A mutation creating an upstream translation initiation codon in SLC22A5 5′UTR is a frequent cause of primary carnitine deficiency

Abstract Primary carnitine deficiency is caused by a defect in the active cellular uptake of carnitine by Na+‐dependent organic cation transporter novel 2 (OCTN2). Genetic diagnostic yield for this metabolic disorder has been relatively low, suggesting that disease‐causing variants are missed. We Sanger sequenced the 5′ untranslated region (UTR) of SLC22A5 in individuals with possible primary carnitine deficiency in whom no or only one mutant allele had been found. We identified a novel 5′‐UTR c.‐149G>A variant which we characterized by expression studies with reporter constructs in HeLa cells and by carnitine‐transport measurements in fibroblasts using a newly developed sensitive assay based on tandem mass spectrometry. This variant, which we identified in 57 of 236 individuals of our cohort, introduces a functional upstream out‐of‐frame translation initiation codon. We show that the codon suppresses translation from the wild‐type ATG of SLC22A5, resulting in reduced OCTN2 protein levels and concomitantly lower transport activity. With an allele frequency of 24.2% the c.‐149G>A variant is the most frequent cause of primary carnitine deficiency in our cohort and may explain other reported cases with an incomplete genetic diagnosis. Individuals carrying this variant should be clinically re‐evaluated and monitored to determine if this variant has clinical consequences.

Pasquali, 2016). The clinical presentation of primary carnitine deficiency shows great variability ranging from hypoketotic hypoglycemia and hepatic encephalopathy early in life, skeletal-and cardiomyopathy later in life, sudden death from cardiac arrhythmia to only fatigue, or no clinical symptoms at all (see for review (Longo, 2016;Magoulas & El-Hattab, 2012)). The first individuals identified with primary carnitine deficiency all presented with severe clinical symptoms, including acute metabolic decompensation or cardiomyopathy, but because the disorder has been included in many neonatal screening programs worldwide, an increasing number of individuals with milder or no clinical symptoms have been identified. These include asymptomatic mothers with this condition, identified via neonatal screening of their newborn children, who are carriers and picked up due to maternal carnitine deficiency (Longo, 2016;Schimmenti et al., 2007). Although these mothers are usually asymptomatic, they may be at risk for sudden death from arrhythmia (Longo, 2016;Schimmenti et al., 2007).
Primary carnitine deficiency is an autosomal recessive disorder caused by mutations in the OCTN2-encoding SLC22A5 gene located on chromosome 5q31 (Nezu et al., 1999;Tang et al., 1999). Over 150 different mutations in SLC22A5 have been reported (Frigeni et al., 2017;Li et al., 2010). It has been shown that the frequency of nonsense mutations is significantly increased in symptomatic patients and that the residual carnitine-transport activity is lower in fibroblasts from symptomatic patients than in fibroblasts from asymptomatic women (Rose et al., 2012).
When primary carnitine deficiency is suspected or a newborn is identified via screening, first plasma acylcarnitine analysis, including free carnitine levels, is performed followed by measurements of carnitine-transport activity in fibroblasts and/or genetic analysis of the SLC22A5 gene (Longo, 2016;Magoulas & El-Hattab, 2012). The carnitine-transport activity assay is reliable for confirming the diagnosis but is only performed in a small number of laboratories throughout the world and requires a skin biopsy being taken. For this reason, in most cases, genetic analysis is performed.
Recently, it was shown in a large study including fibroblasts from 358 subjects with possible primary carnitine deficiency, that carnitine-transport activity was reduced to 20% or less of normal in fibroblasts of 140/358 subjects. Subsequent Sanger sequence analysis of 95 of the 140 biochemically proven OCTN2-deficient cells revealed mutations in the coding regions of only 84% of the SLC22A5 alleles (Frigeni et al., 2017).
We also observed a relatively low diagnostic yield with genetic testing of the SLC22A5 gene. Sanger sequencing of the coding regions of the SLC22A5 gene in 236 individuals with possible primary carnitine deficiency revealed bi-allelic mutations in 133 individuals (56.4%). In the remaining individuals, we only identified one heterozygous (69 individuals; 29.2%) or no mutation (34 individuals; 14.4%).
Here we report the identification and characterization of a 5′ untranslated region (UTR) variant in SLC22A5, c.-149G>A, which introduces a functional upstream out-of-frame translation initiation codon. We show that this codon suppresses translation from the wildtype AUG of SLC22A5, resulting in reduced OCTN2 protein levels and concomitantly lower OCTN2 transport activity explaining the carnitine deficiency in patients harboring this variant. We identified this variant in 57 individuals in whom we previously identified only one or no mutation in the coding region of SLC22A5.
Expanded cell colonies were plated in 96 wells plate for 24 hr after which NanoLuc luciferase activity (Nano-Glo Luciferase assay, Promega) was measured according to the manufacturer′s protocol using the spectrophotometer Infinite M200 Pro (Tecan).

| Carnitine transport assay
Control and patient primary fibroblasts (equivalent to approximately 20 µg protein/well) were cultured in quadruplicate in 24-well tissue culture plates (Costar) in HAM F-10 (Gibco) supplemented with 10% Fetal Bovine serum (Bodinco), 100 U/ml penicillin, 100 mg/ml streptomycin (LifeTechnologies), and 250 ng/ml Fungizone (LifeTechnologies) in a humidified atmosphere of 5% CO 2 at 37°C for 1 week. The cells were washed three times with phosphate-buffered saline (PBS; room temperature) and incubated in Dulbecco′s PBS supplemented with 5 mM glucose, 1 mg/ml bovine serum albumin and 5 µM D 3 -carnitine with and without 0.625 mM of mersalyl acid (Sigma Aldrich) at 37°C 5% CO 2 . After 2 hr, the cells were washed with ice-cold PBS and the reaction was stopped by adding ethanol supplemented with 25 pmol D 9 -carnitine per well as an internal standard. Cells were incubated for 1 hr at −20°C. The ethanol was taken from the cells and evaporated under a stream of nitrogen.
The residue was dissolved in 50 µl of 0.1% heptafluorobutyric acid (HFBA; Thermo Scientific, Waltham). Unlabeled, D 3 and D 9 -carnitine were measured using UPLC-MS. The UPLC-MS system consisted of an Acquity SDS liquid chromatography system coupled to a Premier-XE mass spectrometer (Waters).

| Functional assessment of the c.-149G>A variant
To determine if the 5′-UTR c.-149G>A variant introduces a functional translation initiation codon that is recognized and used in vivo, we transfected HeLa cells with different reporter constructs generated from pcDNA3 containing the NanoLuc luciferase ORF preceded by wild-type or mutant 5′UTR sequences and the first 15 or 16 nucleotides of exon 1 (see Figure 2). Cells transfected with the reporter constructs in which either the 5′UTR containing the wild-type AUG or the 5′UTR containing the mutant AUG is in frame with NanoLuc luciferase, both showed luciferase activity, indicating that the mutant AUG can be used for translation initiation (see Figure 2). However, the luciferase activity in the cells transfected with the 5′UTR containing the mutant AUG in frame with NanoLuc luciferase was much lower than the activity in the cells transfected with the wild-type 5′UTR containing the wild-type AUG in frame with NanoLuc luciferase (11% compared with the wild-type AUG).
Moreover, cells transfected with the 5′UTR containing both the mutant and the wild-type AUG and with the wild-type AUG in frame with NanoLuc luciferase also showed reduced luciferase activity.
Taken together, these findings indicate that the novel AUG created by the 5′-UTR c.-149G>A variant creates a functional, but weaker, translation initiation codon, which suppresses translation from the wild-type AUG of SLC22A5.

| Functional consequence of the c.-149G>A variant
The functional expression studies indicate that the presence of the upstream AUG causes a reduction in the synthesis of OCTN2.
Unfortunately, this cannot be assessed by immunoblot analysis in homogenates of patient fibroblasts due to the lack of specific antibodies that recognize OCTN2 in fibroblasts. As an alternative,  mutation, which is known to be very frequent in asymptomatic individuals picked up via NBS (Li et al., 2010;Magoulas & El-Hattab, 2012), has an allele frequency of 21.1% in our cohort.

| DISCUSSION
To study the consequence of DNA variants on OCTN2 activity in patients fibroblasts we developed a novel sensitive assay based on tandem mass spectrometry. This assay is more sensitive than the currently used radiochemical assays and allows accurate F I G U R E 3 Functional consequence of SLC22A5 c.-149G>A variant. The carnitine-transport activity was measured in fibroblasts of control subjects and fibroblasts with different mutations in SLC22A5. The genotype of the different cell lines is indicated and the activity is expressed as a percentage of the mean activity of the two control cell lines measured in the same experiment. All measurements were done in duplicate and all cell lines were analyzed in two independent experiments minimally, except for one cell line (*) which was only analyzed in one experiment measurement of low residual activities. Using this assay, we found that the residual OCTN2 activity in fibroblasts homozygous for the c.-149G>A variant was higher (31%) than the activity in fibroblasts homozygous for the c.136C>T (p.P46S) mutation (17%).
As far as we are aware, individuals carrying the c.-149G>A variant never experienced any severe clinical symptoms that could be related to primary carnitine deficiency, such as severe brain dysfunction, cardiomyopathy, muscle weakness, or hypoglycemia.
However, it will require clinical re-evaluation and long-term monitoring to determine if this variant has any clinical consequences.
Mutations in the 5′UTR as cause of disease has been described for different diseases, including autosomal recessive metabolic disorders (Barbosa, Peixeiro, & Romão, 2013;Fu et al., 2015;Semler et al., 2012;von Bohlen et al., 2017;Willemsen et al., 2017) and thus should be considered in patients with autosomal recessive disease in whom only one heterozygous mutation has been found. Indeed, such disease-causing mutations may be more frequent than previously assumed due to the fact that most clinical DNA testing specifically focuses on the coding regions of genes.