Spectrum of mutations of the LPL gene identified in Italy in patients with severe hypertriglyceridemia

BACKGROUND
Monogenic hypertriglyceridemia (HTG) may result from mutations in some genes which impair the intravascular lipolysis of triglyceride (TG)-rich lipoproteins mediated by the enzyme Lipoprotein lipase (LPL). Mutations in the LPL gene are the most frequent cause of monogenic HTG (familial chylomicronemia) with recessive transmission.


METHODS
The LPL gene was resequenced in 149 patients with severe HTG (TG > 10 mmol/L) and 106 patients with moderate HTG (TG > 4.5 and <10 mmol/L) referred to tertiary Lipid Clinics in Italy.


RESULTS
In the group of severe HTG, 26 patients (17.4%) were homozygotes, 9 patients (6%) were compound heterozygotes and 15 patients (10%) were simple heterozygotes for rare LPL gene variants. Single or multiple episodes of pancreatitis were recorded in 24 (48%) of these patients. There was no difference in plasma TG concentration between patients with or without a positive history of pancreatitis. Among moderate HTG patients, six patients (5.6%) were heterozygotes for rare LPL variants; two of them had suffered from pancreatitis. Overall 36 rare LPL variants were found, 15 of which not reported previously. Systematic analysis of close relatives of mutation carriers led to the identification of 44 simple heterozygotes (plasma TG 3.2 ± 4.1 mmol/L), none of whom had a positive history of pancreatitis.


CONCLUSIONS
The prevalence of rare LPL variants in patients with severe or moderate HTG, referred to tertiary lipid clinics, was 50/149 (33.5%) and 6/106 (5.6%), respectively. Systematic analysis of relatives of mutation carriers is an efficient way to identify heterozygotes who may develop severe HTG.


M A N U S C R I P T
A C C E P T E D ACCEPTED MANUSCRIPT 4 consumption [6] or the presence of variants in other genes affecting TG metabolism [7]. The current view is that cumulative multiple genetic variants can increase the risk of HTG, especially in individuals who are heterozygous carriers of a loss of function mutation in one of the major genes affecting LPL-mediated lipolysis of TG-rich lipoproteins [8][9][10].
To date almost 180 LPL gene variants causing LPL deficiency have been reported (Supplemental Table S. 10 and Supplemental references . In this study we describe the spectrum of mutations in the LPL gene we identified in patients with the clinical diagnosis of familial chylomicronemia or Type IV/V hyperlipidemia, referred to three Italian Lipid Clinics over the last decade.

M A N U S C R I P T
A C C E P T E D ACCEPTED MANUSCRIPT 6

Biochemical analysis
Plasma lipids were measured as previously specified [11]. In some patients LPL activity in postheparin plasma was measured as reported [12] Analysis of the LPL gene Genomic DNA was extracted from peripheral blood leukocytes by a standard procedure. The exons and the promoter region of the LPL gene were amplified by polymerase chain reaction (PCR) and sequenced using appropriate primers [13]. Multiplex ligation-dependent probe amplification (MLPA) (SALSA MLPA P218-B1 LPL probe mix, MRC Holland, Amsterdam, the Netherlands) was used for the detection of major rearrangements of the LPL gene [13]. The PCR products were separated on an ABI PRISM 3100 sequencer and the data analysed by Peak Scanner™ software

Analysis of other candidate genes for HTG
The sequence of other candidate genes involved in monogenic HTG (APOC2, APOA5, GPIHBP1and LMF1) was performed as previously reported [11] in all patients carrying novel LPL variants and in all index patients found to be simple heterozygous for rare LPL variants. Besides rare variants, only the following common SNPs known to have an effect on plasma TG are reported in Tables: 1) Tables 1 and 2 show the list of 35 individuals with severe HTG, found to carry rare biallelic variants of the LPL gene in homozygous (patients  or compound heterozygous state (patients [27][28][29][30][31][32][33][34][35]. Among the unrelated index cases 21 were homozygotes and 9 compound heterozygotes. In the latter the presence of the two variants "in trans" (i.e. on different alleles) was confirmed by sequencing the LPL gene in the parents.
Tables 1 and 2 also show the age at molecular diagnosis and the key clinical features of each patient. In approximately 1/3 of the patients molecular diagnosis was performed before 1 year of age. At the time of molecular diagnosis the history of a single or multiple episodes of acute pancreatitis was recorded in 19 out of 35 patients (54%).
The age of the first episode of pancreatitis (as retrieved from the clinical records) was variable from 7 months to 50 years of age (mean ± SD: 18.6 ± 15.9 years, median 17.0 years). In one patient pancreatitis occurred during pregnancy at 24 years of age. Table 3 shows the mean age and plasma lipids in homozygotes/compound heterozygotes. The mean and median plasma TG levels (recorded at the time of molecular diagnosis) was 48.7 ± 57.7 mmol/L and 28.2 mmol/L, respectively (range 10.3 -326 mmol/L, interquartile range 21.1-56.7 mmol/L). There was no difference in plasma lipids between patients with or without a history of pancreatitis; pancreatitis positive patients were older than pancreatitis negative patients ( Table 4). In children <1 year of age plasma TG level was higher than in the older children or in the adults, but the prevalence of pancreatitis was much lower (9% vs The results of the multiple in silico analyses were consistent in predicting that the two novel missense variants [p.(I109T) and p.(G237D)] and that previously reported p.(G81D) [15] were pathogenic (Supplemental Table S

M A N U S C R I P T
A C C E P T E D ACCEPTED MANUSCRIPT 10 deletion] recently described by our group [13] (Table 2 and Supplemental Tables S.3 Among the novel missense variants the p.(E143D) substitution was predicted to be benign or tolerated by six out of seven algorithms. However, since the genomic change (c.429G>T) underlying this missense variant involved the last nucleotide of exon 3, we thought that this mutation might affect the function of the donor splice site of intron 3. In silico analysis (Human Splicing Finder, NetGene2 and ASSEDA) indicated a substantial decrease of the function of this donor splice site (leaky site) and possibly the activation of an alternative donor site in intron 3.
Thus, it is reasonable to assume that this variant generates an abnormal mRNA.
The p.(C302S) substitution was predicted to be not tolerated or damaging by 5 out of 7 algorithms; however, the cysteine residue at position 302 is highly conserved in various species. The p.(H229R) substitution was predicted to be pathogenic by all algorithms; this histidine residue is also highly conserved (Supplemental Tables S.4B

Heterozygous carriers of rare LPL gene variants
Twenty-one unrelated index cases (patients 36-56), mostly with the diagnosis of type V or type IV hyperlipidemia, were found to be simple heterozygous for rare LPL variants ( and 37) had a positive history of at least one episode of pancreatitis (33%); four of them had a history of recurrent pancreatitis (patients 37, 49, 51, and 56) and one subject suffered from acute pancreatitis during pregnancy (patient 50).
Fifteen different LPL variants were found, mostly reported previously ( Table 5) Tables S.4B, S.4C, S.4D). In addition one subject (patient 36) was found to be a carrier of the novel frameshift mutation (c.651delT), detected in one unrelated homozygote (patient 8 in Table 1 Table 5).
None of the rare LPL gene variants listed in Tables 1, 2 and 5 was found in a sample of 250 normolipidemic individuals of the Italian population.

Carriers of rare LPL gene variants identified in close relatives of index cases
Mutation screening among relatives of index cases (homozygotes/compound heterozygotes and simple heterozygotes) led to the identification of 44 simple heterozygotes, aged from 4 to 75 years (mean ± SD; 41.2 ± 18.0 years, median 39 years) ( Table 6). The number of relatives per family ranged from 1 to 4. Plasma TG levels of these subjects ranged from 0.6 to 19.2 mmol/L (median 1.9 mmol/L, interquartile range 1.5-2.7 mmol/L); only in three subjects was plasma TG level above 10 mmol/L. At the age of molecular diagnosis none of these subjects had a positive history of pancreatitis. Among subjects with moderate TG levels we found 9 heterozygotes and one homozygote for p.(D36N), and 10 heterozygotes for p.(N318S). Their TG levels ranged from 5.1 to 9.9 mmol/L (mean ± SD: 7.0 ± 1.5 mmol/L) (Supplemental Table S

DISCUSSION
In this study we describe rare variants of the LPL gene found in a group of patients with severe or moderate primary hypertriglyceridemia (HTG), referred to three tertiary Lipid Clinics in Italy. The group included 149 patients with plasma TG concentration >10 mmol/L and 106 patients with plasma TG >4.5 and <10 mmol/L. Rare variants of the LPL gene were identified in 51 unrelated patients (21 homozygotes, 9 compound heterozygotes and 21 simple heterozygotes) (overall 20%).
More specifically 11.8% of patients were carriers of two mutant alleles and 8.2% were carriers of one mutant allele.
The prevalence of monogenic LPL deficiency (presence of two mutant alleles) in the general population is estimated to be 1:1.000.000 and that of carriers of one mutant allele 1:500 [5]. The prevalence of subjects with LPL mutations found among patients with severe HTG varies considerably in different surveys. Wang et al. [17] found 6 carriers of rare LPL variants among 110 non diabetic adult patients with severe HTG (mean plasma TG level 32.6 ± 26.5 mmol/L); four of them were heterozygous for variants known to be the cause of LPL deficiency (Supplemental Table   S ; by considering the carriers of at least one mutant allele (homozygotes, compound heterozygotes and simple heterozygotes) (Tables 1, 2 and 5), the percentage of carriers of rare LPL variants among our patients with plasma TG >10 mmol/L was approximately 30%, a figure close to that found by Surendran et al. [20]. The discrepancy between our results and those of Wang et al. [17], Wright et al [18] and Evans et al. [19] is probably due to a clinical selection bias, as our patients (who included both children and adults) have been referred to tertiary Lipid Clinics that usually investigate in depth the most severe cases of dyslipidemias especially in children and young adults. The higher percentage of patients carrying rare LPL variants (55%) reported by Martin-Campos et al. [21] may be explained by the young age of subjects with Type I hyperlipidemia they investigated (among 29 patients 9 were newborns and 11 children/adolescents).
The level of plasma TG showed a considerable variation in our homozygotes and compound heterozygotes, ranging from 10.3 to 326 mmol/L. Interestingly, in children below 1 year of age the mean plasmaTG level was twice that found in older children and in adult patients taken together (Supplemental Table S.1). The reason for this striking difference is not clearly understood. It probably depends on the fact that the stringent feeding schedule of breast or bottle fed infants (and the short time periods between meals) facilitate the progressive accumulation of chylomicrons over time, while in older children and in adults longer fasting periods between meals (like overnight fasting) might delay or mitigate the rate of chylomicron accumulation in plasma.
From the clinical stand point 19 homozygotes/compound heterozygotes (54%) had a positive history of at least one episode of acute pancreatitis. While pancreatitis was recorded only in 1 out of 11 patients <1 year of age, it was documented in 18 out of 24 older children and adults (age range 4-74 years), despite a lower mean level of plasma TG. In simple terms this would suggest that the longer the time of exposure to very high plasma TG levels, the higher is the chance to develop acute pancreatitis.
Among unrelated index cases found to be simple heterozygous for LPL variants ( Table 5)  history of pancreatitis; in five of them plasma TG level was >10 mmol/L. In this group of patients, three (15%) had type 2 diabetes, a condition that may have contributed to increase plasma TG and worsen an otherwise mild HTG related to the presence of one mutant LPL allele [6,7]. It also conceivable that patients with severe HTG were carries of variants of other HTG related genes with a cumulative effect on the phenotype [8][9][10]. Indeed, this appears to be the case as almost 50% of these patients were found to carry SNPs of LPL and other HTG related genes known to have a plasma TG raising effect (Table 5).
We performed LPL mutation screening in family members of index cases (homozygotes/compound heterozygotes and simple heterozygotes) and identified 44 simple heterozygotes (approximately 2 subjects per family). Their age ranged from 4 to 75 years (mean 41.2 ± 18.0; median 39.0); their TG level ranged from 0.6 to 19.2 mmol/L (mean 3.2 ± 4.1, median 1.9, interquartile range 1.5-2.7 mmol/L). The mean plasma TG level of these individuals was much lower than that recorded in the index cases found to be simple heterozygous for LPL mutations (Tables 6, 7), suggesting that in the latter other genetic or non-genetic factors contributed to HTG. None of the simple heterozygotes identified by the analysis of close relatives of index cases had a positive history of pancreatitis, probably because their plasma TG was largely below 10 mmol/L, the alert threshold level for the risk of pancreatitis; in fact, only 3 individuals (two of whom belonging the same family) had plasma TG >10 mmol/L ( Table 6).
Overall, our observations indicate that among unrelated index cases carrying rare LPL gene variants with plasma TG >10 mmol/L the prevalence of pancreatitis (20/45, 44%) appears to be much higher than that (10-20%) reported in some large series of molecularly undefined hypertriglyceridemic subjects with severe HTG [22,23].The prevalence of pancreatitis was even higher (24/50, 48%) if one considers all homozygotes (unrelated as well as related subjects), compound heterozygotes and simple heterozygotes with TG level >10 mmol/L.

M A N U S C R I P T
A C C E P T E D ACCEPTED MANUSCRIPT 16 As expected, we found a great allelic heterogeneity in carriers of rare LPL variants (Figure 1). Most mutations were single base changes in exons leading to 23 missense mutations (8 of which not reported previously). By using a battery of seven in silico algorithms we attempted to define the possible pathogenicity of the novel missense mutations. This analysis indicated that 4 mutations were pathogenic, 3 probably pathogenic, 1 possibly pathogenic and 1 non pathogenic (Supplemental Table S.4B). It should be stressed that seven out of eight variants involve amino acids that are highly conserved among species, suggesting that their substitution is likely to have an impact on the structure/function of the enzyme. One novel missense mutation [p.(E143D)], found in a compound heterozygote (patient 29 in Table 2), turned out to be tolerated/not deleterious. However, since the underlying genomic mutation (c.429G>T) changes the last nucleotide of exon 3, it is likely that this apparent missense mutation is in fact a splicing mutation as predicted by in silico analysis.
We found one novel nonsense mutation as well as one minute deletion and one deletion/insertion predicted to cause a frameshift leading to a premature termination codon. Finally we report four novel intronic mutations affecting splice sites. Two of them (c.250-1G>C; c.1019-2A>T) were found to generate in vitro abnormal mRNAs, predicted to encode truncated proteins. The other two novel mutations (c.88+1G>T and c.88+2T>G) affected the highly conserved dinucleotide "GT" of the donor splice site of intron 1 and were predicted in silico to abolish the function of this site.
One important aspect that emerged from our survey is the relatively large number of infants (<1 year of age) present in our series of homozygotes/compound heterozygotes. This is not surprising as in recent years the resequencing of the LPL gene (and the other monogenic HTG candidate genes) has been extended to neonates with severe HTG, as a rapid and efficient procedure to assess the presence of genetic defects of the lypolytic cascade, (replacing other more laborious assays, such as the measure of LPL activity in post-heparin plasma) [7,13,15,21,[24][25][26][27]. This diagnostic workup in neonates with milky plasma avoids delays in therapeutic interventions (dietary changes, extracorporeal treatment or exchange transfusion) [13,15,[24][25][26][27], directed to reduce plasma TG and to prevent the occurrence of acute pancreatitis.
Finally, on the basis of this survey we are now considering which of the patients listed in Tables 1   and 2 are suitable candidates for the LPL gene therapy (Alipogene-Tiparvovec) [28], recently approved by the European Medicines Agency for the treatment of LPL deficiency. In this context LPL activity and mass will be measured in post-heparin plasma of homozygotes/compound heterozygotes with missense mutations and a positive history of recurrent pancreatitis to select those patients, who in principle, might have the best benefit (and possibly fewer complications) from this new treatment [7].

Conflicts of interest
The authors have no conflict of interest to disclose.            References for reviews are reported in RED  The in silico prediction for novel missense mutations is reported in bold characters. Homo sapiens MANUSCRIPT   10   Sus scrofa  D  T  V  G  I  R  A  S  R  G  I  H  P  G  I  R  D  S  C  N  M  R  L  A   Horse  D  T  V  G  I  R  A  S  R  G  I  H  P  G  I    References for reviews are reported in RED  Children n. 9

M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT
Adolescents n.

c.250 -1G>C (Intron 2)
Construction of the reporter minigene: a 3767 nt fragment of wild type LPL gene, containing exon 2, intron 2 and exon 3 was inserted in pTargeT plasmid vector. The splice site mutation was introduced by site directed mutagenesis. The wild type and mutant plasmid were transfected in COS1 cells. The wild type minigene generated a transcript of 340 nt, while the mutant minigene generated two transcripts of 411 nt and 332 nt, respectively.
In the 332 nt transcript exon 2 joined to exon 3 devoid of the first 7 nt at the 5' end. In the 441 fragment exon 2 was followed by 72 nucleotides 3' of intron 2 (partial intron retention) and by exon 3. The resulting frameshifts led to the insertion of a premutare termination codon. The predicted translation products of the mutant transcript are two truncated proteins p.[(T85Yfs*15, V84Efs*86] expected to be devoid of function.

c.1019 -2A>T (Intron 6)
Construction of the reporter minigene: a 1311 nt fragment of wild type LPL gene spanning from intron 6 to intron 8 was obtained by PCR amplification. To reduce the size of the minigene, intron 6 and intron 7 had been shortened by producing an internal deletion of 3.5 kb and 7.5 kb, respectively. The genomic DNA fragment was inserted in a pTargeT vector and the splice site mutation was introduced by site directed mutagenesis.

M A N U S C R I P T
A C C E P T E D ACCEPTED MANUSCRIPT

18
The wild type and mutant plasmid were transfected in COS1 cells. The wild type minigene generated a 508 nt transcript while the mutant minigene generated a 347 nt transcript. The sequence of the 347 transcripts showed that exon 6 joined directly to exon 8 with the complete skipping of exon 7.
This leads to a frameshift with the formation of a premature termination codon. The product of this transcript is predicted to be a truncated protein p.(V340Gfs*13) devoid of function.