Putative protective role of PCSK9 variants in a multigenerational family with familial hypercholesterolemia

Abstract

Background

Familial hypercholesterolemia (FH) is most frequently caused by pathogenic variants in LDLR, but phenotypic variability suggests the influence of genetic modifiers.

Methods

We investigated a large multigenerational family with FH, combining clinical data, lipid profiles, and genetic analysis with functional studies. LDLR and PCSK9 variants were characterized according to ACMG/ClinGen guidelines. Functional assays in CHO-ldlA7 cells assessed LDLR activity, while plasma PCSK9 levels were quantified by ELISA.

Results

The LDLR c.2479G > A variant was associated with the FH phenotype in the family and showed borderline functional impairment in vitro. The presence of the established loss-of-function PCSK9 c.137G>T variant, together with the rare PCSK9 c.2023del variant with a predicted deleterious effect, was identified in a normocholesterolemic carrier of the LDLR variant who showed markedly low circulating PCSK9 levels.

Conclusions

These findings suggest the potential relevance of investigating gene–gene interactions and variants with opposing effects on the lipid phenotype in FH. Integrating clinical, biochemical, genetic, and functional data may help explain intrafamilial variability and support more precise interpretation of inherited dyslipidemias.

Graphical abstract

Highlights

• •A large family reveals variable familial hypercholesterolemia severity.
• •LDLR p.(Val827Ile) shows borderline reduced low-density lipoprotein uptake.
• •A putative PCSK9 modifier genotype is linked to very low serum PCSK9.
• •Protective PCSK9 variants may attenuate the familial cholesterol phenotype.
• •Extended genetic testing can clarify discordant cholesterol phenotypes.

1. Introduction

Familial hypercholesterolemia (FH) is an autosomal disorder of lipid metabolism characterized by elevated levels of total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-C), which predispose affected individuals to premature atherosclerotic cardiovascular disease (ASCVD). If untreated, FH is associated with an increased risk of myocardial infarction and cardiovascular mortality, frequently manifesting decades earlier than in the general population. Despite being a monogenic condition, it remains underdiagnosed and undertreated worldwide, representing a significant public health challenge [1] [2]. The estimated prevalence of heterozygous FH is approximately 1 in 200–250 individuals, making it one of the most common inherited metabolic disorders, whereas homozygous FH is considerably rarer, affecting about 1 in 160,000–300,000 individuals [3].

Clinical diagnosis of FH is based on standardized criteria such as the Dutch Lipid Clinic Network (DLCN) score, which incorporates lipid values, clinical signs (xanthomas, corneal arcus), family history of hypercholesterolemia or premature ASCVD, and the presence of a pathogenic genetic variant [4]. Nevertheless, the FH phenotype can be variable and may be subtle or even absent in childhood or early adulthood. Early diagnosis is essential, as timely initiation of statins and other lipid-lowering therapies can restore near-normal life expectancy for most. Consequently, cascade genetic screening of relatives is recommended once an index case is identified [5]. The incorporation of genetic testing into FH diagnostic workflows significantly increases the detection rate, facilitates the identification of affected relatives, and enables more precise cardiovascular risk stratification.

Approximately 90% of genetically confirmed FH cases are caused by pathogenic loss-of-function (LOF) variants in the low-density lipoprotein receptor gene (LDLR, MIM 606945) [6]. Additional monogenic causes include deleterious variants in the apolipoprotein B gene (APOB, MIM 107730), that impair the ability of LDL-C particles to bind LDL receptor (LDLR) [7], and gain-of-function (GOF) variants in proprotein convertase subtilisin/kexin type 9 (PCSK9, MIM 607786), which accelerate LDLR degradation [8]. Rare biallelic LOF variants in LDL receptor adaptor protein 1 (LDLRAP1, MIM 605747) cause the autosomal recessive form of FH, frequently characterized by extremely increased LDL-C levels and early onset ASCVD(9).

Although FH is considered a monogenic disease, its clinical expression is highly heterogeneous. Individuals carrying the same pathogenic variant, even within a single family, may exhibit markedly differences in LDL-C levels, age at onset of cardiovascular events, or response to lipid lowering therapy. This variability suggests the influence of genetic, epigenetic, and environmental modifiers. The identification of genetic modifiers is particularly relevant, as they may explain “resilience” or unexpected mild phenotypes in certain carriers of pathogenic variants and may also provide insights into novel therapeutic approaches.

The PCSK9 gene encodes proprotein convertase subtilisin/kexin type 9 (PCSK9), a secreted serine protease that plays a central role in cholesterol homeostasis. PCSK9 regulates the number of LDLR on the hepatocyte surface by binding to LDLR and targeting it for lysosomal degradation. Consequently, increased PCSK9 activity leads to higher plasma LDL-C levels, whereas reduced activity promotes LDL clearance. Since its identification as an FH gene in 2003(8), PCSK9has attracted considerable attention both for its role in monogenic disease and as a therapeutic target. Monoclonal antibody inhibitors of PCSK9 are now widely used in clinical practice, and small interfering RNA therapies are emerging as long-acting alternatives.

Remarkably, PCSK9 also represents one of the best-characterized examples of naturally occurring protective genetic variants. Carriers of LOF variants in PCSK9 exhibit lifelong lower LDL-C levels and significantly lower risk of coronary artery disease without apparent adverse effects [9]. These observations firmly establish that PCSK9 activity directly influences cardiovascular risk. Furthermore, the coexistence of variants in PCSK9 and other FH-related genes raises the potential for gene–gene interactions that may modulate disease severity. However, reports of such interactions remain limited, and functional validation is seldom available.

Herein, we report a family with FH carrying a controversial LDLR variant, and describe functional studies alongside a critical appraisal of the available evidence to discuss its potential pathogenicity. Additionally, we describe a rare PCSK9 genotype, including an established LDL-C-lowering variant and a second variant with a predicted deleterious effect, that may modify the phenotypic expression of the disease within this family.

2. Patients and methods

2.1. Patients and ethical approval

The index case was a male in his fifties with a clinical diagnosis of FH according to the Dutch Lipid Clinic Network (DLCN) criteria, who was referred for genetic testing. Following the index case results, cascade genetic testing was offered to all available relatives. All participating family members provided written informed consent for genetic analysis in accordance with the Declaration of Helsinki, and the study was approved by the institutional ethics committee. Clinical data was collected, including lipid profiles, cardiovascular history, and physical examination (xanthomas, corneal arcus).

2.2. Genetic analysis

Peripheral blood samples were collected in EDTA tubes, and genomic DNA was extracted using the QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany) following the manufacturer's instructions. DNA concentration and purity were assessed by NanoDrop™ spectrophotometry and Qubit fluorometry (Thermo Fisher Scientific).

Genetic testing was performed by targeted next-generation sequencing (NGS) using a custom panel of 198 genes related to lipid metabolism. The analysis included the coding regions of the genes included in the panel, the adjacent intronic boundaries (approximately ±20 bp from exon–intron junctions), and promoter regions covered by the design. Therefore, canonical splice-site and near-exonic intronic variants were evaluated. Deep intronic LDLR variants located outside the regions captured by the panel were not systematically assessed. Library preparation and exome enrichment were carried out according to the manufacturer's protocol (Roche Nimblegen, Madison, WI, USA), and sequencing was performed on a MiSeq system (Illumina, San Diego, CA, USA). For this study, analyses focused on the four major FH genes: LDLR, APOB, PCSK9, and LDLRAP1 plus FH related genes (APOE, ABCG5, ABCG8, LIPA).

The bioinformatics analysis was performed using customized algorithms. Briefly, sequences were mapped to the GRCh37/hg19 human reference sequence. Variants were interpreted using in silico pathogenicity predictors (CADD, PolyPhen-2, MutAssessor, FATHMM, VEST), conservation scores (GERP2, PhastCons, PhyloP), and splicing predictors (MaxEntScan, NNSplice, GeneSplicer, Human Splicing Finder). Variants with minor allele frequency (MAF) > 1% in population databases (gnomAD, 1000 Genomes) were excluded. BAM files were analyzed using Alamut Visual (SOPHiA Genetics, Lausanne, Switzerland). Variant classification followed ACMG/AMP and ClinGen guidelines [10] [11].

Large rearrangements in LDLR were screened by multiplex ligation-dependent probe amplification (MLPA; MRC-Holland, Amsterdam, The Netherlands). Variants of interest and segregation analysis were confirmed by Sanger sequencing using BigDye Terminator v3.1 Cycle Sequencing Kit and analyzed on an ABI 3500 Genetic Analyzer (Applied Biosystems).

2.3. Functional studies of the LDLR variant

2.3.1. Cell culture and transfection

LDLR-deficient CHO-ldlA7 cells (kindly provided by Dr. Monty Krieger, Massachusetts Institute of Technology, Cambridge, MA, USA) [12] were maintained in Ham's F-12 medium (Gibco, Life Technologies, Carlsbad, CA, USA) supplemented with 5% heat-inactivated fetal bovine serum (FBS), 100 U/mL penicillin, 100 μg/mL streptomycin, 2 mmol/L L-glutamine, and 50 mg/mL Normocin (InvivoGen, Toulouse, France). Cells were cultured at 37 °C in a humidified atmosphere with 5% CO₂ and passaged every 2-3 days.

Human LDLR cDNA constructs were generated using a GFP-tagged plasmid vector (C-GFPSpark®, NM_000527; Sino Biological, Beijing, China). Site-directed mutagenesis was performed (QuikChange Lightning kit; Agilent Technologies, Santa Clara, CA, USA) to introduce the variant of interest. The previously functionally characterized LDLR p.(Gly549Asp) variant was included as a pathogenic control to validate the assay performance and dynamic range [13]. Plasmid integrity and the presence of the intended mutation were confirmed by Sanger sequencing.

Transient transfections were performed in 24-well plates (1.5 × 10⁵ cells/well) using Lipofectamine 3000 (Invitrogen™, Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions. Cells were incubated for 48 h before downstream analyses.

2.3.2. LDLR activity by flow cytometry

LDL was isolated from a healthy donor and labeled with DiI (1,1′-dioctadecyl-3,3,3,3′-tetramethylindocarbocyanine perchlorate; Molecular Probes, Life Technologies). Transfected CHO-ldlA7 cells were incubated with 20 μg/mL DiI-LDL for 4 h at either 4 °C to measure LDL binding, or 37 °C to assess LDL uptake/internalization. After incubation, cells were washed with PBS-2% BSA and resuspended in PBS with 0.1% DAPI to exclude dead cells. Flow cytometry was performed using a BD FACSCanto™ II cytometer (BD Biosciences, San Jose, CA, USA), acquiring at least 10⁴ events per sample.

Experiments were performed in three independent biological replicates, each corresponding to an independent cell culture, seeding, and transient transfection performed on different days. Within each biological replicate, all experimental conditions were analyzed in three technical replicates, defined as three wells per construct derived from the same transfection mixture. CHO-ldlA7 cells transfected with the empty vector were included in each experiment as a negative control to account for background autofluorescence and non-specific fluorescence during flow cytometry analysis. Fluorescence values were corrected using the empty-vector control and then normalized to wild-type LDLR activity, which was set at 100% in each independent experiment. Data are presented as the mean ± standard deviation (SD) of three independent biological replicates. Statistical comparisons between groups were performed using Student's t-test, as previously described for LDLR functional characterization assays (Rodriguez Jiménez 2019).

2.3.3. 3D modelling of PCSK9 protein structure

A 3D model of PCSK9 (PDB ID: 2PMW) was obtained from the RCSB [14] protein data bank and PCSK9 mutant structure was generated by SWISS-MODEL (workspace homology modelling using

AF-Q8NBP7-F1 PCSK9 model from AlphaFold as a template [15]. 3D structures were visualized by UCSF-Chimera [16].

2.4. Quantification of serum PCSK9

Serum PCSK9 protein levels were quantified using a commercial enzyme-linked immunosorbent assay (Human PCSK9 ELISA Kit, Cell Biolabs Inc., San Diego, CA, USA) according to the manufacturer's specifications. Samples were measured in duplicate. Absorbance was read at 450 nm with background correction at 570 nm. PCSK9 concentration was calculated against a standard curve generated with recombinant human PCSK9. Control serum samples were obtained from anonymized residual clinical laboratory samples. Two control groups were generated: 9 adult controls and 16 pediatric controls. For each group, serum pools were prepared by combining 15 independent serum samples selected from individuals with lipid profiles within the reference ranges established by our clinical laboratory. Samples were selected anonymously, and no identifiable clinical information was available beyond age group and confirmation of a normal lipid profile. These pooled sera were used as reference controls for the quantification of circulating PCSK9 levels by ELISA.

3. Results

3.1. Clinical characteristics of the proband

The index case was a male patient clinically diagnosed with FH in early adulthood and subsequently followed regularly in a endocrinology outpatient clinic. His maximum recorded total cholesterol and LDL cholesterol levels were 335 mg/dL and 265 mg/dL, respectively. In his forties', while receiving atorvastatin 10 mg daily and with total cholesterol and LDL cholesterol levels of 113 mg/dL and 50 mg/dL, respectively, he experienced several episodes of retrosternal chest pain radiating to the left shoulder during physical exercise. Coronary angiography revealed a single-vessel obstructive lesion involving the circumflex coronary artery. He was therefore diagnosed with chronic ischemic heart disease manifesting as stable effort angina. Treatment included coronary stent implantation and intensification of lipid-lowering therapy to atorvastatin 40 mg daily. Genetic testing was performed in the patient, and the results are presented below.

3.2. Proband's molecular analysis

Molecular analysis revealed the presence of the following heterozygous variants: LDLR (NM_000527.5) c.2479G>A p.(Val827Ile), PCSK9 (NM_174936.4) c.137G>T p.(Arg46Leu) and APOB (NM_000384.3) c.2630C>T p.(Pro877Leu).

The LDLR c.2479G>A consists of a guanine to adenine substitution that predicts the replacement of amino acid valine to isoleucine at position 827 of the protein, located in the C-terminal cytosolic domain of the LDL receptor. It is present at a minor frequency (MAF) of 1,83% in Ashkenazi Jewish exomes-genomes at the Genome Aggregation Database (gnomAD-v4.1.0), with four homozygotes individuals reported. Although earlier ClinVar submissions and previous reports described this variant in the context of familial hypercholesterolemia, including classifications as likely pathogenic, the current ClinGen Familial Hypercholesterolemia Variant Curation Expert Panel (VCEP) classification for LDLR NM_000527.5:c.2479G>A p.(Val827Ile) is variant of uncertain significance (VUS). According to the ACMG/ClinGen FH-specific LDLR curation specifications, the currently applicable evidence includes segregation with the FH phenotype (PP1) and in silico prediction supporting a deleterious effect (PP3). Therefore, this variant was considered a borderline variant requiring integration of population frequency, segregation, functional data, and clinical context.

The PCSK9 c.137G>T consists of a guanine to thymine nucleotide substitution that predicts the replacement of amino acid arginine to leucine at position 46 of the protein, located in the prodomain of PCSK9. It is present at a global MAF of 1,48% at the Genome Aggregation Database (gnomAD-v4.1.0), 226 homozygotes reported, and it has previously been described as a LOF variant associated with lower LDL-C levels [17].

The APOB c.2630C>T variant consists of a cytosine to thymine substitution that predicts the replacement of an amino acid of proline to leucine at position 877 of the ApoB-100 protein. It is present at a global MAF of 0,04789% at the Genome Aggregation Database (gnomAD-v4.1.0), three homozygotes reported, and it is classified as a VUS in ClinVar. In silico predictors are not conclusive towards the pathogenicity of the variant, that is located within the N-terminal βα1 domain of ApoB-100, a region not directly involved in the canonical LDL receptor (LDLR)–binding domain.

Following the initial findings in the index case, genetic analysis was extended to available family members to better characterize the identified variants and to carry out cascade screening.

3.3. Segregation analysis

The index case (III.9) had three offspring (IV.4, IV.5 and IV.6). The eldest one, IV.4 carried the same genotype as her father and was clinically diagnosed of FH. The youngest one, IV.6, did not carry the LDLR variant but carried the APOB c.2630C>T p.(Pro877Leu) and the PCSK9 c.137G>T p.(Arg46Leu) variants in heterozygosity, she showed normal LDL-C. Interestingly, the individual IV.5 inherited LDLR c.2479G>A p.Val827Ile, his lipid levels remained within the normal range across serial measurements, with total cholesterol ranging from 135 to 170 mg/dL and LDL-C from 70 to 98 mg/mL. Additionally, he showed two more variants, both in PCSK9 c.137G>T p.(Arg46Leu) in homozygosity and the very rare variant, c.2023del p.(Val675Leufs*130) in heterozygosity. The variant c.2023del p.(Val675Leufs*130) was inherited from his mother (III.8) who had a normal lipid profile with total cholesterol of 194 mg/mL and LDL-C of 119 mg/dL.

Three relatives of the index case (III.10-12) did not carry the LDLR variant and showed normal lipid profile. Two of them carried the APOB c.2630C>T p.(Pro877Leu), suggesting that this variant may be not related to FH. The individual II.1 had hypercholesterolemia and died prematurely due to IAM in his sixties. Molecular analysis of his offspring revealed that III.1 and III.3, both diagnosed with hypercholesterolemia, carried the LDLR c.2479G>A p.(Val827Ile), PCSK9 c.137G>T p.(Arg46Leu) and APOB c.2630C>T p.(Pro877Leu) variants in heterozygosity. III.2 had normal lipid profile and only carried the PCSK9 c.137G>T p.(Arg46Leu) in heterozygosity. The individual II.2 carried LDLR c.2479G>A p.(Val827Ile), PCSK9 c.137G>T p.(Arg46Leu) and APOB c.2630C>T p.(Pro877Leu) in heterozygosity and had hypercholesterolemia. Her descendant, III.4 inherited the LDLR variant and showed hypercholesterolemia. The complete familial pedigree is shown in Fig. 1.

Fig. 1.

Pedigree of the family showing segregation of LDLR, PCSK9 and APOB variants. Squares represent males, circles represent females, and diamonds indicate family members whose sex is not specified. Individuals fulfilling the clinical diagnosis criteria for familial hypercholesterolemia (FH) are indicated by half-filled symbols. The first line below each symbol indicates the individual identifier, followed by age and the genotypes at LDLR, PCSK9 and APOB. Next line indicates the lipid phenotype, classified as normocholesterolemia or hypercholesterolemia according to clinical data and laboratory-defined reference ranges for total cholesterol and LDL-C (TC < 200 mg/dL; LDL-C <130 mg/dL). When available, the last line shows baseline TC and LDL-C levels (mg/dL). An asterisk (*) preceding these values indicates that the measurements were obtained while the individual was receiving statin treatment. Abbreviations: MI: myocardial infarction; Chol: cholesterol; WT: wild-type; hmz: homozygous; TC: total cholesterol; LDL-C: low-density lipoprotein cholesterol; nd: no data available.

Pedigree analysis showed cosegregation of LDLR c.2479G>A p.(Val827Ile) with the FH phenotype in twelve informative meioses. These informative meioses corresponded to relatives carrying the LDLR variant who also fulfilled the clinical diagnosis of hypercholesterolemia. According to the ClinGen FH VCEP specifications, segregation in ≥6 informative meioses supports PP1 at strong strength.

Two LDLR variant-positive individuals, IV.5 and IV.2, had normal cholesterol levels and were therefore not counted as meioses supporting cosegregation. These individuals were considered exceptions to the overall segregation pattern. Of note, individual IV.5 also carried two additional PCSK9 variants, including one extremely rare variant, which may have modified the lipid phenotype. The in silico characterization of this PCSK9 variant is described below.

3.4. Functional characterization of the LDLR c.2479G > A p.(Val827Ile)

3.4.1. Functional activity

Flow cytometry analysis showed a significant reduction in LDL uptake in CHO-ldlA7 cells expressing LDLR p.(Val827Ile) compared with wild-type LDLR, with residual uptake of approximately 74% of wild-type activity. In contrast, no significant differences were observed in LDL binding or LDLR expression (Fig. 2). These findings are consistent with a moderate/borderline impairment of LDLR internalization in this heterologous model. Given the borderline magnitude of the functional effect, this result should be interpreted in combination with population frequency, segregation data, and other available evidence.

Fig. 2.

Functional characterization of LDLR c.2479G > A; (A) Flow cytometry histogram overlay showing DiI-LDL binding; (B) Quantification of LDL uptake relative to wild-type (WT); (C) LDLR cell-surface expression. The previously functionally characterized LDLR p.(Gly549Asp) variant was included as a pathogenic control. The dashed line indicates the PS3 threshold (70% of WT activity), and the upper continuous line indicates the BS3 threshold (90% of WT activity), as defined by the ClinGen FH VCEP specifications for LDLR functional assays. * <0.05; **** <0.0001.

3.5. In silico analysis of PCSK9 c.2023del p.(Val675Leufs*130)

The variant PCSK9 c.2023del p.(Val675Leufs*130) consists of a single-nucleotide deletion of a guanine that change the reading frame of the gene extending the translation of PCSK9 by 130 amino acids beyond the reference stop codon (130/692 ∼18%). It is present at a global MAF of 0,0003099% at the Genome Aggregation Database (gnomAD-v4.1.0), no homozygotes reported. It is classified in ClinVar Database as VUS by three submitters (February 2026) and to our knowledge, it has not previously been reported in individuals affected with PCSK9-related disorders in the literature.

This variant affects the C-terminal region of PCSK9, also known as V domain or C-terminal Cys/His-rich domain (CHRD), which is required for PCSK9-mediated degradation of LDLR [18]. This region has a cylindrical shape and consists of three subunits built from six antiparallel β strands connected by internal disulphide bonds that ensure the proper folding of the protein (Fig. 3). It also contains a large number of histidine residues, which may enhance PCSK9-LDLR binding under the acidic pH conditions of endosomes [19].

Fig. 3.

PCSK9 model. A. Structure of native PCSK9 protein (PBD:2PMVW). PCSK9 comprises a prodomain (light blue, aa 31-152), a catalytic domain (green, aa 153-421) and a CHRD domain (aa 453-692) that is composed of three modules. M1: pink, aa 453-529; M2: yellow, aa530-603; M3: red, aa 604-692). B. Structure of mutant PCSK9 protein (Alphafold). The mutated model predicts an extended protein that fails to form the terminal β-sheet present in the wild-type structure, resulting in loss of the native C-terminal folding. C. Loss of Cys678 and Cys679. The mutant sequence is showed in white. Lost residues Cys678 and Cys679 are labeled and the loss of both bisulfide bonds is showed in dark pink.

In silico structural analysis of p.(Val675Leufs*130) predicts an altered protein that exhibits a marked increase in structural disorder at the C-terminal domain (Fig. 3) and the loss of Cys678 and Cys679, two highly conserved residues of the subdomain 3. Both cysteines are critical for beta-sheet stabilization: Cys678 from strand 6 forms a disulfide bond with Cys626 from β strand 2, whereas Cys679 from β strand 6 forms a disulfide bond with Cys608 from β strand 1.

Loss of terminal cysteines of the C-terminal domain has previously been reported as a cause of aberrant folding of PCSK9. Zhao et al. (2006) proved that variant p.(Cys679X) caused aberrant folding of the protein, as evidenced by abnormal migration patterns on SDS-PAGE gel after treatment with dithiothreitol (DTT). Furthermore, expression studies in HEK-293 cells revealed that the Cys679X variant impairs PCSK9 secretion by retaining the abnormal protein in the endoplasmic reticulum (ER) [20]. Taking this information into consideration, together with the predicted destabilization caused by loss of the sheet-like native conformation of subdomain 3 and steric hindrance from protein elongation, as supported by the structural model, we suggest the p.(Val675Leufs*130) variant reported in this study is likely to have the same effect.

3.6. Serum PCSK9 levels

Quantitative analysis of serum PCSK9 was performed using an ELISA assay. The assay included serum samples from the son (IV.5), additional family members and, a subset of normolipidemic controls. Individual IV.5, who carried the PCSK9:[c.137G>T; c.2023del]+[c.137G>T], showed markedly lower circulating PCSK9 protein levels compared with other family members and the control group (Fig. 4). This finding is compatible with reduced circulating PCSK9 associate with this rare genotype; however, it cannot be considered direct functional evidence of loss function for the variant c.2023del.

Fig. 4.

Quantitative analysis of serum PCSK9 levels (ng/mL). Numbers correspond to family members according with Fig. 1. Dashed bars indicate adult controls (A Control) and child controls (C Control), each derived from pooled serum samples from individuals with lipid values within the laboratory-established reference range. Dashed bars indicate individuals carrying the PCSK9 c.137G>T p.(Arg46Leu), whereas empty bars indicate individuals without PCSK9 variants. The solid bar corresponds to individual IV.5, who carried PCSK9 [c.137G>T; c.2023del]+[c.137G>T]. The table below the graphic shows the PCSK9 genotype, serum PCSK9 concentration and statin treatment status for each individual. All variants identified in the individuals are heterozygous, except where homozygosity is explicitly indicated. Abbreviations: hmz: homozygous, WT: wild-type.

3.7. Classification of LDLR c.2479G>A variant integrating available data

LDLR c.2479G>A variant is classified by the ClinGen FH VCEP expert panel as a variant of uncertain significance according to the ClinGen FH VCEP specifications for the LDLR gene, based on the following evidence: (i) PP1_Supporting, variant segregates with the FH phenotype in 3 informative meioses and (ii) PP3_Supporting, REVEL = 0,77, which exceeds the recommended threshold (≥0.75).

However, the combined segregation data and functional analyses performed in the present study indicate that LDLR c.2479G>A variant represents a borderline variant between VUS and likely pathogenic. Under strict application of the ClinGen FH VCEP specifications for LDLR, PS3 was not applied at any strength, because LDL uptake was 74% of wild-type activity and therefore did not meet the <70% threshold required for functional evidence of pathogenicity. This result was therefore interpreted as borderline functional evidence and not as formal PS3 evidence. Segregation analysis showed cosegregation of the variant with the FH phenotype in 12 informative meioses, supporting PP1_Strong. In silico prediction supported a deleterious effect, with a REVEL score of 0.77, supporting PP3_Supporting. The PM2 criterion was not applied because of the variant frequency in population databases. Thus, according to strict ClinGen FH VCEP criteria, the variant remains VUS because does not fulfil formal requirements for likely pathogenic classification. Nevertheless, considering the functional trend, segregation pattern, protein location, and clinical context of this family, we interpret LDLR c.2479G > A variant as a likely pathogenic candidate from an integrated clinical-utility perspective.

4. Discussion

Herein, we describe a family carrying the long-standing controversial LDLR variant, c.2479G>A p.(Val827Ile) which has undergone multiple reclassifications over time. This variant has previously been reported as associated with FH and was historically referred to as “FH New York-5” [21]. Current public database evidence remains heterogeneous: ClinVar aggregates conflicting interpretations across submitters, and a substantial fraction of laboratories still classify the variant as a VUS [22]. According to ClinGen FH Variant Curation Expert Panel LDLR-specific ACMG/AMP specifications, this variant has also been evaluated as VUS, typically supported by limited segregation, computational evidence and high population frequency (e.g., PP1/PP3 depending on dataset strength) [23].

In large reference datasets, LDLR c.2479G>A p.(Val827Ile) variant is rare, with a global MAF of 0,046%, but it is markedly enriched in specific genetic ancestry groups such as Ashkenazi Jews (GnomAD subpopulation frequency reported at ∼0.018, with homozygotes) [24]. The high prevalence of this variant may reflect a founder effect, consistent with studies highlighting that some Jewish subpopulations exhibit an FH prevalence 7 times higher than that of general Caucasian populations. This fact illustrate why allele-frequency evidence for p.(Val827Ile) should be interpreted in the context of ancestry and local FH epidemiology rather than using global prevalence assumptions alone [25].

Moreover, the reported functional consequences of this variant vary across studies, further contributing to uncertainty regarding its pathogenicity. In our CHO-ldlA7 model, LDLR p.(Val827Ile) retained ∼74% of wild-type LDL uptake consistent with a borderline internalization defect rather than a severe loss of function. This result is consistent with a borderline internalization defect and align with previous studies reporting either mild impairment to near-normal behavior depending on assay design and cellular context. For example, a systematic cell-based phenotyping study evaluating rare LDLR missense alleles reported that many variants exhibit context-dependent functional penetrance, and functional readouts can refine cardiovascular risk inference in cohorts [26]. More recently, massively parallel functional mapping of nearly all LDLR missense substitutions generated quantitative sequence–function maps for both LDLR abundance and LDL uptake, providing an additional framework in which some conservative substitutions may score near-normal in specific heterologous assay settings [27].

The c.2479G>A p.(Val827Ile) substitution lies within the FxNPxY internalization motif in the LDLR cytosolic tail, a signal required for efficient packaging of LDLR into clathrin-coated pits through binding to PTB-domain endocytic adaptors [5]. Structural studies have shown that the ARH/LDLRAP1 PTB domain recognizes not only the core FxNPxY residues but also an extended LDLR tail segment (I(−7)xF(−5)xNPxY(0)QK(+2)) that adopts a distinctive “hook-like” conformation, providing a plausible mechanistic basis by which even conservative substitutions at non-core positions could modulate adaptor engagement and endocytic efficiency [28]. Conceptually, variants in the LDLR cytosolic domain can produce class 4 LDLR defects, characterized by preserved receptor surface expression but impairment internalization [6].

An explanation for these discrepancies is the tissue-specific hierarchy and redundancy of LDLR endocytic adaptors. Genetic and cellular studies established that ARH/LDLRAP1 has a tissue-restricted, physiologically critical role in hepatic LDL clearance, whereas LDLR internalization can be maintained in other cell types through alternative adaptor proteins [29]. In non-hepatic contexts, Dab2 can act as a major LDLR sorting adaptor (including HeLa cells and fibroblasts), and in certain settings can function independently of AP-2 and ARH [30]. Moreover, mechanistic dissection of clathrin-mediated uptake demonstrated that LDLR's FxNPxY-type signal is recognized redundantly by Dab2 and ARH, and that simultaneous depletion of both adaptors leads to marked LDLR surface accumulation. Importantly, Dab2 expression is exceptionally low in hepatocytes, which explains why ARH loss produces severe hypercholesterolemia in vivo [31]. CHO-ldlA7 cells (and other non-hepatic models) may systematically underestimate the functional impact of subtle NPxY-motif variants that primary disrupt ARH-dependent endocytic sorting in hepatocytes, leading to false negative or borderline in vitro readouts despite clinically relevant impairment of hepatic LDL clearance. Therefore, given the adaptor redundancy outside the liver and the dominant role of ARH in hepatic LDL clearance, functional interpretation of NPxY-motif variants should prioritize context-aware models rather than relying on a single heterologous system as definitive evidence.

In summary, our data support a modest but consistent functional impact of the LDLR c.2479G>A variant in CHO-ldlA7 cells, and provide an explanation for the differences in retained LDLR activity reported across studies. Functional studies, together with segregation analysis in family members, provide supportive evidence that LDLR c.2479G>A variant may contribute to the FH phenotype observed in this family.

Individuals IV.5 and IV.2 represented apparent exceptions in the segregation analysis, as both carried the LDLR c.2479G>A variant but did not display hypercholesterolemia. Although normolipidemia among individuals with a molecular diagnosis of hypercholesterolemia has been reported to be uncommon (1,6%) [32], incomplete penetrance and substantial inter-individual variability in LDL-C levels are well recognized among carriers of variants in LDLR, APOB or PCSK9 [22,33]. Phenotypic severity is also influenced by variant type, with LDLR null variants generally associated with higher LDL-C concentrations and more severe clinical manifestations than receptor-defective variants [34]. In addition, protective modifiers may attenuate the biochemical phenotype, including co-inheritance of LDLR-lowering variants, such as loss-of-function in APOB or PCSK9 [17], a favourable polygenic background characterized by low burden of LDL-C-raising alleles) [35], and environmental or lifestyle factors that may modulate lipid levels and cardiovascular risk [36].

In individual IV.5, the absence of hypercholesterolemia despite carrying the LDLR variant can be plausibly explained by the concomitant presence of a rare PCSK9 genotype, comprising the established LDL-C-lowering variant c.137G>T p.(Arg46Leu) in homozygosity and the rare c.2023del p.(Val675Leufs*130) variant in heterozygosity, whose effect is predicted to be deleterious but remains inferential. PCSK9 p.(Arg46Leu) variant is a well-established LOF variant that has been consistently associated with lower LDL-cholesterol concentrations, cardiovascular disease and reduced circulating PCSK9 levels [17] [37] [38]. Moreover, immunoassay-based studies support an allele-dose effect, with lower circulating PCSK9 concentrations as the number of p.(Arg46Leu) alleles increases [39]. In IV.5, the co-occurrence of the additional PCSK9 frameshift variant, may further reduces functional PCSK9 availability, providing a plausible explanation for the markedly low serum PCSK9 levels observed. Mechanistically, this combined genetic background would be expected to attenuate LDLR degradation and promote LDLR recycling, thereby contributing to normal LDL-cholesterol concentrations despite the presence of an FH-causing variant. Nevertheless, circulating PCSK9 levels should be interpreted with caution, as they may vary according to genetic background, age, metabolic status, and lipid-lowering therapy [39]. Therefore, although the markedly reduced PCSK9 level observed in IV.5 supports a functional effect of the PCSK9 genotype, comparisons across family members and controls should take these potential confounders into account.

By contrast, in individual IV.2, who carries the familial LDLR variant but not PCSK9 c.2023del, the absence of hypercholesterolemia remains unexplained by the variants identified. This individual also carries APOB c.2630C>T p.(Pro877Leu), a variant previously reported in trans with a pathogenic LDLR variant, where it was hypothesized to attenuate the severity of hypercholesterolemia [40]. Based on the segregation data and available lipid profiles in our family, we cannot exclude a mild LDL-lowering effect of APOB p.(Pro877Leu). However, other relatives carrying both the familial LDLR variant and APOB p.(Pro877Leu) showed hypercholesterolemia, indicating that any potential LDL-lowering effect of this APOB variant is not sufficient, by itself, to counterbalance the hypercholesterolemic effect associated with the LDLR variant. Therefore, the normolipidemic phenotype of IV.2 likely reflects the contribution of additional unmeasured factors, such as a favourable polygenic background, environmental influences, or other genetic modifiers not captured by our analysis.

Therefore, while variant classification must remain evidence-based and interpreted with appropriate caution from a strictly genetic standpoint, we consider that LDLR c.2479G>A p.(Val827Ile) represents a borderline variant that does not strictly meet all thresholds established by the ClinGen FH VCEP, but for which the overall balance of evidence supports a likely pathogenic clinical interpretation. This conclusion is based on the convergence of several lines of evidence: its location within the cytoplasmic internalization domain of LDLR, the borderline but consistent reduction in LDL uptake observed in our functional assay, and the segregation pattern identified in this family. In the context of familial hypercholesterolemia, a preventable disorder in which early identification, treatment adherence, and cascade testing are central to reducing cardiovascular risk, this interpretation also has important clinical implications. Retaining a borderline variant as a VUS may limit its clinical actionability, as this classification can be difficult for patients to interpret and may lead to underestimation of genetic risk, reduced adherence to preventive measures, and lower motivation to communicate risk to relatives. In addition, in some healthcare settings, classification as a VUS may restrict access to advanced lipid-lowering therapies, particularly in patients with statin intolerance, suboptimal response to conventional treatment, or persistent high cardiovascular risk.

5. Conclusion

In conclusion, while cascade testing of the familial pathogenic variant in first-degree relatives remains the recommended and cost-effective strategy for cardiovascular prevention, our findings support extending genetic analysis—particularly when the lipid phenotype is discordant with the expected genotype. Broader NGS-based approaches can uncover potential modifier variants that may help explain the clinical heterogeneity of familial hypercholesterolemia and refine risk estimation, since protective alleles present in an apparently unaffected carrier may not be inherited by descendants, potentially unmasking hypercholesterolemia. This case exemplifies the interpretative challenges introduced by broad sequencing approaches and underscores the need to integrate clinical, biochemical, genomic, and functional evidence to reach robust conclusions. Collectively, our data are compatible with the hypothesis that a rare PCSK9 genotype may act as a putative disease modifier of the phenotypic impact of an LDLR variant that may contribute to the FH phenotype. These findings reinforce the relevance of considering potential genetic modifiers to improve counselling and advance precision medicine in dyslipidemia.

Authors contributions

SRN, analyzed the NGS data, design the study and write the manuscript; PMH contributed to clinical management, and follow-up of the patients, IHM and AHC performed the in silico analysis of variants and review the manuscript, ACA and NGO performed molecular analysis, CRJ and MDA performed biochemical analysis, CRJ performed functional analysis. All authors have approved the final article.

Ethics declaration

Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed.

This study included organ or tissue donors. This study includes human biological material and consent was obtained by donors, or their next of kin or legal representatives, for use in this study and for publication of the article. The samples used in this research were not sourced from executed prisoners or prisoners of conscience.

This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. This study was approved by the COMITÉ DE ÉTICA DE LA INVESTIGACIÓN CON MEDICAMENTOS-HOSPITAL UNIVERSITARIO LA PAZ, MADRID, ESPAÑA. (Approval No. References: 07/639862.9/22; PI 3615)

Financial support

This study has been funded by Instituto de Salud Carlos III (ISCIII), Spain , through the project "PI21/01239″ and PI18/00917 and co-funded by the European Union.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.