3-methylcrotonyl-CoA carboxylase deficiency: Clinical, biochemical, enzymatic and molecular studies in 88 individuals

Eighty-eight subjects with MCC deficiency from 78 families (10 sib-pairs) were included in this study (Tables1,2,3 and4). Cultured fibroblasts (n = 69) or genomic DNA (n = 16) from 85 individuals were sent to our laboratory for confirmation of MCC deficiency. In the remaining 3 subjects mutation analysis was performed in a genetic laboratory in the USA. Forty-five subjects were male, 40 were female; the sex of 3 individuals was not reported. Of the 88 individuals 45 (51%) were Caucasian, 27 (31%) Turkish, 8 (9%) Arab, 6 (7%) Asian, one (1%) African-American, and one patient (1%) was of mixed African Caucasian ancestry.

Clinical, biochemical, enzymatic or mutation data of 32 individuals have been reported earlier [proband 20-32a, 33-34, 36-43a, 44 and 46[11], 30[23], 35a and 35b[30], 42[37], 44[38], 50[28], 80[9], 81[39], 96a and 96c[10], and 136[26]].

A questionnaire was designed and sent out to the treating physicians. This questionnaire specifically addressed the mode of diagnosis, clinical symptoms, the psychomotor development, biochemical markers and long-term treatment regimens. Additionally, in some cases medical reports that were sent in together with the diagnostic samples were available and used for clinical data collection.

Fibroblasts were cultured in a culture medium containing 10% foetal calf serum, and the activities of PCC and MCC were assayed simultaneously in crude fibroblast homogenates by measuring the incorporation of ¹⁴C-bicarbonate into acid-non-volatile products as described earlier[40].

After obtaining informed consent mutation analysis was performed in 83 individuals. The 5 individuals in whom no mutation analysis was performed were siblings of index cases and presented with a metabolite profile typical for MCC deficiency. In probands from whom RNA and DNA were available RT-PCR amplification and sequencing of the entire MCCC2 ORF was first performed and if no clearly pathogenic coding alterations were detected in MCCC2, the entire MCCC1 ORF was also analyzed. Identified mutations were confirmed by PCR amplification of genomic DNA. In individuals from whom only DNA was available amplification of all MCCC1 and MCCC2 exons and flanking intronic sequences from genomic DNA followed by direct sequencing was performed.

RNA and genomic DNA were extracted from cultured skin fibroblasts or peripheral blood leukocytes using the QIAamp® RNeasy or QIAamp® RNA Blood Mini Kit and the QIAamp® DNA Mini Kit (Qiagen AG, Basel, Switzerland), respectively. The RT-PCR reaction was performed using the One-Step RT-PCR kit (Qiagen AG, Basel, Switzerland) following the manufacturer’s instructions. First-strand MCCC1 and MCCC2 cDNA was amplified as described[2]. PCR products were sequenced in a thermocycler and analyzed with an ABI Prism 3100 Avant using the dye-terminator method (Applied Biosystem, Rotkreuz, Switzerland) according to the manufacturer’s instructions. To confirm mutations identified in RT-PCR products, a genomic fragment containing the corresponding exon was amplified using flanking intronic primers, and the PCR product was sequenced directly. In cases where only one of the two alleles could be identified in the standard RT-PCR product all exons and flanking intronic regions were sequenced. The sequences of all primers are available upon request.

To exclude that the identified missense mutations are common polymorphisms we amplified the relevant exons from genomic DNA of 100 controls (200 alleles).

The following mutations were introduced by site-directed mutagenesis into the existing wildtype MCCC1 and MCCC2 pCR Blunt II TOPO vector (Invitrogen, Basel, Switzerland): MCCC1 p.E288G, p.G379D, p.I434M and MCCC2 p.S39F, p.G118del, p.Y146N, p.H282R, p.V434L, p.A456V, p.G475R, p.S523G. All constructs were then transferred into the mammalian expression vector pTracer-CMV2 (Invitrogen, Basel, Switzerland) and seq-uenced for validation. For expression studies the constructs were transiently transfected into transformed cultured fibroblasts deficient in either MCCC1 (homozygous for c.1264_1265insG/p.Q422Rfs*10 or compound heterozygous for c.1264_1265insG/p.Q422Rfs*10 and c.1682-3A > G/p.N561Kfs*10) or MCCC2 (homozygous for c.127 C > T/p.Q43*) by electroporation as described[2]. The cells were harvested 48 hours later and assayed for MCC and PCC activity.

Western blot analysis of proteins extracted from cells harvested 48 h after transfection was performed as described earlier[41]. For immunostaining of MCCC2 a commercially available antibody was used (Abnova). Antiserum for MCCC1 was produced by inoculating rabbits with peptides corresponding to the hydrophilic strech of the last 19 C-terminal amino acids (RHTPLVEFEEEESDKRESE) of human MCCC1 conjugated to keyhole limpet hemocyanin (Covance, Denver, Colorado). β-Actin was stained as control.

The biotin-containing MCC and PCC α-subunits and PC were also stained using streptavidin-alkaline phosphatase followed by colorimetric detection (Transcend™ Non-Radioactive Translation Detection Systems Kit, Promega, Dübendorf, Switzerland).

Twenty-six individuals (30%) were diagnosed by selective metabolic screening (SMS) due to clinical symptoms (n = 18) or a positive family history (n = 8) while 53 individuals (60%) were identified by expanded NBS. Additionally, 9 mothers (10%) were diagnosed following a positive NBS result of their healthy offspring. In patients in whom a metabolic work-up was initiated due to clinical symptoms age at diagnosis ranged between one week and 13 years (median 1.5 years) (Table2). Patients identified by family screening had a median age at diagnosis of 3.75 years (range 1.5-17 years) (Table3), and mothers were diagnosed at a median age of 30.5 years (range 24-38 years) (Table4).

In 34 (43%) of 80 subjects clinical symptoms were reported ranging from acute metabolic decompensation with ketoacidosis, hypoglycemia and encephalopathy to neuromuscular symptoms, mental retardation or attention deficit hyperactivity disorders (Figure1). Twelve patients, 5 of which were diagnosed by newborn screening, had at least one acute metabolic decompensation. The most common clinical symptoms of acute crises were vomiting and encephalopathy with impaired conciousness. Neurologic symptoms like seizures, metabolic stroke, hemiparesis and cerebral edema were less frequent. The most common laboratory findings were acidosis and hypoglycaemia. Among chronic symptoms mental retardation including speech retardation were the most common findings followed by seizures, muscular hypotonia, muscle weakness, muscle pain and failure to thrive. In 5 patients an attention deficit hyperactivity disorder was reported.

Thirty-five out of 61 (57%) living individuals of whom recent follow-up information was available or who had been followed for at least until the age of three years have remained asymptomatic. Notably, 25 (69%) of the 36 subjects identified by NBS of whom either recent information or follow-up data until the age of at least three years were available, have remained without symptoms.

Sixty-nine out of 75 individuals of whom information on treatment was available received dietary and/or medical therapy. Forty individuals were given a protein-restricted or leucine-restricted diet (which was stopped later in life in some cases) and in 23 a leucine-free amino acid mixture was administered at least temporarily. In 30 subjects oral biotin was given on a trial basis. 63 patients were supplemented with oral carnitine, and 10 patients received oral glycine.

Presence of C5OH in blood or dried blood spots and presence of elevated urinary excretion of 3-HIVA and 3-MCG at the time of diagnosis are shown in Tables1,2,3 and4. A biochemical phenotype characteristic for MCC deficiency with elevated excretion of 3-HIVA and/or 3-MCG, defined as more than twice the upper normal value, was found in 85% (68/80) of individuals. In 14% (11/80) only mildly elevated excretion was detected. In one patient (1%) no elevated excretion of 3-HIVA and 3-MCG was detected. Notably, 5 individuals including one patient reported earlier[39] did not excrete 3-MCG, the pathognomonic metabolite of MCC deficiency. However, one of these individuals showed massive excretion of 3-MCG when re-evaluated 6 months later.

C5OH in dried blood spots was highly elevated in 85% (66/78) and slightly (less than twice the upper normal value) elevated in 15% (12/78) of individuals. Low C5OH concentrations were not always linked with low urinary excretion of metabolites and vice versa.

In 43/68 individuals (63%) a secondary carnitine deficiency was present. Remarkably, 24 (60%) of those 40 children identified through NBS of whom information on free carnitine concentrations was available had decreased free carnitine levels already at the time of diagnosis or within the neonatal period. In 15 of them free carnitine concentration was below 5 μmol/l.

Of the 83 individuals in whom mutation analysis was performed, 31 had mutations in MCCC1 and 52 in MCCC2 (Tables1,2,3 and4). Forty-eight probands were found to be homozygous (12 for MCCC1 mutations and 36 for MCCC2 mutations) while 28 were compound heterozygous (15 for MCCC1 mutations and 13 for MCCC2 mutations). In 5 of these patients RT-PCR results showed exon skipping either for one or both alleles, but the underlying genomic mutation could not be identified. In the remaining 7 subjects (4 MCCC1 and 3 MCCC2) it was not possible to detect a second mutation in spite of sequencing all exons and flanking intronic sequences. However, the mutant allele identified appeared to be homozygous in RT-PCR, but was clearly heterozygous in genomic DNA. This suggests that the steady state level of mRNA from the second allele was not detectable as would be the case for a promoter mutation or an intragenic deletion or insertion missed by genomic PCR.

We identified a total of 15 novel MCCC1 and 16 novel MCCC2 mutations (shown in bold in Tables5 and6). The 15 novel MCCC1 mutations comprise 7 missense, 2 nonsense, 1 splice site and 5 frameshift mutations (5 due to small deletions and one due to a small insertion). The 16 novel MCCC2 variants include 11 missense and 4 splice site mutations and 1 deletion of a single amino acid.

The functional consequences of three MCCC1 (p.E288G, p.G379D, p.I434M) and 8 MCCC2 (p.S39F, p.G118del, p.Y146N, p.H282R, p.V434L, p.A456V, p.G475R and p.S523G) mutations were investigated by expression studies (Table7). The MCCC1 p.E288G and p.G379D mutations showed no residual activity while the p.I434M mutant allele yielded on average 46% of MCC1 wildtype activity.

Only one of the MCCC2 mutations, p.A456V, showed virtually no enzyme activity whereas seven mutations were found to have residual activity ranging on average from 9 to 84% of MCC2 wildtype activity. Western blot analysis of the expressed proteins revealed virtually wildtype levels for the MCCC1 p.I434M protein while the levels of MCCC1 p.E288G and MCCC1 p.G379D proteins were severely reduced or not detectable (Figure2). Five of the 8 expressed MCCC2 proteins (p.Y146N, p.H282R, p.A456V, p.G475R and p.S523G) were detected at normal or only slightly reduced levels, while the p.S39F and p.V434L protein levels were severely reduced and no protein was detectable after transfection with the p.G118del construct (Figure3).

MCC deficiency has been described as a genetic condition with low clinical penetrance[6]. Results of a comparative analysis of case reports with NBS data reported by Stadler and co-workers[6] suggest that less than 10% of affected individuals ever develop minor symptoms and only less than 1 to 2% might have a risk for severe adverse outcome. In their study all 14 individuals diagnosed by NBS remained asymptomatic during a follow-up period of 1.75 to 6.5 years. In contrast, in our study only 69% of the 36 subjects identified by NBS with a follow-up of at least 3 years have stayed completely asymptomatic while the remainder (31% =11) developed clinical symptoms including various neurologic symptoms as well as at least one acute metabolic decompensation in 5 children. This indicates that early diagnosis with concomitant early initiation of therapy and counselling of parents cannot prevent a clinical manifestation in all cases. However, our data have to be interpreted with caution, since the source of patients being those sent to a diagnostic referral laboratory may lead to a selection bias in favour of symptomatic individuals since it can be assumed that samples for confirmatory diagnosis are more likely to be obtained from symptomatic subjects than from asymptomatic individuals. Additionally, the clinical phenotype of MCC deficiency is still not well-defined.

In our study population the most common clinical features were acute episodes of metabolic acidosis with vomiting, hypoglycemia and acidosis, muscular symptoms such as muscular hypotonia, muscle weakness and muscle pain and neurological abnormalities including developmental delay and seizures as well as attention deficit hyperactivity disorders (Tables1,2,3 and4). The acute metabolic crises reported in 12 patients are likely to be caused by the underlying metabolic defect. However, the causative attribution of all other clinical signs, especially of unspecific chronic neurologic symptoms such as mental retardation, attention deficit disorders and fatigue to MCC deficiency remains questionable. It could be speculated that these symptoms might as well be caused by undiagnosed genetic defects other than MCC deficiency. Such an additional genetic disorder –though unprobable- would be more likely in individuals that are the product of a consanguineous union; thus a higher share of symptomatic individuals would be expected in this subgroup when compared to subjects whose parents are reported not to be consanguineous. However, no significant difference in the clinical manifestation rate between the two subgroups could be shown (41% symptomatic patients in both subgroups, 12 of 29 individuals with reported parental consanguinity versus 18 of 44 subjects with parents reported not to be consanguineous).

Altogether, though the share of individuals with clinical symptoms was higher in our study population, our data underline the observation of Stadler and co-workers[6] that compared to other organic acidemias, individuals with MCC deficiency appear to have a significantly higher tolerance toward metabolic stress; even complete absence of MCC activity seems to cause clinical manifestations only in association with environmental triggering factors in a rather small subgroup of individuals. Dietetic treatment is usually not required. However, considering the frequency of carnitine deficiency in our study population, regular monitoring of free carnitine concentrations and – if necessary - oral carnitine substitution seems to be warranted. Also, an emergency regimen during intercurrent illness may be advisable.

As shown in Tables1,2,3 and4 the vast majority of individuals displayed a typical biochemical phenotype with accumulation of metabolites characteristic for MCC deficiency in both blood and urine at the time of diagnosis. Mild elevations of one or more metabolites were the exception and were not more common in asymptomatic individuals. A completely unremarkable urine organic acid pattern was detected in only one woman. However, the concentration of C5OH in her blood was clearly elevated at the same time.

The phenomenon that mutations may cause a clear biochemical phenotype in otherwise asymptomatic individuals is well known from other inborn errors of metabolism implemented in expanded NBS such as isovaleric acidemia[42] or medium-chain acyl-CoA dehydrogenase deficiency[43]. From the data available in this study we were able to confirm earlier observations that in MCC deficiency there is no apparent association between the biochemical and the clinical phenotype and that a mild biochemical phenotype does not seem to be a predictor of a mild clinical expression[2, 6, 11]. Furthermore, we have recently shown that also individuals who are carriers of a single mutation at the MCCC1 locus and have residual MCC activity greater than 20% of control may present with a mild biochemical phenotype characteristic for MCC deficiency[12].

In all but two individuals MCC activity in fibroblasts was severely reduced combined with normal activity of PCC. As for the biochemical phenotype, we were not able to demonstrate a correlation between the residual enzyme activity and the clinical phenotype. When individuals with a less severe deficiency of MCC as indicated by a PCC/MCC ratio of < 50 were compared to a subgroup with a more severe deficiency and a ratio of > 50, the manifestation of clinical symptoms was not significantly more common in the latter group (40% (6/15) versus 47% (22/47), respectively).

Molecular genetic analysis of 83 subjects enrolled in the current study revealed a total of 31 new MCCC1 (n = 15) and MCCC2 (n = 16) mutations considered to be causative of MCC deficiency. This brings up the total of mutations published to date to 64 for MCCC1 and 68 for MCCC2 (Tables5 and6)[2‐12].

In our study cohort MCCC2 mutations were 1.7 times more common than MCCC1 mutations (63% versus 37%, respectively).

Tables5 and6 illustrate a broad genetic heterogeneity at both the MCCC1 and MCCC2 locus. Mutations are distributed along almost the entire coding regions of both genes with the exception of exon 2 of both the MCCC1 and MCCC2 gene which do not host any mutations.

The majority of mutations are private. Notably, in our cohort of 78 families only 5 MCCC1 and 10 MCCC2 mutations have been found in more than one family. The most common mutation was the p.R385S mutation in MCCC1 which was found in 8 individuals and 9 alleles. This missense mutation has been shown to have a dominant negative effect in the presence of the wild type allele and may lead to biochemical and clinical abnormalities in heterozygous individuals[44]. However, in agreement with earlier reports p.R385S appears not to be a predictor of a particular phenotype and has been found in severely affected patients as well as in asymptomatic individuals (this study,[2, 3, 44].

Another recurring mutation was c.1526delG (p.C509Sfs*14) which was the only mutation found in all 3 individuals from the Faroe Islands suggesting that this is a founder mutation.

Among the novel mutations identified in this study (shown in bold in Tables5 and6) the most common were missense mutations (n = 18). Frameshift (n = 5) and splice site mutations (n = 5) were more frequent than nonsense mutations (n = 2) and small deletions (of a single amino acid) (n = 1).

We assume deleterious functional consequences for the frameshift mutations MCCC1 c.658_662delTCAGA (p.S220Tfs*15), c.1193_1194delTG (p.V398Gfs*19), c.1820delG (p.S607Ifs*5), c.2009_2043del35 (p.A670Dfs*34), c.2123dupA (p.H708Qfs*8)], the splice site mutations MCCC1 c.639 + 2 T > A (p.S164Rfs*3); MCCC2 c.903 + 6_903 + 9delTACG (p.?), c.1073-12 C > G (p.G358Vfs*6 + p.G358Afs*12), c.1149 + 1 G > T (p.?), c.1149 + 5 G > C (p.?)] and the nonsense mutations MCCC1 c.1114 C > T (p.Q372*), c.1882 G > T (p.E628*)] because they result in truncated proteins lacking functionally important domains such as the BC (in MCCC1) or CT (in MCCC2) domains[2, 45].

Among the novel MCCC1 missense mutations all variants [p.M65L, p.G180V, p.A268D, p.E288G, p.G379D, p.I434M, p.V439M] change residues within the BC domain, while all novel MCCC2 missense mutations [p.C131F, p.Y146N, p.K152T, p.Y169D, p.I200N, p.G220E, p.P224L, p.G237D, p.V434L, p.G475R, p.G475R] affect the CT domain of the MCCC2 protein. During the final stage of this manuscript preparation, the crystal structure of P. aeruginosa MCC holoenzyme, with >50% sequence identity to the human counterpart (Additional file1: Figure S1 Additional file2: Figure S2), was reported[46]. The structure reveals a markedly different holoenzyme architecture compared to other biotin-dependent enzymes, hence providing an unprecedented opportunity to understand MCCC1/2 missense mutations in the protein context. We rationalize that the reported variants are very likely to affect protein function for the following reasons: 1) No other alterations in the MCCC1 and MCCC2 gene have been found despite sequencing of the complete coding regions of both genes in all individuals carrying one of those mutations; 2) Amino acid sequence alignments revealed that all but one MCCC2 c.599 T > A (p.I200N)] of the nucleotide changes affect highly conserved residues across species (PolyPhen2,http://​genetics.​bwh.​harvard.​edu/​pph/​), and many are also conserved among other biotin-dependent enzymes (Additional file1: Figure S1, Additional file2: Figure S2); 3) A majority of the mutations are mapped onto the catalytic core of the MCCC1 BC domain and the substrate binding region of the MCCC2 CT domain (yellow spheres in Figure4A), suggesting their functional importance; 4) Most of the missense mutations alter the physicochemical properties of the amino acid position (e.g. replacing small with bulky, unipolar with polar, or uncharged with charged residues) and hence are likely to affect the local molecular environment; 5) None of these 19 variants is present in the 1000 Genomes Project dataset (http://​www.​1000genomes.​org/​home); 6) It is of note that some MCCC1/2 missense mutations, affecting residues at the inter-subunit interface (Figure4B and C), may affect not only the corresponding domain fold but also disrupt the assembly of the α and β subunits into the functional hetero-dodecamer.

So far, functional consequences of only 4 MCCC1 and 8 MCCC2 mutant alleles have been proven by expression in a mammalian expression system[2, 11]. In this study we expressed 3 further MCCC1 and 8 MCCC2 mutations (Table7) including those expected to show residual enzyme activity based on studies in fibroblasts. Expression of two of the 3 MCCC1 [p.E288G and p.G379D] and one of the 8 MCCC2 [p.G118del] mutants resulted in severely reduced MCC activity and protein levels on Western blot analysis confirming deleterious consequences of these mutations on enzyme function and protein stability. Glu288 and Gly379 in MCCC1 are highly conserved residues in the BC domains among biotin-dependent enzymes (Additional file1: Figure S1). In the P. aeruginosa MCC holoenzyme structure, Glu288 forms an ionic pair interaction (Glu288-Arg444) to hold two secondary structure elements together, as observed in the homologous structures of ACC1 (Glu427-Arg604; PDB code 2YL2), ACC2 (Glu533-Arg710; 3GLK) and PCCα (Glu302-Arg459; 3N6R). Gly379 is expected to be at an invariant position at the beginning of a nearby β-strand, directly facing the aforementioned arginine residue. Mutations at Glu288 and Gly379 therefore may disrupt the packing of these secondary structure elements (Figure4D).

Deleterious effects of two further MCCC2 mutant alleles [p.H282R and p.A456V] were shown by severely reduced MCC activities of less than 13% of simultaneously expressed wildtype activity. However, in both cases protein levels were normal, and the two affected residues His282 and Ala456 are located at the monomeric surface, indicating that these mutations do not affect the stability of the protein. Expression of all other mutant alleles [MCCC1 p.I434M, MCCC2 p.S39F, p.Y146N, p.V434L, p.G475R and p.S523G] yielded considerable levels of residual MCC activity (Table7). The equivalent of MCCC1 Ile434 in other biotin-dependent carboxylases can be Phe or Leu (Additional file2: Figure S2), suggesting that substitution to a Met (similar size to Phe and Leu) in the p.I434M mutation may well be tolerated. All mutated residues from the 7 MCCC2 missense constructs (p.S39F, p.Y146N, p.282R, p.V434L, p.A456V, p.G475R, p.S523G) are at least partially exposed to the surface of the monomer, and not buried within the enzymatic core. Fibroblasts of 4 individuals (No 69, 80, 126 and 127) each homozygous for one of these mutations also showed residual enzyme activity, which was, however, much lower (6.2 – 31% of the median control value) than the activities of the expressed mutant alleles (39% - 77% of simultaneously expressed wildtype activity). Inspite of the high MCC activities after expression Western blot analysis revealed severely reduced protein levels in 2 cases (MCCC2 p.S39F, p.V434L). Normal levels of protein were expressed by two other mutant alleles (MCCC2 p.S523G, p.G475R). Even more drastic differences were found between MCC activity of the expressed protein (45%-89%) and that of fibroblasts (3.2% - 7.6%) that are compound heterozygous for the MCCC1 mutation p.I434M and MCCC2 mutations p.Y146N and p.S523G (individuals No 56, 66 and 74). Thus, expression studies may not demonstrate/identify the specific functional abnormalities for at least some of the mutant alleles with residual enzyme activity. However, no other mutations were found despite sequencing of the complete coding region of both MCC genes.

Our data confirm previous studies reporting no clear genotype-phenotype correlation in MCC deficiency[3, 6, 11] suggesting that factors other than the genotype at the MCC loci have a major influence on the clinical phenotype[11]. In line with this we identified clinically asymptomatic female adults carrying null mutations in homozygosity and siblings of which one was asymptomatic and the other showed neurologic symptoms compatible with influence of environmental factors on the clinical outcome of affected individuals.

None of the mutations found in asymptomatic individuals with MCC deficiency detected by NBS was prevalent in this group, which is in contrast to other inborn errors of metabolism such as isovaleric acidemia[42] or medium-chain acyl-CoA dehydrogenase deficiency[43]. Consequently, genotyping still appears to be of no help in predicting the clinical outcome of individuals with MCC deficiency.