Translational accuracy depends on the correct formation of aminoacyl-tRNAs, which, in the majority of cases, are produced by specific aminoacyl-tRNA synthetases that ligate each amino acid to its cognate isoaceptor tRNA. Aminoacylation of tRNAGln, however, is performed by various mechanisms in different systems. Since no mitochondrial glutaminyl-tRNA synthetase has been identified to date in mammalian mitochondria, Gln-tRNAGln has to be formed by an indirect mechanism in the organelle. It has been demonstrated that human mitochondria contain a non-discriminating glutamyl-tRNA synthetase and the heterotrimeric enzyme GatCAB (where Gat is glutamyl-tRNAGln amidotransferase), which are able to catalyse the formation of Gln-tRNAGlnin vitro. In the present paper we demonstrate that mgatA (mouse GatA) interference in mouse cells produces a strong defect in mitochondrial translation without affecting the stability of the newly synthesized proteins. As a result, interfered cells present an impairment of the oxidative phosphorylation system and a significant increase in ROS (reactive oxygen species) levels. MS analysis of mitochondrial proteins revealed no glutamic acid found in the position of glutamines, strongly suggesting that misaminoacylated Glu-tRNAGln is rejected from the translational apparatus to maintain the fidelity of mitochondrial protein synthesis in mammals.
Protein synthesis relies on the correct unravelling of the genetic code in a two-step process, appropriate aminoacylation of tRNAs and specific codon/anticodon base pairing. The formation of the aa-tRNAs (aminoacyl-tRNAs) is an essential process consisting in the faithful pairing between the amino acid with its cognate tRNA by an ATP-dependent reaction mediated by specific aaRSs (aa-tRNA synthetases) . However, a subset of aa-tRNAs do not form by the direct and conventional pathway. The absence of the enzyme GlnRS (glutaminyl-tRNA synthetase) in the vast majority of bacteria and archaea force them to use a two-step alternative pathway for the correct formation of Gln-tRNAGln . In these organisms lacking GlnRS, tRNAGln is first mischarged with glutamic acid by an ND-GluRS [non-discriminating GluRS (glutamyl-tRNA synthetase)] . The mischarged Glu-tRNAGln is then transamidated by a Glu-AdT (glutamyl-tRNA amidotransferase) in an ATP-dependent reaction using glutamine as the amide donor . Likewise, many bacteria and archaea lack the AsnRS (asparaginyl-tRNA synthetase), with aspartic acid being first ligated to tRNAAsn by an ND-AspRS (non-discriminating aspartyl-RS), and then amidated to asparagine by an Asp-AdT (aspartyl-tRNA amidotransferase) . In prokaryotes, two different tRNA-dependent amidotransferases have been found, the heterotrimeric enzyme GatCAB (where Gat is glutamyl-tRNAGln amidotransferase)  and the heterodimeric enzyme GatDE . The former, found in bacteria and archaea, can function as both Glu-AdT and Asp-AdT. To date, GatDE has been found exclusively in archaea where it functions as Glu-AdT . GatE and GatB are paralogous subunits and belong to a protein family with tRNA-dependent kinase activity, whereas GatD and GatA appear to be the catalytic subunits with asparaginase and amidase activities respectively. Different studies, including structural analysis, strongly suggests that GatC is required for the proper folding of the catalytic A and B subunits of the GatCAB enzyme, working as a molecular belt to maintain the structure of the complex [4,6].
Mitochondrial DNA encodes the full set of tRNAs necessary for mitochondrial translation as well as two rRNAs (12S and 16S). Therefore the majority of the components of the translational apparatus are encoded in the nucleus and imported into mitochondria, including aaRSs. The prokaryotic origin of the mitochondria and the absence of GlnRS activity in purified plant and mouse mitochondria pointed to the existence of the transamidation pathway within the organelle more than 20 years ago . This idea was reinforced by whole genome sequencing analysis, which showed that vast majority of eukaryotic genomes sequenced to date lack the gene encoding a mitochondrial GlnRS , but encode GatB and GatA orthologues with putative mitochondrial localizations. Description of a dual-targeted tRNA-dependent amidotransferase in chloroplast and mitochondria from Arabidopsis thaliana further supports the existence of the transamidation route in these organelles . However, several studies carried out in yeast and mammals have shown that Gln-tRNAGln synthesis can be achieved by several mechanisms. Rinehart et al.  showed the presence of cytoplasmic GlnRS and cytoplasmic tRNAGln in Saccharomyces cerevisiae mitochondria, indicating that yeast mitochondria are endowed with a potential mechanism for the direct aminoacylation of tRNAGln fully imported from the cytoplasm. Nevertheless, this idea is argued in a later study by Frechin et al.  who reported that S. cerevisiae cytoplasmic GluRS is a non-discriminating enzyme that is also imported to the organelle, where, together with a new tRNA amidotransferase, GatFAB, it synthesizes Gln-tRNAGln. The characterization of the gene encoding the new GatF subunit demonstrated not only its essential role in the GatAB function, but also the relevance of the transamidation pathway for mitochondrial glutaminyl-tRNAGln synthesis . Finally, the identification and characterization of the human orthologues of the bacterial gatA (QRSL1), gatB (PET112) and gatC (GATC) demonstrated their mitochondrial location, identified the misaminoacylated Glu-tRNAGln intermediate by knock-down assays, and reconstituted the heterotrimeric enzyme and its transamidation capacity in vitro . All of these data strongly suggest the existence and spreading of the transamidation route within mitochondria in most living organisms, but have also demonstrated the existence of alternative and redundant pathways. Furthermore, the vast majority of the evidence in mammals has been obtained in vitro, the functional relevance of the transamidation pathway for the mitochondrial protein biosynthesis in vivo remains to be elucidated.
With the aim to study the functional role of the transamidation synthesis route of Gln-tRNAGln on mammalian mitochondrial translation in vivo, we have interfered with the gene gatA in mouse and human cultured cells. Here we demonstrate that gatA knockdown reduces the levels of the trimeric enzyme GatCAB, strongly impairs mitochondrial translation in vivo, induces an OXPHOS (oxidative phosphorylation) defect and increases ROS (reactive oxygen species) production. Our results firmly establish that GatCAB is essential for the synthesis of Gln-tRNAGln, and therefore protein synthesis, in mammalian mitochondria.
MATERIALS AND METHODS
Cloning and mutagenesis
To generate cell lines stably interfering mgatA (mouse GatA; also known as Qrsl1) in an inducible manner, a specific shRNA-expressing construct was created. The complementary oligonucleotides BgmgatA-primer (5′-GATCCCCCTGTC-TCTCCCTCATCAAGCTTGATGAGGGAGA-GACAGTTTTTGGAA-3′; sense target sequence is underlined, antisense target sequence is in italic and nucleotides in the loop are boxed) and HdmgatA-primer (5′-AGCTTTTCCAA-AAACTGTCTCTCCCTCATCAAGTCTCTTGAACTTGATG-AGGGCGAGACAGGGG-3′), 63 and 64 nt respectively in size, were designed, annealed and cloned into pSuperior.puro (Oligoengine) according to manufacturer's recommendations. Transcription from the vector of the annealed oligonucleotides renders a BgmgatA-primer-like RNA molecule that, when folded, yielded an shRNA whose processing generates a specific mgatA siRNA.
Cloning of mgatA, mgatB and mgatC cDNAs were carried out by a reverse transcription reaction of mouse liver RNA (Invitrogen SuperScript® III Reverse Transcriptase kit) followed by PCR amplifications with the primers mgatA-fw (forward), 5′-AT-GCTGGGCCGGACTCTCCG-3′, mgatA-rev (reverse), 5′-TCA-TCTCTGGGTCAGAGA-3′, mgatB-fw, 5′-GGGACTTCGG-CGACGCTATGG-3′, mgatB-rev, 5′-GAGGTGGGGACGTG-AACTTTA-3′, mgatC-fw, 5′-ATGTTGTCACGGGCCGCGA-GT-3′ and mgatC-rev, 5′-ATGTTGTCACGGGCCGCGAGT-3′. PCR products were directly cloned into the pGEMT-Easy vector (Promega). To express mGatA, mGatB and mGatC fused to eGFP, their coding regions lacking the stop codon were amplified by PCR using the corresponding constructs described above as template and the following oligonucleotides as primers: mgatA-EGFP-fw, 5′-CCCTCGAGATGCTGGGCCGGA-3′, mgatA-EGFP-rev, 5′-CCGGATCCCTCTGGGTCAGAGA-3′, mgatB-EGFP-fw, 5′-CCCTCGAGATGGCGGCGCCCAT-3′, mgatB-EGFP-rev, 5′-CCGGATCCAATGACAGCTTCCTCT-3′, mgatC-EGFP-fw, 5′-CCCTCGAGATGTTGTCACGGGCCGCGAGT-3′ and mgatC-EGFP-rev, 5′-CCGGATCCTCTGCTGTGGAAGA-AGGAAT-3′ (XhoI is underlined and BamHI is in italic). After XhoI/BamHI digestion, amplicons were cloned into the XhoI/BamHI-digested pEGFP-N1 vector (Clontech).
To generate GST-fusion constructs, mgatA, mgatB and mgatC cDNAs were cloned into NcoI/NotI restriction sites of the vector pGEX-2tKN (GE Healthcare). Thus using cDNA constructs with pGEMT-Easy as a template, we carried out PCR amplification from the primers, including recognition sites for NcoI (fw, underlined) or NotI (rev, italics), 5′-CCC-CCATGGTGGGCCGGACTCTC-3′ and 5′-TGTGCGGCCGC-TCATCTCTGGGT-3′ (mgatA-pGEX), 5′-GTTGGATCCATGG-CGGCGCCC-3′ and 5′-GTCGCGGCCGCTCACAATGACAG-3′ (gatB-pGEX), and 5′-CCCCCATGGGGTCACGGGCCGCGA-3′ and 5′-TGTGCGGCCGCTCACTCTGCTGT-3′ (gatC-pGEX).
In order to express transgenic mgatA avoiding RNAi, we made a construct to express a version of mgatA including an altered shRNA complementary sequence without affecting its amino acid composition. Thus we carried out a multiplex PCR using the mgatA-pGEMT-Easy construct as a template and the primers mgatA-fw (5′-CCAAGCTTATGCTGGGC-CGGACTCTCCGAGAAG-3′; HindIII recognition site is under-lined), mgatAmut-rev (5′-CATTTAGATACTTGGTTTTTTTA-ATCAATGACAAGCAGTTCTTGCAGAGCTC-3′; differences from wild-type nucleotides are in bold and underlined), mgatAmut-fw (5′-GAGCTCTGCAAGAACTGCTTGTCATT-GATTAAAAAAACCAAGTATCTAAATG-3′; differences from wild-type nucleotides are in bold and underlined) and mgatA-rev (5′-GGGGATCCTCATCTCTGGGTCAGAGAAGCAG-3′; BamHI recognition site is underlined). After HindIII/BamHI digestion, this modified version of mgatA (mgatA*) was cloned into the HindIII/BamHI restriction sites of the inducible pCDNA4.TO expression vector yielding the mgatA*-pCDNA4.TO plasmid. The fidelity of all constructions was confirmed by DNA sequencing.
Cell culture and RNAi assays
Murine MCA3D and C2C12 and HEK (human embryonic kidney)-293T and HeLa cells were maintained at subconfluence in DMEM (Dulbecco's modified Eagle's medium; Life Technologies) supplemented with 10% FBS, 50 μg/ml uridine, penicillin and streptomycin at 37°C in a humidified atmosphere containing 5% CO2. To generate control and mgatA RNAi stable cell lines, MCA3D cells were co-transfected with pcDNA6/TR and either with pSUPERIOR or its shRNA-expressing derivative vectors using Lipofectamine™ 2000 (Life Technologies) following the manufacturer's instructions. Stable transfectants were selected with 1.5 μg/ml puromycin and 5 μg/ml blasticidin. A total of 40 clones were analysed and two representative clones were selected. To concomitantly express mgatA shRNA and modified mgatA*, the MCA3D cell line stably transfected with plasmid expressing mgatA shRNA was then transfected as above with the mgatA*-pCDNA4.TO construct, and stable transfectants were selected with 500 μg/ml zeocin. In the same manner, control cell lines were made transfecting the pCDNA4.TO vector.
Mouse and human derivative cell lines C2C12, HEK-293T and HeLa respectively were transiently transfected with 30 nM siRNA using Lipofectamine™ 2000 following the manufacturer's instructions. We used three Silencer Select pre-designed siRNAs against mgatA (S94457, S94458 and S94459) and hgatA (human GatA; s30618, s30619 and s30620). As a control we used siRNA Silencer negative control #2 (Ki #2). For Taqman assays we used the probes Mm01185699m1 (for mgatA), Hs00733915_m1 (for hgatA) and 4308329 (for 18S rRNA). All products were purchased from Applied Biosystems. For the mgatA and hgatA transient silencing experiments, cells were analysed 96 h after transfection.
In order to compare the ability of the cells to grow under mitochondrial-stress conditions, 24 h after transfection cells were transferred to medium containing either glucose or galactose as a carbon source as described previously . Cell growth was calculated 4 days after transfection as the number of cells in glucose medium compared with those in galactose medium.
Subcellular localization of mGatA, mGatB and mGatC
To analyse the subcellular localization of the mGatA–, mGatB–and mGatC–EGFP fusion proteins, HEK-293T cells were transiently transfected with 2 μg of the corresponding expressing plasmid. At 24 h after transfection, cells were incubated with 50 nM TMRE (tetramethylrhodamine ethyl ester; Molecular Probes), a mitochondrion-selective dye. Cells were visualized with a TCS SP2 laser-scanning confocal microscope (Leica) using a ×63 magnification 1.32 numerical aperture oil-immersion UV lens. GFP-fusion proteins were imaged using a 488 nm excitation wavelength and an emission collection wavelength of 510 nm (green). Mitochondria were imaged by TMRE fluorescence using a 543 nm laser excitation wavelength and an emission collection wavelength window of 560–580 nm (red).
RNA extraction and qPCR (quantitative real-time PCR)
Total RNA was extracted from cells using the TRIzol® reagent (Sigma) following the manufacturer's instructions. RNA concentration was measured using a Nanodrop spectrophotometer. For cDNA synthesis, the High-Capacity cDNA Reverse Transcription kit (Life Technologies) with 1 μg of each mRNA was used. Relative qPCR was carried out using a 7900HT Fast Real-Time PCR system (Applied Biosystems).
Generation of anti-mGatA, anti-mGatB and anti-mGatC polyclonal antibodies
mGatA–GST, mGatB–GST and mGatC–GST fusion proteins were overexpressed in the Escherichia coli strain BL21 codon plus (Stratagene). Briefly, bacteria harbouring each construct were grown at 37°C in LB medium containing 200 μg/ml ampicillin and 30 μg/ml chloramphenicol to a D600 value of 0.6. Overexpression of the fusion proteins was then induced by adding 1 mM IPTG for 2 h at 37°C. Cells were harvested and resuspended in 2 ml of PBS and disrupted by sonication on ice. The lysate was centrifuged at 300 g for 5 min at 4°C and the supernatant was then centrifuged again at 16000 g for 15 min at 4°C. Since the overexpressed proteins formed inclusion bodies, the resulting pellets were separated by SDS/PAGE (10% gels) and the corresponding bands were cut and proteins purified by electroelution (Model 422 Electro-eluter; Bio-Rad Laboratories) following the manufacturer's instructions. Polyclonal antibodies were obtained by immunization of two New Zealand White rabbits with each purified protein following standard procedures. The rabbits used in the present study were maintained according to the law (R.D. 1201/2005, October 10th, BOE, October 21st, 2005) and housed in the animal facility of Universidad Autónoma de Madrid, Register number ES-28079-0000097, and approved by the Ethical committee for Research from the Universidad Autónoma de Madrid (Certificate Code C13–201). Rabbits were killed with CO2 followed by cervical dislocation, a method that has been authorized by DG XI European Commission 2010/63/UE, published 22 September 2010, annex IV.
Cellular pellets were resuspended in RIPA buffer [1% NP40 (Nonidet P40), 0.1% SDS, 0.5% deoxycholate, 150 mM NaCl, 2 mM EDTA and 50 mM Tris/HCl (pH 8) with Protease Inhibitor Cocktail (Sigma)] and incubated for 20 min on ice. Whole protein extracts were subsequently centrifuged for 10 min at 4°C at 13000 g to remove cell debris. Protein concentration was measured using Bio-Rad Laboratories DC Protein Assay. Proteins were subjected to SDS/PAGE (10% gels) and then transferred on to PVDF membranes (Immobilon-P; Millipore). Membranes were blocked overnight at 4°C with 5% (w/v) non-fat dried skimmed milk powder in TBST (TBS containing 0.1% Tween 20), followed by a 1 h incubation with the corresponding primary antibody. The primary antibodies used were anti-COXI (cytochrome c oxidase 1; Life Technologies), anti-Mn-SOD (manganese superoxide dismutase; Millipore), anti-catalase (Sigma–Aldrich), anti-β-actin (Sigma–Aldrich), anti-HSP60 (heat-shock protein 60; Sigma–Aldrich), and anti-AFG3L1 [AFG3-like AAA (ATPase associated with various cellular activities) ATPase 1] and anti-AFG3L2 (from Professor Thomas Langer, University of Cologne, Cologne, Germany) . Horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG (1:3000 dilution; Invitrogen) were used as secondary antibodies. Protein bands were visualized using the ECL Western Blotting Detection system (GE Healthcare).
In vivo labelling of mitochondrial translation products
In vivo labelling of mitochondrial translation products was conducted as described previously . Briefly, 4 days after interference induction with 2 μg/ml doxycycline, 2.5×105 cells were labelled for 30, 60 or 90 min in methionine/cysteine-free DMEM containing 200 μCi/ml [35S]methionine (PerkinElmer) and a 100 μg/ml concentration of an inhibitor of cytoplasmic protein synthesis (cycloheximide for pulse–chase experiments or emetine for pulse experiments). After the incubation cells were harvested (pulse experiments) or incubated in regular DMEM for 6 and 24 h (chase) after a 90 min pulse. In transient interfering experiments, cells were pulse-labelled 96 h after siRNA transfection. Finally, 50 μg of total protein extracts were subjected to SDS/PAGE (15–20% gradient gels). The labelled mitochondrial translation products were detected through direct autoradiography and quantified using a Typhoon PhosphorImager system (GE Healthcare).
ROS measurements by flow cytometry
Cytoplasmic and mitochondrial ROS were determined using DCFH-DA [5(6)-carboxy-2′,7′-dichlorodihydrofluorescein diacetate; Invitrogen] and DHR123 (dihydrorhodamine 123) respectively. Mitochondrial superoxide anion was measured using MitoSOX™ (Molecular Probes). Briefly, cells were induced with 2 μg/ml doxycycline for 4 days. DCFH-DA, DHR123 and MitoSOX™ were added to 250000 cells to a final concentration of 10 μM and incubated at 37°C for 30 min. Cells were then collected and the fluorescence analysed using a BD FACScanto II (Becton Dickinson) flow cytometer.
Cellular oxygen consumption was measured by high resolution respirometry with the OROBOROS oxygraph-2k in a standard configuration, with 2 ml of DMEM (Gibco) per chamber, at 37°C and a 750 rev./min stirrer speed. The software DatLab 4 (OROBOROS Instruments) was used for data acquisition (1 s time intervals) and analysis, including two points calibration of the polarographic oxygen sensors and online calculation of the time derivative of oxygen concentration. Respiration was automatically corrected for contributions of the polarographic oxygen sensor and of oxygen back-diffusion to total apparent respiration. A simple phosphorylation control titration regime was performed, to obtain several defined states of mitochondrial respiration in intact cells. For this purpose, exponentially growing cells were collected by trypsinization, washed, counted and resuspended. Endogenous, leaking (with oligomycin added at 49 nM) and uncoupled [with FCCP (carbonyl cyanide p-trifluoromethoxyphenylhydrazone) added at 1.2 mM] respiration analyses were performed. To correct for the oxygen consumption that was not due to the electron transport system, inhibition of complex I (with rotenone) and complex III (with antimycin A) was performed.
Mitochondria were isolated as described in . Mitochondrial proteins (200 μg) were subjected to SDS/PAGE (15–20% gradient gel). Individual gel bands corresponding to the molecular mass of MT-ATP8 (mitochondrially encoded ATP synthase 8) were cut and digested with trypsin and, after reduction with DTT (10 mM) and alkylation of cysteine groups with iodoacetamide (50 mM), modified porcine trypsin (Promega) was added at a final ratio of 1:50 (trypsin/protein). Digestion proceeded overnight at 37°C. After digestion, samples were vacuum-dried and finally dissolved in 1% acetic acid for LC–MS analysis.
The resulting tryptic peptide mixtures were subjected to nano-LC–MS for protein identification. Peptides were injected on to a C18 RP (reverse-phase) nanocolumn [100 μm I.D. (internal diameter) and 12 cm; Mediterranea™ sea; Teknokroma] and analysed in a continuous acetonitrile gradient of 0–43% B (95% acetonitrile and 0.5% acetic acid) over 90 min and 50–90% B over 1 min. A flow rate of approximately 300 nl/min was used to elute peptides from the RP nanocolumn to an emitter nanospray needle for real-time ionization and peptide fragmentation on an LTQ-Orbitrap mass spectrometer (Thermo Fisher). An enhanced FT (Fourier transform)-resolution spectrum (resolution, 60000) followed by the MS/MS spectra from most intense five parent ions were analysed along the chromatographic run (130 min). Dynamic exclusion was set at 0.5 min.
For protein identification, MS/MS were extracted and the charge state deconvoluted by Proteome Discoverer 1.0 build 43 (Thermo Fisher Scientific). All MS/MS samples were analysed using SEQUEST™ (version 18.104.22.168; Thermo Fisher Scientific). SEQUEST™ was set up to search MouseMitoprots.fasta (1.0, 14 entries), performed assuming the full trypsin digestion. Two mixed cleavages were allowed, and an error of 15 p.p.m. or 0.8 Da was set for full MS or MS/MS spectra searches respectively. A decoy database was used to identify the for FDR (false discovery rate), which was set at 0.05 by applying the corresponding filter (score against charge state). Oxidation of methionine residues, phosphorylation in serine or threonine residues, and deamidation in glutamine or asparagine residues were selected as dynamic modifications.
Results are means±S.D. for at least three independent measurements. One-way or two-way ANOVA was performed using GraphPad Prism version 5.03 for Windows. Statistical significance was assumed at P<0.05.
Identification and subcellular localization of the mouse GatCAB subunits
In order to identify the mouse orthologues of the bacterial gatA, gatB and gatC genes, we searched in the mouse non-redundant NCBI databases for similar proteins to those described in Bacillus sp. using BlastP. The results obtained predicted the candidates NCBI Reference Sequence NM_001081054.2 as mgatA, NCBI Reference Sequence NM_144896.4 as mgatB and NCBI Reference Sequence NM_029645.3 as mgatC. The in silico analysis of the putative mitochondrial localization of the mGatA, mGatB and mGatC proteins, using the prediction tools MITOPROT (http://ihg.gsf.de/ihg/mitoprot.html), TargetP 1.1 (http://www.cbs.dtu.dk/services/TargetP/), PSORT II (http://psort.hgc.jp/form2.html) and Predotar (http://urgi.versailles.inra.fr/predotar/predotar.html), showed a total agreement for a high probability of a mitochondrial location for mGatB and mGatC and a low probability for mGatA, a similar result to that found for the human orthologue proteins . In order to confirm in vivo the subcellular location of the three subunits of the mouse GatCAB heterotrimeric enzyme, we made C-terminal EGFP-fusion proteins of each of them, which were transiently expressed in HEK-293T cells. As shown in Figure 1, mGatA, mGatB and mGatC co-localized with the mitochondrial net showed by the specific mitochondrial marker TMRE, strongly suggesting the mitochondrial location for the three subunits.
Subcellular localization of mouse GatCAB
The stability of mGatB requires mGatA
The in vitro synthesis of human mitochondrial Gln-tRNAGln using recombinant human mtGluRS (mitochondrial GluRS) and hGatCAB strongly suggested a functional amidotranferase pathway in human mitochondria . However, due to the alternative ways to synthesize Gln-tRNAGln described above, it is still unknown if the amidotransferase pathway is essential for the mammalian mitochondrial protein biosynthesis in vivo. In order to address this question, we functionally studied the role of the heterotrimeric enzyme mGatCAB in mouse mitochondria using RNAi. Thus a 19-nt-long fragment complementary to a sequence in the coding sequence of mouse gatA mRNA was cloned into the inducible vector pSUPERIOR so that, when transcribed, it generated a shRNA molecule which was processed to a mature mgatA siRNA. Stable co-transfection into MCA3D cells and double selection for this construct together with the pCDNA6-TetR plasmid let us obtain cells able to interfere with mgatA in an inducible manner. A total of 40 individual clones were isolated and screened using qPCR to check for mgatA knockdown efficiency after 96 h of induction with doxycycline (2 μg/ml). Two different clones (1 and 16) that showed a high interfering efficiency (>90%) were used in all of the subsequent studies (Figure 2A). A slight decrease in mgatA expression was found in non-induced RNAi clone 1 and 16 cells, probably due to a small fraction of RNA polymerase II that may elude the repressor action.
mgatA interference in the MCA3D cell line
To analyse the levels of each subunit of the complex mGatCAB, we generated polyclonal antibodies as described in the Materials and methods section. We then performed SDS/PAGE followed by inmunoblot analysis with anti-mGatA and anti-mGatB inmunoserums to analyse the protein extracts obtained from the control and interfered cells. We detected a substantial decrease in mGatA levels in both clones (Figure 2B, upper panel) that parallel the reduction in the amount of mgatA mRNA. Interestingly, depletion of mGatA provoked a significant reduction in mGatB (Figure 2B, lower panel) without changes in the mgatB mRNA steady-state levels (Figure 2C). This is probably due to the instability of the individual subunits when they are not incorporated in the GatCAB enzyme.
Unfortunately the anti-mGatC inmunoserum was unable to detect the endogenous protein, and therefore we have not been able to quantify it in the mgatA-interfered cells. However, our results clearly show that the amount of mGatA and mGatB in the interfered cells was strongly decreased and therefore the remaining levels of fully assembled enzyme would be very low.
mgatA interference strongly impairs mitochondrial protein synthesis
In order to test if the transamidation pathway for Gln-tRNAGln formation plays a key functional role in vivo in mouse mitochondria, we evaluated mitochondrial protein synthesis in mgatA-interfered cells. We incubated control and mgatA-interfered cells with [35S]methionine for 30, 60 and 90 min in the presence of an inhibitor of cytoplasmic protein synthesis, and analysed the labelled mitochondrial translation products by SDS/PAGE, as described in the Materials and methods section. The results show a clear and generalized mitochondrial protein synthesis defect, with a reduction in labelling of up to 80% after a 90-min pulse in mgatA-interfered cells when compared with the controls (Figures 3A and 3B). This result strongly suggests that mGatCAB is an essential enzyme for mitochondrial translation in vivo.
Depletion of GatA severely compromises mammalian mitochondrial protein synthesis
In order to confirm these results, we carried out GatA interference in other mammalian cell lines. Three specific siRNAs against mgatA and hgatA and a scrambled negative control siRNA were transiently transfected into both murine C2C12 and HEK-293T cells. At 4 days after transfection, mitochondrial translation products were pulse-labelled with [35S]methionine for 90 min in the presence of emetine. As shown in Figure 3(C), the interfered C2C12 and HEK-293T cells reproduced the mitochondrial protein synthesis defect showed by MCA3D cell line in which mgatA had been interfered with.
To ensure that the defect observed in the interfered cells was exclusively due to the gatA knockdown, we overexpressed mGatA from the plasmid pcDNA4-TO in control and in RNAi-inducible clones 1 and 16. mgatA* is a modified version of mgatA whose transcription, also regulated by doxycycline, generates an mRNA that is not recognized by the induced siRNA and encodes mGatA with the same amino acid composition (mGatA expressed from the plasmid will be referred to as mGatA* in order to be distinguished from the endogenous mGatA). Thus treatment with doxycycline (2 μg/ml) induces concomitant mgatA interference and mGatA* expression. As shown in Figure 4(A), mGatA* overexpression yields levels of the protein mildly higher than the endogenous levels both in the control and in mgatA-interfered cells. In addition, the levels of mGatB were also restored, confirming the previous observation pointing to the requirement of GatCAB complex assembly to avoid degradation of the individual subunits. In addition, pulse-labelling of mitochondrial proteins of mgatA-interfered clone 1, with and without expression of mgatA*, showed that mGatA* was able to rescue the mitochondrial translation defect caused by mgatA interference (Figure 4B). These results confirmed that the observed phenotype is directly caused by the lack of mGatA.
Overexpression of GatA restores GatB levels and rescues the mitochondrial translation defect in interfered cells
mgatA interference does not affect mitochondrial protein composition and stability
The significant reduction in in vivo labelling detected in mgatA-interfered cells increased with time during incubation (Figure 3B), a result that may be due to a defect in protein synthesis combined with a decrease in protein stability. In this case, if the mischarged tRNA is not completely rejected from the translational machinery, some Glu-tRNAGln could override the quality control mechanism and the mitochondrial proteins would incorporate glutamic acid instead of glutamine, affecting their stability. To further investigate this hypothesis, we carried out pulse–chase experiments in control and mgatA-interfered cell lines (Figure 5A). The results clearly show that the stability of the newly synthesized peptides was similar in cells with reduced levels of GatCAB and the control (Figure 5B). Since an increase in misfolded mitochondrial proteins could correlate with an increase in the levels of the quality control mitochondrial proteases, we used Western blotting to analyse the levels of AFG3L1 and AFG3L2, two subunits of the AAA protease complexes involved in mitochondrial protein degradation. Interestingly, although no changes in AFG3L2 levels were detected, the levels of AFG3L1 were slightly increased, possibly due to a putative increase in the amount of truncated mitochondrial proteins provoked by a decrease in the availability of Gln-tRNAGln (Figure 5C).
Reduction in mGatA does not affect the stability and composition of mitochondrial proteins
Finally, to firmly rule out a potential misincorporation of glutamic acid into the mitochondrial proteins in mgatA-interfered cells, MS analysis of the MT-ATP8 polypeptide, which has the higher relative number of glutamine residues (4.3%) among the 13 proteins encoded by mitochondrial DNA, was carried out in both control and mgatA-interfered cells.
The results show that neither different amidation levels in MT-ATP8 (Figures 5D and 5E) nor, therefore, the presence of glutamic acid in glutamine positions was altered by mgatA depletion. Taken together, these data are consistent with the existence of a quality control mechanism functioning in vivo to maintain the accuracy of mitochondrial protein synthesis, most probably mediated by mtEF-Tu (mitochondrial elongation factor Tu) as previously suggested by in vitro analysis [17,18].
Reduction of mGatA impairs mitochondrial OXPHOS and significantly increases ROS production
The vast majority of mitochondrial DNA metabolism dysfunctions impairs the activity of one or more complexes of the respiratory chain, and therefore induces functional mitochondrial defects. In order to test alterations of mitochondrial respiration produced by mgatA depletion and the associated lowered mitochondrial translation rates, we measured coupled respiratory rates in mgatA-interfered cells by high-resolution respirometry (Figure 6A). The results reveal that oxygen consumption was decreased by 30% in the interfered cells when compared with the controls, confirming the functional effects of mgatA knockdown.
Depletion of mGatA partially affects mitochondrial respiration and results in increased ROS production
In addition, we analysed if this impairment in oxygen consumption could be in parallel with enhanced ROS production in mitochondria . Three different fluorescent dyes were used to quantify changes in cellular and mitochondrial ROS, DCFH-DA, a general cytoplasmic oxidative stress indicator, DHR123, passively diffuses into mitochondria of living cells and is oxidized by peroxides and peroxynitrites to rhodamine 123, and MitoSOX™, a compound widely used to evaluate mitochondrial superoxide (O2•−) levels. The results obtained using DHR123 and DCFH-DA show more than a 2-fold increase in ROS levels in mgatA-interfered cells (Figures 6B and 6C), indicating a high oxidative stress in cells with a defect in mitochondrial translation, probably as a consequence of impairment in the respiratory chain. However, no changes in superoxide levels could be observed using MitoSOX™ (Figure 6D). These results may suggest, in addition to the increased levels of cytoplasmic and mitochondrial ROS, that the superoxide may be efficiently detoxified. In order to corroborate if the observed increase in ROS was directly provoked by mgatA interference, cytoplasmic and mitochondrial ROS were measured in control and interfered cells expressing mGatA*. ROS production was found to be totally or partially reversed when mGatA* was overexpressed (Figures 6E and 6F).
To determine whether increased ROS levels would result in a compensatory up-regulation of the mitochondrial antioxidant defences, the steady-state levels of Mn-SOD, the mitochondrial enzyme responsible of eliminating O2•−-producing H2O2) and catalase (which degrades H2O2 to H2O) were measured. Western blot analysis of whole-cell protein extracts showed an up-regulation of Mn-SOD, whereas catalase levels remained unchanged (Figure 6G). This result is in agreement with previous results that show control levels of superoxide in mgatA-interfered cells.
hgatA interference reduces the mitochondrial-dependent growth in human cells
Galactose is not efficiently used as a glycolytic substrate since its conversion into glucose 1-phosphate is slower than that from glucose. Thus only cells with a fully functional oxidative metabolism can grow using galactose as a carbon source. To further investigate the cellular phenotype caused by the reduction in GatA in mammalian cells, we measured the relative ability of human HeLa cells to proliferate in glucose compared with galactose. Thus wild-type cells were transfected with either hgatA siRNA or silencer Ki#2 as a negative control. At 96 h after transfection, cells were harvested and counted (Figure 7A) and hgatA mRNA levels were measured using qPCR (Figure 7B). Interestingly, hgatA knockdown reduced the capability of the control human HeLa cells to grow on galactose, strongly suggesting the involvement of hgatA in mitochondrial function in human cells.
hgatA knockdown reduces the capability of human HeLa cells to grow on galactose
The fidelity of protein synthesis depends on the correct attachment of specific amino acids to its cognate isoaceptor tRNA, a process catalysed by the aaRS . Mitochondrial protein synthesis depends on 13 mRNAs, 22 tRNAs and 12S and 16S rRNAs, all encoded in the mitochondrial genome, and a high number of factors encoded in the nuclear genome, including mitochondrial aaRS. In mammalian genome databases, 19 species of mitochondrial aaRS genes are annotated, i.e. the full set with the exception of GlnRS. Therefore to synthesize Gln-tRNAGln mammalian mitochondria need to use alternative routes to the canonical aaRS-catalysed aminoacylation. Previous studies have demonstrated that human mitochondria contain a ND-GluRS and the tRNA amidotransferase GatCAB, and that both recombinant enzymes in vitro can link Gln with its cognate tRNAGln .
Using a RNAi approach in mouse and human cells, we have interfered with the expression of the gene encoding the GatA subunit, which contains the transamidase activity of the GatCAB enzyme. Interfering with mgatA in murine MCA3D cells significantly reduced the levels of not only the mGatA protein, but also of mGatB, suggesting that GatA depletion prevents the correct assembly of the enzyme and therefore induces the degradation of all subunits. Through in vivo labelling of the proteins encoded in the mitochondrial genome, we have shown that mitochondrial translation is dramatically affected after mGatA depletion, revealing an essential role for the GatCAB enzyme in the process of protein biosynthesis in mammalian mitochondria. The translation defect is reproduced when gatA was interfered with in two other mammalian systems confirming the functional role of the transamidation pathway in mammalian mitochondrial protein synthesis. The observed defect induced impairment of the OXPHOS system in MCA3D mgatA-interfered cells, as shown by a 30% oxygen consumption decrease identified using high-resolution respirometry. Additionally, there was an increase in the total amount of cellular ROS and an induction of antioxidant enzymes, indicating that the defect in the respiratory chain produced oxidative stress. We further confirmed the mitochondrial defective phenotype provoked by the absence of gatA in human cells by means of a deficient ability to grow when galactose was used as a carbon source.
Although several lines of evidence have demonstrated the presence of the transamidation pathway in mitochondria, the data have mostly been obtained using in vitro approaches and the functional relevance of this mechanism for mitochondrial protein biosynthesis in mammals is essentially unknown. In addition, the generation of Gln-tRNAGln is the most divergent route of aminoacyl-tRNA synthesis found in Nature, suggesting that several independent mechanisms could function redundantly in mitochondria. For example, previous studies have demonstrated that yeast mitochondria import cytoplasmic tRNAGln and an ND-GluRS . However, it has also been demonstrated that yeast Gln-tRNAGln is synthesized by a transamidase mechanism mediated by the enzyme GatFAB . The present study provides the first in vivo evidence of the transamidation route in mammalian mitochondria for the synthesis of Gln-tRNAGln, showing its essential role in the translation of mitochondria encoded polypeptides.
Translation fidelity is one of the basic and most important needs for the deciphering of the genetic code. It is known that the translation factor mtEF-Tu is able to discriminate mischarged tRNAs in vivo [12,17,18]; however, several studies have shown that this mechanism of quality control could be bypassed when mischarged tRNAs are present in excess [20,21]. The results of the present study clearly show that the stability of the newly synthesized proteins remains unchanged in cells with low levels of the GatCAB enzyme. Consistently, the levels of AFG3L2, a subunit of m-AAA protease complexes involved in mitochondrial protein turnover , are not increased in mgatA-interfered cells. Surprisingly, the levels of AFG3L1, another subunit of m-AAA protease complexes, were slightly increased. One possible explanation could be a more important involvement of AFG3L1 than AFG3L2 in the degradation of the putative higher amount of truncated mitochondrial peptides caused by a decrease in the availability of Gln-tRNAGln.
These observations are consistent with MS data, which showed no difference between the interfered cells and controls in the amidation level of MT-ATP8 peptide, the mitochondrial protein which has the highest percentage of glutamine residues among the 13 proteins encoded by mitochondrial DNA. Our results also support the previously described low-binding affinity of mtEF-Tu with mischarged tRNAs [17,18], and strongly suggest that the misaminoacylated form of tRNAGln is rejected from the translational apparatus in order to maintain the accuracy of mitochondrial translation.
ATPase associated with various cellular activities
AFG3-like AAA ATPase
Dulbecco’s modified Eagle’s medium
human embryonic kidney
heat-shock protein 60
manganese superoxide dismutase
mitochondrially encoded ATP synthase 8
mitochondrial elongation factor Tu
quantitative real-time PCR
reactive oxygen species
tetramethylrhodamine ethyl ester
Lucía Echevarría and Miguel Fernández-Moreno carried out the experiments. Paula Clemente, Rosana Hernández-Sierra and María Esther Gallardo provided experimental assistance or reagents throughout the course of the work. Lucía Echevarría, Miguel Fernández-Moreno and Rafael Garesse wrote the paper. Miguel Fernández-Moreno and Rafael Garesse directed the research.
We thank Professor Thomas Langer for the anti-AFG3L1 and anti-AFG3L2 antibodies.
This work was supported by the Center for Biomedical Research on Rare Diseases (CIBERER), the Instituto de Salud Carlos III [grant numbers PI 07/0167 and PI 10/0703 (to R.G.)] and the Comunidad de Madrid [grant number S2010/BMD-2402 (to R.G.)]. L.E. was funded by the Ministerio de Ciencia e Innovación (Spain) and P.C. by the Comunidad de Madrid.
Present address: University of Versailles-Saint Quentin en Yvelines (UVSQ), UFR des sciences de la santé Simone Veil, 78180 Montigny, France
Present address: Department of Laboratory Medicine, Karolinska Institutet, Stockholm, Sweden