We have conducted a genome-wide search for shared homozygosity in affected sibs in order to identify the mutation responsible for the severe HTG in a consanguineous Chilean family. There are three strategies commonly used to search for rare causative mutations in severe monogenic diseases in families: a) genome or exome sequencing, b) linkage analysis followed by sequencing of candidate genes within the genomic area significant for linkage, and c) resequencing all candidate genes related to the disease. Exome/genome sequencing (a) is becoming with no doubt a key tool to find disease-causing mutations
. The use of exome sequencing approach would have been clearly successful in the identification of the genetic cause of HTG in our family. However, in a more general context, there are still some challenges associated with the exome sequencing strategy. It has been estimated that each individual exome harbors around 20,000 sequence variants in comparison with the human reference genome, including 5,000 variants affecting amino acid sequence. After filtering with common genetic variation found in databases, still many variants are left as candidate causal, even if we consider recessive defects
[31, 32]. Linkage analysis (b) is a classical genetic strategy to search for causal loci responsible of genetic diseases which is based on the use of a map of genetic markers (microsatellite markers or SNP) across the genome. A special case of linkage analysis is the so-called homozygosity mapping strategy, which has been extensively used in the study of genetic causes underlying diseases with recessive effect in consanguineous families. In these types of studies, the objective is to identify homozygous markers shared by the affected cases that pinpoint genomic regions that may contain the responsible gene and mutation
. It is important to mention that the combination of genome/exome sequencing and homozygosity mapping is an appealing approach that has been successfully applied in the identification of mendelian diseases
. Direct sequencing of candidate genes (c) is conceptually the most straightforward approach to find causal mutations. However, there are some important considerations that should be taken into account regarding the laboratory effort required for resequencing studies. Sequencing the five most plausible candidate genes for severe HTG (LPL, APOC2, APOA5, LMF1 and GPIHBP1) would include the analysis of ten exons of the LPL gene (main candidate gene), four exons of the APOA5 gene, thirteen exons of the LMF1 gene, four exons of the GPIHBP1 gene and five exons of the APOC2 gene (a total of 36 exons). If no mutation is found initially in these genes, next logical steps would be the search for novel mutations in genes with common variants that have been associated with multifactorial HTG through genome-wide association studies such as for example GCKR (19 exons), and possibly testing genes with loss-of-function mutations causing hypotriglyceridemia (for example, 6 exons of the ANGPTL3, 29 exons of the APOB gene, among others). Therefore, resequencing genes with traditional techniques may represent an important investment in terms of time, especially if the causative mutation is not found in the initial steps. In contrast, the timeframe we have employed in this homozygosity mapping study in submitting DNA to genome-wide analysis, locating the causative gene and discovering the mutation was of approximately 6 months. This was possible because the homozygosity mapping directly led us to the chromosomal region in which the causal mutation is located. As stated before, another advantage of the homozygosity mapping strategy is that limits the search for candidate mutations provided by exome sequencing in consanguineous families
, and provides the possibility to identify new genes not previously described as responsible for extreme phenotypes. On the contrary, the main limitation of homozygosity mapping is that it requires consanguineous families. It is also important to notice that sometimes (especially in small families), this method pinpoints a very large chromosomal region, in which hundreds of genes could be responsible of the causal mutation. It is also important to mention that diseases caused by compound heterozygous are not detectable using this strategy. In our study, homozygosity mapping correctly pinpointed the region carrying the gene responsible for the disease
[9, 28] in 11q23 that contains APOA5/A4/C3/A1 gene cluster. After Sanger sequencing, the Q97X mutation of the APOA5 gene was identified in a homozygous state in the two affected sisters and in a heterozygous state in other family members. Although statistical power for finding significant LOD scores is always increased with larger number of affected and unaffected relatives, it is noteworthy to mention that the discovery of mutations is possible using only a single family
. Given the arguments presented above, we believe that homozygosity mapping is a valid and fast approach to locate causal recessive mutations in consanguineous families. This power of this strategy is enhanced when used in combination with resequencing of candidate genes or exome sequencing techniques.
The APOA5 gene is located on chromosome 11q23, within the APOA1/C3/A4/A5 gene cluster, a well-studied region that had previously been associated with variations in plasma TG
. Apo A-V protein was discovered in 2001 by two research teams simultaneously
[35, 36]. This apolipoprotein is synthesized in the liver and is associated with very low density lipoprotein (VLDL), chylomicrons (QM) and high-density lipoprotein (HDL). Its concentration in plasma is very low (0.1-0.4 μg/ml), which is approximately 300-fold lower than Apo C-III and roughly 2000-fold lower than Apo A-I
[9, 28, 37]. It has been proposed that Apo A-V influences plasma TG levels primarily by increasing the catabolism of TG-rich lipoproteins through stimulation of LPL
[38–41] or alternatively by inhibiting the rate of production of VLDL
. It was demonstrated that adenoviral overexpression of Apo A-V in mice is related to a 70% reduction in TG levels compared to wild-type mice, while Apoa5 knock-out mice showed a four-fold plasma levels relative to controls
[43, 44]. In humans, an inverse association between plasma Apo A-V and TG levels has been reported in cases of homozygous APOA5 nonsense mutations
, whereas other reports indicate higher ApoA-V plasma concentrations in HTG patients
. Interestingly, it has been shown that intravenous injection of ApoA-V into apoa5 knock-out mice induces a 60% decrease in plasma TG after four hours
An excess of rare APOA5 variants have been found in HTG patients, providing evidence for the importance of APOA5 in this disorder
. Mutations in the APOA5 gene have been described in cases of severe chylomicronemia, such as the Q148X mutation, Q139X mutation, IVS3+3G>C and three heterozygous missense variants
[48–51]. The phenotypic expression of these mutations is highly variable, especially in heterozygous subjects, who usually require additional adverse environmental triggers (e.g. obesity, diabetes) or the presence of other susceptibility genetic variants. The Q97X mutation found in this study was described for the first time by Oliva et al.
 and subsequently by Charriere et al.
. After the removal of the endoplasmic reticulum signal peptide of 23 amino acids, this mutation would cause truncation at residue 74 of the mature protein (343 amino acids)
, creating a predicted 10 kDa peptide instead of the 39 kDa complete protein. This peptide, if synthesized (see discussion on NMD below), would lack the essential domain for lipid binding and LPL activation (residues 192 to 238) and therefore is expected to be non-functional
[41, 55] (Additional file
1: Figure S4 shows our own prediction model of the Apo A-V protein structure). In these previous reports on homozygous patients carrying the Q97X mutation, neither wild-type nor truncated Apo A-V were detected in plasma, thus suggesting a complete Apo A-V deficiency
There is a remarkable variability shown by affected HTG cases that are carriers of the Q97X mutation in different studies, especially in terms of drug requirements, responsiveness to treatment, age of onset, TG plasma levels and other clinical manifestations
[52, 53] (See Additional file
1: Table S4). For example, heterozygous Q97X carriers reported by Oliva et al. showed phenotypes ranging from severe hyperchylomicronemia to normotriglyceridemia
. It is possible that partial Apo A-V deficiency by itself is not sufficient to trigger severe HTG unless adverse environmental stimuli overload the lipolytic system
[52, 53]. Similar observations have been reported for HTG cases with heterozygous genotypes in other APOA5 mutations (Q139X, Q148X)
[9, 48, 49, 56]. In our study, two heterozygous carriers of the mutation (subjects 7 and 9) are both diabetic and overweight, showing an intermediate phenotype of HTG. They developed severe HTG although never reached TG values as high as the two homozygous 97X sisters, with no history of acute pancreatitis. On the other hand, the two homozygous affected sisters showed triglycerides levels much higher than in the cases previously described (112 mmol/L compared to 11 mmol/L in the case reported by Oliva and 40 mmol/L in the case reported by Charriere)
[52, 53]. It is possible that our patients carry other alleles that contribute to this more severe phenotype, such as common or rare gene variants described in the APOE, LPL or APOB genes
[16, 21, 22].
The fact that heterozygous carriers of this autosomic recessive disease also show some pathologic manifestations indicates either haploinsufficiency or a dominant negative effect of the truncated protein. In the first case, LPL may play its role properly in adequate conditions maintaining plasma TG levels within a normal range. However, if the system is subjected to an overload due to adverse environmental factors, the enzyme is no longer able to maintain normal TG metabolism, and HTG appears. In the second case, it can be postulated that the truncated peptide has a negative effect by itself. In this context, NMD (Nonsense Mediated mRNA Decay) is a cellular mechanism that eliminates mRNA that carries a premature stop codon in order to prevent the synthesis of truncated proteins. It has been proposed that NMD is not effective on premature stop codons located in the last exon of the gene or less than 50 nucleotides upstream of the last splice junction
. Thus, it is possible that the NMD mechanism does not occur in the Q97X premature stop codon, and the truncated protein is synthesized and itself interferes with the mature Apo A-V. Although previous reports of Q97X mutations did not find the truncated peptide plasma
[52, 53], one report indicates that the mature protein did not properly bind to TG-rich lipoproteins in Q97X heterozygotes
, suggesting that truncated peptide may be synthesized but not secreted to the plasma.
APOA5 is a relatively polymorphic gene showing common variants in the promoter region (−1131T>C) and in the signal peptide (S19W) that have been previously associated with elevated plasma TG levels
[28, 29, 58]. Recently, APOA5 -1131T>C polymorphism has been associated with higher plasma TG levels in a large epidemiologic study (16% increase for every C allele inherited)
. The presence of susceptibility alleles for HTG in APOA5 -1131T>C and S19W might be important in heterozygous carriers of Q97X when the mutation is located in a different haplotype than the susceptibility common variants APOA5 -1131C and 19W. In the French pedigree described by Charriere et al., most of the Q97X heterozygotes carried another susceptibility haplotype on the second chromosome
. Similar results were observed in the report by Marcais et al.
. In our family, only one of the three Q97X heterozygous subjects showed the 19W susceptibility allelic variant (subject 9), while the other heterozygous subjects showed the wild-type genotype in the -1131T>C variant (Figure
1). It is remarkable that even though subject 9 also carries the 19W susceptibility variant, he has not developed a more severe disease compared with heterozygous subject 7, who shows wild type S19 homozygous genotype. Nevertheless, we consider that it is not possible to assess the influence of S19W common variants in Q97X heterozygous subjects, given the limited number of subjects in this family.
Regarding Apo C-III levels, both the mother (subject 7) and one of the affected sisters of the proband (subject 12) in our study showed elevated levels of Apo C-III. It is remarkable that both were the only ones displaying severe HTG (>5.6 mmol/L) at the time the test was performed in 2006, when even the proband (who was under treatment) showed normal TG levels and also normal levels of this apolipoprotein (Figure
1). In previous reports on Q97X and other APOA5 mutations that lead to Apo A-V deficiency, higher levels of Apo C-III in the mutation carriers have also been described
[48, 49, 52, 53]. In fact, a 2.7-fold increase of Apo C-III was observed in Apo A-V deficient plasma compared to control plasma
. Furthermore, the S19W and −1131 T>C variants in the APOA5 gene have been associated with higher levels of Apo C-III
. On the other hand, it has also been described that hypertriglyceridemic patients display higher levels of this apolipoprotein compared to controls
. All these observations suggest a complex relation between Apo A-V and Apo C-III. Therefore, it is possible that the impaired TG hydrolysis in Apo A-V deficient plasma is a consequence of the lack of an LPL activator (Apo A-V), as well as the presence of an inhibitor (Apo C-III)
. The proband in our study, who is homozygous for the Q97X mutation and should have a deficiency of Apo A-V in plasma, did not show elevated levels of Apo C-III. Therefore, the observation of normal Apo C-III in the proband may suggest that, in our patients, plasma Apo C-III levels are mainly determined by HTG itself and not by the Apo A-V deficiency.