Functional and Structural Analysis of Five Mutations Identified in Methylmalonic Aciduria cbIB Type

Abstract

ATP cob(I)alamin adenosyltransferase (ATR, E.C.2.5.1.17) converts reduced cob(I)alamin to the adenosylcobalamin cofactor. Mutations in the MMAB gene encoding ATR are responsible for the cblB type methylmalonic aciduria. Here we report the functional analysis of five cblB mutations to determine the underlying molecular basis of the dysfunction. The transcriptional profile along with minigenes analysis revealed that c.584G>A, c.349-1G>C and c.290G>A affect the splicing process. Wild-type ATR and the p.I96T (c.287T>C) and p.R191W (c.571C>T) mutant proteins were expressed in a prokaryote and a eukaryotic expression systems. The p.I96T protein was enzymatically active with a K_M for ATP and K_D for cob(I)alamin similar to wild-type enzyme, but exhibited a 40% reduction in specific activity. Both p.I96T and p.R191W mutant proteins are less stable than the wild-type protein, with increased stability when expressed under permissive folding conditions. Analysis of the oligomeric state of both mutants showed a structural defect for p.I96T and also a significant impact on the amount of recovered mutant protein that was more pronounced for p.R191W that, along with the structural analysis, suggest they might be misfolded. These results could serve as a basis for the implementation of pharmacological therapies aimed at increasing the residual activity of this type of mutations.

INTRODUCTION

ATP:cob(I)alamin adenosyltransferase (ATR, E.C.2.5.1.17) is an enzyme involved in vitamin B₁₂ metabolism that converts reduced cob(I)alamin to adenosylcobalamin, the active cofactor of methylmalonyl-CoA mutase (MUT, EC 5.4.99.2), which catalyzes the reversible rearrangement of methylmalonyl-CoA to succinyl-CoA in the catabolism of branched-chain amino acids, odd-chain fatty acids and cholesterol. Mutations in the human MMAB gene encoding the ATR enzyme are responsible for the cblB type methylmalonic aciduria (MIM# 607568) (Dobson et al., 2002; Leal et al., 2003). CblB patients present either a severe form with neonatal ketoacidosis, lethargy, failure to thrive and encephalopathy or a milder, chronic form with less impaired neurological outcome. A cellular in vitro response to hydroxocobalamin (OHCbl) has been reported in certain cases but this response appears to be unclear in vivo.

The human MMAB gene product, the ATR enzyme, is a member of the PduO family of cobalamin adenosyltransferases, which catalyzes the transfer of adenosine from ATP to cobalamin generating AdoCbl. To date, 24 mutations have been identified in the MMAB gene in cblB type patients (HGMD^® Professional Release 2009.3). Recently, the crystal structures of the homologous ATR proteins from T. acidophilum and L. reuteri along with the human enzyme with ATP bound have been determined (Saridakis et al., 2004; Schubert and Hill, 2006; St Maurice et al., 2007), allowing for evaluation of the structural impact of pathogenic mutations in ATR. The enzyme is a homotrimer, each subunit composed of a five-helix bundle and an active site located at the subunit interface. Several mutations identified in patients have been mimicked in prokaryotic orthologs and their kinetic parameters determined. They revealed the existence of mutants with reduced affinity for the substrate and cofactor (p.R191W) and with negligible activity and presumed instability in vivo (p.R186W, p.R190H and p.E193K) (Zhang et al., 2006). In another study, a Salmonella ATR-deficient strain was used to express mutations generated by random mutagenesis in the human ATR coding sequence, allowing the delineation of the active site of the enzyme and the putative residues implicated in cob(II)alamin reduction (Fan and Bobik, 2008).

In this study, we report the genetic analysis of four methylmalonic aciduria cblB type patients and the functional analysis of the identified mutations. In order to determine the molecular basis of in vitro B₁₂ responsiveness we have functionally analyzed two missense changes, p.I96T and p.R191W, identified in three B₁₂ responsive patient-derived cell lines, in prokaryotic and eukaryotic in vitro expression systems. We have also included the functional analysis of the three mutations affecting splicing (c.290G>A, c.584G>A and c.349-1G>C) in an ex vivo assay using minigenes.

MATERIALS AND METHODS

Genetic and biochemical analysis in patients’ fibroblasts

Patient 1 (P1), previously described in (Merinero et al., 2008), had a neonatal presentation of the disease. Patients 2 (P2) and 3 (P3) are siblings, P2 developed a late onset of the disease (at 4 years) and died shortly afterwards. P3 was genetically diagnosed without presenting clinical symptoms and to date, he is clinically normal. Patient 4 (P4) was referred to our laboratory for genetic analysis (Table 1). Since diagnosis, both the asymptomatic patients P3 and P4 are under therapy with protein restriction, carnitine supplementation, OHCbl administration, oral (5ml/d) in P3, and intramuscular (5mg/15d) in P4 and metronidazol (only P4). During therapy the patients show differences at the biochemical level: P3 shows milder plasma C3 levels (~10μM) compared to P4 (80μM); P3 has normal plasma odd-chain fatty-acid levels (OLCFA) (<0.4%) while P4 has increased ones (~4%). Urine MMA varies in P3 (<200–6000 mmol/mol creat), while there is no data available from P4. P3 has a perfect clinical development according to the clinician. P4 has a normal neurological and cardiological evaluation with normal renal function, weight (percentile 25–50) and height (percentile10) and he is doing well at school. Due to reduction of bone density in lumbar spine, alandronate, calcium and vitamin D were administered with improvement in growth and bone density. During his life he has showed some other metabolic decompensations associated with infections, vomiting, metabolic ketoacidosis etc. without requiring hospital admission. In these crises the patient required nutritional adaptation. It is worth noting that the patient has excellent dietary compliance. This work has been approved by our institutional ethics committee and informed consent has been obtained from the patients and their legal care-givers.

Table 1.

Genotype, clinical and biochemical phenotype in four methylmalonic aciduria patients cblB type

Patient	MUT activity** (nmol/min/mg)	[1-14C]-propionate −/+ ***	Allele 1	Allele 2	Onset	Outcome
P1	0.99	0.06/0.25	p.I96T (c.287T>C)	p.R191W (c.571C>T)	Neonatal (4 days)	7y/severe encephalopathy
P2*	0.76	0.26/1.01	p.I96T (c.287T>C)	p.S174fs (c.584G>A)	Late onset (4 years)	Died at 4y
P3*	NS	NS	p.I96T (c.287T>C)	p.S174fs (c.584G>A)	studied due to previous affected sister (3 months)	5y/asymptomatic
P4	0.55	0.08/0.07	p.I117_Q118del (c.349-1G>C)	p.G66fs (c.290G>A)	Neonatal (4 days)	12y/asymptomatic

^*Patients 2 (P2) and 3 (P3) are siblings

^**MUT activity in control fibroblasts with 36μM AdoCbl was 0.94±0.40 nmol/min/mg protein

^***[1-¹⁴C]-Propionate uptake (nmol/10h/mg protein) in control fibroblasts without/with (−/+) hydroxocobalamin in the culture medium was 1.90±1.18/2.34±1.61.

NS: not studied.

Mutation nomenclature follows the recommendations of HGVS and cDNA numbering is based on GenBank accession number NM_052845.3

Fibroblast cell lines from patients were cultured under standard conditions in MEM supplemented with 10% fetal bovine serum, 200 mM glutamine and antibiotics. Incorporation of radioactivity from [1-¹⁴C] propionate into acid-precipitable material was assayed in intact fibroblasts grown in basal medium ± 1 μg/mL OHCbl as previously described (Perez-Cerda et al., 1989). All data are summarized in Table 1.

To identify mutations in the MMAB gene, cDNA sequence analysis was performed. The identified changes were confirmed by sequencing the corresponding genomic DNA region. Genotype analysis was performed using the primers and conditions described previously (Martinez et al., 2005). Fibroblast cell lines were used as a source of DNA and RNA, which were isolated using the MagnaPure system following the manufacturer’s protocol (Roche Applied Sciences, Indianapolis, In). Mutation nomenclature follows the recommended guidelines of the Human Genome Variation Society (www.HGVS.org). Nucleotide numbering is based on cDNA reference sequences GenBank accession number NM_052845.3.

Functional analysis of the splicing mutations

For evaluation of in vitro splicing, the pSPL3 vector (Life Technologies, Gibco, BRL, kindly provided by Dr. B. Andresen, Aarhus University, Denmark) was used. Gene fragments corresponding to exons and flanking intronic regions were amplified from patients P2 and P4 and from a DNA control and cloned into the TOPO vector (Invitrogen, Carlsbad, CA) as previously described (Rincon et al., 2007). 2 μg of the wild-type or mutant minigenes were transfected into the following cell lines: Hep3B, HEK293 or COS7 using JetPEI, following the manufacturer’s recommendations. 24–48 h postransfection, cells were harvested, RNA was purified and RT-PCR analysis was performed using the pSPL3-specific primers SD6 and SA2 (Exon Trapping System, Gibco, BRL). The identity of amplified bands was determined by sequence analysis.

Prokaryotic and eukaryotic expression analysis of missense changes

Mutation p.I96T (c.287T>C) and p.R191W (c.571C>T) were introduced using the QuikChange Site Directed Mutagenesis kit (Stratagene, Cedar Drive, TX). The oligonucleotides used for mutagenesis were: MMAB I96Tsense (5′AGTTCAGCTACTGGGTTTGCTCTG3′), MMAB I96T antisense (5′CAGAGCAAACCCAGTAGCTGAACT3′), MMABR 191W sense (5′CCGTGTGCCGCTGGGCCGAGAGAC3′) and MMAB R191W antisense (5′GTCTCTCGGCCCAGCGGCACACGG3′).

For prokaryotic expression analysis we used the NL173 plasmid, based on pET41a, which contains the human ATR sequence without the mitochondrial targeting sequence. Human ATR protein was expressed in E. coli (BL21Star^™DE3 One Shot Cells) and purified following the method described by (Leal et al., 2004). Bacterial cells were transformed with the NL173 plasmid encoding either the normal or mutant (p.I96T and p.R191W) ATRs. Protein expression was induced with 1 mM IPTG; 5 h after induction, cells were collected by centrifugation and frozen at −70°C until further use. Cell pellets were resuspended at 4°C in 50 mM potassium phosphate buffer, pH 8, 100 mM NaCl, 1 mM DTT, 1 mM protease inhibitor phenylmethylsulphonyl fluoride, 1 mM EDTA, 0.05 mg/ml DNase, 0.3 mg/ml lysozyme. After 1 h of incubation, the cell suspension was sonicated and centrifuged to obtain the crude soluble cell extract which was subjected to ammonium sulphate precipitation and subsequent anion-exchange chromatography using a Sepharose Q column, as previously described (Leal et al., 2004). In all purification steps, protein concentration was determined by the Bradford assay. Fractions eluted from the Sepharose Q column were subjected to SDS-PAGE to identify those enriched with the ATR protein. In some cases before ATR enzymatic assay, wild-type and mutant ATR proteins were further purified by hydroxyapatite chromatography at 4°C using as mobile phase a gradient of 10–400 mM potassium phosphate buffer pH 8, and 5 mM KCl. Fractions containing ATR of the highest purity were pooled and concentrated using an Amicon concentrator to a final concentration of 10 mg/ml.

Full-length cDNA of wild-type MMAB was generated from total RNA fibroblasts by RT-PCR and cloned into pGEM-T vector (Promega, Madison, WI). Subsequently, MMAB cDNA was further excised and inserted into HindIII/EcoRV sites of pFLAG-CMV-5c expression vector (Sigma, St. Louis, MO) to obtain MMAB cDNA coupled with FLAG-tag at the C-terminus (pFLAG-MMAB).

Eukaryotic expression analysis was performed in the P4 fibroblast cell line stably transformed with pBABE kindly provided by Dr. JA Enriquez (University of Zaragoza, Spain). A total of 4×10⁵ transformed cells were transfected using 2.5 μg of normal or mutant pFLAG-MMAB constructs using lipofectamine 2000 (Invitrogen, Carlsbad, CA); then the cells were grown at 27 or 37°C. Harvested cells were used to determine ATR activity which was indirectly measured by the analysis of [¹⁴C]-propionate incorporation into acid-precipitable material in intact cells grown in basal medium (Perez-Cerda et al., 1989) Propionate oxidation is catalyzed by two mitochondrial enzymes, the biotin-dependent propionyl-CoA carboxylase and the AdoCbl-dependent methylmalonyl-CoA mutase. AdoCbl is synthesized by ATR and the patients with defects in the MMAB gene exhibit deficient propionate incorporation. Statistical analysis was performed using an F-test analysis of variance followed by a 1-tailed paired t-test. P values less than 0.05 were considered significant.

Enzyme activity assays

ATR activity was measured as previously published (Johnson et al., 2001) with minor modifications. Assays contained 200 mM Tris-HCl pH 8.0, 1.6 mM KH₂PO4, 2.8 mM MgCl₂, 100 mM KCl, 0.4 mM ATP and 80 μM OHCbl (Sigma Aldrich, St. Louis, MO). 800 μl aliquots of the reaction mixture were dispensed into cuvettes kept at 37oC. Titanium (III) citrate (10 μl) was added, the reaction initiated by the addition of 10–100 μg of purified protein, and the decrease in absorbance at 388 nm was monitored. Anaerobic procedures were used to prepare the reaction mix, the protein solution and titanium(III)citrate.

Measurement of dissociation constants

The equilibrium dissociation constants (Kd) for cob(II)alamin analogs were determined fluorimetrically as described by (Chowdhury and Banerjee, 1999) with modifications. To determine the Kd values, 40–150 μg of protein in 50 mM Tris-HCl buffer, pH 8.0, was added to a quartz cuvette and successive aliquots (1–10 μl) of the cobalamin analogue (2–7 mM) were added then the fluorescence (excitation wavelength: 280 nm, emission wavelength 340 nm) was measured after each addition. The Kd for cob(II)alamin was determined under anaerobic conditions by monitoring the shift in the UV-visible spectrum from 300 nm to 750 nm using 10 μM cob(II)alamin and varying concentrations of ATR as described previously (Yamanishi et al., 2005). The kinetic data were analyzed using the IGOR Pro6 software (WaveMetrics Inc, Lake Oswego, OR).

Western blot analysis

Protein concentration in fibroblast and bacterial cell extracts was determined using the BIO-RAD protein assay (Bio-Rad, Bio-Rad Laboratories, Munchen, Germany) following the manufacturer’s protocol. Equal amounts of total protein for bacterial cell extracts were loaded onto a 10% Laemmli SDS-PAGE System. After SDS-PAGE or native gel electrophoresis, proteins were transferred to a nitrocellulose transfer membrane (GE Healthcare, Buckinghamshire, UK) using a BIO-RAD apparatus in Tris (25 mM)-glycine (192 mM)-methanol (20%) without SDS for 1 h. Poinceau staining was used to monitor equal loading of protein. Immunodetection was carried out using commercially available anti-ATR antibodies (ProteinTech Group Inc, Chicago, IL) as primary antibodies diluted 1:1000. The secondary antibodies used were conjugated goat-anti mouse IgG-horseradish peroxidase (1:10000) (Santa Cruz Biotechnology Inc, Santa Cruz, CA) and were detected using the Enhanced Chemiluminescence System (GE Healthcare, Buckinghamshire, UK). Relative protein amounts were determined using a calibrated densitometer GS-800 (BioRad).

Oligomeric state of ATR

Size-exclusion chromatography using a Superdex 200 (1.2 cm × 92 cm) column was performed to assess the oligomeric state of the wild-type and mutant ATR proteins. 100 μg of protein purified by ammonium sulphate precipitation and anion-exchange chromatography were applied to the column in 50 mM Tris HCl buffer pH 8.0, 100 mM KCl and a flow rate of 2.0 ml min⁻¹. Molecular mass value estimations were obtained using a calibration curve generated with molecular weight standards (Bio-Rad).

Blue native electrophoresis was performed using 4–16% Native PAGE gels, Native Marker and Native running buffer from Invitrogen. Samples were prepared with 4x Native sample buffer with G-250 sample additive (Invitrogen). The electrophoretic separation was performed at 4°C for one hour at 150V followed by another hour at 250V. Lanes containing purified ATR and molecular weight marker were stained using 2% Coomassie Brilliant Blue R-250 (Bio-Rad) solution in acetic acid and destained in acetic acid: methanol: water (0.5:2:2). The remaining lanes were transferred to a PVDF membrane using the iBLOT system from Invitrogen. Immunodetection was carried out as described above.

Protein stability studies

The stability of the wild-type and mutant p.I96T and p.R191W proteins expressed in the prokaryotic system was investigated by incubating the transformed cells at 37°C or 27°C following IPTG induction. Soluble supernatant was incubated at 37 or 27°C, aliquots were removed at different time points and the cell extracts were subjected to SDS-PAGE. Western blot analysis was performed as described above. Following densitometric analysis, relative protein levels were expressed as percentage referred to time 0.

Structural analysis and molecular dynamics

Visualization of the crystal structure of ATR and localization of the mutant residue I96 was performed using DSViewerPro5.0 (Accelrys Inc, San Diego, CA) with the PDB coordinates file 2IDX.

Human ATP:cobalamin adenosyltransferase (hATR) monomers were modelled using as template, the B and C chains from the PDB structure 2IDX (Schubert and Hill, 2006), and the program MODELLER9v7 (Eswar et al., 2006). The ATP binding loop ‘KIYTK’ was kept in the bound conformation from chain B, while the unresolved loop ‘SSAREAHLKYT’ in B chain was taken from C chain. Trimer reconstruction was performed by superposition of the modelled monomer over 2IDX structure with the help of the MAMMOTH program (Ortiz et al., 2002). All three active sites were modelled as occupied by an ATP molecule.

Models of mutants p.R191W and p.I96T were generated with the program PyMOL (DeLano, 1998–2003) using the modelled trimer as template. Protonation states of ionizable groups at pH 6.5 for the three systems were calculated using the H++ server (Gordon et al., 2005). The positions of hydrogen atoms, standard atomic charges and radii for all the atoms were assigned according to the ff03 force field (Duan et al., 2003). The complexes were immersed in cubic boxes of TIP3P water molecules (William et al., 1983) large enough to guarantee that the shortest distance between the solute and the edge of the box was larger than 13 A. Counter ions were also added to maintain electro neutrality. Three consecutive minimizations were performed involving: i) only hydrogen atoms; ii) only the water molecules and ions; and iii) the entire system.

Simulation details

Starting minimized structures, prepared as stated above, were simulated in the NPT ensemble using Periodic Boundary Conditions and Particle Mesh Ewald to treat long-range electrostatic interactions. The systems were then heated and equilibrated in two steps: i) 200 ps of molecular dynamics (MD) heating the whole system from 100 K to 300 K, and ii) equilibration of the entire system during 1.0 ns at 300 K. The equilibrated structures were the starting points for the 15 ns MD simulations at constant temperature (300 K) and pressure (1 atm). SHAKE algorithm was used to keep bonds involving H atoms at their equilibrium length, allowing a 2 fs time step for the integration of Newton’s equations of motion. ff03 and TIP3P force fields, as implemented in AMBER 10 package (D.A. Case, 2008), were used to describe the proteins, the peptides and the water molecules, respectively. Sample frames at 10 ps intervals from the molecular dynamics trajectory were subsequently used for the analysis.

In silico stability analysis

Absolute free energies of the hATR complex and its mutants (p.R191W and p.I96T) were estimated using the MM-GBSA (Still et al., 2002) approach as implemented in the AMBER10 package. MM-GBSA method approaches the free energy of the complexes as a sum of a molecular mechanic (MM) interaction term, a solvation contribution through a generalized Born (GB) model, and a surface area (SA) contribution to account for the non-polar part of solvation. Distributions of the observed free energies, measured every 10 ps snapshots along with the molecular dynamics simulations, were plotted using the R package (Gentleman R., 1997).

In addition, to better characterize the changes produced by the mutated residues, an energy decomposition analysis in a pairwise fashion (between the residues surrounding the mutations) was performed using a cut off of 5 Å from the mutated residues. Polar contribution to solvation free energies were calculated with GB, whereas non-polar contributions were estimated to be proportional to the area lost upon binding using the LCPO method to calculate accessible surface areas (Weiser et al., 1999).

These calculations were performed, for each 10 ps snapshot from the simulations, using the appropriate module within AMBER 10 package. Free energy decomposition interaction matrix was represented in an energy dependent colour gradient using the program matrix2png (Pavlidis and Noble, 2003).

RESULTS

P1, P2 and P4 patients showed normal MUT activity and P1 and P2 exhibited in vitro responsiveness to B₁₂ (Table 1). The age of onset and the clinical outcome of these patients are summarized in Table 1. The p.I96T (c.287T>C) sequence change was identified in three of the patients (Patients P1, P2 and P3). This mutation was found in combination with the missense change p.R191W (c.571C>T) and in two siblings in combination with c.584G>A mutation, affecting the last nucleotide of exon 7 in the MMAB gene. This last mutation results in exon 7 skipping (r.520_584del) in patient fibroblasts, which would result in a premature stop codon (p.S174fs) (Figure 1A).

Figure 1.

Schematic representation of the mutations mapped in genomic DNA and RT-PCR pattern from fibroblast and minigenes assay. A) Patient P2 carries one nucleotide change located on exon 7 (c.584G>A) and a missense change on exon 3 (c.287C>T; p.I96T). The transcriptional profile shows three bands corresponding to the skipping of exon 7 transcribed from the allele bearing the c.584G>A change, a normal band from the allele bearing the c.287 T>C and a heteroduplex extra band (upper band). B) Patient 4 carries a change located on the last nucleotide of exon 3 (c.290 G>A) and an intronic change (c.349-1G>C) in intron 4. The RT-PCR pattern shows a smaller band without exon 3 resulting from the c.290G>A change and a larger one with an in-frame deletion of the first six nucleotides of exon 5 from the allele bearing c.349-1G>C.

The RT-PCR pattern and subsequent sequence analysis of the amplified products obtained from patient P4 fibroblasts, showed the presence of two bands, one of them corresponding to the skipping of exon 3 (r.197_290del), which is predicted to generate a truncated protein and the other one corresponding to an in-frame deletion of two amino acids (r.349_354del6, p.I117_Q118del) due to activation of a cryptic splice site inside exon 5. The genomic DNA analysis showed two nucleotide changes c.290G>A and c.349-1G>C in the last nucleotide of exon 3 and in the 3′ splice site of intron 4, respectively (Figure 1B).

Functional analysis of the splicing mutations

In this work, we have investigated the functional effect of the three nucleotide changes affecting splicing (c.290G>A, c.349-1G>C and c.584G>A) using a cell based splicing assay. In the patient fibroblasts, only transcripts resulting from the skipping of exon 7 and exon 3 derived from the c.584G>A and c.290G>A alleles are detected. However, if correct splicing occurred occasionally, a mutant protein with the p.R195H and p.G97E changes would result. A minigene functional splicing assay was employed to investigate this possibility. All splicing mutations were cloned in minigenes and transfected in Hep3B, HEK293 or COS cells. All the mutants resulted in aberrant splicing. In addition of the strong 3′and 5′ splice sites, the pSPL3 vector has after the multiple cloning site, cryptic 3′ and 5′ splice sites. In the case of the c.290G>A and c.584G>A mutant constructs, both splicing sequences were used, the classical and the cryptic one and therefore the fragment amplified contained a vector sequence generated from use of the weaker cryptic spice sites (Figure 2) In the case of c.584G>A a small but detectable amount of the correctly spliced transcript (resulting in the p.R195H change) was observed.

Figure 2.

Splicing cellular minigene analysis. RT-PCR pattern obtained after transfection with the corresponding wild-type and mutant minigene constructs and schematic representation of each minigene construct. The PCR from the c.290G>A and c.349-1G>A minigenes was performed using minigene-specific primers while in the c.584G>A, a specific forward primer located at the junction of the vector sequence and exon 6 was used to prevent different bands obtained when minigene-specific primers are used. All fragments were sequenced and the schematic representation of the resultant transcripts is indicated. Gene fragments corresponding to exon 3 (A) or exon 5 (B) and flanking 3′and 5′ intronic regions were amplified from patients and cloned into the pSPL3 vector (grey line and box). For the c.584G>A minigene, the amplified fragment included exon 6, intron 6 and exon 7 and also the intronic sequence adjacent to both exons. The exonic sequences of the pSPL3 vector are highlighted in black boxes (V-boxes: vector sequence) and striped boxes refers to the sequences involved when the cryptic splice sites are used (A and C). The sequence of the RT-PCR fragment obtained with the normal minigene (c.290G, c.349-1G and c.584G) revealed the presence of a normal transcript with the corresponding exon included. In the two former cases we have also detected a slight band corresponding to a complete skipping of exon 3 and 5 that cannot be amplified in the third case. The sequence of the fragment obtained after transfection with the mutant construction containing the c.290A or c.584A changes revealed the presence of a vector sequence of 119bp inserted in the aberrant transcript due to activation of a 3′and 5′ cryptic splice site located in the vector sequence following the poly-cloning site.

Functional and structural analysis of the missense mutations

Since the three patients responsive to B₁₂ in vitro share the p.I96T mutation described previously (Merinero et al., 2008), we sought to analyze the kinetic stability properties of the mutant protein. Residue I96 is located in helix α1 (Figure 3A), distant from the ATP binding site while residue R191 projects into the central cavity of the trimeric protein. Given the location of the I96 residue in α1 helix, the mutation could change the stability of the protein and/or perturb binding of the substrates, ATP and cobalamin but is not predicted to be involved in trimer formation.

Figure 3.

Structural analysis of missense mutation. A) Location of the p.I96T and p.R191W mutations in the crystal structure of human ATR. The enzyme is a homotrimer, with each subunit (shown in a different shade of blue) composed of a five-helix bundle and with the active site located at the subunit interfaces. Residue I96 is located in helix α1, distant from the ATP binding site while residue R191 projects into the central cavity of the trimeric protein. ATP (red) and the residues, I96 (blue) and R191 (gold) are shown in stick representation. This figure was generated from the PDB file 2idx. B) Wild-type hATR trimer core (represented as cartoon) showing hydrogen bonds (dashed yellow lines) between residues E91, R190 and R191 (represented as sticks). C) R191W mutant hATR trimer core (represented as cartoons) showing hydrogen bonds (dashed yellow lines) between residues E91, R190 and R191 (represented as sticks).

Absolute free energies for wild-type hATR and the mutants, p.R191W and p.I96T, were estimated by the MM-GBSA method and predicted a decrease in stability for the p.R191W mutant (−15026.93 ± 68.35 kcal/mol) compared to the wild-type ATR (−15436.88 ± 72.76 kcal/mol) or the p.I96T mutant (−15441.07 ± 67.06 kcal/mol). In fact, a significant difference was not found between wild-type ATR and the p.I96T mutant (Supp. Figure S1).

In addition, pair wise interaction energies were calculated for the residues surrounding the mutations. In p.R191W, a network of hydrogen bonds at the trimer centre involving R191 and E91 residues was detected, linking the three monomers (Figure 3B). The equivalent interaction in p.R191W mutant is substantially diminished (Figure 3C). However, this energy loss is compensated, at least in part, by a new inter subunit hydrogen bond pattern established between the R190 and E91 residues. In addition, the effect of the stacked conformation adopted by the mutant W191 residue creates a distortion in the trimer core that is reflected by other changes involving the interaction profile of E84, especially those related to R194 and R195, and the interaction between R190 and bound ATP (data not shown). On the contrary, a significant difference in the pair wise interaction energy when comparing I96T mutant to the wild-type protein was not observed.

Wild-type and mutant (p.I96T) ATR were expressed in E. coli and purified. The degree of purification was estimated to be 95% after the first step and ~99% after the final step. The specific activity and kinetic parameters for wild-type and p.I96T ATR were measured in parallel (Table 2) and similar results were obtained with and without the last hydroxyapatite purification step. The p.I96T mutant exhibits a slight reduction in specific activity, retaining 60% of wild-type levels with no significant differences in the Kds for ATP or cobalamins, with the exception of CNCbl which showed a 1.7-fold increase in Kd.

Table 2.

Comparison of kinetic parameters for wild-type and p.I96T ATR

	wild-type	p.I96T
Specific activity (nmol min−1 mg−1)	200 ± 10	123 ± 19
KM (μM) ATP	6.2 ± 1.3	7.8 ± 1.9
Kd (μM) AdoCbl	1.7 ± 0.4	2.58 ± 0.62
Kd (μM) OHCbl	8.5 ± 1.1	7.8 ± 1.6
Kd (μM) CNCbl	7.8 ± 0.5	13.3 ± 0.3
Kd (μM) Cob(II)alamin	3.6± 1.6	1.37 ± 0.33

Data are the mean of at least two independent experiments each performed in duplicate

The oligomeric profile of the p.I96T mutant protein was analyzed by size exclusion chromatography following ammonium sulphate precipitation and anion-exchange chromatography. Under these conditions, the protein exists mainly as a trimeric, similar to wild-type ATR (data not shown). Electrophoresis under nondenaturing conditions detected aggregates of ATR in the crude cell lysates (supernatants or pellets) with either wild-type or mutant ATR, although the size distributions were different (Figure 4). Wild-type ATR showed lower molecular weight aggregates (~250 kDa), while the majority of the protein was present in a trimeric form. The p.I96T mutant showed both trimeric and monomeric forms. Both supernatant and crude extracts of mutant proteins showed a significant reduction in the amount of protein recovered, more pronounced in the case of p.R191W that could not be detected under any conditions (Figure 4).

Figure 4.

Oligomeric state of ATR. Blue native gel electrophoresis of cell extracts subjected to immunodetection for ATR. Lane 1, 1 μg of purified wild-type hATR; lane 2, total extract from non-transformed BL21 bacteria; lanes 3–5, whole cell extracts from uninduced bacteria transformed with wild-type ATR plasmid (lane 3), p.I96T ATR (lane 4) and p.R191W ATR (lane 5); Lanes 6–14 induced transformed bacteria. Lanes 6, 7 and 8, whole cell extracts of wild-type (1 μg), p.I96T (2.5 μg) or p.R191W (20 μg); lanes 9, 10 and 11, supernatant of wild-type (1 μg), p.I96T (2.5 μg) and p.R191W (20 μg) after centrifugation of sonicated cell extracts; lanes 12, 13 and 14, pellets from wild-type (1 μg), p.I96T (2.5 μg) and p.R191W (20 μg) obtained after centrifugation.

Next, we investigated the relative stabilities of the p.I96T and p.R191W mutants expressed in bacteria grown at 37°C or 27°C (Figure 5). The p.I96T mutant was found to have a significantly decreased half life (Figure 5A) compared to wild-type protein at 37°C (34 versus 85 h). At lower temperature (27°C) the half-lives for both proteins increased (Figure 5B).

Figure 5.

Degradation time course for wild-type p.I96T and p.R191W ATR at 37°C and 27°C. Crude cell extracts from cultures expressing wild-type or p.I96T or p.R191W ATR were incubated at 37°C (A) or 27°C (B) and aliquots were removed at different time points at 37 and 27°C respectively and subjected to SDS-PAGE and Western blot analysis as described in methods. Protein amounts were quantified by laser densitometry. The data are the mean ± SD and represent the percent density of each protein relative to its density at time 0. The half-life of each protein is noted.

R191 projects into the central cavity of the trimeric ATR structure (Figure 3A); interactions between this residue from adjacent monomers might play a role in stabilizing the quaternary structure of the protein. In fact, the p.R191W mutant was highly unstable at 37°C (Figure 5A) with a half-life of 1.3 h. The expression level of the mutant increased when the cells were grown at 27°C with a corresponding increase in its half-life to 26.5 h (Figure 5B).

Wild-type and mutant ATRs were also expressed as fusion proteins with a FLAG tag in a cblB fibroblast cell line (P4). Cells were cultured at 37o and 27°C and the [¹⁴C]-propionate incorporation levels were determined as an indirect measure of ATR activity in the transfected cells. At 37oC, the mutant p.I96T and p.R191W proteins showed a statistically significant reduction in [¹⁴C]-propionate incorporation: 65% and 52% of wild-type levels respectively. When the cells were grown at 27°C, [¹⁴C]-propionate incorporation levels of p.I96T and p.R191W mutants increased to levels close to wild-type ones (Figure 6).

Figure 6.

Functional analysis of the p.I96T and p.R191W mutants in a cblB null fibroblast cell line. A fibroblast cell line was transfected with a vector encoding wild-type or mutant (p.I96T or p.R191W) MMAB cDNA and the culture was incubated at 37 or 27°C. The cells were harvested followed by incorporation of [¹⁴C]-propionate into trichloroacetic acid-precipitable material in the absence of OHCbl that was quantified as an indirect measure of ATR activity. The data are the mean ± SD of the percentage of [¹⁴C]-propionate incorporation for each transfection versus the wild-type transfection (n=4). The p-values were calculated for the pairs indicated in the figure (*** = p-value <0.001).

DISCUSSION

The focus of this work was the functional analysis of five cblB mutations and the investigation of the underlying molecular mechanism of in vitro B₁₂ responsiveness observed in three MMA patients belonging to the cblB class. The results obtained indicate that all five nucleotide changes are disease-causing mutations, opening the possibility for genetic counselling, carrier identification and prenatal diagnosis in at-risk families, and allowing determination of possible phenotype-genotype correlations and mutation-specific therapies.

Regarding the functional analysis of the exonic and intronic nucleotide changes affecting splicing, our results suggest that there are negligible levels of normal transcript associated with the mutations studied, from fibroblasts from patients and in cell lines (Hep3B, Hek293 and COS cells) transfected with the corresponding minigenes. Based on our data, the in vitro responsiveness in these patients might be associated with the missense changes, p.I96T or p.R191W. In relation to patient P4 who bears two splicing mutations, no stimulation of [¹⁴C]-propionate was observed when cells were supplemented with OHCbl. However, given that the patient is clinically asymptomatic, the in-frame deletion of two amino acids resulting from the c.349-1G>C allele might retain certain degree of activity. Addressing this will be the aim of future investigations.

The three in vitro B₁₂ responsive patients (P1, P2 and P3) share the missense mutation p.I96T, which was expressed in a prokaryotic system in order to analyze its structural and kinetic properties. The mutant protein showed a modest reduction in specific activity but normal binding affinity for ATP and cob(II)alamin substrates. Inconsistent results were obtained in evaluating the effect of the mutations on the oligomeric structure of the protein. Thus, size exclusion chromatography revealed a normal oligomeric profile, while native gel electrophoresis indicated an aberrant profile. In the former case, the purification of ATR prior to Superdex chromatography might have led to the separation of monomers, while the soluble or crude extracts loaded on the native gel allowed detection of all the ATR forms and involved only minor manipulation. However, the significantly lower levels of mutant protein observed by native gel electrophoresis, and the lower stability suggests that loss of biological function in the p196T mutant might be due to abnormal folding rather than a change in oligomeric state. Impaired protein folding can enhance degradation (e.g. in Gaucher and lysosomal storage disorders), lead to deposition of misfolded protein (e.g. in Alzheimer’s disease) or result in an inactive form (Majtan et al. in press; Gamez et al., 2000).

When both mutant proteins are expressed in E. coli under permissive low temperature conditions, the steady-state levels of ATR and its half-life are increased. Similar results were obtained using a eukaryotic expression system where the level of ¹⁴C-propionate incorporation was similar to control values when the transfected cells were grown at 27°C. Expression studies of the p.R191W mutation in E. coli have been previously described using a GST fusion protein in which its stability was not mentioned (Zhang et al., 2006). In contrast, in Salmonella enterica, the p.R191W mutant was expressed at lower levels than the control, indicating that it impairs proper folding and leads to degradation mediated by cellular proteases (Fan and Bobik, 2008). The residual activity associated with the pR191W mutant was reported to be 30% in the E. coli system and undetectable in S. enterica (Fan and Bobik, 2008; Zhang et al., 2006). Mutations affecting residues close to the 191 position, e.g. R190H and p.E193K are also unstable in all expression systems (Fan and Bobik, 2008; Zhang et al., 2006) and might represent a region critical for stability.

Our data suggest that p.R191W and p.196T mutants result in severe folding problems although when expressed as fusion proteins with the Flag peptide in an eukaryotic expression system, residual activity was observed. However, since transient expression was performed using the P4 cell line, it is possible that in addition to FLAG stabilization, complementation with the endogenous ATR might have contributed to the rather high residual activity of these mutants.

The responsiveness to in vitro OHCbl supplementation could be explained partially by a stabilizing effect of the cofactor which could act as a natural chaperone, as has been described for other cofactors, e.g. tetrahydrobiopterin in phenylalanine hydroxylase deficiency (Erlandsen et al., 2004; Perez et al., 2005; Pey et al., 2004). Other mechanisms may also be operating in the OHCbl-responsive cell line, P1, since the p.R191W mutant (albeit with a GST tag) exhibits a 16-fold higher K_M for cobalamin (Zhang et al., 2006). Defects in ATR might also affect its interaction with MMAA or MUT proteins, assuming that the three enzymes operate in a complex and increased levels of B₁₂ might have an stabilizing effect (Banerjee, 2006; Banerjee et al., 2009).

In addition to the genetic diagnosis of the disease, knowledge of the mutational spectrum in each population provides insights into genotype-phenotype correlations. Based on the clinical outcome, patients bearing the p.I96T mutation exhibit a highly variable phenotype. Thus, while one of them is clinically asymptomatic, the other two are either dead or severely neurologically impaired. These findings also suggest that the missense changes, p.I96T and p.R191W, might be categorized as a loss-of-function folding mutations that are usually associated with variable genotype-phenotype correlation (Andresen et al., 1997; Pey et al., 2003; Pey et al., 2007). Inter-individual differences in gene expression of the protein quality control and proteasomal proteins have been described as phenotypic modifiers in several misfolding diseases (Dipple and McCabe, 2000a; Dipple and McCabe, 2000b).

In conclusion, functional analysis of the mutations described in this study is relevant to our understanding of the pathogenic mechanisms of MMA and also for the design of mutation-specific therapies. In several genetic disorders such as phenylketonuria or Gaucher’s disease (Bernier et al., 2004; Mu et al., 2008; Pey et al., 2008) pharmacological rescue of folding mutants or up-regulation of gene expression for mutants with residual activity is currently envisaged as a possible option for correcting, at least partially, the phenotype. In the case of the p.I96T and p.R191W mutant proteins, pharmacological chaperones or compounds such as statins (Murphy et al., 2007) which may result in activation of MMAB gene expression, should be explored as potential strategies to treat the disease if increased steady-state levels and activity of the protein can be achieved.