Corresponding author: Steve E. Humphries, Centre for Cardiovascular Genetics, British Heart Foundation Laboratories, The Rayne Building, Royal Free and University College London Medical School, London WC1E 6JJ, UK. Tel: +44-020-7679-6962; Fax: +44-020-7679-6212; E-mail: email@example.com
Familial hypercholesterolemia (FH) is caused by mutations in the genes for LDLR, APOB or PCSK9, and identification of the causative mutation provides definitive diagnosis so that the patient can be treated, their relatives tested and, therefore, premature heart disease prevented.
DNA of eight unrelated individuals with clinically diagnosed FH were analyzed using a High-Resolution Melting method (HRM) for the LDLR gene (coding region, promoter and intron/exon boundaries), the APOB gene (part exon 26) and the PCSK9 gene (exon7). Variations found were sequenced and the effect on function of confirmed variants examined using predictive algorithms. Gross deletions and insertions were analysed using MLPA.
Three novel LDLR variants were found, p.(S470C), p.(C698R) and c.2312–2A>C. All were predicted to be pathogenic using predictive algorithms. Three previously reported disease-causing mutations were identified (p.(G20R), p.(N272T) and p.(S286R); the latter was also carried by a hypercholesterolaemic relative. One patient carried the pathogenic APOB variant p.(R3527Q). No large LDLR deletions nor insertions were found, neither were any PCSK9 variants identified.
HRM is a sensitive method for screening for mutations. While the causative mutation has been identified in 88% of these clinically defined FH patients, there appears to be a high degree of allelic heterogeneity in Croatian patients.
Familial hypercholesterolemia (FH) is a common autosomal dominant disorder characterised by elevated Low Density Lipoprotein-cholesterol (LDL-C) and premature coronary heart disease (CHD) (Marks et al., 2003.) The main pathologic mechanism is an impaired removal of LDL-C particles from the plasma. Mutations in the gene coding for the LDL receptor (LDLR) are the most common cause of FH (Varret et al., 2008). The gene for the LDL-receptor is situated on chromosome 19, with 18 exons and is 45kb long (Brown & Goldstein, 1986). Less frequently mutations in the APOB gene, coding for apolipoprotein B the major structural protein of LDL-C particles, or in the proprotein convertase subtilisin/kexin type 9 (PCSK9) gene can also cause FH (Brown & Goldstein 1986; Humphries & Talmud 1995; Abifadel et al., 2003; Timms et al., 2004). The high plasma LDL-C levels result in deposition of LDL cholesterol in tissues (xanthelasmas, tendinous xanthomas, arcus corneae), most importantly in arteries, forming atherosclerotic plaques. This results in premature cardiovascular morbidity and mortality that can be effectively reduced by statin treatment (Simon Broome Register Group, 1991). Today more than 1100 mutations in LDLR are described (http://www.ucl.ac.uk/fh) (Leigh et al., 2008; Usifo et al., 2012). There are several different systems used worldwide for the clinical diagnosis of FH, but today genetic testing provides the possibility of making a definitive diagnosis based on pathogenic variations in the LDLR, APOB and PCSK9 genes. Several consensus guidelines for the diagnosis and management of FH have been published recently (summarised in Humphries, 2012), and all recommend the use of “cascade testing” of relatives by testing for the mutation found in the proband.
The prevalence of heterozygous FH is about 1:500 (Usifo et al., 2012) suggesting that there are likely to be at least 9000 FH patients in Croatia, most of them unrecognised and untreated. In recent years the number of clinical and genetic studies to diagnose FH has increased worldwide but this disease is still under-diagnosed in Croatia. There are only two published studies on Croatian FH patients using genetic analysis, one reporting a novel missense LDLR mutation, p.(C127R) (Humphries, 2012), and the second a novel nonsense mutation p.(Q92X) in a homozygous patient (Simon Broome Register Group, 1991; Whittall et al., 2010; Dumic et al., 2007). Therefore the aim of this study was to define genetic variations in Croatian FH patients, as a basis for future genetic testing of probands and cascade testing in their relatives.
Patient Selection Criteria
Eight unrelated Croatian FH heterozygous patients were recruited after being diagnosed and treated at the outpatient clinic, Department for Metabolic Diseases, University Hospital Center, Zagreb. Definition of FH was based on the UK Simon Broome criteria (Simon Broome Register Group, 1991). Coronary Artery Disease (CAD) was defined as documented evidence of a myocardial infarction, angina pectoris, intervention on coronary arteries (stenting, coronary artery bypass grafting) or positive result of stress test. Molecular genetic analysis was performed in the Center for Cardiovascular Genetics, UCL, London, UK. DNA was isolated from whole blood samples using standard methods. Polymerase chain reaction and high-resolution melt were performed using the Rotor-Gene 6000 (Qiagen Ltd, Crawley, West Sussex, UK). The promoter and coding regions of the LDLR together with the p.(R3527Q) mutation (part of exon 26) of APOB and p.(D374Y) (exon 7) of the PCSK9 gene were included for screening. Further steps in analysis were performed as described (Whittall et al., 2010). Samples showing a different HRM melting profile were subsequently sequenced. Sequencing was carried out using fresh PCR products or directly from the HRM-PCR by Source BioScience LifeSciences UK Ltd., Nottingham UK. APOE genotype for the two SNPS rs429358 and rs7412 detecting the Ε2/3/4 protein isoforms was performed as described (Smart et al., 2010) using TaqMan technology (Applied Biosciences, ABI, Warrington, UK). Reactions were performed on 384-well microplates and analysed using ABI TaqMan 7900HT software. Primers and Minor Groove Binder (MGB) probes are available upon request.
Detected variants were designated as pathogenic or nonpathogenic using criteria as previously described (Cotton & Scriver, 1998). Multiplex Ligation-dependent Probe Amplification (MLPA) was used to search for large deletions and insertions in the LDLR gene as described (Taylor et al., 2009). Mutation nucleotide numbers were designated using the LDLR sequence reported [http://www.ucl.ac.uk/fh], with cDNA numbering beginning with A of the initiating ATG = 1. Mutations were designated according to the Human Genome Variation Society recommendations (http://www.hgvs.org/mutnomen/). Variations found were compared with online databases at http://www.ucl.ac.uk/fh, the 1000 Genomes Project and the Exome variant server (Leigh et al., 2008) and their potential clinical impact was assessed using algorithms on pathogenicity predictive websites: SIFT, PolyPhen and Mutation-t@ster (Schwarz et al., 2010; Kumar et al., 2009; Adzhubei et al., 2010), for amino acid variations. The effects of synonymous substitutions and intronic variants on splicing were assessed using NNSplice, Splice Port and NetGene2 Server (Hebsgaard et al., 1996; Reese et al., 1997; Dogan et al., 2007). These programs give predictive scores for splice acceptor and donor sequences for wild type and variant sequences. Alterations in the scores between wild type and variants allow an assessment to be made regarding whether or not normal exon splicing will be affected.
The characteristics of the eight patients with a clinical diagnosis of FH (7 female, 1 male), are shown in Table 1. Overall the mean serum total cholesterol was 7.95 mmol/l (min 6.1 mmol/l, max 10.7 mmol/l), and the mean serum LDL-C was 5.96 mmol/l (minimum 4.4 mmol/l, maximum 8.4 mmol/l). Three patients had a personal or family history of tendon xanthomas, and six had a personal or family history of CAD. One individual, Za6, was included with an LDL-C of only 4.4 mmol/l, because their premature CAD developed at the age of 45 years. Table 1 also shows the APOE genotype of all examined patients, and as expected from the allele frequency in the general population of Europe, 6/8 patients (75%) have the common APOE E3E3 genotype and 2/8 are E4E4.
Table 1. Characteristics and Lipid Levels of Examined Subjects
LDL% age reduction
Gender F, female; M, male; TC, total cholesterol; HDL, high-density lipoprotein cholesterol; LDL, low-density lipoprotein cholesterol;
TG, triglycerides; Y, yes; N, no, FH, family history; CAD, cardiovascular disease.
HRM and sequencing was used to screen for mutations in the promoter and all coding exons of the LDLR gene, part of exon 26 in APOB covering the region coding for p.(R3527Q) and exon 7 of PCSK9 covering the region coding for p.(D374Y). Samples showing any difference in HRM profile were sequenced to identify the underlying variation. As well as the expected previously reported polymorphisms, three mutations previously found in FH patients were identified, p.(G20R), p.(N272T) and p.(S286R), as shown in Table 2. Three novel mutations were found p.(S470C), (shown in Fig. 1), p.(C698R) and c.2312–2A>C at the intron 15/exon 16 junction (shown in Fig. S1). Their disease causing potential was supported by bioinformatic prediction algorithms (Table 2) along with cross-species conservation analysis. The hypercholesterolaemic daughter of Za11 (LDL-C 6.90 mmol/l at 25 years age) also carried p.(S286R). One patient, Za6, was found to carry the APOB p.(R3527Q) mutation and her lipid levels were; total cholesterol 6.1 mmol/l, LDL cholesterol 4.40 mmol/l.
Table 2. Mutations found in the Croatian Samples and their Predicted Effect
Key to SAAP results: Binding: Native residue was involved in specific H-bond or packing interaction with another protein or ligand. Interface: Native residue was in an interface. Clash: Variant introduced an amino acid which clashes with its surroundings. Impact: Site of variant is significantly conserved. Buried Charge: Variant introduced charged residue into protein core. CorePhilic: Variant introduced a hydrophilic residue into protein core. SProtFT: Site of variant is annotated as a feature in UniprotKB/SwissProt. SSGeom: Native residue was a Cys involved in a disulphide bond. The number refers to the number of structures in which the effect is observed.
scores are for wt donor splice site in intron1 
Samples where no FH-causing mutation had been identified were examined for gross deletions and rearrangements using MLPA, but none were identified (data not shown). Therefore, in the eight unrelated patients, an FH-causing mutation has been identified in seven subjects for an overall detection rate of 88%.
The aim of this study was to analyze the mutation spectrum in patients from Croatia with clinically diagnosed FH. Data on the incidence and prevalence of FH in Croatia are virtually nonexistent (Rukavina et al., 2001) so this genetic analysis was conducted to confirm the clinical diagnosis of patients and to describe possible variations. There are several recommendations for diagnosing FH (Marks et al., 2003; Wierzbicki et al., 2008; Goldberg et al., 2011; Descamps et al., 2011; Watts et al., 2011) and all of them now include genetic analysis as the key basis for confirming the diagnosis. In this study we found seven different mutations in eight unrelated patients, suggesting that the FH group is genetically heterogeneous. The high overall detection rate of 88% indicates that our patients were clinically well characterised and selected, and that the laboratory methods used are of appropriate sensitivity (Whittall et al., 2010). However the sample of FH patients available here is too small to make a definitive statement about whether or not any of these three newly identified LDLR mutations might be common in patients with severe hypercholesterolaemia in Croatia, but at this time additional samples of such patients are not available for testing.
Not surprisingly, given the geographical and historical location of Croatia, several of the identified mutations have been previously identified in neighbouring countries. The p.(S286R) mutation was described as a disease-causing Greek mutation (Traeger-Synodinos et al., 1998; Diakou et al., 2011). Two of our subjects, a father (Za11, Table 2) and his hyperlipidaemic daughter, carried the mutation. Their origin is the Adriatic coast region, the Mediterranean area of Croatia, where in ancient times some Greek colonies were founded. The variation, p.(G20R), has been reported five times in the LDLR database (Leigh et al., 2008) first described by Amsellem et al. (2002) and later by Laurie et al. (2004) who defined it as a European mutation. It was also reported in a study of Turkish patients (Sözen et al., 2005). The Turkish invasions of Croatia several centuries ago and the fact that Croatia had a relatively long border with the Turkish Empire until the 20th century, could explain the presence of this variation. This variant occurs in the signal peptide sequence, and prediction algorithms do not suggest this change to be pathogenic, although as shown in Figure 2, a comparison of the conservation in the signal peptide sequence across ten species identifies no species with arginine or any charged amino acid at this site, and even subtle changes in the hydrophobic properties of this sequence may influence insertion of the peptide into the endoplasmic reticulum membrane and the correct transport of the protein to the cell surface (Talmud et al., 1996). The patient carrying this variant does not have an extensive family, and we are unaware of any family studies that have demonstrated co-segregation of this variant with hypercholesterolaemia to confirm the functional impact of this variant. The variant p.(N272T), was reported in a patient of Czech origin (Dušková et al., 2011) and is also recorded in the LDLR database (Leigh et al., 2008). Finally, one patient carried the common APOB p.(R3527Q) mutation, which occurs in 3–5% of FH cases in the UK (Humphries et al., 2006) and frequently throughout Europe. Carriers of this mutation usually have only moderately increased levels of LDL-C (Humphries et al., 2006) which was the case in our patient Za6 (Table 1).
We also found three novel LDLR variants: p.(S470C) (HRM and sequencing shown in Fig. 1) in the second repeat of the β-propeller domain, p.(C698R) in the third epidermal growth factor (EGF)-like domain and c.2312–2A>C in the splice acceptor site of intron 15 (see Fig. S1). Both p.(S470C) and p.(C698R) lie within the EGF-like domain responsible for the dissociation of LDL from the LDLR in the endosome at low pH, and recycling of the receptor to the cell membrane (Beglova & Blacklow, 2005). Although the serine at position 470 is only moderately conserved (Scorecons 0.569, Table 2), introduction of a novel cysteine residue at this position by the variant p.(S470C) is likely to be pathogenic, as it is 18 residues away from the only cysteine (position 452) in the LDL-R that is not already involved in a disulphide bond. If such a novel bond were to form, the secondary structure of the second repeat in the β-propeller domain would be significantly disrupted, which could in turn interfere with release of LDL from LDLR during receptor recycling. The cysteine residue substituted in variant p.(C698R) participates in the final disulphide bond of the mature LDLR molecule which closes the last hairpin loop in the EGF-like domain. Removal of this bond is predicted to disrupt the secondary structure of the peptide. Furthermore, it has been shown in the LDL-receptor (Djordjevic et al., 1996) and in other peptides, including αIIbβ3 integrin which also has EGF-like domains that such changes can result in incorrect folding and retention of the peptide in the endoplasmic reticulum and concomitant reduced secretion of the peptide from the cell (Gorr et al., 1999; Mor-Cohen et al., 2007). The consequent reduced secretion would be likely to have a pathogenic effect.
The c.2312–2A>C mutation is predicted to ablate the 3’ splice acceptor in intron 15 (Table 2) resulting in incorrect splicing as shown in Figure 3. As the intron15 acceptor splice site is destroyed (score from 0.99 wild type to 0 (NNSplice-9: Berkeley Drosophila Genome Project (BDGP (Hebsgaard et al., 1996))) in the variant) an alternative cryptic splice site may be used. The alternative acceptor splice site could be within the exon so that part of exon 16 is lost or within intron 15 so that part of the intron is now included in the mRNA. The outcome of the use of these alternative splice sites would depend on whether the reading frame of the mRNA is maintained, or on the consequence of the loss of amino acids or the insertion of novel amino acids to the LDL-R protein. It is difficult to predict which cryptic splice site may be used. In intron 15 the nearest site with a score greater than 0.01 (NNSplice-9: BDGP is at c.2312–33_34 with a score of 0.71. If this site is used then it would result in the insertion of 11 novel amino acids and a frame shift leading to a premature stop codon after 16 aberrant amino acids. There are eight putative acceptor splice sites (AG dinucleotides) in exon 16. However, they have predicted scores of less than 0.01 (BDGP), much lower than the acceptor splice site of exon 17 (0.57). Therefore, they are unlikely to be favored over the exon 17 splice site. However, the use of any of these would result, at best, in the loss of amino acids from exon 16 and at worst a frame shift and premature stop codon. If a cryptic splice site is not used the use of the next wild type acceptor splice site (exon 17) would lead to skipping of exon 16 resulting in the splicing of exon 15 to exon 17. In this case the reading frame is maintained but the amino acids coded by exon 16 are missing. Exon 16 includes part of the transmembrane domain of the LDLR. A similar mutation, c.2312–1G>T at the same splice site, first described in Italy, was shown to lead to skipping of exon 16 which codes for part of the membrane spanning domain (Liguori et al., 2003). Loss of these amino acids would lead to the protein not being embedded in the membrane correctly. All three variants were predicted, by all the relevant algorithms used, to be disease causing.
One limitation of our study is that we have only screened exon 7 of the PCSK9 gene in these individuals to look for the presence of causative mutations. In the UK around 2% of patients with FH have mutation in this exon of the PCSK9 gene, but throughout Europe mutations causing FH in PCSK9 are extremely rare, and as far as we are aware, no patients with a PCSK9 mutation have been reported in neighbouring countries to Croatia such as Italy or Greece. We have also not screened the LDLRAP1 gene for possible mutations because mutations in this gene cause an autosomal recessive form of hypercholesterolaemia (Garcia et al., 2001, Eden et al., 2002) and in all the patients examined here the pattern of inheritance was autosomal dominant, making it extremely unlikely that a mutation in the LDLRAP1 could be causal of their hypercholesterolaemia.
This study provides novel data on FH in Croatia and contributes to the worldwide effort in identifying variations in the LDLR gene that can cause FH. Morbidity and mortality trends for CVD in Croatia are not positive, therefore every effort to enable early diagnosis in patients who are at risk of CVD are more than welcome (Reiner et al., 2006). This is especially true for patients with FH whose genetic background in combination with poor lifestyle predispose for early cardiovascular morbidity and mortality if undiagnosed and untreated. There is a need for further genetic analysis and increasing awareness of FH in lipid clinics, public presentations, and cascade testing, as has been carried out so successfully in Holland (Umans-Eckenhausen et al., 2001), Scandinavia (Leren et al., 2008) and Spain (Palacios et al., 2012).
SEH and RW are funded by the British Heart Foundation. SEH is a BHF Professor and is funded by PG08/008. MF is funded by a MRC CASE award. IP was supported by grant of the International Atherosclerosis Society.