor TC change (+15%; Table 1). In contrast, heterozygous TgWThigh mice overexpressing approximately 20-fold the endogenous level of the PCSK9 protein (data not shown) showed 57% higher TC levels with an approximately 5-fold increase in LDL cholesterol. Intriguingly, although LDLR was quasi-absent in TgWThigh liver extracts (Fig. 3), this increase represented only 35% of the approximately 15-fold increase seen in Ldlr−/− mice (Fig. 4C), indicating that high circulating levels of PCSK9 could not fully reproduce the LDLR deficiency phenotype. Lagace et al. obtained comparable results.21 Extrahepatic LDLR clearance activity is likely to be responsible for this discrepancy. Moreover, circulating PCSK9 was shown to be active on liver LDLR of recipient wild-type mice only at very high concentrations, obtained by parabiosis21 or injection of the purified protein.24 This suggests that, at physiological concentrations, wild-type PCSK9 acts locally.

The PCSK9 deficiency in an Ldlr−/− background did not further modify plasma cholesterol levels (Table 1 and Fig. 4D). This indicates that PCSK9 (1) acts on circulating cholesterol exclusively via its effect on the LDLR and (2) does not directly regulate apolipoprotein B synthesis and/or release in plasma. We, however, observed that the absence or overexpression of PCSK9, but not the absence of LDLR (Fig. 4C),21 modulated HDL cholesterol levels (−38% and +28% in Pcsk9−/− and TgWThigh mice, respectively) (Fig. 4A,C).

The high level of PCSK9 in the liver and its up-regulation during liver regeneration1 prompted us to assess if PCSK9 played a role in this process through LDLR and/or other targets. We thus compared wild-type and total knockout regenerating livers. After PHx, (1) all wild-type and heterozygous mice were healthy, whereas 4 of 22 knockout mice died, and the rest were weak (Fig. 5B1-B3); (2) liver regeneration was delayed in Pcsk9−/− mice (Fig. 7); (3) the regenerating knockout livers were approximately 25% smaller at 72 hours post-PHx and exhibited multiple necrotic lesions, still visible 2 weeks later (Figs. 5 and 6); (4) the lesions were prevented by a high-cholesterol diet (Figs. 5-7); and (5) preliminary data indicated that necrotic lesions were also absent in the hypercholesterolemic single LDLR and double LDLR/PCSK9 knockout mice, as well as in LDLR-deficient TgWThigh mice (data not shown). Whether necrotic lesions arise from severe hypocholesterolemia, not observed in the latter mice, or from the presence of an excess of cell surface LDLR in Pcsk9−/− mice, is still not clear. Nevertheless, low circulating cholesterol is a critical factor. Indeed, the levels of HMG-CoA reductase mRNA, encoding the rate-limiting enzyme in cholesterol synthesis, were further increased in partially hepatectomized knockout mice, but not in wild-type ones (Fig. 5C), indicating that, in knockout mice, cholesterol dropped below a critical threshold following PHx.

Surprisingly, although exhibiting higher LDLR concentrations, Pcsk9−/− regenerating livers showed a higher rate of cellular cholesterol synthesis (Fig. 5C), and poor accumulation of intracellular lipids. Resistance to steatosis may be due to a more efficient metabolism of cellular cholesterol, possibly through elimination in bile. Preliminary data indicated that this phenomenon is LDLR-independent. Indeed, in an Ldlr−/− background, Pcsk9−/− regenerating livers still exhibited resistance to steatosis, while TgWThigh ones showed enhanced steatosis (data not shown). Nevertheless, the absence of lesions at 72 hours post-PHx in cholesterol-fed Pcsk9−/− mice reflects the critical importance of maintaining a minimum level of cholesterol to prevent the formation of necrotic foci. In this context, complications or death were reported in post-hepatectomized patients that exhibited a higher drop in plasma cholesterol in the first 3 days after surgery.25 At this stage, we cannot exclude a cholesterol-independent role of PCSK9 in liver regeneration. However, to date, PCSK9 is only known to trigger the degradation of LDLR in the liver, although it can also target ex vivo two other LDLR family members, VLDL receptor and apoER2,26 but not LRP1.12, 27

PCSK9 is presently considered as a likely safe target for cholesterol-lowering therapy.28 The present study demonstrates that, unlike Pcsk9−/− mice, Pcsk9+/− mice did not show any impairment in liver regeneration, suggesting that a 50% loss of PCSK9 is not deleterious. Two women lacking functional PCSK9 did not exhibit any complications or overt phenotypes other than hypocholesterolemia.19 Liver regeneration is not impaired in donor hepatectomy subjects even if their liver has steatosis.20, 29 The present results, however, should raise concern for liver donors who have hypocholesterolemia, because it could be secondary to a PCSK9 loss of function mutation. This raises the possibility that liver regeneration might be impaired in such cases and that a cholesterol-rich diet could avoid adverse consequences. This may apply to other conditions where liver regeneration is an issue (for example, drug-induced or CCl4 liver necrosis and alcoholic cirrhosis). It will be of the utmost importance in the future to determine the role of PCSK9 in normal liver physiology and in conditions of liver injury. Liver integrity could be PCSK9-dependent, and PCSK9-deficient patients could be at risk upon hepatic damage.


We acknowledge the secretarial help of Brigitte Mary. We thank Claudia Toulouse for excellent animal care and Dominic Fillion and Qinzhang Zhu for technical assistance.