An overview of combined D-2- and L-2-hydroxyglutaric aciduria: functional analysis of CIC variants

Inclusion criteria for this study were the presence of SLC25A1 variants and combined D/L-2-HGA; 26 individuals were evaluated. Clinical data was collected from referring physicians using questionnaires. For previously published case reports (Muntau et al. 2000; Read et al. 2005; Edvardson et al. 2013; Chaouch et al. 2014; Mühlhausen et al. 2014; Prasun et al. 2015; Smith et al. 2016), data was completed by two of our authors (MW, AP) based on published information. No clinical data could be obtained for patient nos. 2 and 26 (sibling of patient no. 3). The D/L-2-HGA biochemical diagnosis was, in most cases (N = 22), established by our laboratory using liquid chromatography tandem mass spectroscopy (LC-MS/MS) measurements of D-2-HG and L-2-HG in urine, as previously described (Struys et al. 2004), or the information was collected via questionnaires (N = 1) or from literature (N = 3; two of three were reported not to have increased urinary D-2-HG and L-2-HG) (Chaouch et al. 2014).

For seven of the unpublished D/L-2-HGA-affected individuals, all exons and adjacent splice sites of the SLC25A1 coding region were amplified by polymerase chain reaction (PCR), as previously described (Nota et al. 2013). Sequencing analysis was performed using an ABI 3130xl genetic analyzer (Applied Biosystems, Nieuwekerk a/d Ijssel, NL), and data was interpreted using Mutation Surveyor (Softgenetics, PA, USA). Whole-exome sequencing, followed by direct Sanger sequencing, resulted in the genetic diagnosis in another affected individual.

The coding sequence of the SLC25A1 gene was recloned from pCMV6-AC-GFP (Origene, Rockville, MD, USA) into the pEGFP-N1 vector (Clontech). Subsequently, the enhanced green fluorescent protein (EGFP) was removed from the vector, as it interfered with protein function. For each of the 17 missense mutations included in this study, recombinant plasmids were generated by site-directed mutagenesis, as previously described (Betsalel et al. 2012). Successful mutagenesis and absence of PCR artifacts was confirmed by full-length sequencing of the SLC25A1 coding sequence.

SLC25A1 ^−/− fibroblasts from patient 9, homozygous for c.18_24dup; p.Ala9Profs*82 mutation, were transfected with wild-type SLC25A1 (wt), empty vector or mock transfected, by electroporation using 4D–Nucleofector™ system and P2 primary cell kit (Lonza, Cologne, Germany), following the manufacturer’s guidelines. Thereafter, the 17 SLC25A1 constructs were transiently transfected in SLC25A1 ^−/− fibroblasts using 1.5 million cells per condition. The pEGFP-N1 was cotransfected with the CIC-expressing vectors (with or without the introduced variants) in a ratio of 1 to 100. All experiments were performed in triplicate.

Twenty-four hours after transfection, cells were incubated for 48 h with Dulbecco’s modified Eagles medium (DMEM) enriched with [U-¹³C₆] glucose. Thereafter, culture media and cell pellets were collected and stored at −20 °C and −80 °C, respectively. To assess restoration of the CIC function in wild-type transfectants and activity of mutated proteins, [¹³C₂] citrate levels in cell culture media of transfected fibroblasts were used, as previously described (Nota et al. 2013). The percentage of residual CIC activity is expressed to the activities of wild-type transfectants, which were arbitrarily set at 100% in each experiment.

Transfected cells were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis. First, cells were lysed in urea lysis buffer (8 M urea, 100 mM NaCl, 10 mM Tris-HCl, pH 8.0) and sheared through an insulin syringe needle for DNA disruption. Protein content was determined using a bicinchoninic acid protein assay (Sigma-Aldrich, St Louis, MO, USA). Proteins (17 μg) were size-separated in a 12% NuPAGE® Bis-Tris precast gel (Invitrogen, Carlsbad, CA, USA) and transferred to a polyvinylidene fluoride (PVDF) membrane using the iBlot® Dry Blotting System (Invitrogen). Immunodetection of the SLC25A1 protein was carried out using rabbit polyclonal anti-SLC25A1 primary antibody (Proteintech, 15,235–1-AP), polyclonal goat anti-rabbit immunoglobulins/horseradish peroxidase (HRP) secondary antibody (Dako, P 0448) and enhanced chemiluminescent substrates (Lumi-Light plus Western blotting substrate; Roche Applied Science, Indianapolis, IN, USA). Images were acquired with the ChemiDoc MP imager (Bio-Rad Laboratories) and analyzed using Image Lab software. Actin was used as an internal loading control.

Using the scoring systems developed by Pierri et al. 2014, functional and/or structural importance of mutated residues was evaluated. The residue-specific score (RS) estimates the strength of the evolutionary selection on each amino acid residue of an individual mitochondrial carrier. Considering all 216 residues previously scored for SLC25A1 (Pierri et al. 2014), a residue with an RS > 4.68 is above the median value and therefore considered functionally and/or structurally important in the human CIC. A transversal score (TS) was calculated by averaging the 53 residue-specific scores corresponding to equivalent positions identified by the multialignment of the 53 members of the human mitochondrial carrier protein family (MCF), using the Bos taurus protein BtAAC1 structure as reference. A residue with a TS > 3.79 is above the median value and considered functionally and/or structurally important in the common MCF structure and transport mechanism.

The homology model of human CIC (residues 23–299) was made with MODELER (Fiser and Sali 2003) based on the X-ray structure of the bovine adenosine diphosphate/adenosine triphosphate (ADP/ATP) carrier (Pebay-Peyroula et al. 2003).

An overview of all 22 mutations found in the SLC25A1 gene, including six novel, is given in Fig. 1. Most of these mutations are private, and only seven mutations were found in more than one patient. The most frequent mutation is p.Ala9Profs*82, detected in five patients (four apparently unrelated families).

Presenting symptoms are summarized in Table 1. Dysmorphic features seen more than once are prominent forehead–frontal bossing in four patients, bitemporal hypoplasia–midface hypoplasia in three, hypertelorism in four, and down-slanting eyes in three. A low or flat nasal bridge was seen in four, low-set or rotated ears in three, abnormal thumbs in two, micrognathia in three, and retrognathia in two. Head circumference was normal in seven; six patients were microcephalic and one macrocephalic. Weight was normal in ten, and four patients had growth retardation. Delayed motor milestones were reported in 16 patients, with no development at all in two patients. Motor function evaluation was not determined or given in seven cases. Cognitive development was impaired in 13 patients, and no development was observed in three. No specific tests were performed to evaluate cognition. Other specific abnormalities are optic atrophy, seen in patient 1, a feature described previously by Edvardson et al., in patient 22.

At follow-up, epilepsy becomes a more prominent feature. Seizures were seen clinically or on electroencephalogram (EEG) in 20 patients, apnea was reported as seizure in three cases. Four patients showed no seizures, but this was on the grounds of a normal EEG during apneic attacks. Radiographic information from magnetic resonance imaging (MRI), computed tomography (CT scan), or brain ultrasound (US) investigations showed brain atrophy and/or ventriculomegaly in 14 of 20 patients and corpus callosum hypoplasia or dysplasia in 12 of 20. Germinolytic cysts or germinal matrix cysts were seen in two patients. Bilateral frontal lobe cysts, evidence of prenatal intraventricular bleeding, and cerebellar hypoplasia and gliotic hyperintensity of frontal white matter were seen separately in individual cases. Brain MRI of patient 24 was normal (Chaouch et al. 2014). Initial laboratory findings revealed lactic acidosis in 14 of 17 cases. Urinary levels of Krebs metabolites of seven of eight unpublished patients are shown in Supplementary Table 1.

Transfection of SLC25A1 ^−/− fibroblasts with wild-type SLC25A1 not only restored SLC25A1 levels, as detected by immunoblotting (Fig. 2a), but also resulted in the expected increase of citrate efflux (by > 3-fold compared with cells transfected with an empty vector) (Fig. 2b), accompanied by decreased intracellular D-2-HG and L-2-HG levels (Fig. 2c).

Residual activities of studied missense variants varied from almost 0 to 85% of that of wild-type transfectants. Patients with a severe clinical outcome—death before 1 year and/or intensive medical care—had missense variants with residual activities up to 23% and/or severe truncating mutations (i.e., nonsense or frameshift), considered to result in no CIC activity. Therefore, by using the 25% residual activity as cutoff value, we divided the D/L-2-HGA patient cohort into three groups: (1) patients with two severely affected CIC alleles (truncating or missense mutations with < 25% residual activity); (2) patients compound heterozygous for a severely and a mildly affected allele; (3) patients with two relatively mildly affected CIC alleles (with residual activity > 25%).

In group 1, we included four patients carrying two truncating mutations and eight (nos. 3, 6, 8, 16, 17, 18, 22, and 23) with missense variants showing severe impairment of the CIC function (< 25%; Table 1). Five of these missense variants were detected in patients in homozygous form: c.844C > T; p.Arg282Cys (patient 3), (c.139G > A; p.Glu47Lys (patient 16); c.278G > A; p.Gly93Asp (patient 17); c.430G > C; p.Glu144Gln (patient 8), and c.593G > A; p.Arg198His (patient 18). The first four of these five patients presented with a rapidly progressing, severe phenotype associated with early death before 1 year of age (Table 1). The fifth patient (patient 18), homozygous for c.593G > A; p.Arg198His that had an early onset with severe clinical signs, was completely care dependent but was alive at the age of 5 years (Smith et al. 2016). Three of the eight group 1 patients with missense variants were compound heterozygous: c18_24dup; p.Ala9Profs*82 and c134C > T; p.Pro45Leu (sibling patients 6 and 23); c.389G > A; p.Gly130Asp and c.845G > A; p.Arg282His (patient 22). The latter (also described by Edvardson et al. 2013) presented with a severe phenotype as well, being the first CIC patient described with agenesis of corpus callosum and optic nerve hypoplasia. The four patients (nos. 7, 9, 13, and 14) with truncating mutations also had a severe phenotype and died before 1 year of age. Patients 2 and 26 (sibling of patient 3) should also be included in group 1, as they were homozygous for the severe c.844C > G; p.Arg282Gly and c.844C > T; p.Arg282Cys missense mutation, respectively. However, no clinical data could be obtained to confirm this.

The four patients of the second group of SLC25A1-deficiency (nos. 5, 19, 21, and 12) were all compound heterozygous for a severe and a mild allele. Patient 5 was compound heterozygous for missense substitution c.499G > A; p.Gly167Arg (49% activity) and the p.Ala9Profs*82 mutation, which could explain the relatively severe phenotype of the patient, which was associated with death before 2 years of age. Patient 19 was compound heterozygous for a novel frame shift mutation, c.648_655del; p.Met218Serfs*25 and the c.119 T > A; p.Ile40Asn (34% activity). Despite the poor clinical condition of a persistent vegetative state and being dependent on mechanical ventilation and tube feeding, patient 19 was alive (23 months) at the time of the questionnaire; citrate treatment was started at 3 months of age at 800 mg/kg per day, and since then, fewer cerebral attacks were registered. The c.844C > T; p.Arg282Cys missense variant (5% residual activity) was not only detected in the homozygous form in patient 3 (patient group 1) but also in patient 21, compound heterozygous with the c.605 T > C; p.Met202Thr variant (66% activity). The latter was also heterozygous in patient 10 of group 3. Patient 21 had a milder phenotype than patient 3, probably explained by the fact that the second allele of patient 21 harbored a missense variant with high residual CIC activity. Interestingly, this patient also presented clinical features of a myasthenia crisis during respiratory arrest events, precipitated by intercurrent illness. Supplementation with citrate at 800 mg/kg per day prevented any further need for hospitalization during febrile illnesses. The last patient in this group, patient 12, was compound heterozygous for a frameshift mutation and the c.821C > T; r.[821c > t, 820_821del] mutation, involving a splice site. This splice mutation, which mainly results in expression of the frameshift transcript (r. 820_821del; p.Ala274Ilefs*24), may also result in some authentic spliced transcript containing a missense variant (p.Ala274Val; Nota et al. 2013, unpublished data). To investigate the activity of this potential minor transcript r.821c > t (p.Ala274Val), we studied this variant as well, which had 85% residual activity. Patient 12 is the first patient for which the effects of citrate treatment on the clinical course are described in detail (Mühlhausen et al. 2014).

The third group of D/L-2-HGA patients included six cases (five unrelated families), of which four were homozygous (nos. 1, 15, 24, and 25) and two compound heterozygous (nos. 10 and 20). We detected six missense variants with residual activities ranging from 30 to 71% of the activity of wild-type transfectants. The c.578C > G; p.Ser193Trp (31% activity) was detected in two unrelated patients: in homozygous form in patient 1 and heterozygous form in patient 20. Patient 1 was 21 months old at the date of receiving the completed questionnaire, and had early symptom onset at 10 days of age. The patient underwent intermittent respiratory support via continuous positive airway pressure (CPAP) through tracheostomy. Patient 20, aged 3 years, was compound heterozygous for the c.578C > G; p.Ser193Trp and the c.82G > A; p.Ala28Thr variant (71% activity). By comparison, the phenotype of patient 1, homozygous for the c.578C > G; p.Ser193Trp variant, was more severe than that of patient 20. Based on this observation, we speculate that compound heterozygosity of this variant with a higher residual activity missense variant represents an advantage for clinical course and severity of the disease. A similar correlation was observed for patient 10, who was compound heterozygous for the c.605 T > C; p.Met202Thr (66% activity) and c.890A > G; p.Tyr297Cys (30% activity); this patient died at the age of 11 years with typical manifestations of D/L-2HGA. Patients 10 and 20 had a milder clinical phenotypes and were not dependent on mechanical ventilation; tube feeding was only occasionally needed. For patient 20, citrate treatment was initiated at 9 months of age at a dosage of 1500 mg/kg per day and resulted in developmental gains, improved muscle tone, and absence of clinical seizures and apneic events. The effect of citrate treatment in patient 10 cannot be properly assessed because of the very late start in the disease course (at 10.5 years of age) and the very short duration of the trial (1 or 2 weeks, respectively). The homozygous variant, c.740G > A; p.Arg247Gln (52% activity), was present in an adult sibling pair (patients 24 and 25) with a very mild, atypical phenotype, suggestive of congenital myasthenia syndrome (Chaouch et al. 2014). Patient 15 was homozygous for c.784 T > C; p.Cys262Arg variant, which displays 48% residual activity. Intriguingly, despite the relatively high residual activity, this patient presented a severe clinical phenotype of CIC deficiency, with abnormal brain MRI findings and recently developed clinical seizures. This patient (2 years of age) was dependent on intensive medical support. However, he was able to have breathing sprints off the ventilator for 8 h per day, during which time his pulse oximetry remained > 93% on room air. Otherwise, he was ventilator-dependent without the need of supplemental oxygen. In addition to the patients discussed above, two apparently unrelated patients (nos. 4 and 11) were included in this group. In both, the above described c.821C > T mutation (r.[821c > t, 820_821del]) was detected in homozygous form.

Positions of missense mutations were mapped onto the homology model of CIC (Fig. 3). It is noteworthy that all residues affected by these missense substitutions have a TS above the threshold of 3.79 (Supplementary Table 2), with the exception of alanine 28, with a TS of 3.40; it also has the highest residual activity of all the disease-causing CIC mutations (Table 1). This observation, together with the fact that various inactivating mutations have been reported at corresponding sites in other MC (Abrams et al. 2015; Ma et al. 2007; Palmieri 2014; Shamseldin et al. 2016; Thompson et al. 2016); (Supplementary Table 2), indicates that these mutations affect residues important in the common structural and functional features of MCF and potentially lead to impaired transport activity.

The single mutations of patients in group 1 are found in functionally important locations within the CIC protein: (1) in the area of the substrate-binding site, consisting of residues in the contact points on each of the three even-numbered transmembrane helices (Robinson and Kunji 2006) and residues in the vicinity of the contact points that also contribute to determine substrate specificity (Marobbio et al. 2008; Monné et al. 2012); (2) in the 3-fold repeated and highly conserved signature motif sequence PX[D/E]XX[K/R]X[K/R]X20–30[D/E]GXXXX[W/Y/F][K/R]G (PROSITE PS50920, PFAM PF00153 and IPR00193), found in members of the mitochondrial carrier family (Palmieri 1994). Here, prolines and first glycines are located in the odd- and even-numbered transmembrane helices, respectively, on the matrix side of the substrate-binding site, constituting the so-called proline–glycine (PG) level 2, implicated in opening and closing the matrix gate of the central substrate translocation pore (Palmieri and Pierri 2010a, b); (3) at PG level 1, where prolines and glycines are found in even- and odd-numbered transmembrane helices on the intermembrane space side, and are suggested to participate in opening and closing the cytoplasmic gate of the central substrate translocation pore (Palmieri and Pierri 2010a, b). It is also noteworthy that all residues affected by these missense substitutions are located in positions with high RS. All RSs of group 1 patients ranging from 5.04 (Glu144Gln) to 6.07 (Arg198His) are above the threshold of 4.68, indicating that these mutations affect residues important in the CIC (Table 2), in keeping with the observed functional data.

All residues affected by missense substitutions of groups 2 and 3 are found within the CIC protein: (1) inside the carrier cavity but not in the substrate binding area, with one exception; i.e., isoleucine 40; (2) outside the substrate translocation pore toward the membrane and at PG level 1; (3) on the outside of the matrix gate. They all have RSs above the threshold of 4.68, except for alanine 28, with the lowest RS (3.34), which corresponds to the highest residual activity of all disease-associated CIC mutations (71% of WT); arginine 247 with a RS of 3.38, and isoleucine 40 with a RS of 3.50. The latter residues display substantial residual transport activity (52% and 34%, respectively) (Table 2).