Advertisement
Canadian Journal of Cardiology

Whole-Gene Duplication of PCSK9 as a Novel Genetic Mechanism for Severe Familial Hypercholesterolemia

Open AccessPublished:August 03, 2018DOI:https://doi.org/10.1016/j.cjca.2018.07.479

      Abstract

      Background

      Familial hypercholesterolemia (FH) is a common genetic disorder of severely elevated low-density lipoprotein (LDL) cholesterol, characterized by premature atherosclerotic cardiovascular disease. Although copy number variations (CNVs) are a large-scale mutation-type capable of explaining FH cases, they have been, to date, assessed only in the LDLR gene. Here, we performed novel CNV screening in additional FH-associated genes using a next-generation sequencing–based approach.

      Methods

      In 704 patients with FH, we sequenced FH-associated genes APOB, PCSK9, LDLRAP1, APOE, STAP1, LIPA, and ABCG5/8 using our LipidSeq targeted next-generation sequencing panel. Bioinformatic tools were applied to LipidSeq data for CNV screening, and identified CNVs were validated using whole-exome sequencing and microarray-based copy number analyses.

      Results

      We identified a whole-gene duplication of PCSK9 in 2 unrelated Canadian FH index cases; this PCSK9 CNV was also found to cosegregate with affected status in family members. Features in affected individuals included severely elevated LDL cholesterol levels that were refractory to intensive statin therapy, pronounced clinical stigmata, premature cardiovascular events, and a plasma PCSK9 of approximately 5000 ng/mL in 1 index case. We found no CNVs in APOB, LDLRAP1, APOE, STAP1, LIPA, and ABCG5/8 in our cohort of 704 FH individuals.

      Conclusions

      Here, we report the first description of a CNV affecting the PCSK9 gene in FH. This finding is associated with a profound FH phenotype and the highest known plasma PCSK9 level reported in a human. This finding also has therapeutic relevance, as elevated PCSK9 levels may limit the efficacy of high-dose statin therapy and also PCSK9 inhibition.

      Résumé

      Contexte

      L’hypercholestérolémie familiale (HF) est un trouble génétique courant caractérisé par une élévation marquée du taux de cholestérol LDL, qui se traduit par une atteinte cardiovasculaire athéroscléreuse prématurée. Bien que les variations du nombre de copies (VNC) constituent un type de mutation à grande échelle pouvant expliquer les cas d’HF, elles n’ont été étudiées à ce jour que pour le gène du LDLR (récepteur des LDL). Nous avons procédé au repérage de nouvelles VNC d’autres gènes liés à l’HF, au moyen d’une méthode de séquençage à haut débit.

      Méthodes

      Chez 704 patients atteints d’HF, nous avons séquencé, à l’aide de notre panel de séquençage ciblé à haut débit LipidSeq, les gènes suivants liés à l’HF : APOB, PCSK9, LDLRAP1, APOE, STAP1, LIPA et ABCG5/8. Le repérage des VNC a été réalisé sur les données issues du séquençage à l’aide d’outils bio-informatiques. Les VNC ainsi mises en évidence ont été validées au moyen du séquençage de l’exome complet et d’analyses du nombre de copies utilisant des puces à ADN.

      Résultats

      Nous avons constaté une duplication du gène PCSK9 entier chez 2 proposants Canadiens non apparentés atteints d’HF; nous avons également observé une coségrégation de cette VNC du gène PCSK9 avec l’atteinte chez les membres de la famille. Les caractéristiques des sujets atteints comprenaient un taux très élevé de cholestérol LDL réfractaire à un traitement intensif par statine, des signes cliniques prononcés, des événements cardiovasculaires prématurés et une concentration plasmatique de proprotéine convertase subtilisine/kexine-9 (PCSK9) d’environ 5000 ng/ml chez un proposant. Nous n’avons trouvé aucune VNC des gènes APOB, LDLRAP1, APOE, STAP1, LIPA et ABCG5/8 dans notre cohorte de 704 sujets atteints d’HF.

      Conclusions

      Notre étude fait état du premier cas de VNC touchant le gène PCSK9 dans l’HF. Cette observation est associée à un phénotype d’HF grave et à la concentration plasmatique de PCSK9 la plus élevée qui ait été relevée à ce jour chez l’humain. Cette observation a aussi une portée thérapeutique, étant donné qu’à des taux élevés, la PCSK9 peut limiter l’efficacité d’un traitement par statine à forte dose ainsi que sa propre inhibition.
      Familial hypercholesterolemia (FH) is an inherited disorder characterized by elevated low-density lipoprotein cholesterol (LDL-C) that is estimated to affect approximately 1 in 250 individuals worldwide.
      • Akioyamen L.E.
      • Genest J.
      • Shan S.D.
      • et al.
      Estimating the prevalence of heterozygous familial hypercholesterolaemia: a systematic review and meta-analysis.
      Lifelong exposure to elevated LDL-C in patients with FH is associated with accelerated atherosclerosis and an increased risk for premature cardiovascular events.
      • Nordestgaard B.G.
      • Chapman M.J.
      • Humphries S.E.
      • et al.
      Familial hypercholesterolaemia is underdiagnosed and undertreated in the general population: guidance for clinicians to prevent coronary heart disease: consensus statement of the European Atherosclerosis Society.
      FH is heterogeneous at the molecular level;
      • Defesche J.C.
      • Gidding S.S.
      • Harada-Shiba M.
      • et al.
      Familial hypercholesterolaemia.
      most cases result from inactivating mutations in the LDLR gene but several other genes have also been implicated, including APOB and PCSK9, and less frequently LDLRAP1,
      • Garcia C.K.
      • Wilund K.
      • Arca M.
      • et al.
      Autosomal recessive hypercholesterolemia caused by mutations in a putative LDL receptor adaptor protein.
      APOE,
      • Marduel M.
      • Ouguerram K.
      • Serre V.
      • et al.
      Description of a large family with autosomal dominant hypercholesterolemia associated with the APOE p.Leu167del mutation.
      STAP1,
      • Fouchier S.W.
      • Dallinga-Thie G.M.
      • Meijers J.C.M.
      • et al.
      Mutations in STAP1 are associated with autosomal dominant hypercholesterolemia.
      LIPA,
      • Chora J.R.
      • Alves A.C.
      • Medeiros A.M.
      • et al.
      Lysosomal acid lipase deficiency: a hidden disease among cohorts of familial hypercholesterolemia?.
      and ABCG5/8.
      • Hansel B.
      • Carrié A.
      • Brun-Druc N.
      • et al.
      Premature atherosclerosis is not systematic in phytosterolemic patients: severe hypercholesterolemia as a confounding factor in five subjects.
      Within these genes, a spectrum of mutation types has been described; these are predominantly small in scale, ranging from missense, to frameshift, to splicing mutations.
      DNA copy number variations (CNVs) are genomic structural variants that include deletions and duplications larger than 50 base pairs in size.
      • Zarrei M.
      • MacDonald J.R.
      • Merico D.
      • Scherer S.W.
      A copy number variation map of the human genome.
      In FH, CNVs have to date been systematically assessed only in the LDLR gene. Within LDLR, > 80 unique CNV events have been described in patients with FH, and are responsible for approximately 10% of molecularly defined cases in many cohorts.
      • Bourbon M.
      • Alves A.C.
      • Sijbrands E.J.
      Low-density lipoprotein receptor mutational analysis in diagnosis of familial hypercholesterolemia.
      • Iacocca M.A.
      • Hegele R.A.
      Role of DNA copy number variation in dyslipidemias.
      CNVs have been traditionally evaluated using a specialized dedicated laboratory method, namely multiplex ligation-dependent probe amplification (MLPA) of LDLR.
      • Wang J.
      • Ban M.R.
      • Hegele R.A.
      Multiplex ligation-dependent probe amplification of LDLR enhances molecular diagnosis of familial hypercholesterolemia.
      We recently showed that LDLR CNVs could be detected using bioinformatic tools applied to targeted next-generation sequencing (NGS) data; we reported 100% concordance in LDLR CNV detection between NGS bioinformatics and MLPA.
      • Iacocca M.A.
      • Wang J.
      • Dron J.S.
      • et al.
      Use of next-generation sequencing to detect LDLR gene copy number variation in familial hypercholesterolemia.
      Because our targeted NGS panel evaluates other FH-associated genes,
      • Iacocca M.A.
      • Hegele R.A.
      Recent advances in genetic testing for familial hypercholesterolemia.
      • Johansen C.T.
      • Dube J.B.
      • Loyzer M.N.
      • et al.
      LipidSeq: a next-generation clinical resequencing panel for monogenic dyslipidemias.
      we next performed novel CNV screening in these genes in a cohort of 704 individuals with FH. We now report a large-scale duplication encompassing the entire PCSK9 gene in 2 FH index cases and their affected family members.

      Materials and Methods

      Study subjects

      A total of 704 patient samples were screened for CNVs in this study. All patients had at least “possible” FH according to validated clinical criteria.
      • Iacocca M.A.
      • Wang J.
      • Dron J.S.
      • et al.
      Use of next-generation sequencing to detect LDLR gene copy number variation in familial hypercholesterolemia.
      • Wang J.
      • Dron J.S.
      • Ban M.R.
      • et al.
      Polygenic versus monogenic causes of hypercholesterolemia ascertained clinically.
      This cohort included 429 samples from individuals referred to London Health Sciences Centre, University Hospital (London, ON) for the treatment of severe hypercholesterolemia, plus 275 samples sent by collaborating physicians for genetic analyses. Our protocol was approved by the Western University Research Ethics Board (No. 07920E), and all individuals provided informed consent for genetic analyses.

      Targeted NGS

      Targeted NGS was performed using our LipidSeq panel, comprising 73 lipid metabolism-related genes including all specified non-LDLR FH-associated genes, namely, APOB, PCSK9, LDLRAP1, APOE, STAP1, LIPA, and ABCG5/8. Genomic DNA was isolated from whole blood using the Puregene DNA Blood Kit (Gentra Systems; Qiagen, Mississauga, ON). Libraries were prepared using the Nextera Rapid Capture Custom Enrichment Kit (Illumina, San Diego, CA), and enriched samples were sequenced on a MiSeq personal sequencer (Illumina) using 2 × 150 bp paired-end chemistry as described.
      • Iacocca M.A.
      • Wang J.
      • Dron J.S.
      • et al.
      Use of next-generation sequencing to detect LDLR gene copy number variation in familial hypercholesterolemia.
      • Johansen C.T.
      • Dube J.B.
      • Loyzer M.N.
      • et al.
      LipidSeq: a next-generation clinical resequencing panel for monogenic dyslipidemias.
      A custom automated workflow in Genomics Workbench version 11.0.1 (CLC Bio, Aarhus, Denmark) was used for bioinformatic processing of raw sequencing data, generating .BAM (target region coverage data) and .VCF (variant calling data) files necessary for subsequent CNV analysis. Our LipidSeq method has an average depth of coverage (DOC) of 300-fold per base.

      Whole-exome NGS

      Whole-exome sequencing (WES) was performed in 2 CNV-positive index cases at the London Regional Genomics Centre (London, ON). Library preparation was performed using the TruSeq Rapid Exome Kit (Illumina), and enriched samples were sequenced on the Illumina NextSeq500 using 2 × 150 bp paired-end chemistry. Bioinformatic analysis of raw sequencing data was performed using a custom automated workflow in CLC Genomics Workbench, as described above. This WES method has an average DOC of 125-fold per base.

      NGS CNV analysis

      CNV screening of NGS data was performed using the bioinformatic tool CNV Caller, an application within the VarSeq v1.4.3 variant annotation software (Golden Helix, Bozeman, MT). VarSeq CNV Caller uses DOC analysis to identify probable CNV events, comparing normalized DOC across targeted genomic regions in the sample of interest to the mean normalized DOC in ≥ 30 matched reference controls. The criteria used to call CNVs in genes of interest have been previously described in detail.
      • Iacocca M.A.
      • Wang J.
      • Dron J.S.
      • et al.
      Use of next-generation sequencing to detect LDLR gene copy number variation in familial hypercholesterolemia.
      VarSeq CNV Caller was applied to both our targeted NGS (LipidSeq) and WES data.

      CNV confirmation by microarray analysis

      Confirmation in 2 CNV-positive index cases was performed using the CytoScan HD Array (ThermoFisher Scientific, Waltham, MA). The array has > 1.9 million nonpolymorphic probes and > 750,000 single nucleotide polymorphism (SNP) probes. Only the CNV-containing genomic region was evaluated in each sample. The microarray was performed at Victoria Hospital (London, ON) in accordance with the manufacturer’s protocol, as described.
      • Dron J.S.
      • Wang J.
      • Berberich A.J.
      • et al.
      Large-scale deletions of the ABCA1 gene in patients with hypoalphalipoproteinemia.
      Data were analysed using the Chromosome Analysis Suite version 3.2 (ThermoFisher Scientific). The signal patterns were compared with normal in silico reference data built in the Chromosome Analysis Suite software. Copy number loss or gain was visualized by log2 ratio (sample intensity/expected reference intensity).

      Plasma PCSK9 analysis

      Plasma PCSK9 levels were assessed in CNV-positive index case A by both an enzyme-linked immunosorbent (ELISA) and immunoprecipitation assay. ELISA (CircuLex) was performed on ethylenediaminetetraacetic acid (EDTA) plasma according to the manufacturer’s protocol (MBL International, Woburn, MA), and repeated in triplicate. Immunoprecipitation was performed on 10 μL of EDTA plasma (case) or 50 μL (normal control) aliquots that were diluted into 1 mL of buffer A (20 mM Hepes-KOH, pH 7.4, 100 mM NaCl, 1.5 mM MgCl2, 1 mM CaCl2, 1% NP-40) containing 2 mM PefaBloc (Roche, Basel, Switzerland) and precipitated overnight at 4°C using an in-house rabbit polyclonal anti-serum (Ab 1697) raised against full-length recombinant human PCSK9 and captured using goat antirabbit IgG-conjugated agarose beads (Rockland Immunochemicals, Limerick, PA). The beads were washed 3 times with buffer A and eluted in SDS loading buffer (50 mM Tris-HCl, pH 6.8; 1% SDS; 5% glycerol; 10 mM EDTA; 0.0032% bromophenol blue, 2.5% (v/v) 2-mercaptoethanol). Immunoprecipitated proteins were subjected to 8% SDS-PAGE, transferred to nitrocellulose membrane (Bio-Rad, Hercules, CA), and incubated with a primary anti-PCSK9 mouse monoclonal antibody (15A6). An infrared dye (IRDye-800)-labelled secondary antibody was used for detection on a LI-COR Odyssey infrared system (LI-COR Biosciences, Lincoln, NE). ELISA and immunoprecipitation experiments were performed after index case A had ceased PCSK9 inhibitor therapy for approximately 4 weeks; however, this patient was still taking atorvastatin 40 mg plus ezetimibe 10 mg daily when samples were obtained for measurements.

      Statistical analysis

      Analyses of demographic features were performed in SAS version 9.1 (SAS Institute, Cary, NC). Discrete traits were compared using χ2 analysis, typically 2 × 2 contingency analyses, whereas quantitative traits were compared using unpaired t tests. The nominal level of statistical significance was set at P < 0.05.

      Results

      Study sample demographics

      Baseline clinical and biochemical traits of our cohort individuals are described in Supplemental Table S1.

      Case presentations

      Index case A is a male of Northern European descent who was first treated for severe hypercholesterolemia at age 37 in 2013. He presented with an untreated LDL-C of 14.9 mmol/L, tendon xanthomata, and extensive atherosclerosis with CCS class III angina symptoms. He was found to have severe multivessel coronary artery disease, with 95%, 99%, and 60% occlusions in left main, circumflex, and right coronary arteries, respectively, and 100% occlusion of the first diagonal branch of the left anterior descending coronary artery. He underwent urgent 3-vessel coronary arterial bypass graft (CABG) surgery. His initial response to high-intensity statin therapy was poor, with < 20% LDL-C reduction from baseline values. The addition of ezetimibe 10 mg daily to atorvastatin 80 mg daily reduced the LDL-C to 9.7 mmol/L (ie, a 34.8% reduction from baseline). Serial biweekly plasmapheresis treatments were more effective, with mean postapheresis total cholesterol of 2.8 mmol/L; however, these treatments were discontinued after several months because of poor venous access. The addition of alirocumab 150 mg subcutaneously every 2 weeks reduced LDL-C to 6.8 mmol/L (ie, an incremental 29.9% reduction from the value on statin plus ezetimibe), and a similar response was noted when alirocumab was switched to evolocumab 140 mg subcutaneously every 2 weeks. Most recently, evolocumab dose was increased to 420 mg subcutaneously every 2 weeks, reducing LDL-C to 5.5 mmol/L (a further 19.1% reduction, or a 43.3% reduction compared with statin plus ezetimibe treatment alone). In 2015, initial LipidSeq NGS analysis followed by CNV analysis in LDLR by MLPA showed no causative monogenic mutations. This patient had a high polygenic risk score (2.27; in the 99th percentile) based on a weighted LDL-C 10-SNP evaluation, as previously described;
      • Wang J.
      • Dron J.S.
      • Ban M.R.
      • et al.
      Polygenic versus monogenic causes of hypercholesterolemia ascertained clinically.
      however, this was not sufficient to explain his extremely elevated LDL-C levels.
      His father had a historical untreated total cholesterol level of approximately 15 mmol/L with a similar attenuated response to statin treatment. He underwent CABG at age 50 years. Index case A also reported premature cardiovascular disease in second-degree paternal relatives. His mother’s untreated total cholesterol was 6.0 mmol/L. His asymptomatic 13-year-old daughter had serum total cholesterol of 9.5 mmol/L, triglycerides of 0.6 mmol/L, LDL-C of 7.5 mmol/L, and HDL-C of 1.78 mmol/L.
      Index case B is a male of Northern European descent, not known to be related to index case A, who was referred at age 40 with refractory, severe hypercholesterolemia, which was first diagnosed at age 25 years. His historical untreated LDL-C was 14.5 mmol/L. He was asymptomatic from cardiovascular and metabolic perspectives. He had diffuse xanthomatosis, involving finger extensor, Achilles, and plantar flexor tendons bilaterally. With rosuvastatin 40 mg daily and ezetimibe 10 mg daily, his lowest recorded LDL-C level was 4.32 mmol/L, but typically this level ranged between 5.5 and 7.0 mmol/L on treatment. PCSK9 inhibition was never initiated before he was lost to follow-up because of work-related relocation. His family history was strongly positive for hyperlipidemia. His father suffered a stroke at age 55 and had bilateral lower limb amputations in the seventh decade of life. His mother had hypercholesterolemia and underwent 4-vessel CABG at age 62. His maternal grandfather died at age 40 of a myocardial infarction. His 10-year-old son was reported to have hypercholesterolemia.
      His older sister, younger brother, and younger sister all had severe hypercholesterolemia; all received high-intensity statin and ezetimibe. His younger sister was assessed at age 38 after having been diagnosed with hypercholesterolemia at age 31. Her highest recorded untreated LDL-C level was 11.4 mmol/L. A lifelong cigarette smoker, she continued to smoke 1 pack daily even after her hypercholesterolemia diagnosis. On examination, she had bilateral xanthelasmas and diffuse pronounced xanthomatosis, involving finger extensor and Achilles tendons bilaterally. At age 35, she developed left lower limb claudication, with diffuse femoral atherosclerosis demonstrated angiographically. With rosuvastatin 40 mg daily and ezetimibe 10 mg daily, her lowest recorded LDL-C level was 6.12 mmol/L, but typically this level ranged between 7 and 8 mmol/L on treatment. Her 11-year-old son was also reported to have hypercholesterolemia. Before PCSK9 inhibitors became available, she died at age 42 of a myocardial infarction. Initial LipidSeq NGS analysis of both siblings in 2014 followed by CNV analysis in LDLR using MLPA found no causative mutations to explain their phenotype; both also had a low 10-SNP polygenic risk score (1.0; in the 5th percentile).

      Targeted NGS data CNV analysis

      In 2017, we rescreened the LipidSeq data in our patients with FH to search for CNVs in FH-associated genes outside of LDLR. We detected a large-scale duplication encompassing the entire PCSK9 gene in index cases A and B, described above. Located on chromosome 1p32, the human PCSK9 gene is approximately 25 kilobases long and comprises 12 exons. Sample outputs for each index case are shown in Figures 1A and 2A . We detected no CNVs in APOB, LDLRAP1, APOE, STAP1, LIPA, and ABCG5/8 in this cohort of 704 individuals with FH.
      Figure thumbnail gr1
      Figure 1Next-generation sequencing (NGS)-based detection of a PCSK9 copy number variation (CNV) in a patient with familial hypercholesterolemia (index case A). (A) Targeted NGS output: duplication of all 12 exons of the PCSK9 gene, plus rs11206510 probe 8655 bases upstream of PCSK9. Different regions of the output are as follows: (i) Normalized ratio metric computed for each NGS probe target region in PCSK9; depth of sequence coverage comparative to reference controls. (ii) Normalized z-score metric; number of standard deviations the depth of coverage is from the reference control mean coverage. (iii) Called CNV state per probe target region, determined by ratio and z-score metrics together with supporting evidence from variant allele frequencies (not shown). (iv) Multiple affected target regions merged by segmentation analysis to call a contiguous duplication event. (v) Exon map of the PCSK9 gene. (vi) LipidSeq probe target regions. (B) Whole-exome sequencing (WES) output: validation of PCSK9 whole-gene duplication, plus flanking genes BSND (5′) and USP24 (3′) unaffected (diploid). Panel regions (i)-(iv) are as in (A). (v) Exon map of PCSK9 and flanking genes. (vi) WES probe target regions.
      Figure thumbnail gr2
      Figure 2Next-generation sequencing (NGS)-based detection of a PCSK9 copy number variation in a patient with familial hypercholesterolemia (index case B). (A) Targeted NGS output: duplication of all 12 exons of the PCSK9 gene, plus rs11206510 probe 8655 bases upstream of PCSK9. (B) Whole-exome sequencing output: validation of PCSK9 whole-gene duplication, plus flanking genes BSND (5′) and USP24 (3′) unaffected (diploid). All panel regions are as in .
      We next obtained DNA samples from family members of index case A and performed targeted NGS-based CNV analysis. Both the affected father and affected daughter of index case A were positive for this PCSK9 duplication, whereas the unaffected mother was CNV negative. The above-described affected sister of index case B was also one of our patients with FH; her DNA was available for analysis and was found to be CNV positive. No additional family members of index case B were available for analysis. Pedigrees are shown in Figure 3. Sample outputs for index case family members are shown in Supplemental Figure S1, A-D.
      Figure thumbnail gr3
      Figure 3Family pedigree of 2 familial hypercholesterolemia index cases with a whole-gene duplication of PCSK9. Males and females are represented as squares and circles, respectively, whereas black-shaded and unshaded represent individuals with reported severe hypercholesterolemia and normal lipid profiles, respectively. Enlarged shapes refer to individuals where a DNA sample was possible to obtain and analyse for the presence (+) or absence (−) of a PCSK9 copy number variation (CNV). Grey diagonal lines indicate deceased. Roman numerals I-IV indicate generation. CABG, coronary arterial bypass graft; LDL-C, low-density lipoprotein cholesterol; MI, myocardial infarction; TC, total cholesterol.

      Whole-exome CNV analysis

      To determine whether the large-scale duplication encompassing PCSK9 extended beyond the PCSK9 locus and into flanking genes we performed WES in both index cases and applied subsequent bioinformatic CNV analysis. In index cases A and B, the genes flanking PCSK9—5′ BSND (upstream) and 3′ USP24 (downstream)—were unaffected. Sample outputs are shown in Figures 1B and 2B.

      CNV confirmation

      Microarray-based CNV analysis performed in FH index cases A and B confirmed the presence of whole-gene PCSK9 duplications, whereas adjacent genes were unaffected. The array allowed for further fine mapping; the total size of this CNV duplication was predicted to be approximately 35 kilobases. Sample outputs are shown in Supplemental Figure S2.

      Plasma PCSK9 levels

      Plasma PCSK9 in index case A was approximately 5000 ng/mL as determined by ELISA; this was a 21-fold increase compared with a normal control (Fig. 4A). Immunoprecipitation and immunoblot analysis confirmed that the observed increase corresponded to full-length PCSK9 (Fig. 4B) and not a furin-cleaved inactive form.
      • Benjannet S.
      • Rhainds D.
      • Hamelin J.
      • Nassoury N.
      • Seidah N.G.
      The proprotein convertase (PC) PCSK9 is inactivated by furin and/or PC5/6A: functional consequences of natural mutations and post-translational modifications.
      • Han B.
      • Eacho P.I.
      • Knierman M.D.
      • et al.
      Isolation and characterization of the circulating truncated form of PCSK9.
      Plasma samples from index case B or affected relatives were not available for analysis.
      Figure thumbnail gr4
      Figure 4Plasma PCSK9 level in a patient with familial hypercholesterolemia (FH) (index case A) with a PCSK9 whole-gene duplication. (A) Plasma PCSK9 enzyme-linked immunosorbent assay measurement (repeat n = 3). (B) Plasma PCSK9 immunoprecipitation. PCSK9 was immunoprecipitated from ethylenediaminetetraacetic acid (EDTA) plasma using a rabbit polyclonal antibody raised against full-length recombinant human PCSK9 and detected with a monoclonal antibody (15A6). For comparison with normolipidemic control, plasma from a patient with FH (index case A) was diluted 5-fold before immunoprecipitation analysis.

      Discussion

      We report a whole-gene duplication of PCSK9 in 2 unrelated FH index cases and several of their affected family members, which we identified through bioinformatic analysis of targeted NGS data. To our knowledge, this is the first description in FH of a CNV affecting the PCSK9 gene. Additional features in affected individuals included severely elevated LDL-C levels that were refractory to treatment, pronounced clinical stigmata, premature cardiovascular events, and a plasma PCSK9 of approximately 5000 ng/mL in 1 index case, which is the highest known level ever reported in a human.
      PCSK9 is a serine protease that governs net LDL receptor (LDLR) activity. Secreted mainly by the liver, PCSK9 binds the LDLR at the cell surface, and after endocytosis of the LDLR-PCSK9 complex, diverts LDLR toward lysosomes for degradation, thus short circuiting the normal recycling of the receptor to the cell surface.
      • Zhang D.-W.
      • Lagace T.A.
      • Garuti R.
      • et al.
      Binding of proprotein convertase subtilisin/kexin type 9 to epidermal growth factor-like repeat A of low density lipoprotein receptor decreases receptor recycling and increases degradation.
      Mutations causing a gain of function (GOF) in PCSK9 enhance LDLR degradation, resulting in elevated plasma LDL-C. Genetic analysis of patients with atypical FH initially led to the discovery of PCSK9’s role in LDLR recycling and cholesterol homeostasis; in 2003, Abifadel et al.
      • Abifadel M.
      • Varret M.
      • Rabès J.-P.
      • et al.
      Mutations in PCSK9 cause autosomal dominant hypercholesterolemia.
      identified 2 GOF missense variants, p.Ser127Arg and p.Phe216Leu, in 3 French families with autosomal dominant FH. Since then, approximately 30 different PCSK9 mutations, many with distinct GOF mechanisms, have been described throughout all domains of the protein.
      • Dron J.S.
      • Hegele R.A.
      Complexity of mechanisms among human proprotein convertase subtilisin-kexin type 9 variants.
      To date, however, these mutations have all been small-scale variants—namely 24 missense, 2 splicing, 2 trinucleotide indels, and a 5′untranslated region substitution.
      • Dron J.S.
      • Hegele R.A.
      Complexity of mechanisms among human proprotein convertase subtilisin-kexin type 9 variants.
      The large-scale whole-gene duplication identified here, causing an increase in gene dosage, constitutes a novel GOF mechanism for PCSK9 in FH.
      The severity of the FH phenotype in the index cases and their affected relatives is notable. In particular, untreated LDL-C levels ranged between 11.0 and 15.0 mmol/L, with prominent xanthomatosis and atherosclerotic cardiovascular disease presenting in the fourth decade of life, specifically 4-vessel CABG in index case A and myocardial infarction in index case B relatives. LDL-C levels elevated to this degree are more characteristic of homozygous FH; however, both NGS-based and microarray-based CNV analysis confirm that only a single PCSK9 allele was affected in both families, with overall copy number increasing from 2 (diploid) to 3.
      Thus, a single extra copy of PCSK9 seems to profoundly affect LDL-C homeostasis, underlying a severe form of FH. However, the phenotypic outcome of any gene duplication depends on several factors, including the location and orientation of the duplicated genomic material. In other disorders, there is evidence to suggest that most large-scale duplications occur in tandem;
      • Newman S.
      • Hermetz K.E.
      • Weckselblatt B.
      • Rudd M.K.
      Next-generation sequencing of duplication CNVs reveals that most are tandem and some create fusion genes at breakpoints.
      however, it is possible that the duplicated material is present elsewhere in the genome. The expression of a duplicated gene depends on genomic location and its epigenetic regulation. One caveat of using exome-based NGS CNV analysis, as well as microarray-based CNV analysis, is that although duplicated material can be detected, its precise location and orientation are not always defined.
      This PCSK9 CNV is associated with a plasma PCSK9 measurement of approximately 5000 ng/mL, a 21-fold increase compared with a normal control. Although variability in ELISA-based protocols makes comparisons with values from other studies difficult, this is nonetheless the highest known human level reported. This finding supports the functionality of this particular CNV event—that is, the duplicated material is actively expressed.
      Other factors might have influenced plasma PCSK9 levels in index case A, including high baseline LDL-C levels and statin treatment. For instance, plasma PCSK9 levels positively correlate with LDL-C; for LDL-C levels typically seen in homozygous FH (ie, LDL-C > 13.0 mmol/L, as seen in the index cases reported here), baseline plasma PCSK9 levels are 2- to 3-fold higher than normolipidemic controls.
      • Drouin-Chartier J.-P.
      • Tremblay A.J.
      • Hogue J.-C.
      • et al.
      The contribution of PCSK9 levels to the phenotypic severity of familial hypercholesterolemia is independent of LDL receptor genotype.
      • Raal F.
      • Panz V.
      • Immelman A.
      • Pilcher G.
      Elevated PCSK9 levels in untreated patients with heterozygous or homozygous familial hypercholesterolemia and the response to high-dose statin therapy.
      • Cameron J.
      • Bogsrud M.P.
      • Tveten K.
      • et al.
      Serum levels of proprotein convertase subtilisin/kexin type 9 in subjects with familial hypercholesterolemia indicate that proprotein convertase subtilisin/kexin type 9 is cleared from plasma by low-density lipoprotein receptor–independent pathways.
      Also, statins upregulate PCSK9 expression; the PCSK9 promoter contains a sterol regulator element site and is co-expressed with LDLR after nuclear translocalization of the sterol regulator element binding protein-2 in response to low intracellular cholesterol.
      • Dubuc G.
      • Chamberland A.
      • Wassef H.
      • et al.
      Statins upregulate PCSK9, the gene encoding the proprotein convertase neural apoptosis-regulated convertase-1 implicated in familial hypercholesterolemia.
      Index case A remained on rosuvastatin 40 mg daily while plasma was taken for PCSK9 determination. Typically, administration of a high-intensity statin is associated with a 25% to 50% increase in plasma PCSK9 levels.
      • Nozue T.
      Lipid lowering therapy and circulating PCSK9 concentration.
      Raal et al.
      • Raal F.
      • Panz V.
      • Immelman A.
      • Pilcher G.
      Elevated PCSK9 levels in untreated patients with heterozygous or homozygous familial hypercholesterolemia and the response to high-dose statin therapy.
      showed that rosuvastatin 40 mg daily resulted in a 37% increase in plasma PCSK9 levels. The effect of statins on PCSK9 expression might be amplified when an extra copy of PCSK9 is present. However, despite both the high background LDL-C and high-intensity statin therapy in index case A, the observed 21-fold increase in plasma PCSK9 is still disproportionately high. It is possible that the duplicated PCSK9 gene in this patient may be located elsewhere in the genome, perhaps driven by an unknown enhancer element, or resides in a genomic region where transcription is continually active (ie, in a euchromatic state), resulting in an increased rate of expression and thus high PCSK9 levels.
      These findings have therapeutic implications. As statin-induced upregulation of PCSK9 may be accentuated in patients with an extra copy of PCSK9, high-dose statin therapy may have only limited efficacy. Indeed, there was resistance to intensive statin therapy in both index cases and in several family members. With plasma PCSK9 levels increased, index case A also appeared to require a high dose of a PCSK9 inhibitor.
      In addition, we found no CNVs in APOB, LDLRAP1, APOE, STAP1, LIPA, and ABCG5/8 in this cohort of 704 individuals with FH. It is possible that pathogenic CNVs in these other FH-related genes exist, but may require larger FH cohorts to be detected. The potential of finding CNVs in these genes is of interest because approximately 40% of patients with suspected FH in many clinical cohorts have no apparent “typical” or obvious mutations underlying their phenotype. Evaluating SNPs to define a possible polygenic basis for hypercholesterolemia may explain approximately an additional 20% of clinically ascertained FH cases,
      • Wang J.
      • Dron J.S.
      • Ban M.R.
      • et al.
      Polygenic versus monogenic causes of hypercholesterolemia ascertained clinically.
      but still leaves a substantial number of “unexplained” cases. Systematic screening in additional populations could help evaluate the possibility that CNVs in other FH-related genes may be present in some subjects with FH.
      Given the CNV-detection methodologies used here, the exact location of the PCSK9 duplication within the genome is not known. This limits the ability to determine whether the same ancestral CNV event is present in both families, and also to speculate on possible gene expression influences that may explain the disproportionately high PCSK9 level detected.

      Conclusions

      In conclusion, we performed novel CNV screening in FH-associated genes in a large cohort of 704 individuals with FH and identified a whole-gene PCSK9 duplication in 2 FH index cases and their affected family members. This is the first report of a PCSK9 CNV associated with a severe FH phenotype and profoundly elevated plasma PCSK9. The grossly elevated PCSK9 level may limit the efficacy of intensive statin therapy and perhaps also the efficacy of PCSK9 inhibition. These findings also highlight the potential for finding novel disease-causing variants when CNV screening is extended beyond the commonly studied LDLR gene, and may help to further avoid false-negative genetic diagnoses and direct treatment strategy.

      Funding Sources

      R.A. Hegele is supported by the Jacob J. Wolfe Distinguished Medical Research Chair , the Edith Schulich Vinet Research Chair in Human Genetics , and the Martha G. Blackburn Chair in Cardiovascular Research . R.A. Hegele has received operating grants from the Canadian Institutes of Health Research (Foundation Grant), the Heart and Stroke Foundation of Ontario (HSF G-18-0022147 ), and Genome Canada through Genome Quebec (award 4530).

      Disclosures

      R.A. Hegele has received honoraria for membership on advisory boards and speakers’ bureaus for Aegerion, Amgen, Gemphire, Lilly, Merck, Pfizer, Regeneron, Sanofi, and Valeant, all unrelated to the topic of this manuscript. The rest of the authors have no conflicts of interest to disclose.

      Supplementary Material

      References

        • Akioyamen L.E.
        • Genest J.
        • Shan S.D.
        • et al.
        Estimating the prevalence of heterozygous familial hypercholesterolaemia: a systematic review and meta-analysis.
        BMJ Open. 2017; 7e016461
        • Nordestgaard B.G.
        • Chapman M.J.
        • Humphries S.E.
        • et al.
        Familial hypercholesterolaemia is underdiagnosed and undertreated in the general population: guidance for clinicians to prevent coronary heart disease: consensus statement of the European Atherosclerosis Society.
        Eur Heart J. 2013; 34: 3478-3490
        • Defesche J.C.
        • Gidding S.S.
        • Harada-Shiba M.
        • et al.
        Familial hypercholesterolaemia.
        Nat Rev Dis Prim. 2017; 3: 17093
        • Garcia C.K.
        • Wilund K.
        • Arca M.
        • et al.
        Autosomal recessive hypercholesterolemia caused by mutations in a putative LDL receptor adaptor protein.
        Science. 2001; 292: 1394-1398
        • Marduel M.
        • Ouguerram K.
        • Serre V.
        • et al.
        Description of a large family with autosomal dominant hypercholesterolemia associated with the APOE p.Leu167del mutation.
        Hum Mutat. 2013; 34: 83-87
        • Fouchier S.W.
        • Dallinga-Thie G.M.
        • Meijers J.C.M.
        • et al.
        Mutations in STAP1 are associated with autosomal dominant hypercholesterolemia.
        Circ Res. 2014; 115: 552-555
        • Chora J.R.
        • Alves A.C.
        • Medeiros A.M.
        • et al.
        Lysosomal acid lipase deficiency: a hidden disease among cohorts of familial hypercholesterolemia?.
        J Clin Lipidol. 2017; 11: 477-484
        • Hansel B.
        • Carrié A.
        • Brun-Druc N.
        • et al.
        Premature atherosclerosis is not systematic in phytosterolemic patients: severe hypercholesterolemia as a confounding factor in five subjects.
        Atherosclerosis. 2014; 234: 162-168
        • Zarrei M.
        • MacDonald J.R.
        • Merico D.
        • Scherer S.W.
        A copy number variation map of the human genome.
        Nat Rev Genet. 2015; 16: 172-183
        • Bourbon M.
        • Alves A.C.
        • Sijbrands E.J.
        Low-density lipoprotein receptor mutational analysis in diagnosis of familial hypercholesterolemia.
        Curr Opin Lipidol. 2017; 28: 120-129
        • Iacocca M.A.
        • Hegele R.A.
        Role of DNA copy number variation in dyslipidemias.
        Curr Opin Lipidol. 2018; 29: 125-132
        • Wang J.
        • Ban M.R.
        • Hegele R.A.
        Multiplex ligation-dependent probe amplification of LDLR enhances molecular diagnosis of familial hypercholesterolemia.
        J Lipid Res. 2005; 46: 366-372
        • Iacocca M.A.
        • Wang J.
        • Dron J.S.
        • et al.
        Use of next-generation sequencing to detect LDLR gene copy number variation in familial hypercholesterolemia.
        J Lipid Res. 2017; 58: 2202-2209
        • Iacocca M.A.
        • Hegele R.A.
        Recent advances in genetic testing for familial hypercholesterolemia.
        Expert Rev Mol Diagn. 2017; 17: 641-651
        • Johansen C.T.
        • Dube J.B.
        • Loyzer M.N.
        • et al.
        LipidSeq: a next-generation clinical resequencing panel for monogenic dyslipidemias.
        J Lipid Res. 2014; 55: 765-772
        • Wang J.
        • Dron J.S.
        • Ban M.R.
        • et al.
        Polygenic versus monogenic causes of hypercholesterolemia ascertained clinically.
        Arterioscler Thromb Vasc Biol. 2016; 36: 2439-2445
        • Dron J.S.
        • Wang J.
        • Berberich A.J.
        • et al.
        Large-scale deletions of the ABCA1 gene in patients with hypoalphalipoproteinemia.
        J Lipid Res. 2018; 59: 1529-1535
        • Benjannet S.
        • Rhainds D.
        • Hamelin J.
        • Nassoury N.
        • Seidah N.G.
        The proprotein convertase (PC) PCSK9 is inactivated by furin and/or PC5/6A: functional consequences of natural mutations and post-translational modifications.
        J Biol Chem. 2006; 281: 30561-30572
        • Han B.
        • Eacho P.I.
        • Knierman M.D.
        • et al.
        Isolation and characterization of the circulating truncated form of PCSK9.
        J Lipid Res. 2014; 55: 1505-1514
        • Zhang D.-W.
        • Lagace T.A.
        • Garuti R.
        • et al.
        Binding of proprotein convertase subtilisin/kexin type 9 to epidermal growth factor-like repeat A of low density lipoprotein receptor decreases receptor recycling and increases degradation.
        J Biol Chem. 2007; 282: 18602-18612
        • Abifadel M.
        • Varret M.
        • Rabès J.-P.
        • et al.
        Mutations in PCSK9 cause autosomal dominant hypercholesterolemia.
        Nat Genet. 2003; 34: 154-156
        • Dron J.S.
        • Hegele R.A.
        Complexity of mechanisms among human proprotein convertase subtilisin-kexin type 9 variants.
        Curr Opin Lipidol. 2017; 28: 161-169
        • Newman S.
        • Hermetz K.E.
        • Weckselblatt B.
        • Rudd M.K.
        Next-generation sequencing of duplication CNVs reveals that most are tandem and some create fusion genes at breakpoints.
        Am J Hum Genet. 2015; 96: 208-220
        • Drouin-Chartier J.-P.
        • Tremblay A.J.
        • Hogue J.-C.
        • et al.
        The contribution of PCSK9 levels to the phenotypic severity of familial hypercholesterolemia is independent of LDL receptor genotype.
        Metabolism. 2015; 64: 1541-1547
        • Raal F.
        • Panz V.
        • Immelman A.
        • Pilcher G.
        Elevated PCSK9 levels in untreated patients with heterozygous or homozygous familial hypercholesterolemia and the response to high-dose statin therapy.
        J Am Heart Assoc. 2013; 2e000028
        • Cameron J.
        • Bogsrud M.P.
        • Tveten K.
        • et al.
        Serum levels of proprotein convertase subtilisin/kexin type 9 in subjects with familial hypercholesterolemia indicate that proprotein convertase subtilisin/kexin type 9 is cleared from plasma by low-density lipoprotein receptor–independent pathways.
        Transl Res. 2012; 160: 125-130
        • Dubuc G.
        • Chamberland A.
        • Wassef H.
        • et al.
        Statins upregulate PCSK9, the gene encoding the proprotein convertase neural apoptosis-regulated convertase-1 implicated in familial hypercholesterolemia.
        Arterioscler Thromb Vasc Biol. 2004; 24: 1454-1459
        • Nozue T.
        Lipid lowering therapy and circulating PCSK9 concentration.
        J Atheroscler Thromb. 2017; 24: 895-907