The brain-specific compound NAA (N-acetylaspartate) occurs almost exclusively in neurons, where its concentration reaches approx. 20 mM. Its abundance is determined in patients by MRS (magnetic resonance spectroscopy) to assess neuronal density and health. The molecular identity of the NAT (N-acetyltransferase) that catalyses NAA synthesis has remained unknown, because the enzyme is membrane-bound and difficult to purify. Database searches indicated that among putative NATs (i.e. proteins homologous with known NATs, but with uncharacterized catalytic activity) encoded by the human and mouse genomes two were almost exclusively expressed in brain, NAT8L and NAT14. Transfection studies in HEK-293T [human embryonic kidney-293 cells expressing the large T-antigen of SV40 (simian virus 40)] indicated that NAT8L, but not NAT14, catalysed the synthesis of NAA from L-aspartate and acetyl-CoA. The specificity of NAT8L, its Km for aspartate and its sensitivity to detergents are similar to those described for brain Asp-NAT. Confocal microscopy analysis of CHO (Chinese-hamster ovary) cells and neurons expressing recombinant NAT8L indicates that it is associated with the ER (endoplasmic reticulum), but not with mitochondria. A mutation search in the NAT8L gene of the only patient known to be deficient in NAA disclosed the presence of a homozygous 19 bp deletion, resulting in a change in reading frame and the absence of production of a functional protein. We conclude that NAT8L, a neuron-specific protein, is responsible for NAA synthesis and is mutated in primary NAA deficiency (hypoacetylaspartia). The molecular identification of this enzyme will lead to new perspectives in the clarification of the function of this most abundant amino acid derivative in neurons and for the diagnosis of hypoacetylaspartia in other patients.
NAA (N-acetylaspartate), a brain-specific molecule discovered more than 50 years ago, is the second most abundant metabolite in brain after glutamate. It is essentially present in neurons, where its concentration reaches approx. 20 mM. Its abundance can be determined by MRS (magnetic resonance spectroscopy) to monitor neuronal density and health in neurological disorders, including Alzheimer's disease, multiple sclerosis and AIDS encephalopathy . NAA is formed from aspartate and acetyl-CoA by Asp-NAT (aspartate N-acetyltransferase), a membrane-bound enzyme [2–4], which is unstable in the presence of detergents  and could therefore not be purified. Its subcellular localization has been variously reported as mitochondrial [6,7] or microsomal [4,8]. NAA is not degraded in neurons, but it is transferred, by an undefined mechanism, to oligodendrocytes, the myelin-forming cells, where it is hydrolysed by ASPA (aspartoacylase, also known as aminoacylase II). The acetate released in this process serves as a precursor for the synthesis of fatty acids, which are required for the formation of myelin lipids . NAA serves also as a precursor for N-acetylaspartylglutamate [10,11], the most abundant dipeptide present in brain and possibly a neurotransmitter . Other proposed roles for NAA are osmoregulation or disposal of amino groups resulting from glutamate transamination . It is accepted that NAA does not appear to be a neurotransmitter , but to date no agreement exists on the relative importance of the other potential roles.
NAA metabolism is disrupted in two inborn errors of metabolism. Deficiency of aspartoacylase, caused by mutations in the ASPA gene, gives rise to a leucodystrophy known as Canavan disease [OMIM (Online Mendelian Inheritance in Man) accession number 271900], which is characterized by onset in early infancy, dystonia, severe mental defects, blindness, megalencephaly and death, on average, by 18 months . NAA concentration is raised in brain, CSF (cerebrospinal fluid) and urine. The ‘opposite’ disorder, i.e. the absence of brain NAA, has been described in only one patient, with truncal ataxia, marked developmental delay, seizures and secondary microcephaly [15–17]. This disorder, designated ‘hypoacetylaspartia’, is presumably due to a deficiency of Asp-NAT, but this has not been demonstrated.
A better definition of the role of NAA and the identification of the defect underlying hypoacetylaspartia would greatly benefit from the molecular identification of Asp-NAT. This identification was the purpose of the present work.
Preparation of expression vectors
Human and mouse total brain cDNA were synthesized with random primers and 1.5–2.5 μg of total RNA with the M-MuLV (Moloney murine leukaemia virus) reverse transcriptase (Fermentas) according to the manufacturer's instructions. FirstChoice® human brain total RNA was from Applied Biosystems, and mouse total brain RNA was prepared as described in .
The open reading frames of mouse NAT14 (GenBank® reference sequence NM_201355) and mouse NAT8L (GenBank® reference sequence NM_001001985) were PCR-amplified using 2.5 units of Pwo (Pyrococcus woesei) polymerase (Roche Applied Science) in the presence of 1.5 M betaine (Sigma–Aldrich) with mouse total brain cDNA as a template. For the untagged constructions, a 5′-primer (for primer sequences see Supplementary Table S1 available at http://www.BiochemJ.org/bj/425/bj4250127add.htm), containing the putative ATG preceded by a perfect Kozak sequence and a KpnI site, and a 3′-primer, containing the putative stop codon flanked by a XbaI site, were used. For the C-terminally tagged constructs, the 5′-primers were the same, but the putative stop codon in the 3′-primers was mutated and flanked by a XbaI site. For the N-terminally tagged constructs, the 5′-primers were devoid of Kozak sequence and the 3′-primers were the same as for the wild-type amplifications. The PCR amplifications started with an initial denaturation step at 96 °C for 3 min, followed by 35 cycles for 30 s at 96 °C, 1 min at 60 °C and 1 min at 72 °C, with a final extension step at 72 °C for 5 min. All PCRs reported in the present study were performed with this five-step program except for site-directed mutagenesis (see below). PCR products of the expected size were obtained, subcloned into the pBlueScript plasmid (Invitrogen) and checked by sequencing. The KpnI/XbaI-digested products were prepared from the Escherichia coli XL1blue strain (C-terminally tagged constructs) or from E. coli JM110 strain (wild-type and N-terminally tagged constructs) and inserted in pCMV5, pEF6MycHisA and pEF6HisB expression vectors (Invitrogen) for the wild-type, C-terminally tagged and N-terminally tagged constructs respectively. For human NAT8L, amplifications were performed with 2.5 units of Pfu (Pyrococcus furiosus) polymerase (Fermentas), in the presence of 1 M betaine, with 2 μl of human total brain cDNA as a template. The three types of construct were prepared as for the mouse sequences and checked by sequencing.
pBlueScript vector (50 ng) containing the human NAT8L sequence, designed to be expressed with a His6 tag at its C-terminus, was used as a template in a 50 μl PCR mixture [25 pmol of each primer (see Supplementary Table S1), 0.2 mM dNTP, 1 M betaine and 2.5 units of Pfu polymerase in 1× Pfu buffer]. All these primers were phosphorylated at their 5′ end. This mixture was subjected to thermal cycling at 96 °C for 30 s for denaturation, followed by 30 s at 55 °C for annealing and 5 min at 72 °C for extension. A total of 20 cycles was used and the procedure was ended by 7 min at 72 °C. For both mutants, three identical reactions were performed and the products were pooled at the purification step in 35 μl of 10 mM Tris/HCl, pH 8.5. The purified PCR product (30 μl) was incubated overnight at 16 °C with 5 units of T4 DNA ligase (Fermentas) in the presence of 5% (v/v) PEG [poly(ethylene glycol)] 4000 in 50 μl of 1× ligase buffer (Fermentas). The ligation product was then digested for 2 h at 37 °C with 20 units of DpnI in order to eliminate the wild-type template. Aliquots (15 μl) of this mixture were then used to transform 100 μl of E. coli XL1blue bacteria. The complete NAT8L coding region of all plasmids was sequenced to verify the presence of the introduced deletion and the absence of random mutations. The KpnI/XbaI fragment was then subcloned into the pEF6MycHisA vector and both mutants were expressed as His6- and Myc-tagged fusion proteins.
Transfections of HEK-293T cells and Western blot analysis
HEK-293T cells [human embroynic kidney-293 cells expressing the large T-antigen of SV40 (simian virus 40)] were cultured and transfected essentially as described by Rzem et al.  using the jetPEI™ procedure. After 48 h at 37 °C, the cells were washed once with 5 ml of cold PBS. They were scraped into 0.8 ml of extraction buffer (20 mM Hepes, pH 7.1, 5 µg/ml leupeptin and 5 µg/ml antipain), frozen in liquid nitrogen, thawed and lysed by vortex-mixing. The extracts were incubated with 100 units/ml DNase I prepared in 5 mM MgSO4 (Sigma–Aldrich) for 1 h on ice before use. Protein concentration was determined with the Bradford assay , using γ-globulin as a standard. Western blots were performed as described previously  with a monoclonal anti-His6 antibody (PentaHis antibody from Qiagen) diluted 1:2000 in PBS containing 1% (w/v) BSA. After washing, membranes were probed for 1 h at room temperature (22 °C) with peroxidase-conjugated anti-(mouse Ig) antibody (Sigma–Aldrich). Detection was performed using the ECL (enhanced chemiluminescence) Western blot analysis system from GE Healthcare according to the manufacturer's protocol.
NAA measurements by isotope dilution GC–MS
Quantification of NAA in cell medium was performed by GC–MS as described by Kelley and Stamas  with N-[2H3]acetylaspartate (provided by Dr H.J. ten Brink, VU Medical Center, Amsterdam, The Netherlands). Samples (0.2 ml) were treated with hydroxylamine hydrochloride in the presence of N-[2H3]-acetylaspartate, as the internal standard, and extracted with ethyl acetate followed by diethyl ether. After evaporation of the combined extracts under a stream of nitrogen, the residue was resuspended in acetonitrile and derivatization was achieved with a BSTFA/TMCS [N,O-bis(trimethylsilyl)trifluoroacetamide/trimethylchlorosilane; 99:1] mixture. Analysis was performed by GC using a Hewlett-Packard 6890 series GC system equipped with a 30 m×0.25 mm fused silica capillary column (CP-SIL 8CB; Varian), with helium as the carrier gas, coupled with a Hewlett-Packard 5973 mass-selective detector under electron-impact fragmentation. Mass spectrometric data were collected in the selected-ion mode at m/z=274 for native NAA and m/z=277 for the synthetic isotopic NAA. Calibration curves were constructed by mixing 4.5 μg of the isotopic NAA with standard solution containing 0–6 μg of NAA, followed by drying under nitrogen and direct derivatization with the silylating agent.
Assay of Asp-NAT
Asp-NAT was assayed radiochemically in a mixture (200 μl final volume) comprising, unless otherwise stated, 10 mM potassium phosphate, 20 mM Hepes, pH 7.1, 1 mM MgCl2, 50 μM L-aspartate, 105 c.p.m. L-[U-14C]aspartate (GE Healthcare), 200 μM acetyl-CoA (Sigma–Aldrich) and 10 μl of a cell extract containing approx. 5–10 mg of protein/ml. After 20–40 min at 30 °C, the reaction was stopped by a 5-min incubation at 80 °C and 1 ml of 5 mM Hepes, pH 7.1, was added. The sample was centrifuged for 5 min at 16000 g and the supernatant was applied on to a 1 ml Dowex AG1X8 column (Cl− form, 100–200 mesh; Acros Organics) prepared in a Pasteur capillary pipette. The latter was washed with 2 ml of 5 mM Hepes, pH 7.1, followed by 5 ml of 150 mM NaCl in the same buffer, to elute unreacted aspartate, and 5 ml of 300 mM NaCl, to elute NAA. Radioactivity was counted in the presence of Ultima Gold™ (PerkinElmer) in a liquid-scintillation counter. The kinetic studies shown in Figure 3 and Supplementary Figures S2–S4 were performed using washed pellets, obtained by centrifuging extracts for 20 min at 16000 g, resuspending the pellets in the initial volume of extraction buffer, recentrifuging the preparation under similar conditions, and resuspending the final pellets in the original volume of extraction buffer.
All experiments with human DNA have been performed in accordance with the Declaration of Helsinki (2000) of the World Medical Association and have been approved by the Ethical Committee of the Faculty of Medicine of University of Louvain. Genomic DNA analysis of the child with hypoacetylaspartia  was performed with informed consent of the legal parents. The three exons of the NAT8L gene were PCR-amplified using 0.5 μM forward and reverse primers (see Supplementary Table S1) with 75 ng of genomic DNA as a template. These reactions (50 μl) were performed using Biotools DNA polymerase (Kordia Life Sciences) in the provided buffer containing 0.2 mM dNTPs and 1 M betaine. PCR products were purified and sequenced directly on both strands by use of a CEQ2000 sequencer (Beckman) or by the Macrogen sequencing service. For the screening of the 118 healthy controls, the region surrounding the deletion was amplified from genomic DNA with a specific pair of primers (see Supplementary Table S1). PCR conditions were the same except that a concentration of 1.5 M betaine was used. The PCR products were directly loaded on a 2% (w/v) agarose gel containing 0.5 μg/ml ethidium bromide to distinguish wild-type (223 bp) and mutant (204 bp) alleles.
Neuronal cultures and infection
All experiments with animals have been performed in accordance with Belgian law and with the agreement of the Ethical Committee for Animal Experiments of the Faculty of Medicine of the University of Louvain, Brussels, Belgium. Primary cultures of cortical neurons were prepared from 17–18-day-old Wistar rat embryos as described previously . Brains were dissected, transferred to HBSS (Hanks balanced salt solution) buffer medium (1 mM sodium pyruvate and 10 mM Hepes, without Ca2+ and Mg2+), dissociated in the same medium containing 1.26 mM Ca2+ and Mg2+ (0.50 mM MgCl2 and 0.41 mM MgSO4) and centrifuged for 10 min at 200 g at room temperature. Cells were plated in culture dishes pre-treated with 10 μg/ml poly-L-lysine in PBS and cultured for 7 days in neurobasal medium (Invitrogen) supplemented with 2% (v/v) B-27 medium (Invitrogen) and 0.5 mM L-glutamine prior to infection with recombinant lentivirus. The NAT8L–pEF6MycHisA plasmid was utilized for construction of recombinant lentivirus, as previously described . After 7 days of culture, neurons were incubated in the presence of the lentivirus for 3 days before analysis.
Immunofluorescence in transfected CHO (Chinese-hamster ovary) cells and infected primary neurons
CHO cells were propagated in a DMEM (Dulbecco's modified Eagle's medium)/Ham's F12 mixture (1:1) supplemented with 10% (v/v) FBS (fetal bovine serum) and antibiotics (100 units/ml penicillin and 100 μg/ml streptomycin). Cells were seeded at 50000 cells/cm2 in 24-well plates on glass coverslips coated with 25 μg/μl fibronectin (in PBS) and grown for 1 day (approx. 80% confluency). Transfection with the NAT8L-pEF6MycHisA plasmid or the PDIb [PDI (protein disulfide isomerase) domain b]–HA (haemagglutinin) construct  was carried out with 0.5 μg of total DNA and 1 μl of Lipofectamine/cm2 at 37 °C for 5 h followed by reincubation overnight in DMEM/Ham's F-12 with 10% (v/v) FBS without antibiotics. After 24 h, cells were processed for immunofluorescence as described by Mettlen et al. .
For immunofluorescence experiments, neurons were seeded at 105 cells/cm2 on glass coverslips, fixed with 4% (v/v) paraformaldehyde for 15 min at room temperature and then permeabilized with 1% (v/v) Triton X-100 in PBS for 15 min. Non-specific immunostaining was prevented by blocking for 1 h in a PBS solution containing 3% (w/v) non-fat dried milk powder at room temperature.
For primary antibodies, we used a mouse IgG2a anti-Myc (9B11) monoclonal antibody (Cell Signaling) at a dilution of 1:500, a mouse IgG1 anti-His6 monoclonal antibody (Qiagen) at a dilution of 1:50, a rabbit affinity-purified specific antibody against NAT8L at a dilution of 1:100 (see below), a rabbit anti-HA antibody at a dilution of 1:300 (Roche) or a rat monoclonal antibody recognizing KDEL-bearing proteins (Abcam) at a dilution of 1:300, either alone or combined with a mouse IgG2b monoclonal antibody to PDI (Abcam) at a dilution of 1:50 or a rabbit MAP-2 (microtubule-associated protein-2) primary antibody (provided by J.-P. Brion, Université Libre de Bruxelles, Belgium; serum was used at a dilution of 1:1000). Alexa Fluor®-labelled secondary antibodies (Molecular Probes) were each used at 5 μg/ml. For live mitochondria labelling, cells were incubated with 250 nM MitoTracker Red (CMXRos; Molecular Probes) in DMEM/Ham's F-12 at 37 °C for 30 min followed by a chase in MitoTracker-free medium for 30 min, before fixation and permeabilization. All preparations were mounted in Mowiol in the dark overnight and examined with a LSM 510 META confocal microscope (Zeiss) using a Plan-Apochromat 63×/1.4 oil DIC (differential interference contrast) objective.
Specific antibodies against NAT8L
Antibodies were raised against two peptides of mouse NAT8L (C)NTAFRGLRQHPRTQLL (designated P44) and (C)MSVDSRFRGKGIAKALG (designated P45) that have previously been shown to be suitable for immunization . These peptides were conjugated to keyhole-limpet haemocyanin (Eurogentec) and injected together into rabbits four times at 2-week intervals. Serum was taken from the rabbits 1 month after the final injection of these peptides. IgGs were bound to the column in PBS and were isolated from the sera by affinity chromatography on HiTrap™ Protein G HP columns (GE Healthcare). IgGs were eluted with 50 mM glycine/HCl, pH 2.7, containing 150 mM NaCl. Collected fractions were immediately neutralized with 1 M Tris/HCl, pH 8. Antibodies specific to the P44 and P45 NAT8L peptides were then purified by affinity chromatography on columns coupled with the P44 or P45 peptide. Peptides were coupled through their N-terminal cysteine residue to iodoacetamide immobilized on a cross-linked agarose support (Sulfolink; Thermo Scientific Pierce). IgG fractions were loaded on to the peptide-coupled column in PBS. Antibodies bound to the column were eluted with 50 mM glycine/HCl, pH 2.7, containing 150 mM NaCl, and were immediately neutralized with 1 M Tris/HCl, pH 8.
The characteristics of Asp-NAT are probably shared by only a limited number of mammalian gene products. First, this enzyme most likely belongs to a family of NATs; secondly it is mainly, if not exclusively, expressed in brain; thirdly it is membrane-bound; and finally, it should have close homologues only in species that synthesize NAA. We decided therefore to search mammalian genomes for genes encoding proteins with these characteristics.
A search in the human and mouse genomes with N-acetyltransferase as a query yielded about 30 genes. These comprised a few well-identified sequences (e.g. NATs acting on arylamines, glucosamine 6-phosphate, glutamate and spermine/spermidine) but a higher number of putative NATs (i.e. hypothetical proteins showing sequence similarity to the well-characterized NATs with undefined specificity). mRNA expression data from the Symatlas database  indicated that two of these genes were mainly expressed in brain tissue, NAT14 and NAT8L. NAT14, also called Klp1 (K562 cell-derived leucine-zipper-like protein 1) is thought to be a transcription factor regulating the expression of coproporphyrinogen oxidase by binding to a promoter regulatory element . NAT8L, also designated Shati, is overexpressed in brain after administration of metamphetamine to mice . Both proteins showed at least one putative transmembrane domain. Close homologues (of over 50% sequence identity) of NAT8L and NAT14 were only found in vertebrates. Taken together, these findings indicated that both NAT8L and NAT14 were good candidates for the Asp-NAT.
Gene structure, tissue distribution and protein sequence analysis of NAT8L
NAT8L is encoded by a three-exon gene that is on mouse chromosome 5 and human chromosome 4 (band 4p16.3). Previous PCR results obtained with mouse NAT8L indicated that the mRNA encoding this protein is expressed not only in the brain, but also in the liver and kidneys . We therefore performed quantitative PCR amplification (as described in ) on mouse tissue cDNA with pairs of primers situated in exons 1 and 2, or in exons 2 and 3. Our findings indicated that NAT8L was amplified from brain cDNA and to a lesser extent from thymus or spleen cDNA (results not shown). The abundance of the mRNA was calculated to be approx. 10-fold lower in both these tissues compared with the brain. No specific amplification was observed with liver, kidney, lung, skeletal muscle, testis and heart cDNA (results not shown).
The NAT8L cDNA was also amplified from human brain cDNA using primers situated in exons 1 and 3. Sequencing of the amplified product confirmed the predicted protein as the sequence registered under NCBI accession number EAW82541. No amplification could be observed in a further experiment with primers corresponding to a cDNA sequence encoding a protein (SwissProt accession number Q8N9F0) that differs from human and mouse NAT8L in the approximately first 90 residues. This is not surprising as this cDNA in fact contains an exon from an upstream human gene (LOC401115 in human) fused to exons 2 and 3 of the NAT8L gene (results not shown).
The NAT8L sequences of various species are aligned in Figure 1. As noted previously , the C-terminal region is homologous with NATs. The NAT8L proteins also contain a conserved N-terminal sequence of 15 residues, a proline- and alanine-rich region of variable length (residues 40–73 in the human sequence) and a highly hydrophobic region of approx. 30 amino acids that just precedes the NAT domain.
Multiple alignment of NAT8L sequences
Targeting prediction programs (TargetP  and PSort2 ) did not suggest that NAT8L had a mitochondrial propeptide or a signal peptide. It should also be noted that exon 1 of human, mouse and rat NAT8L genes contain an in-frame stop codon that precedes the initiator ATG codon. This indicates that the protein sequence truly starts with the methionine shown as Met1 in the alignment. We demonstrate in the next section that NAT14 does not correspond to Asp-NAT so we have not described its structure in detail.
Identification of Asp-NAT as NAT8L
Both mouse NAT8L and NAT14 were overexpressed in HEK-293T cells, either as native proteins or as fusion proteins with a His6 tag at the C- or the N-terminus. GC–MS analysis  of the medium collected after 48 h indicated the presence of a product co-eluting with authentic NAA in medium from the cells transfected with all NAT8L constructs (see Supplementary Figure S1 available at http://www.BiochemJ.org/bj/425/bj4250127add.htm), whereas this compound was below the detection limit of 5 μM in control cells (see Supplementary Figure S1A) and those transfected with NAT14 constructs. The fragmentation pattern of this product (see inset of Supplementary Figure S1B) was identical with that of authentic NAA, with major fragments having m/z ratios of 73, 147, 184, 245, 274 and 376 . Quantification of NAA by isotope dilution with the trideuterated compound (Figure 2B) indicated that the concentration of NAA in the medium amounted to 200–250 μM in cells transfected with the different NAT8L constructs, corresponding to approx. 300 nmol per mg of cell protein. Analysis of cell extracts indicated that this compound was also present intracellularly at concentrations of approx. 60 nmol/mg of protein (results not shown).
Effect of transfection of HEK-293T cells with mouse NAT8L or NAT14 on the NAA concentration in the medium and the Asp-NAT activity in cell extracts
Asp-NAT activity was assayed in cell extracts by monitoring the acetyl-CoA-dependent conversion of L-[14C]aspartate into a more acidic product (Figure 2C). Consistent with the results above, this activity was observed in extracts of cells transfected with the three types of NAT8L constructs, but not in other extracts. Slightly lower activities were observed with the N-terminally tagged and the C-terminally tagged forms when compared with untagged NAT8L (59% and 75% respectively). Western blot analysis with an anti-His6 antibody (Figure 2A) revealed that all His6-tagged versions of NAT8L and NAT14 were expressed. This indicated that the absence of Asp-NAT activity in cells transfected with NAT14 was not due to lack of expression of this protein.
Kinetic properties of NAT8L
The Asp-NAT activity was characterized further in extracts of cells expressing wild-type mouse NAT8L. The enzymatic activity was almost completely (>90%) recovered in the pellet after a 20-min centrifugation of frozen/thawed extracts at 16000 g (results not shown), indicating that NAT8L is membrane-bound. The kinetic studies were therefore performed on washed pellet to eliminate any soluble and potentially interfering enzymes. Detergents showed a biphasic effect on the activity (Figure 3A), increasing it at low concentrations, presumably by allowing access of the catalytic site to substrate, and being inhibitory at higher concentrations. Double-reciprocal plots of the saturation curve for aspartate were linear both in the absence and in the presence of 0.5 mM CHAPS, which increased Vmax by approx. 30% without affecting the Km (approx. 90 μM) for aspartate (see Supplementary Figure S2 available at http://www.BiochemJ.org/bj/425/bj4250127add.htm). With acetyl-CoA as the varying substrate, the double-reciprocal plots were non-linear in the absence of CHAPS, but linear in its presence, indicating that part of the enzyme was poorly accessible to acetyl-CoA in the absence of detergent (Figure 3B). The Km values were approx. 9 and 13 μM in the presence of 0.5 and 1 mM CHAPS respectively.
Kinetic properties of the Asp-NAT activity of NAT8L
[14C]Glutamate could also act as a substrate for the enzyme, with an approx. 50-fold lower affinity (Km of approx. 5 mM) and a similar Vmax compared with aspartate (results not shown), which is in agreement with previous results [31,32]. Accordingly, the incorporation of [14C]aspartate into NAA was inhibited by glutamate competitively with respect to aspartate with a Ki of approx. 6 mM (Figure 3C and Supplementary Figure S3A available at http://www.BiochemJ.org/bj/425/bj4250127add.htm). No inhibition was observed with other amino acids, indicating that the enzyme is specific (results not shown). The enzyme was also inhibited by its reaction product, NAA. This inhibition was essentially competitive with respect to aspartate (Ki = 0.56 mM; Figure 3D and Supplementary Figure S3B) and uncompetitive compared with acetyl-CoA (see Supplementary Figure S4 available at http://www.BiochemJ.org/bj/425/bj4250127add.htm).
Subcellular localization of NAT8L
As the subcellular localization of Asp-NAT was controversial, it was of interest to determine the localization of NAT8L in transfected cells. Confocal microscopy was therefore used to analyse CHO cells, transfected with C-terminally Myc-tagged, N-terminally His6-tagged and untagged NAT8L, and neurons in primary culture, infected with a lentiviral vector allowing the expression of Myc-tagged NAT8L. Immunolabelling of Myc-tagged NAT8L expressed in CHO cells or in primary neurons delineated the nuclear envelope (arrows in Figures 4 and 5) and produced a reticular pattern strongly suggestive of the ER (endoplasmic reticulum). This was confirmed by the substantial, although not complete, co-localization of Myc-tagged NAT8L with KDEL-bearing proteins of the ER (arrowheads in Figures 4 and 5). Analysis of the colour intensity profiles in the cytoplasm showed that the peaks of the NAT8L labelling mostly coincided with the peaks of ER-marker labelling (Figures 4D and 5C and Supplementary Figure S5C available at http://www.BiochemJ.org/bj/425/bj4250127add.htm). This type of co-localization is in keeping with the well-known heterogeneity among ER domains [33,34]. In contrast, the Myc-tagged NAT8L signal was fully segregated from the Golgi complex, recognized by co-transfection with the PDIb–HA construct  (Figure 4B), and from mitochondria labelled in vivo by MitoTracker Red (Figures 4C and 4E). This latter conclusion was confirmed by analysis of the intensity profiles (Figure 5B). Targeting to the nuclear envelope and the ER was independent of epitope tagging, as NAT8L fused to a Myc tag at its C-terminus, or to a His6 tag at its N-terminus, showed a similar localization pattern to native NAT8L recognized with affinity-purified antibodies (see Supplementary Figure S5A).
Co-localization of Myc-tagged NAT8L with the ER and complete segregation from the Golgi complex and mitochondria
Co-localization of Myc-tagged NAT8L with ER and its segregation from mitochondria in neurons
Mutation of NAT8L in a patient with hypoacetylaspartia
To determine whether NAT8L was mutated in the patient deficient in NAA, we amplified the three exons of the NAT8L gene and screened them for mutations. A 19-bp deletion in the homozygous form was observed in the first exon (Figure 6A). This mutation leads to a change in the reading frame after amino acid 69 and therefore it would be expected that no functional protein is produced. To rule out the possibility that this mutation would still allow the production of an active protein by codon-slipping or by reinitiation at a later AUG codon, we prepared an expression vector for this mutated form (with a C-terminal His6 tag) and transfected it into HEK-293T cells. Extracts prepared from these cells did not contain any detectable Asp-NAT activity or material reacting with the anti-His6 antibody, whereas extracts of cells transfected with the unmutated human cDNA were enzymatically active and contained an immunoreactive band of the appropriate size (Figures 6B and 6C).
Analysis of the mutation found in the patient with hypoacetylaspartia and its effect on the activity and expression of NAT8L
Several human NAT8L protein entries (e.g. Ensembl accession number BAC04426.1) comprise only the last 134 amino acids of NAT8L. The sequence of this putative shorter protein(s) is unaffected by the mutation and it was therefore important to verify whether it displayed Asp-NAT activity. Transfection of a vector allowing the expression of this shorter form resulted in the expression of a polypeptide of the expected size (approx. 18 kDa) but with no detectable enzymatic activity (Figures 6B and 6C). Taken together, these findings indicated that the mutation found in the patient indeed corresponds to a null mutation.
The patient is an adopted child and no material was therefore available to ascertain the presence of the mutation in the biological parents. The mutation found in the patient could be detected by PCR amplification of genomic DNA and gel electrophoresis. This assay allowed also the detection of the mutation in the heterozygous state, as indicated by amplifications performed on samples containing equal amounts of DNA from the patient and a control. The 19 bp deletion was not found in 118 control DNA samples either in the homozygous or the heterozygous state (results not shown).
Molecular identification of Asp-NAT
Despite its discovery about 50 years ago , Asp-NAT had not yet been identified molecularly due to the difficulty of purifying this membrane-bound enzyme . The availability of databases comprising virtually all proteins encoded by the human and mouse genomes, and those providing information on their tissue distribution, enabled us to identify two candidate proteins for this enzyme. Transfection studies in HEK-293T cells showed that NAT8L-expressing cells synthesize NAA and that extracts of these cells display Asp-NAT activity. This activity had properties similar to those of brain Asp-NAT, including its Km value for aspartate, its specificity for this amino acid, its inhibition by the reaction product, its association with membranes and its inhibition by detergents [4,5,7,32]. These findings therefore indicate that NAT8L protein is responsible for NAA synthesis in brain. This conclusion is strongly supported by the finding that the NAT8L gene is mutated in the only patient reported to have undetectable levels of NAA in brain. Considering that NAA is restricted to neurons, where it is synthesized, our identification is also consistent with immunohistochemical studies showing that mouse NAT8L (the Shati protein) is exclusively present in neurons .
As expected for Asp-NAT, close homologues of NAT8L were found in vertebrates, whose brain contains NAA, but not in insects and Caenorhabditis elegans, where NAA has never been reported. Unfortunately, the genome of the crayfish, in which NAA has been described , has not yet been sequenced.
Molecular basis of hypoacetylaspartia
The finding that the NAT8L gene is mutated in the single known case of hypoacetylaspartia [15–17] indicates that this disorder, in which no NAA is detectable in brain, is due to a lack of synthesis of this compound in brain. This excludes other potential causes, such as an increased degradation by a hyperactive aspartoacylase, or a deficiency in the transport or the formation of aspartate and/or acetyl-CoA in the compartment in which Asp-NAT is present. The finding that this mutation induces a change in reading frame indicates a null mutation. Accordingly, no NAA synthase activity could be detected in cells that had been transfected with a plasmid carrying this mutated allele.
The homozygous mutation found in the patient suggests that he may be the offspring of consanguineous parents. Unfortunately, no information is available on the biological parents. Therefore one has to be very cautious before concluding that the phenotype of this unique patient is representative of hypoacetylaspartia, as he may suffer from other recessive genetic diseases.
It is likely that hypoacetylaspartia is under-diagnosed for two reasons. First, MRS is not systematically performed to investigate neurological problems in young children. Secondly, decreased levels of NAA are usually interpreted as being secondary to a loss of neuronal integrity, rather than as a primary cause of neuronal dysfunction. However, point mutations in the NAT8L gene that would not lead to a complete loss of enzymatic activity could result in low, yet detectable, levels of NAA and be associated with milder forms of the disease than that described in the only patient known so far. It would therefore be of interest to determine more systematically the brain NAA level in young patients with mental retardation, epilepsy of unknown origin and/or ataxia, and to search the NAT8L gene for mutations if the NAA level is decreased.
Subcellular localization of Asp-NAT
As mentioned in the Introduction, the subcellular localization of Asp-NAT is still a matter of controversy. Fractionation studies of brain extracts have variously concluded a microsomal  or mitochondrial localization of Asp-NAT [6,7]. However, one of the studies concluding a mitochondrial localization in fact shows that the specific activity of Asp-NAT is higher in the microsomal than in the mitochondrial fraction , which argues for a microsomal localization. The finding that purified brain mitochondria produce NAA when incubated with glutamate (a source of aspartate) and pyruvate (a source of acetyl-CoA in the mitochondrial matrix) is taken as evidence that intact brain mitochondria contain Asp-NAT . One has to note, however, that the assay used by Patel and Clark  for measuring the NAA produced by mitochondria employed a rather crude (7-fold purified) preparation of kidney aspartoacylase from pig to convert NAA into aspartate. Therefore another metabolite (asparagine, for instance) may have been measured. These contradictory findings are most probably the result of the inherent difficulty of studying the subcellular distribution by fractionation in such a heterogeneous tissue as brain.
Recent tracer studies comparing the incorporation of radioactivity into NAA from L-aspartate and L-malate in a neuroblastoma cell line have led Arun et al.  to conclude that mitochondria are a major site of NAA synthesis. However, the TLC technique used to isolate radiolabelled NAA and L-aspartate does not allow one to separate these metabolites from other radioactive metabolites, which certainly form when cells are incubated for 3 h with a radioactive precursor, meaning the results of this work have to be interpreted with caution.
Low levels of NAA have been reported in the brains of mice deficient in Aralar (also known as Slc25a12, solute carrier family 25 member 12) , a glutamate/aspartate exchanger serving to export aspartate from brain mitochondria. This was taken to suggest that aspartate has to be transferred from the mitochondria to the cytosol before being acetylated by Asp-NAT. This argument, although suggestive against a mitochondrial localization of Asp-NAT, is not decisive because one may not exclude that a defect in Aralar also causes a decrease in the mitochondrial concentration of aspartate.
The results from present study on the localization of NAT8L indicate that this protein does not co-localize with mitochondria, consistent with the absence of a mitochondrial propeptide in the NAT8L sequence. The pattern of the expression with labelling of the nuclear membrane and the, at least partial, co-localization with ER markers, indicate instead that NAT8L is associated with the ER. This localization is independent of the tagging used and is observed not only in CHO cells, but also in neurons in primary cultures. Therefore these findings lead us to conclude that Asp-NAT is associated with the ER.
The results from the present study opens up the possibility of developing new experimental models to better define the role of NAA. None of the roles proposed so far accounts for the very high concentration of NAA in neurons, as NAA does not appear to be a neurotransmitter  and its role as an osmolyte is at best very minor compared with that of other solutes present in neural cells . The role of NAA in myelination may even be questioned as this process is only moderately delayed in the patient with NAA deficiency [16,17]. The creation of a mouse knockout model may also help define which of the symptoms found in the patient with hypoacetyaspartia truly are due to the NAA deficiency. The availability of the Asp-NAT sequence will also allow for a better definition of the mechanisms that control the concentration of this compound, including the effect of pathological conditions  and antipsychotic drugs . In this respect, the finding that NAT8L/Shati is up-regulated following metamphetamine treatment suggests that NAA plays a role, presumably an indirect one, in dopaminergic transmission [26,41]. The present study also suggests that it might be more widely necessary to perform MRS in children with developmental problems, especially when microcephaly is present, to check for NAA deficiency.
Dulbecco's modified Eagle's medium
fetal bovine serum
human embryonic kidney-293 cells expressing the large T-antigen of SV40 (simian virus 40)
magnetic resonance spectroscopy
peptide 44 of mouse NAT8L
peptide 45 of mouse NAT8L
protein disulfide isomerase
PDI domain b
Elsa Wiame, Donatienne Tyteca, Miikka Vikkula Jean-Noël Octave, Marie-Françoise Vincent, Pierre Courtoy and Emile Van Schaftingen designed the research. Elsa Wiame, Donatienne Tyteca, Nathalie Pierrot, François Collard, Gaëtane Noel and Jonathan Desmedt performed research. All authors contributed to data analysis and final revision of the manuscript, which was written by Elsa Wiame, Donatienne Tyteca, Pierre Courtoy and Emile Van Schaftingen.
We thank Ms L. Thanh Lac for her excellent technical assistance and Didier Colau for his help in antibody purification. The PDIb–HA construct was a gift from L. Ruddock (University of Oulu, Finland) and the anti-MAP2 serum a gift from J.-P. Brion (Université Libre de Bruxelles, Belgium).
This work was supported by the Interuniversity Attraction Poles Program-Belgian Science Policy [grant numbers Network P6/05 and Network P6/28]; the Désordres Inflammatoires dans les Affections Neurologiques Centre of Excellence programme of the Région Wallonne; the Région Bruxelloise; the Fonds de la Recherche Scientifique Médicale; the Fondation Saint-Luc; and by ASCO industries. E.V.W. and F.C. are Chargés de Recherche of the Fonds National de la Recherche Scientifique.