Advertisement

Intronic variant screening with targeted next-generation sequencing reveals first pseudoexon in LDLR in familial hypercholesterolemia

Open AccessPublished:February 08, 2021DOI:https://doi.org/10.1016/j.atherosclerosis.2021.02.003

      Highlights

      • Intronic variants in low-density lipoprotein receptor gene (LDLR) can cause familial hypercholesterolemia (FH) but are often neglected.
      • A novel approach to detect these FH-causing variants revealed c.2141-218G > A in LDLR.
      • This variant causes the first ever described occurrence of a pseudo-exon in LDLR.
      • c.2141-218G > A is the deepest known FH-causing variant to date.
      • This emphasizes the need to consider whole LDLR gene analysis in FH.

      Abstract

      Background and aims

      Familial hypercholesterolemia (FH) is caused by pathogenic variants in LDLR, APOB, or PCSK9 genes (designated FH+). However, a significant number of clinical FH patients do not carry these variants (designated FH-). Here, we investigated whether variants in intronic regions of LDLR attribute to FH by affecting pre-mRNA splicing.

      Methods

      LDLR introns are partly covered in routine sequencing of clinical FH patients using next-generation sequencing. Deep intronic variants, >20 bp from intron-exon boundary, were considered of interest once (a) present in FH- patients (n = 909) with LDL-C >7 mmol/L (severe FH-) or after in silico analysis in patients with LDL-C >5 mmol/L (moderate FH-) and b) absent in FH + patients (control group). cDNA analysis and co-segregation analysis were performed to assess pathogenicity of the identified variants.

      Results

      Three unique variants were present in the severe FH- group. One of these was the previously described likely pathogenic variant c.2140+103G>T. Three additional variants were selected based on in silico analyses in the moderate FH- group. One of these variants, c.2141-218G>A, was found to result in a pseudo-exon inclusion, producing a premature stop codon. This variant co-segregated with the hypercholesterolemic phenotype.

      Conclusions

      Through a screening approach, we identified a deep intronic variant causal for FH. This finding indicates that filtering intronic variants in FH- patients for the absence in FH + patients might enrich for true FH-causing variants and suggests that intronic regions of LDLR need to be considered for sequencing in FH- patients.

      Graphical abstract

      Keywords

      1. Introduction

      Familial hypercholesterolemia (FH) is a common autosomal genetic disorder characterized by high levels of low-density lipoprotein cholesterol (LDL-C) resulting in an increased risk for premature cardiovascular disease. FH is caused by pathogenic variants in either the low-density lipoprotein receptor gene (LDLR), apolipoprotein B100 gene (APOB), or proprotein convertase substilisin kexin type 9 gene (PCSK9). However, depending on the severity of the phenotype, an FH-causing mutation is not found in 12–60% of clinical FH patients [
      • Wang J.
      • Dron J.S.
      • Ban M.R.
      • Robinson J.F.
      • McIntyre A.D.
      • Alazzam M.
      • Zhao P.J.
      • Dilliott A.A.
      • Cao H.
      • Huff M.W.
      • Rhainds D.
      • Low-Kam C.
      • Dubé M.-P.
      • Lettre G.
      • Tardif J.-C.
      • Hegele R.A.
      Polygenic versus monogenic causes of hypercholesterolemia ascertained clinically.
      ,
      • Reeskamp L.F.
      • Tromp T.R.
      • Defesche J.C.
      • Grefhorst A.
      • Stroes E.S.
      • Hovingh G.K.
      • Zuurbier L.
      Next-generation sequencing to confirm clinical familial hypercholesterolemia.
      ].
      Multiple causes for the elevated LDL-C levels in these clinical FH variant-negative (FH-) patients have been suggested. For example, high levels of lipoprotein (a) or a polygenic predisposition for high LDL-C are known to mimic FH [
      • Langsted A.
      • Kamstrup P.R.
      • Benn M.
      • Tybjærg-Hansen A.
      • Nordestgaard B.G.
      High lipoprotein(a) as a possible cause of clinical familial hypercholesterolaemia: a prospective cohort study.
      ,
      • Talmud P.J.
      • Shah S.
      • Whittall R.
      • Futema M.
      • Howard P.
      • Cooper J.A.
      • Harrison S.C.
      • Li K.
      • Drenos F.
      • Karpe F.
      • Neil Ha W.
      • Descamps O.S.
      • Langenberg C.
      • Lench N.
      • Kivimaki M.
      • Whittaker J.
      • Hingorani A.D.
      • Kumari M.
      • Humphries S.E.
      Use of low-density lipoprotein cholesterol gene score to distinguish patients with polygenic and monogenic familial hypercholesterolaemia: a case-control study.
      ]. A third possible explanation for the missing inheritable part of FH is the existence of undiscovered genetic variants in known or novel genes involved in LDL-C metabolism. Despite advances in sequencing techniques that allow for large scale genomic analysis in many individuals, this has not yet resulted in the identification of novel FH candidate genes [
      • Berberich A.J.
      • Hegele R.A.
      The complex molecular genetics of familial hypercholesterolaemia.
      ].
      In this context, sequencing of the usually neglected non-coding intronic regions of LDLR with second generation techniques (e.g. whole genome sequencing) has resulted in the identification of new deep intronic variants causal for FH [
      • Reeskamp L.F.
      • Hartgers M.L.
      • Peter J.
      • Dallinga-Thie G.M.
      • Zuurbier L.
      • Defesche J.C.
      • Grefhorst A.
      • Hovingh G.K.
      A deep intronic variant in LDLR in familial hypercholesterolemia.
      ,
      • Kulseth M.A.
      • Berge K.E.
      • Bogsrud M.P.
      • Leren T.P.
      Analysis of LDLR mRNA in patients with familial hypercholesterolemia revealed a novel mutation in intron 14, which activates a cryptic splice site.
      ]. Although intronic variants reside in non-coding regions of the LDLR gene, they can affect regulatory processes such as splicing of LDLR pre-messenger RNA (pre-mRNA). This, in turn, can lead to (partial) exon skipping or (partial) intron retention. In most cases, this results in a shift of the 3-letter coded reading frame that is used for translation of mRNA into proteins. In almost all cases, a frameshift ultimately results in a premature stop codon, leading to shorter, most often nonfunctional LDLR proteins or mRNA that is rapidly degraded by nonsense-mediated decay.
      To further explore the role of deep intronic variants in FH causality, we developed a diagnostic algorithm that takes advantage of the available covered intron sequences generated during diagnostic targeted next-generation sequencing of FH patients. We selected rare deep intronic variants that were present in FH- patients with markedly elevated LDL-C levels and were absent in FH patients with a pathogenic variant in LDLR, APOB, or PCSK9 (FH+). Next, we applied clinical and in silico filtering tools to select those intronic variants that were most likely to cause FH and investigated their effect on mRNA splicing ex vivo and co-segregation with the hypercholesterolemic phenotype in the family of the proband.

      2. Patients and methods

      2.1 Study population and design

      The Amsterdam UMC, location AMC, is the national referral center for DNA diagnostics in patients with different forms of dyslipidemia. For this study, we analyzed the DNA from patients in whom the referring physician requested molecular FH analysis based on their clinical judgement. A molecular diagnosis was made with a targeted next-generation sequencing (NGS) capture (SeqCap easy choice version v. DLv2, Roche NimbleGen Inc., Pleasanton, USA) that covers 29 genes, including FH genes LDLR, APOB, and PCSK9 (see Supplementary Table 1 for full covered gene list). In addition, copy number variant (CNV) analysis was performed for all 29 genes. Patients carrying an FH-causing CNV or variant in LDLR, APOB, or PCSK9 (FH+ patients) formed the control group in this study. The patient group of interest consisted of clinical FH patients in whom no CNV or heterozygous or homozygous likely pathogenic (class 4) or pathogenic (class 5) variants in these genes were found (FH- patients). Supplemental Table 1 depicts whether variants needed to be heterozygous (dominant) or homozygous or compound heterozygous (recessive) for pathogenicity. Variants in CYP7A1 and SCLO1B1 were not taken into account since these are assumed not to affect lipid levels.
      Additional exclusion criteria for the subjects in the FH- group were lipid values obtained during lipid-lowering treatment, LDL-C levels <5 mmol/L, triglycerides ≥3 mmol/L, or missing LDL-C or triglyceride values. Next, all eligible FH-patients were categorized into two groups. Patients with a severe FH phenotype (LDL-C levels ≥7 mmol/L and triglycerides <1.5 mmol/L) were defined as “severe FH” and all other clinical FH patients (LDL-C ≥5 mmol/L and triglycerides <3 mmol/L) were defined as “moderate FH”. The complete study design is summarized in Fig. 1. All included patients gave written informed consent for use of their clinical and genetic data for research purposes. The Medical Ethics Review Committee of the Amsterdam UMC, location AMC, provided a waiver for the re-use of the patients’ clinical and genetic data in the current study (reference ID: W20_490 # 20.542).
      Fig. 1
      Fig. 1Flowchart of study design.
      Intronic variants present in FH patients without an FH-causing variant (FH-) that were not present in FH patients with an FH-causing variant (FH+) were filtered for having a minor allele frequency (MAF) of ≤0.3% in reference cohorts (e.g., gnomAD). Next, two groups of patients were selected, those with severe FH- (low-density lipoprotein cholesterol [LDL-C] ≥7 mmol/L and triglycerides [TG] levels ≤1.5 mmol/L) and those with moderate FH- (LDL-C ≥5 mmol/L and TG <3 mmol/L). Intronic variants in these two groups were further analyzed in silico and/or with cDNA analysis.

      2.2 Genetic analysis

      DNA was isolated from whole blood using EDTA blood withdrawal tubes and the Gentra Puregene kit (Qiagen, Hilden, Germany) according to the manufacturers’ protocols. Variants in exons and flanking nucleotides (20 base pairs [bp]) of 27 dyslipidemia-causing genes of the NGS panel were assessed for pathogenicity according to the American College of Medical Genetics and Genomics (ACMG) guidelines by two trained geneticists [
      • Richards S.
      • Aziz N.
      • Bale S.
      • Bick D.
      • Das S.
      • Gastier-Foster J.
      • Grody W.W.
      • Hegde M.
      • Lyon E.
      • Spector E.
      • Voelkerding K.
      • Rehm H.L.
      Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of medical genetics and genomics and the association for molecular pathology.
      ]. For LDLR, reference genome NM_000527.4 (GrCH37) was used. Intronic regions of LDLR were partly covered by the NGS capture and a complete overview of the covered regions is provided in Supplementary Table 2 and Supplementary Figure 1.

      2.3 Identification of variants of interest

      Variants in the LDLR intronic regions (>20 bp from exon) that were present in FH- patients, but not in the control group of FH+ patients, were selected for further analysis. We selected variants with a minor allele frequency (MAF) ≤0.3% [
      • Richards S.
      • Aziz N.
      • Bale S.
      • Bick D.
      • Das S.
      • Gastier-Foster J.
      • Grody W.W.
      • Hegde M.
      • Lyon E.
      • Spector E.
      • Voelkerding K.
      • Rehm H.L.
      Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of medical genetics and genomics and the association for molecular pathology.
      ] in all ethnic populations shown in the publicly available reference databases gnomAD (Both Exome and Genome datasets, release 2.1; gnomad. broadinstitute.org [
      • Karczewski K.J.
      • Francioli L.C.
      • Tiao G.
      • Cummings B.B.
      • Alföldi J.
      • Wang Q.
      • Collins R.L.
      • Laricchia K.M.
      • Ganna A.
      • Birnbaum D.P.
      • Gauthier L.D.
      • Brand H.
      • Solomonson M.
      • Watts N.A.
      • Rhodes D.
      • Singer-Berk M.
      • England E.M.
      • Seaby E.G.
      • Kosmicki J.A.
      • Walters R.K.
      • Tashman K.
      • Farjoun Y.
      • Banks E.
      • Poterba T.
      • Wang A.
      • Seed C.
      • Whiffin N.
      • Chong J.X.
      • Samocha K.E.
      • Pierce-Hoffman E.
      • Zappala Z.
      • O'Donnell-Luria A.H.
      • Minikel E.V.
      • Weisburd B.
      • Lek M.
      • Ware J.S.
      • Vittal C.
      • Armean I.M.
      • Bergelson L.
      • Cibulskis K.
      • Connolly K.M.
      • Covarrubias M.
      • Donnelly S.
      • Ferriera S.
      • Gabriel S.
      • Gentry J.
      • Gupta N.
      • Jeandet T.
      • Kaplan D.
      • Llanwarne C.
      • Munshi R.
      • Novod S.
      • Petrillo N.
      • Roazen D.
      • Ruano-Rubio V.
      • Saltzman A.
      • Schleicher M.
      • Soto J.
      • Tibbetts K.
      • Tolonen C.
      • Wade G.
      • Talkowski M.E.
      • Neale B.M.
      • Daly M.J.
      • MacArthur D.G.
      The mutational constraint spectrum quantified from variation in 141,456 humans.
      ]) and GoNL (5th release [
      • Francioli L.C.
      • Menelaou A.
      • Pulit S.L.
      • Van Dijk F.
      • Palamara P.F.
      • Elbers C.C.
      • Neerincx P.B.T.
      • Ye K.
      • Guryev V.
      • Kloosterman W.P.
      • Deelen P.
      • Abdellaoui A.
      • Van Leeuwen E.M.
      • Van Oven M.
      • Vermaat M.
      • Li M.
      • Laros J.F.J.
      • Karssen L.C.
      • Kanterakis A.
      • Amin N.
      • Hottenga J.J.
      • Lameijer E.W.
      • Kattenberg M.
      • Dijkstra M.
      • Byelas H.
      • Van Setten J.
      • Van Schaik B.D.C.
      • Bot J.
      • Nijman I.J.
      • Renkens I.
      • Marschall T.
      • Schönhuth A.
      • Hehir-Kwa J.Y.
      • Handsaker R.E.
      • Polak P.
      • Sohail M.
      • Vuzman D.
      • Hormozdiari F.
      • Van Enckevort D.
      • Mei H.
      • Koval V.
      • Moed M.H.
      • Van Der Velde Kj
      • Rivadeneira F.
      • Estrada K.
      • Medina-Gomez C.
      • Isaacs A.
      • McCarroll S.A.
      • Beekman M.
      • Mde Craen A.J.
      • Suchiman H.E.D.
      • Hofman A.
      • Oostra B.
      • Uitterlinden A.G.
      • Willemsen G.
      • Platteel M.
      • Veldink J.H.
      • Van Den Berg L.H.
      • Pitts S.J.
      • Potluri S.
      • Sundar P.
      • Cox D.R.
      • Sunyaev S.R.
      • Den Dunnen J.T.
      • Stoneking M.
      • De Knijff P.
      • Kayser M.
      • Li Q.
      • Li Y.
      • Du Y.
      • Chen R.
      • Cao H.
      • Li N.
      • Cao S.
      • Wang J.
      • Bovenberg J.A.
      • Pe’er I.
      • Slagboom P.E.
      • Van Duijn C.M.
      • Boomsma D.I.
      • Van Ommen G.J.B.
      • De Bakker P.I.W.
      • Swertz M.A.
      • Wijmenga C.
      Whole-genome sequence variation, population structure and demographic history of the Dutch population.
      ]). Variants passing internal sequencing quality filters were considered for this analysis, as were variants with low coverage (<30x coverage). In the severe FH- group, all identified unique intronic variants were candidates for follow-up, irrespective of in silico splicing prediction. In the moderate FH- group, identified variants were first subjected to in silico splicing analyses and only variants that were strongly predicted to affect splicing were selected for cDNA analysis (Fig. 1). One FH+ patient carrying the previously identified deep intronic LDLR variant (c.2140+103G>T) was annotated as being FH- to serve as positive control for our variant selection algorithm [
      • Reeskamp L.F.
      • Hartgers M.L.
      • Peter J.
      • Dallinga-Thie G.M.
      • Zuurbier L.
      • Defesche J.C.
      • Grefhorst A.
      • Hovingh G.K.
      A deep intronic variant in LDLR in familial hypercholesterolemia.
      ].

      2.4 In silico splicing assessment

      The identified intronic LDLR variants that were unique for the FH- group were assessed in silico for their effect on splicing of LDLR pre-mRNA using seven tools: SpliceSiteFinder-like (SSF) [
      • Zhang M.Q.
      Statistical features of human exons and their flanking regions.
      ], MaxEntScan (ME) [
      • Yeo G.
      • Burge C.B.
      Maximum entropy modeling of short sequence motifs with applications to RNA splicing signals.
      ], NNSPLICE (NN) [
      • Reese M.G.
      • Eeckman F.H.
      • Kulp D.
      • Haussler D.
      Improved splice site detection in Genie.
      ], and GeneSplicer (GS) [
      • Pertea M.
      • Lin X.
      • Salzberg S.L.
      GeneSplicer: a new computational method for splice site prediction.
      ], available in the Alamut Visual Software package (version 2.11; Interactive Biosoftware, Rouen, France), and TraP (Version 2) [
      • Gelfman S.
      • Wang Q.
      • McSweeney K.M.
      • Ren Z.
      • La Carpia F.
      • Halvorsen M.
      • Schoch K.
      • Ratzon F.
      • Heinzen E.L.
      • Boland M.J.
      • Petrovski S.
      • Goldstein D.B.
      Annotating pathogenic non-coding variants in genic regions.
      ], SPANR [
      • Xiong H.Y.
      • Alipanahi B.
      • Lee L.J.
      • Bretschneider H.
      • Merico D.
      • Yuen R.K.C.
      • Hua Y.
      • Gueroussov S.
      • Najafabadi H.S.
      • Hughes T.R.
      • Morris Q.
      • Barash Y.
      • Krainer A.R.
      • Jojic N.
      • Scherer S.W.
      • Blencowe B.J.
      • Frey B.J.
      The human splicing code reveals new insights into the genetic determinants of disease.
      ], and SpliceAI [
      • Jaganathan K.
      • Kyriazopoulou Panagiotopoulou S.
      • McRae J.F.
      • Darbandi S.F.
      • Knowles D.
      • Li Y.I.
      • Kosmicki J.A.
      • Arbelaez J.
      • Cui W.
      • Schwartz G.B.
      • Chow E.D.
      • Kanterakis E.
      • Gao H.
      • Kia A.
      • Batzoglou S.
      • Sanders S.J.
      • Farh K.K.-H.
      Predicting splicing from primary sequence with deep learning.
      ]. The SpliceAI in silico assessment was based on comparison of our identified variants with a predefined list of variants and scores provided by Illumina (San Diego, CA, USA; obtained through personal communication). A variant was annotated as being of interest when at least two of the four prediction tools in Alamut showed a score of ≥75% of the score range of that specific tool (SSF: 0–100; ME: 0–16; NN 0–1; GS 0–21, for acceptor splice sites and SSF: 0–100; ME: 0–12; NN 0–1; GS 0–24, for donor splice sites) and an increase ≥2% compared to the wild type signal. A variant was also considered of interest when the splice score was above the given cut-offs of TraP (≥0.459), SPANR (≥5 for absolute difference of percentage spliced), or SpliceAI (≥0.2).

      2.5 Splicing confirmation

      RNA was isolated from whole blood using PAXgene tubes (BD Diagnostics, Franklin Lakes, NJ USA) and the PAXgene RNA isolation kit according to the manufacture protocol (BD Diagnostics, Franklin Lakes, NJ USA) to assess the effect of a variant on splicing. Subsequently, cDNA was generated using the SensiFAST™ cDNA Synthesis Kit (Bioline, London, United Kingdom) or with the Superscript III first-strand synthesis system for RT-PCR (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. LDLR PCR products were amplified using specific forward and reverse primers (see Supplementary Table 2) after which the splicing product was visualized on a 1% agarose gel. Bands of interest were excised, purified (QIAquick Gel Extraction Kit [Qiagen, Hilden, Germany]; BigDye™ Terminator v1.1 Cycle Sequencing Kit [Applied Biosystems, Foster City, CA, USA]) and Sanger sequenced.

      2.6 Statistical analysis

      All normally distributed data are presented as mean ± SD, all non-normally distributed data as median [Inter Quartile Range], and counts as number (%). The mean number of variants per patient group was compared using a two sample t-test or one-way ANOVA. A p-value <0.05 was considered statistically significant. All statistical analyses were performed in R (version 3.6.1; The R Foundation, Viena, Austria) and Rstudio (version February 1, 1335; RStudio, Inc. Boston, MA, USA).

      3. Results

      3.1 Intronic variant detection

      We included a total of 909 FH- patients in our analysis. 38 patients had severe FH (LDL ≥7 and TG <1.5 mmol/L) and 871 patients had moderate FH (LDL ≥5 and TG <3 mmol/L). The control group consisted of 690 patients who carried a likely pathogenic or pathogenic variant in either LDLR, APOB, or PCSK9. Clinical characteristics of all patients are depicted in Table 1.
      Table 1Characteristics of FH patient groups.
      All FH- subjectsSevere FH- (LDL-C ≥7 & TG < 1.5 mmol/L)Moderate FH- (LDL-C ≥5 & TG < 3 mmol/L)FH+ controls
      No. of patients90938871690
      Females (n (%))560 (62)28 (74)532 (61)391 (57)
      Age, years (mean (SD))54.6 (11.6)56.1 (9.57)54.5 (11.7)41.8 (16.9)
      Total cholesterol, mmol/L (median [IQR])8.2 [7.7–8.9]9.7 [9.2–10.2]8.2 [7.7–8.8]8.7 [7.8–9.9]
      LDL-C, mmol/L (median [IQR])6.0 [5.5–6.6]7.5 [7.2–7.9]6.0 [5.5–6.5]6.7 [5.6–7.8]
      HDL-C, mmol/L (median [IQR])1.7 [1.3–2.2]1.2 [1.1–1.4]1.7 [1.4–2.2]1.2 [1.1–1.5]
      Triglycerides, mmol/L (median [IQR])1.4 [1.2–1.7]1.52 [1.3–1.9]1.4 [1.2–1.7]1.5 [0.9–2.1]
      BMI, kg/m2 (mean (SD))26.4 (4.0)26.8 (5.4)26.4 (3.9)26.3 (5.0)
      History of ASCVD (n (%))190 (20.9)7 (18.4)183 (21.0)80 (11.5)
      Diabetes mellitus (n (%))42 (4.6)1 (2.6)41 (4.7)19 (2.7)
      No. of unique deep intronic variants673640
      ASCVD = atherosclerotic cardiovascular disease (defined as history of myocardial infarction, angina pectoris, percutaneous coronary intervention, coronary artery bypass grafting, peripheral artery disease, stroke).
      On average, 54 ± 33% and 36 ± 29% of the eighteen LDLR introns were covered ≥1 and ≥ 30 times, respectively, by our NGS capture (Supplementary Table 2). The average number of deep intronic LDLR variants per patient was 29 ± 8.2. This number did not differ between FH+ and FH- patients (29.3 ± 8.5 vs 28.7 ± 8, p = 0.185). A total of 64 and 196 different intronic variants were identified in severe FH- and moderate FH- patients, respectively. The average number of intronic variants in patients groups did not differ per FH+ genotype (i.e., LDLR, APOB, PCSK9) or FH- phenotype (i.e., severe FH- and moderate FH-) (Supplementary Table 4 and Supplementary Figure 2). The highest number of intronic variants was identified in intron 11, 12 and 15 (Supplementary Figure 3).

      3.1.1 Unique FH- intronic variant detection

      Three unique intronic variants were detected in the severe FH- group, including the previously described c.2140+103G>T variant, which is considered to cause FH [
      • Reeskamp L.F.
      • Hartgers M.L.
      • Peter J.
      • Dallinga-Thie G.M.
      • Zuurbier L.
      • Defesche J.C.
      • Grefhorst A.
      • Hovingh G.K.
      A deep intronic variant in LDLR in familial hypercholesterolemia.
      ] and served as a positive control in our study. The other two novel variants were in silico predicted to be potentially damaging by only one of the 7 used tools (Table 2) and cDNA analysis confirmed that they had no effect on LDLR mRNA splicing.
      Table 2Intronic variants of interest.
      VariantGroupLocationAllele frequency in total FH mutation-negative cohort (%)Maximum allele frequency reference population (gnomAD)
      Maximum allele frequency reported is the largest reported allele frequency in any ethnic population in gnomAD [9], ExAC [9], or GoNL [14]. FH, familial hypercholesterolemia; GS, GeneSplicer; SSF, SpliceSiteFinder; ME, MaxEntScan; NN, NNSPLICE.
      In silico analysiscDNA sequencing results
      c.313+277C>T (rs971920612)Severe FHIntron 30.05500.0009Predicted 19.5% decrease of donor splice site at c.313 + 227 according to GSNo effect on splicing
      c.694+25C>T (rs199540175)Severe FHIntron 40.11000.00221.4% increase of donor splice signal at c.694 according to GSNo effect on splicing
      c.2140+103G>TSevere FHIntron 140.055063% chance of a donor splice signal at c.2140 + 97 according to SpliceAI.

      7.1–100% increase in donor splice signal at c.2140 + 97 according to SSF, ME, NN, and GS
      97 nucleotides insertion, leading to frameshift and premature stop codon in exon 15 of LDLR [6]
      c.1187-96C>T (rs931988884)Moderate FHIntron 80.05504.3 and 10.9% increase of acceptor splice signal at c.1187-84 according to SSF and NN, respectivelyNo effect on splicing
      c.1587-308C>T (rs866311080)Moderate FHIntron 100.05500.000113.9% increase of donor splice signal at c.1587-310 according to SSF.

      62.4% chance of changing RNA transcript according to TraP
      No participation
      c.2141-218G>A (rs991805047)Moderate FHIntron 140.05503.8–100% increase of donor splice signal at c.2141-221

      34% chance of a donor splice signal at c.2141-221 according to SpliceAI
      132 nucleotides pseudo-exon with premature stop codon created in intron 14
      Identified deep intronic variants in LDLR with potential effect on LDLR mRNA splicing. Three unique variants were identified in the severe FH mutation-negative group and three variants in the moderate FH mutation-negative group.
      a Maximum allele frequency reported is the largest reported allele frequency in any ethnic population in gnomAD [9], ExAC [9], or GoNL [14]. FH, familial hypercholesterolemia; GS, GeneSplicer; SSF, SpliceSiteFinder; ME, MaxEntScan; NN, NNSPLICE.
      Among moderate FH patients, a total of 63 unique variants were identified, of which 3 were identified as being of interest given the observed effect in 2 different in silico analyses (Table 2, Supplementary Table 5). The first variant, c.1187-96C>T, was deemed to affect splicing by two in silico tools by addition of an acceptor splice signal at c.1187–84. We performed cDNA analysis and found no effect on splicing. Unfortunately, the patient carrying the second variant, c.1587-308C>T, did not provide informed consent for RNA isolation for this study. The third variant, c.2141-218G>A, was predicted to have a 34% higher chance of causing a cryptic donor splice signal at position c.2141–221 compared to the wild type variant according to SpliceAI. The in silico prediction tools SSF, ME, NN, GS, reported a 5.5%, 45%, 3.8% and a 100% increase in donor splice signal at c.2141–221, respectively (Table 2). Subsequent electrophoresis of the cDNA derived from the carrier of this variant showed an additional larger band on agarose (Fig. 2A). Sequencing of this band showed an insertion of 132 base pairs between exon 14 and exon 15. This insertion starts at a predicted cryptic acceptor splice site at c.2141–352 and spans up to the predicted cryptic donor splice site that was created by the deep intronic variant at c.2141–221 (Fig. 2B). Thus, the c.2141-218G>A deep intronic variant causes partial intron retention and thereby results in the transcription of a pseudoexon. This insertion included a premature stop codon near the end of the insertion, which likely leads to protein truncation. The patient carrying this variant presented with LDL-C levels of 5.7 mmol/L, corresponding with the 98th percentile for age and gender in the Netherlands (available via www.lipidtools.com) [
      • Balder J.W.
      • de Vries J.K.
      • Nolte I.M.
      • Lansberg P.J.
      • Kuivenhoven J.A.
      • Kamphuisen P.W.
      Lipid and lipoprotein reference values from 133,450 Dutch Lifelines participants: age- and gender-specific baseline lipid values and percentiles.
      ].
      Fig. 2
      Fig. 2Agarose gel with cDNA polymerase chain reaction and schematic overview of pseudo-exon occurrence when c.2141-218>A variant is present.
      (A) cDNA polymerase chain reaction (PCR) products for the patients in the pedigree in . Patient (ID) corresponds to the IDs used in this pedigree. All patients with the FH phenotype (1–3) and carrying the c.2141-218G>A variant show an additional PCR product. Variant c.2141-218G>A creates a donor splice site at c.2141–211 and an acceptor splice site at c.2141–353 in intron 14, resulting in a pseudo-exon inclusion in the mRNA. This pseudo-exon has a length of 132 base pairs and predicts a premature stop codon near exon 15 (B). Patient 4, and a non-related no-FH control, did not carry this intronic variant and did not show an additional PCR product like her family members (A).

      3.2 Co-segregation analysis of the c.2141-218G>A variant

      Next, we evaluated whether other carriers of this variant in the proband's family also presented with autosomal dominant hypercholesterolemia. The brother of the proband and his son were both carrier of the c.2141-218G>A variant and presented a hypercholesterolemic phenotype (i.e., LDL cholesterol above 99th and 92nd percentile for age and gender, respectively; Fig. 3). In contrast, the niece of the proband was not carrier of the identified variant and had no hypercholesterolemia. The proband had no children, and both her parents had deceased, which, unfortunately precluded us from further family analysis. We subsequently checked all dyslipidemic patients that were sequenced with our NGS platform from May 2016 (n = 9675) and, although this variant was very well covered by our platform (minimal coverage of 20 times in all patients and mean ± SD coverage of 56 ± 15 times), we found no other index patients carrying this variant.
      Fig. 3
      Fig. 3Pedigree of family with c.2141-218G>A variant.
      The proband (ID: 1), brother (ID: 2), and nephew (ID: 3) carry the variant c.2141-218G>A and all have the FH phenotype (here defined as low-density lipoprotein levels > 95th percentile for age and sex). Her niece (ID: 4) was not found to carry this variant and did not have hypercholesterolemia.

      4. Discussion

      We used a novel approach to detect potential FH-causing deep intronic variants in LDLR and identified a variant, c.2141-218G>A, which causes the inclusion of a pseudoexon between exon 14 and exon 15 in the LDLR mRNA, which includes a premature stop codon. This variant fully segregated with the hypercholesterolemic phenotype in the small family of the identified proband. To our knowledge, the c.2141-218G>A variant described here is the deepest known FH-causing variant at 218 base pairs from intron 14 – exon 15 boundary in LDLR, and currently the only variant described to result in a pseudoexon in this gene. Although we identified this specific variant in one proband only, our study suggests that intronic regions are of interest in patients with clinical but not yet genetically confirmed FH, as these introns may harbor yet undiscovered FH-causing variants.
      Disruption of the LDLR splicing machinery is a frequently observed pathogenic mechanism in FH (~10% of FH-causing variants [
      • Usifo E.
      • Leigh S.E.A.
      • Whittall R.A.
      • Lench N.
      • Taylor A.
      • Yeats C.
      • Orengo C.A.
      • Martin A.C.R.
      • Celli J.
      • Humphries S.E.
      Low-density lipoprotein receptor gene familial hypercholesterolemia variant database: update and pathological assessment.
      ], with almost all variants located in exons or near exon-intron boundaries [
      • Calandra S.
      • Tarugi P.
      • Bertolini S.
      Altered mRNA splicing in lipoprotein disorders.
      ]). Nonetheless, also deeper intronic variants can affect splicing or pre-mRNA secondary structure. Today, in the HGMD database (professional 2020.3) only two pathogenic variants are known beyond ±20 base pairs, strikingly also in intron 14. Deep variants are, however, hard to investigate because of their size and numerous nonfunctional variants. For example, the size of intronic regions in LDLR is on average 2304 base pairs and, while only achieving a mean coverage of 36%, we already detected 29 variants on average in each patient.
      To select potential deeper intronic pathogenic variants, we used a unique method that took advantage of all available intronic data that is generally not analyzed during diagnostic targeted next-generation sequencing. From these data, we selected intronic variants that were absent in a control group of FH+ patients. This selection method yields several advantages. First, filtering against FH + patients yielded greatly reduced numbers and minimized the number of cDNA analyses needed to be performed. Second, this will be an ever increasing refinement of this method, as the number of confirmed FH patients increases.
      Our choice to define two groups of clinical FH patients (moderate and severe FH-) was based on the fact that those patients with the most severe FH phenotype have the highest chance of carrying an FH-causing variant [
      • Wang J.
      • Dron J.S.
      • Ban M.R.
      • Robinson J.F.
      • McIntyre A.D.
      • Alazzam M.
      • Zhao P.J.
      • Dilliott A.A.
      • Cao H.
      • Huff M.W.
      • Rhainds D.
      • Low-Kam C.
      • Dubé M.-P.
      • Lettre G.
      • Tardif J.-C.
      • Hegele R.A.
      Polygenic versus monogenic causes of hypercholesterolemia ascertained clinically.
      ,
      • Reeskamp L.F.
      • Tromp T.R.
      • Defesche J.C.
      • Grefhorst A.
      • Stroes E.S.
      • Hovingh G.K.
      • Zuurbier L.
      Next-generation sequencing to confirm clinical familial hypercholesterolemia.
      ]. Since in silico tools do not have a 100% sensitivity, we aimed not to miss potentially damaging variants in the severe FH- patients and decided to investigate all identified variants in these patients for their effect on splicing ex vivo. Vice versa, since hypercholesterolemia with LDL-C between 5 and 7 mmol/L is not always caused by pathogenic variants, but by other causes such as polygenic FH [
      • Talmud P.J.
      • Shah S.
      • Whittall R.
      • Futema M.
      • Howard P.
      • Cooper J.A.
      • Harrison S.C.
      • Li K.
      • Drenos F.
      • Karpe F.
      • Neil Ha W.
      • Descamps O.S.
      • Langenberg C.
      • Lench N.
      • Kivimaki M.
      • Whittaker J.
      • Hingorani A.D.
      • Kumari M.
      • Humphries S.E.
      Use of low-density lipoprotein cholesterol gene score to distinguish patients with polygenic and monogenic familial hypercholesterolaemia: a case-control study.
      ], secondary causes [
      • Vodnala D.
      • Rubenfire M.
      • Brook R.D.
      Secondary causes of dyslipidemia.
      ], or high lipoprotein (a) levels [
      • Langsted A.
      • Kamstrup P.R.
      • Benn M.
      • Tybjærg-Hansen A.
      • Nordestgaard B.G.
      High lipoprotein(a) as a possible cause of clinical familial hypercholesterolaemia: a prospective cohort study.
      ], we applied in silico tools to select variants of interest. The utility of these in silico tools was confirmed in our data, as in the severe FH- group, only the previously described probably pathogenic c.2140+103G>T variant [
      • Reeskamp L.F.
      • Hartgers M.L.
      • Peter J.
      • Dallinga-Thie G.M.
      • Zuurbier L.
      • Defesche J.C.
      • Grefhorst A.
      • Hovingh G.K.
      A deep intronic variant in LDLR in familial hypercholesterolemia.
      ] was in silico predicted to lead to aberrant splicing, in contrast to the two other variants (c.313+277C>T and c.694+25C>T).
      Our study has several limitations. First, the stringency of in silico criteria remains elusive. In our study, we used 7 different in silico tools and applied scoring cut-offs to identify variants of interest. This selection may have resulted in exclusion of functional variants. Moreover, these in silico tools may not be optimal for predicting the effect of deep intronic variants, as most tools are validated for variants close to the exon boundaries [
      • Holla Ø.L.
      • Nakken S.
      • Mattingsdal M.
      • Ranheim T.
      • Berge K.E.
      • Defesche J.C.
      • Leren T.P.
      Effects of intronic mutations in the LDLR gene on pre-mRNA splicing: comparison of wet-lab and bioinformatics analyses.
      ,
      • Jian X.
      • Boerwinkle E.
      • Liu X.
      In silico tools for splicing defect prediction: a survey from the viewpoint of end users.
      ]. In our study, the most recent developed tool (SpliceAI [
      • Jaganathan K.
      • Kyriazopoulou Panagiotopoulou S.
      • McRae J.F.
      • Darbandi S.F.
      • Knowles D.
      • Li Y.I.
      • Kosmicki J.A.
      • Arbelaez J.
      • Cui W.
      • Schwartz G.B.
      • Chow E.D.
      • Kanterakis E.
      • Gao H.
      • Kia A.
      • Batzoglou S.
      • Sanders S.J.
      • Farh K.K.-H.
      Predicting splicing from primary sequence with deep learning.
      ]) was the only one to accurately predict the pathogenicity of the c.2141-218G>A variant. Second, it is possible that we discarded variants that caused compound heterozygous FH in the FH+ group. Third, we still lack information on the deeper non-covered parts of the introns (Supplementary Table 2). Fourth, it is possible that our mRNA analysis was not sufficient to detect large pseudoexons or intron retentions. Lastly, we did not examine whether the variant c.2141-218G>A indeed results in the absence or non-function of LDLR proteins. However, this variant was shown to result in a premature stop codon further downstream in the included pseudoexon, before the 5′-end of exon 15 (Fig. 2B) and is therefore believed to result in truncation of the translated LDLR protein [
      • Gent J.
      • Braakman I.
      Low-density lipoprotein receptor structure and folding.
      ] or the LDLR mRNA will be targeted to the nonsense-mediated mRNA decay pathway [
      • Kuzmiak H.A.
      • Maquat L.E.
      Applying nonsense-mediated mRNA decay research to the clinic: progress and challenges.
      ]. The latter is suspected, as we were able to detect only small amounts of the mutated mRNA.
      To assess deep intronic variants in a clinical setting, and also assess currently uncovered regions by NGS of LDLR, NGS panels could be expanded by including all intronic LDLR regions. Next, additional filter steps, such as used in the current study (i.e., filtering for unique variants against FH+ patients and in silico assessment) are needed to reduce the number of variants that need to be assessed by mRNA sequencing (i.e., real time PCR or RNA-seq). This approach may result in identification of deep intronic variants, as well as an assessment of their effect on pre-mRNA splicing, with only significantly increasing costs for RNA isolation and analysis.
      In conclusion, we developed a novel selection method to identify deep intronic LDLR variants that potentially cause FH. We were able to discover a new variant in intron 14 (c.2141-218G>A) and describe for the first time a pseudoexon in the LDLR mRNA that likely results in FH. This finding emphasizes the need to consider more extensive LDLR analysis in patients in whom DNA sequencing fails to identify a molecular basis for the FH phenotype.

      Financial support

      This study was funded by a ZonMW grant (VIDI No. 016.156.445 ) obtained by G.K. Hovingh. The funder (ZonMW) was not involved in the design, data collection, analysis, interpretation or any other aspect of this study.

      CRediT authorship contribution statement

      Laurens F. Reeskamp: Writing - original draft, Formal analysis, Methodology, Conceptualization. Manon Balvers: Writing - original draft, Writing - review & editing, Formal analysis, Methodology. Jorge Peter: Writing - original draft, Formal analysis. Laura van de Kerkhof: Writing - original draft, Formal analysis. Lisette N. Klaaijsen: Writing - original draft, Formal analysis. Mahdi M. Motazacker: Writing - original draft, Methodology, Conceptualization. Aldo Grefhorst: Writing - original draft, Writing - review & editing. Natal A.W. van Riel: Writing - original draft, Writing - review & editing. G. Kees Hovingh: Writing - original draft, Funding acquisition, Writing - review & editing, Conceptualization. Joep C. Defesche: Writing - original draft, Writing - review & editing, Conceptualization. Linda Zuurbier: Conceptualization, Methodology, Writing - original draft.

      Declaration of competing interest

      The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: LFR is co-founder of Lipid Tools. GKH has served as consultant and speaker for biotechnology and pharmaceutical companies that develop molecules that influence lipoprotein metabolism, including Regeneron, Aege-rion Pfizer, Merck, KOWA, Sanofi, and Amgen; has served as principal investigator for clinical trials conducted with a.o. Amgen, Sanofi, Eli Lilly, Novartis, Kowa, Genzyme, Cerenis, Pfizer, Dezima, and AstraZeneca; has received research grants from ZonMW (Vidi grant [016.156.445]), Klinkerpad fonds, the European Union, Amgen, Sanofi, AstraZeneca, Aege-rion, and Synageva; has received honoraria and investigator fees (to the Department of Vascular Medicine) for sponsor-driven studies and lectures for companies with approved lipid-lowering therapy in the Netherlands; and is partly employed by Novo Nordisk AS, Copenhagen, Denmark (0.7FTE) and the Amsterdam UMC, Amsterdam, the Netherlands (0.3FTE).

      Acknowledgements

      We would like to acknowledge J.F. Los for her work as a genetic field worker involved in the expansion of the included family and we would like to thank all participants in this study for their participation. This work was partly funded by a grant from ZonMW (Vidi grant [016.156.445]).

      Appendix A. Supplementary data

      The following are the Supplementary data to this article:

      References

        • Wang J.
        • Dron J.S.
        • Ban M.R.
        • Robinson J.F.
        • McIntyre A.D.
        • Alazzam M.
        • Zhao P.J.
        • Dilliott A.A.
        • Cao H.
        • Huff M.W.
        • Rhainds D.
        • Low-Kam C.
        • Dubé M.-P.
        • Lettre G.
        • Tardif J.-C.
        • Hegele R.A.
        Polygenic versus monogenic causes of hypercholesterolemia ascertained clinically.
        Arterioscler. Thromb. Vasc. Biol. 2016; 36: 2439-2445
        • Reeskamp L.F.
        • Tromp T.R.
        • Defesche J.C.
        • Grefhorst A.
        • Stroes E.S.
        • Hovingh G.K.
        • Zuurbier L.
        Next-generation sequencing to confirm clinical familial hypercholesterolemia.
        Eur J Prev Cardiol. 2020; 204748732094299
        • Langsted A.
        • Kamstrup P.R.
        • Benn M.
        • Tybjærg-Hansen A.
        • Nordestgaard B.G.
        High lipoprotein(a) as a possible cause of clinical familial hypercholesterolaemia: a prospective cohort study.
        Lancet Diabetes Endocrinol. 2016; 4: 577-587
        • Talmud P.J.
        • Shah S.
        • Whittall R.
        • Futema M.
        • Howard P.
        • Cooper J.A.
        • Harrison S.C.
        • Li K.
        • Drenos F.
        • Karpe F.
        • Neil Ha W.
        • Descamps O.S.
        • Langenberg C.
        • Lench N.
        • Kivimaki M.
        • Whittaker J.
        • Hingorani A.D.
        • Kumari M.
        • Humphries S.E.
        Use of low-density lipoprotein cholesterol gene score to distinguish patients with polygenic and monogenic familial hypercholesterolaemia: a case-control study.
        Lancet. 2013; 381: 1293-1301
        • Berberich A.J.
        • Hegele R.A.
        The complex molecular genetics of familial hypercholesterolaemia.
        Nat. Rev. Cardiol. 2018; 250: 1-12
        • Reeskamp L.F.
        • Hartgers M.L.
        • Peter J.
        • Dallinga-Thie G.M.
        • Zuurbier L.
        • Defesche J.C.
        • Grefhorst A.
        • Hovingh G.K.
        A deep intronic variant in LDLR in familial hypercholesterolemia.
        Circ Genomic Precis Med. 2018; 11: 1-6
        • Kulseth M.A.
        • Berge K.E.
        • Bogsrud M.P.
        • Leren T.P.
        Analysis of LDLR mRNA in patients with familial hypercholesterolemia revealed a novel mutation in intron 14, which activates a cryptic splice site.
        J. Hum. Genet. 2010; 55: 676-680
        • Richards S.
        • Aziz N.
        • Bale S.
        • Bick D.
        • Das S.
        • Gastier-Foster J.
        • Grody W.W.
        • Hegde M.
        • Lyon E.
        • Spector E.
        • Voelkerding K.
        • Rehm H.L.
        Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of medical genetics and genomics and the association for molecular pathology.
        Genet. Med. 2015; 17: 405-424
        • Karczewski K.J.
        • Francioli L.C.
        • Tiao G.
        • Cummings B.B.
        • Alföldi J.
        • Wang Q.
        • Collins R.L.
        • Laricchia K.M.
        • Ganna A.
        • Birnbaum D.P.
        • Gauthier L.D.
        • Brand H.
        • Solomonson M.
        • Watts N.A.
        • Rhodes D.
        • Singer-Berk M.
        • England E.M.
        • Seaby E.G.
        • Kosmicki J.A.
        • Walters R.K.
        • Tashman K.
        • Farjoun Y.
        • Banks E.
        • Poterba T.
        • Wang A.
        • Seed C.
        • Whiffin N.
        • Chong J.X.
        • Samocha K.E.
        • Pierce-Hoffman E.
        • Zappala Z.
        • O'Donnell-Luria A.H.
        • Minikel E.V.
        • Weisburd B.
        • Lek M.
        • Ware J.S.
        • Vittal C.
        • Armean I.M.
        • Bergelson L.
        • Cibulskis K.
        • Connolly K.M.
        • Covarrubias M.
        • Donnelly S.
        • Ferriera S.
        • Gabriel S.
        • Gentry J.
        • Gupta N.
        • Jeandet T.
        • Kaplan D.
        • Llanwarne C.
        • Munshi R.
        • Novod S.
        • Petrillo N.
        • Roazen D.
        • Ruano-Rubio V.
        • Saltzman A.
        • Schleicher M.
        • Soto J.
        • Tibbetts K.
        • Tolonen C.
        • Wade G.
        • Talkowski M.E.
        • Neale B.M.
        • Daly M.J.
        • MacArthur D.G.
        The mutational constraint spectrum quantified from variation in 141,456 humans.
        Nature. 2020; 581: 434-443
        • Francioli L.C.
        • Menelaou A.
        • Pulit S.L.
        • Van Dijk F.
        • Palamara P.F.
        • Elbers C.C.
        • Neerincx P.B.T.
        • Ye K.
        • Guryev V.
        • Kloosterman W.P.
        • Deelen P.
        • Abdellaoui A.
        • Van Leeuwen E.M.
        • Van Oven M.
        • Vermaat M.
        • Li M.
        • Laros J.F.J.
        • Karssen L.C.
        • Kanterakis A.
        • Amin N.
        • Hottenga J.J.
        • Lameijer E.W.
        • Kattenberg M.
        • Dijkstra M.
        • Byelas H.
        • Van Setten J.
        • Van Schaik B.D.C.
        • Bot J.
        • Nijman I.J.
        • Renkens I.
        • Marschall T.
        • Schönhuth A.
        • Hehir-Kwa J.Y.
        • Handsaker R.E.
        • Polak P.
        • Sohail M.
        • Vuzman D.
        • Hormozdiari F.
        • Van Enckevort D.
        • Mei H.
        • Koval V.
        • Moed M.H.
        • Van Der Velde Kj
        • Rivadeneira F.
        • Estrada K.
        • Medina-Gomez C.
        • Isaacs A.
        • McCarroll S.A.
        • Beekman M.
        • Mde Craen A.J.
        • Suchiman H.E.D.
        • Hofman A.
        • Oostra B.
        • Uitterlinden A.G.
        • Willemsen G.
        • Platteel M.
        • Veldink J.H.
        • Van Den Berg L.H.
        • Pitts S.J.
        • Potluri S.
        • Sundar P.
        • Cox D.R.
        • Sunyaev S.R.
        • Den Dunnen J.T.
        • Stoneking M.
        • De Knijff P.
        • Kayser M.
        • Li Q.
        • Li Y.
        • Du Y.
        • Chen R.
        • Cao H.
        • Li N.
        • Cao S.
        • Wang J.
        • Bovenberg J.A.
        • Pe’er I.
        • Slagboom P.E.
        • Van Duijn C.M.
        • Boomsma D.I.
        • Van Ommen G.J.B.
        • De Bakker P.I.W.
        • Swertz M.A.
        • Wijmenga C.
        Whole-genome sequence variation, population structure and demographic history of the Dutch population.
        Nat. Genet. 2014; 46: 818-825
        • Zhang M.Q.
        Statistical features of human exons and their flanking regions.
        Hum. Mol. Genet. 1998; 7: 919-932
        • Yeo G.
        • Burge C.B.
        Maximum entropy modeling of short sequence motifs with applications to RNA splicing signals.
        J. Comput. Biol. 2004; 11: 377-394
        • Reese M.G.
        • Eeckman F.H.
        • Kulp D.
        • Haussler D.
        Improved splice site detection in Genie.
        J. Comput. Biol. 1997; 4: 311-323
        • Pertea M.
        • Lin X.
        • Salzberg S.L.
        GeneSplicer: a new computational method for splice site prediction.
        Nucleic Acids Res. 2001; 29: 1185-1190
        • Gelfman S.
        • Wang Q.
        • McSweeney K.M.
        • Ren Z.
        • La Carpia F.
        • Halvorsen M.
        • Schoch K.
        • Ratzon F.
        • Heinzen E.L.
        • Boland M.J.
        • Petrovski S.
        • Goldstein D.B.
        Annotating pathogenic non-coding variants in genic regions.
        Nat. Commun. 2017; 8: 1-10
        • Xiong H.Y.
        • Alipanahi B.
        • Lee L.J.
        • Bretschneider H.
        • Merico D.
        • Yuen R.K.C.
        • Hua Y.
        • Gueroussov S.
        • Najafabadi H.S.
        • Hughes T.R.
        • Morris Q.
        • Barash Y.
        • Krainer A.R.
        • Jojic N.
        • Scherer S.W.
        • Blencowe B.J.
        • Frey B.J.
        The human splicing code reveals new insights into the genetic determinants of disease.
        Science. 2015; (80- ): 347
        • Jaganathan K.
        • Kyriazopoulou Panagiotopoulou S.
        • McRae J.F.
        • Darbandi S.F.
        • Knowles D.
        • Li Y.I.
        • Kosmicki J.A.
        • Arbelaez J.
        • Cui W.
        • Schwartz G.B.
        • Chow E.D.
        • Kanterakis E.
        • Gao H.
        • Kia A.
        • Batzoglou S.
        • Sanders S.J.
        • Farh K.K.-H.
        Predicting splicing from primary sequence with deep learning.
        Cell. 2019; (0): 535-548.e24
        • Balder J.W.
        • de Vries J.K.
        • Nolte I.M.
        • Lansberg P.J.
        • Kuivenhoven J.A.
        • Kamphuisen P.W.
        Lipid and lipoprotein reference values from 133,450 Dutch Lifelines participants: age- and gender-specific baseline lipid values and percentiles.
        J Clin Lipidol. 2017; 11: 1055-1064.e6
        • Calandra S.
        • Tarugi P.
        • Bertolini S.
        Altered mRNA splicing in lipoprotein disorders.
        Curr. Opin. Lipidol. 2011; 22: 93-99
        • Usifo E.
        • Leigh S.E.A.
        • Whittall R.A.
        • Lench N.
        • Taylor A.
        • Yeats C.
        • Orengo C.A.
        • Martin A.C.R.
        • Celli J.
        • Humphries S.E.
        Low-density lipoprotein receptor gene familial hypercholesterolemia variant database: update and pathological assessment.
        Ann. Hum. Genet. 2012; 76: 387-401
        • Vodnala D.
        • Rubenfire M.
        • Brook R.D.
        Secondary causes of dyslipidemia.
        Am. J. Cardiol. 2012; 110: 823-825
        • Holla Ø.L.
        • Nakken S.
        • Mattingsdal M.
        • Ranheim T.
        • Berge K.E.
        • Defesche J.C.
        • Leren T.P.
        Effects of intronic mutations in the LDLR gene on pre-mRNA splicing: comparison of wet-lab and bioinformatics analyses.
        Mol. Genet. Metabol. 2009; 96: 245-252
        • Jian X.
        • Boerwinkle E.
        • Liu X.
        In silico tools for splicing defect prediction: a survey from the viewpoint of end users.
        Genet. Med. 2014; 16: 497-503
        • Gent J.
        • Braakman I.
        Low-density lipoprotein receptor structure and folding.
        Cell. Mol. Life Sci. 2004; 61: 2461-2470
        • Kuzmiak H.A.
        • Maquat L.E.
        Applying nonsense-mediated mRNA decay research to the clinic: progress and challenges.
        Trends Mol. Med. 2006; 12: 306-316