Aminoacylation and translational quality control strategy employed by leucyl-tRNA synthetase from a human pathogen with genetic code ambiguity

Aminoacyl-tRNA synthetases should ensure high accuracy in tRNA aminoacylation. However, the absence of significant structural differences between amino acids always poses a direct challenge for some aminoacyl-tRNA synthetases, such as leucyl-tRNA synthetase (LeuRS), which require editing function to remove mis-activated amino acids. In the cytoplasm of the human pathogen Candida albicans, the CUG codon is translated as both Ser and Leu by a uniquely evolved CatRNASer(CAG). Its cytoplasmic LeuRS (CaLeuRS) is a crucial component for CUG codon ambiguity and harbors only one CUG codon at position 919. Comparison of the activity of CaLeuRS-Ser919 and CaLeuRS-Leu919 revealed yeast LeuRSs have a relaxed tRNA recognition capacity. We also studied the mis-activation and editing of non-cognate amino acids by CaLeuRS. Interestingly, we found that CaLeuRS is naturally deficient in tRNA-dependent pre-transfer editing for non-cognate norvaline while displaying a weak tRNA-dependent pre-transfer editing capacity for non-cognate α-amino butyric acid. We also demonstrated that post-transfer editing of CaLeuRS is not tRNALeu species-specific. In addition, other eukaryotic but not archaeal or bacterial LeuRSs were found to recognize CatRNASer(CAG). Overall, we systematically studied the aminoacylation and editing properties of CaLeuRS and established a characteristic LeuRS model with naturally deficient tRNA-dependent pre-transfer editing, which increases LeuRS types with unique editing patterns.


INTRODUCTION
Aminoacyl-tRNA synthetases (aaRSs) are essential components required to establish the genetic code during protein biosynthesis by coupling specific amino acids with their cognate tRNAs in a two-step aminoacylation reaction (1,2). This process requires amino acid activation by condensation with ATP to form the aminoacyl-adenylate (aa-AMP) and pyrophosphate; the activated amino acid is then transferred to the cognate tRNA to yield the aminoacyl-tRNA (aa-tRNA), which is then transferred to the protein biosynthesis machinery as a building block (1). Aminoacylation of tRNA requires adequate efficiency and accuracy, which requires tightly regulated control of the speed of the aa-tRNA production for the ribosome and the risk of generation of aberrant aa-tRNA pairs (3)(4)(5). Transfer RNA always harbors various identity determinants and/or anti-determinants, facilitating selection of the correct tRNA from a large pool of tRNA species (6). However, the specificity of aaRS is greatly challenged by the presence of various types of amino acids and their analogues and the fact that amino acids differ only in the side-chain. AaRSs that do not show an overall selectivity above 1 in 3000 are predicted to require some form of proofreading (editing) mechanism to maintain sufficient accuracy during aa-tRNA synthesis (5,7,8). Editing activity has evolved in half of the currently identified aaRSs to remove any aberrantly produced aa-AMP (pre-transfer editing) and/or aa-tRNA (post-transfer editing). This is an essential checkpoint that ensures translational fidelity (5). Pre-transfer editing can be further divided into tRNAindependent and tRNA-dependent pre-transfer editing. In tRNA-independent pre-transfer editing, the non-cognate aa-AMP is hydrolyzed into the amino acid and AMP molecules without the presence of cognate tRNA, whereas in tRNA-dependent pre-transfer editing, aa-AMP hydrolysis is triggered by the addition of the cognate tRNA (9)(10)(11).
Mis-translation due to the impairment or loss of editing activity can lead to ambiguity of the proteome, having a seriously negative effect on the cellular function of most organisms and causing neuron-degeneration in a mouse model (12).
Leucyl-tRNA synthetase (LeuRS) is a large multidomain class Ia aaRS with both aminoacylation activity to generate Leu-tRNA Leu and editing activity to clear non-cognate aa-AMP and aa-tRNA (13). It can be divided into bacterial and archaeal/eukaryotic types based on primary sequence and domain location (14). Both types of LeuRSs usually consist of a Rossmannfold domain (for amino acid activation and aminoacylation), an a-helix bundle, a C-terminal domain (for tRNA binding) and a CP1 domain (for editing) (15)(16)(17). Extensive studies of various LeuRS species all found that non-cognate norvaline (Nva) is the most significantly mis-activated amino acid among all the noncognate amino acids tested, including Ile, Val, Met and a-amino butyric acid (ABA). For instance, compared with cognate Leu, Nva is mis-activated by Aquifex aeolicus LeuRS (AaLeuRS) (9), Saccharomyces cerevisiae LeuRS (ScLeuRS) (18), human cytoplasmic LeuRS (hcLeuRS) (19), Mycoplasma mobile LeuRS (MmLeuRS) (20), human mitochondrial LeuRS (hmtLeuRS) (unpublished data) 72-, 105-, 100-, 122-and 180-fold less efficiently, respectively. Nva is a non-proteinogenic amino acid differing from Leu only by the absence of a side-chain methyl group. Nva is naturally present in vivo and is a by-product of the Leu biosynthesis pathway (21). Its synthesis is predominantly related to an imbalance in the synthesis of the branched-chain amino acids under pyruvate-high conditions. In addition, Nva significantly accumulates immediately after a shift from aerobic culture conditions to oxygen limitation at high glucose concentrations (22). Therefore, the amount of Nva is dynamic and varies according to the environment. The incorporation of Nva in proteins at Leu codons has been clearly demonstrated. It has been reported to be a natural component of an antifungal peptide of Bacillus subtilis (23) and can be intentionally inserted into heterologous proteins by culturing Escherichia coli in the presence of Nva (US patent, Nov 7, 1989, 4879223). Accompanied by conditions of an elevated ratio of available Nva to Leu in the medium, increasing mis-incorporation of Nva at Leu codons has been observed in recombinant human hemoglobin produced in E. coli as a result of misaminoacylation of tRNA Leu by E. coli LeuRS (EcLeuRS) (24). It is proposed that Nva replacement may disrupt the correct folding and assembly of hemoglobin and other proteins (24). All this evidence suggests that Nva mis-activation by LeuRS is a non-artificial event that occurs in vivo, and that mis-charged Nva-tRNA Leu can be accommodated and used by the ribosome. Therefore, editing of Nva by LeuRS seems to be essential for the correct functioning of organisms.
Based on significant mis-activation of Nva, editing catalyzed by LeuRS (with a functional CP1 domain) has been shown to be one of the most interesting editing mechanisms. This process is predominantly mediated by three diverse pathways (tRNA-independent, tRNA-dependent pre-transfer and post-transfer editing) (10). Both types of LeuRS critically depend on the editing active site embedded in the CP1 domain to perform post-transfer editing (15)(16)(17)25). However, MmLeuRS harbors only tRNA-independent pre-transfer editing activity owing to its natural lack of the CP1 domain (20). Another example of a unique LeuRS is hmtLeuRS, which possesses a degenerate editing active site in the CP1 domain as well as defunct post-transfer editing (26) and tRNA-dependent pre-transfer editing activities (unpublished data). Combining site-directed mutagenesis and AMP formation methodology, the contribution of different pathways to the overall editing process can be quantified (9,10,19). Strikingly, there are quantitative and species-specific differences in the contribution of a specific pathway to the total editing activity of a LeuRS (9,10,19). To evaluate the significance of each mechanism, we have attempted to generate LeuRSs lacking one or more editing mechanism(s); to date, two types have been successfully established. One type contains LeuRSs with abolished post-transfer editing activity, obtained by introducing mutations at key residues (e.g. EcLeuRS-T252R, AaLeuRS-T273R, AaLeuRS-D373A, ScLeuRS-D419A, hcLeuRS-D399A, Giardia lamblia (Gl) LeuRS-D444A) (9,10,18,19,27) or by the inclusion of a small molecule inhibitor (AN2690) of the CP1 editing domain (10). The second type includes LeuRSs for which both the posttransfer editing and tRNA-dependent pre-transfer editing activities (EcLeuRS-Y330D, AaLeuRS-Y358D) have been abolished (10). Our aim was to determine whether a LeuRS with defective tRNA-dependent pretransfer editing activity but intact post-transfer editing would produce mis-charged tRNAs. However, extensive efforts to establish such a LeuRS model failed.
The protein biosynthesis machinery of Candida albicans is of great interest, not only because it is a human pathogen but also in its cytoplasm, the universal Leu codon CUG is translated as both Ser (97%) and Leu (3%) (28,29). This genetic code alteration is mediated by a uniquely evolved tRNA, which bears a CAG anti-codon [C. albicans tRNA Ser (CAG), CatRNA Ser ] and can be aminoacylated either with Ser by C. albicans seryl-tRNA synthetase (CaSerRS) or with Leu by leucyl-tRNA synthetase (CaLeuRS) (29). Therefore, the proteome of C. albicans is ambiguous with some proteins exhibiting differences in primary sequences. For example, a key player in CUG reassignment, CaSerRS, has two isoforms (SerRS-Leu 197 and SerRS-Ser 197 ). The residue at position 197 is located at the SerRS dimer interface, and replacement of Ser by Leu at this site induces a local structural rearrangement, leading to a slightly higher (27%) activity of SerRS-Leu 197 compared with SerRS-Ser 197 (30). These data indicate that distribution of the CUG codon and its ambiguity is not random and has potential significance. CaLeuRS is another critical molecule in the CUG reassignment in C. albicans, which charges CatRNA Ser with Leu to produce Leu-CatRNA Ser . CaLeuRS comprises 1098 residues and has a molecular mass of 126 kDa. A single CUG codon is present at position of 919 of CaLeuRS, which is located at the C-terminal domain. Thus, CaLeuRS should also have two isoforms, CaLeuRS-Ser 919 and CaLeuRS-Leu 919 . Based on the decoding rule of C. albicans (28,29), CaLeuRS exists mainly as CaLeuRS-Ser 919 ($97%), and this was used here as the wild-type form.
In this study, we compared the activity of two LeuRS isoforms and analyzed the cross-species tRNA Leu recognition and editing capacity of CaLeuRS. Interestingly, we showed that CaLeuRS is naturally deficient in tRNAdependent pre-transfer editing activity but with obvious tRNA-independent pre-transfer editing and efficient post-transfer editing of Nva. However, it harbored a measurable level of tRNA-dependent pre-transfer editing of ABA when specific tRNA was present, although editing of ABA seemed not to be a necessity, as the rejection of ABA was efficient at the aminoacylation active site. Furthermore, post-transfer editing of CaLeuRS was not tRNA Leu species-specific but was functional for mischarged CatRNA Ser (CAG), being recognized by other eukaryotic LeuRSs. The KOD-plus mutagenesis kit was obtained from TOYOBO (Japan). T4 ligase, nuclease S1 and restriction endonucleases were obtained from MBI Fermentas (Pittsburgh, PA, USA). Phusion high-fidelity DNA polymerase was purchased from New England Biolabs (Ipswich, MA, USA). Ni 2+ -NTA Superflow was purchased from Qiagen, Inc. (Germany). Polyethyleneimine cellulose plates were purchased from Merck (Germany). Pyrophosphatase (PPiase) was obtained from Roche Applied Science (China). The dNTP mixture was obtained from TaKaRa (Japan). Oligonucleotide primers were synthesized by Biosune (China). Escherichia coli BL21 (DE3) cells were purchased from Stratagene (USA).

Gene cloning, mutagenesis and protein expression
The C. albicans genome was kindly provided by Prof. Jiang-Ye Chen of our institute and was used as the template for amplifying genes encoding CaLeuRS, C. albicans SerRS (CaSerRS) and C. albicans mitochondrial LeuRS (CamtLeuRS). Gene sequences of CaLeuRS, CaSerRS and CamtLeuRS were obtained from the Candida Genome Database (http://www.candidagenome. org/). CaLeuRS, CaSerRS and CamtLeuRS genes were cloned into pET28a at the NheI and XhoI sites with Nterminal His 6 -tag (the mitochondrial targeting sequence of CamtLeuRS had been removed). Plasmids containing EcLeuRS (10), ScLeuRS (18) and Pyrococcus horikoshii LeuRS (PhLeuRS) (31) were constructed previously. The E. coli tRNA(m 1 G37) methyltransferase (TrmD) gene was amplified from the E. coli genome and inserted between the EcoRI and XhoI of sites of pET28a. The plasmid expressing E. coli tRNA nucleotidyltransferase (CCase) was provided by Dr. Gilbert Eriani (Strasbourg, CNRS, France). Mutation at Asp 422 of the CaLeuRS gene was performed with the KOD-plus mutagenesis kit according to the manufacturer's instructions. Asp 422 corresponds to Asp 373 , Asp 419 , Asp 444 and Asp 399 of AaLeuRS, ScLeuRS, GlLeuRS and hcLeuRS, respectively, which are crucial for post-transfer editing of these LeuRSs (9,10,18,19,27). The CTG and TCG codons at position 919 in the CaLeuRS gene were used to over-express the gene encoding CaLeuRS-Leu 919 and CaLeuRS-Ser 919 , respectively. All constructs were confirmed by DNA sequencing. E. coli BL21 (DE3) was transformed with various constructs. A single colony of each of the transformants was chosen and cultured in 500 ml of 2 Â YT medium at 37 C. When the cells reached mid-log phase (A 600 = 0.6), expression of the recombinant proteins was induced by the addition of 0.2 mM isopropyl-1-thio-b-D-galactopyranoside for 8 h at 22 C. Protein purification was performed according to a previously described method (32).

tRNA gene cloning, transcription and methylation
CatRNA Leu (UAA) and CatRNA Ser (CAG) genes were cloned between the PstI and EcoRI sites of pTrc99b with an N-terminal T7 promoter. Detailed T7 in vitro run-off transcription of CatRNA Leu and CatRNA Ser has been described previously (33). The amino acid accepting activities of CatRNA Leu (UAA) or CatRNA Ser (CAG) are 1390 and 1208 pmol/A 260 , respectively. The methyl group of m 1 G37 of CatRNA Ser is a critical element for recognition by LeuRS (29). The purified CatRNA Ser transcript was methylated at position G37 with E. coli TrmD (34) in a mixture containing 0.1 M Tris-HCl (pH 8.0), 1 mM DTT, 0.1 mM EDTA, 6 mM MgC1 2 , 24 mM NH 4 C1, 7.5 mg of bovine serum albumin, 5 mM CatRNA Ser transcript, 100 mM S-adenosylmethionine, 1 U/ml RNase inhibitor and 10 mM TrmD at 37 C for 1.5 h. Approximately 45% of transcripts were methylated in this reaction as estimated in a control experiment with 3 H-labeled S-adenosylmethionine. m 1 G37-CatRNA Ser was ethanol-precipitated at À20 C after phenol/chloroform extraction (twice) and dissolved in 5 mM MgCl 2 . All CatRNA Ser used in this study refers to m 1 G37-CatRNA Ser . Transcribed or over-expressed E. coli tRNA Leu (GAG) (EctRNA Leu ) and human cytoplasmic tRNA Leu (CAG) (hctRNA Leu ) were obtained according to methods described elsewhere, and their amino acid accepting activity was $1500 pmol/A 260 (19,35). In vitro activity assays ATP-PPi exchange measurement was carried out at 30 C in a reaction mixture containing 60 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 2 mM DTT, 4 mM ATP, 2 mM [ 32 P]tetrasodium pyrophosphate, 1 mM Leu or 50 mM non-cognate ABA, Nva, Val, Ile, Met, Ser and 20 nM CaLeuRS. The kinetics of amino acid activation were measured in the presence of Leu (3-1000 mM) or Nva (0.3-50 mM) or ABA (3-940 mM). Samples of the reaction mixture were removed at specific time-points, added to 200 ml of quenching solution containing 2% activated charcoal, 3.5% HClO 4 and 50 mM tetrasodium pyrophosphate and mixed by vortexing for 20 s. The solution was filtered through a Whatman GF/C filter, followed by washing with 20 ml of 10 mM tetrasodium pyrophosphate solution and 10 ml of 100% ethanol. The filters were dried, and [ 32 P]ATP was counted using a scintillation counter (Beckman Coulter).
Aminoacylation of CatRNA Leu with Leu was performed in a reaction mixture containing 60 mM Tris-HCl (pH 7. Mis-aminoacylation of [ 32 P]CatRNA Leu with Nva or ABA was carried out at 30 C in a reaction mixture containing 60 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 2 mM DTT, 4 mM ATP, 5 mM 'cold' CatRNA Leu , 1mM [ 32 P]CatRNA Leu , 20 mM Nva or 376 mM ABA and 1 mM CaLeuRS or CaLeuRS-D422A. Samples at specific time-points were taken for ethanol precipitation with NaAc (pH 5.2) at À20 C overnight. The precipitated samples were centrifuged (10 000g) at 4 C for 30 min, dried at room temperature for 30 min and digested with 6 ml of nuclease S1 (25 U) for 2 h at 37 C. After treatment with nuclease S1, aminoacyl-[ 32 P]tRNA should produce aminoacyl-[ 32 P]AMP and free [ 32 P]tRNA should produce [ 32 P]AMP. Samples (2 ml) of the digestion mixture were loaded and separated by thin layer chromatography (TLC) in 0.1 M NH 4 Ac and 5% acetic acid. Known amounts of [a-32 P]ATP were diluted and loaded onto the TLC plate for the purposes of quantification. The plates were visualized by phosphorimaging, and the data were analyzed using Multi-Gauge Version 3.0 software (FUJIFILM Preparation of Nva-[ 32 P]CatRNA Leu or Nva-[ 32 P]CatRNA Ser was carried out with editing-deficient ScLeuRS-D419A or CaLeuRS-D422A, respectively, in a reaction mixture, which was identical to that used for misaminoacylation. Post-transfer editing of pre-formed Nva-[ 32 P]CatRNA Leu was performed in a reaction mixture containing 60 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 1 mM Nva-[ 32 P]CatRNA Leu and 30 nM CaLeuRS or CaLeuRS-D422A at 30 C. Post-transfer editing of pre-formed Nva-[ 32 P]CatRNA Ser was performed in a reaction containing 60 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 1 mM Nva-[ 32 P]CatRNA Ser and 100 nM CaLeuRS or CaLeuRS-D422A at 30 C. After nuclease S1digestion, the amount of hydrolyzed mis-charged [ 32 P]tRNAs was assayed by TLC according to the procedure described for misaminoacylation.
The AMP formation assay was carried out at 30 C in a reaction mixture containing 60 mM Tris-HCl (pH 7.

RESULTS
CaLeuRS-Leu 919 is more active than CaLeuRS-Ser 919 Determination of the crystal structure of the PhLeuRS-tRNA Leu complex (Protein Data Bank, PDB 1WZ2) shows that the amino acid at position 919 of archaeal/eukaryotic LeuRSs is located in the a29 helix of the C-terminal domain ( Figure 1A and B). The primary sequence of the 919-containing a29 helix is not conserved; thus, it is difficult to identify its homologous site in the crystal structure of PhLeuRS. The CUG codon in E. coli is uniformly translated as Leu. Therefore, we introduced CTG and TCG codons at this position in the CaLeuRS gene to facilitate expression of CaLeuRS-Leu 919 and CaLeuRS-Ser 919 , respectively, in E. coli.
No differences were observed in amino acid activation by CaLeuRS-Ser 919 and CaLeuRS-Leu 919 ( Figure 1C), indicating that Leu or Ser insertion at this position has no direct effect on the structure or function of the aminoacylation active site located in the Rossmann-fold domain. This is consistent with the fact that residue 919 is spatially distant from the aminoacylation active site (>50 Å in the PhLeuRS-tRNA Leu structure) ( Figure 1B).

Subsequent comparisons of the aminoacylation kinetics of
CaLeuRS-Ser 919 and CaLeuRS-Leu 919 revealed that CaLeuRS-Leu 919 displayed a higher K m (2.91 ± 0.37 mM) and a higher k cat (0.62 ± 0.08 s À1 ) compared with the values determined for CaLeuRS-Ser 919 (K m : 1.87 ± 0.23 mM, k cat : 0.31 ± 0.05 s À1 ). These data indicated that CaLeuRS-Ser 919 has a stronger binding affinity for transcribed CatRNA Leu (UAA) during aminoacylation (Table 1). Based on the structure, we suggested that the presence of Ser in this helix may facilitate binding with the variable stem-loop element of CatRNA Leu (UAA). The catalytic efficiency of CaLeuRS-Leu 919 (213.06 s À1 mM À1 ) is $30% higher than that of CaLeuRS-Ser 919 (165.78 s À1 mM À1 ). This phenomenon is similar to that observed in the case of CaSerRS, for which no differences were observed in the amino acid activation of CaSerRS-Leu 197 and CaSerRS-Ser 197 , whereas CaSerRS-Leu 197 showed a slightly (27%) higher activity than CaSerRS-Ser 197 (30).
According to the decoding rule of C. albicans, CaLeuRS is present in the cytoplasm mainly in the form of CaLeuRS-Ser 919 . Thus, in the following study, we used CaLeuRS-Ser 919 as the wild-type CaLeuRS.
Yeast LeuRSs efficiently recognized bacterial, yeast and human tRNA Leu s Species-specific charging of tRNA is common for some aaRSs systems. The aaRSs from higher organisms often have the capacity to charge tRNA from lower species, whereas aaRSs from lower organisms fail to aminoacylate tRNA from higher ones. It is unclear whether yeast LeuRS is able to recognize various tRNA Leu s from other species. In this study, we investigated the tRNA Leu recognition capacity in detail using CaLeuRS as a model system.
The CatRNA Leu gene could not be over-expressed in E. coli and was obtained by T7 in vitro transcription. We also obtained transcribed and over-expressed EctRNA Leu and hctRNA Leu to reveal any potential role of base modification in recognition. Moreover, as transcribed S. cerevisiae tRNA Leu without modification showed no Leu accepting activity (data not shown), commercial S. cerevisiae yeast total tRNA was used.
CaLeuRS recognized all the available tRNAs. It aminoacylated transcribed or over-expressed EctRNA Leu with similar k cat values (0.474 ± 0.026 and  0.555 ± 0.061 s À1 , respectively), although the K m for transcribed EctRNA Leu (2.68 ± 0.39 mM) was nearly 4-fold greater than that for over-expressed EctRNA Leu (0.74 ± 0.08 mM). Interestingly, CaLeuRS efficiently charged both transcribed and over-expressed hctRNA Leu , which was from a higher organism. A similar recognition pattern as seen with the two EctRNA Leu s was also observed, with comparable k cat values but a smaller K m for over-expressed hctRNA Leu , indicating base modification was important for tRNA recognition. Additionally, CaLeuRS obviously charged yeast total tRNA with K m and k cat values of 0.39 ± 0.05 mM and 0.174 ± 0.019 s À1 , respectively, and with the greatest catalytic efficiency (1486.05 s À1 mM À1 ) for over-expressed hctRNA Leu among all the tested tRNAs ( Table 2).
Owing to recognition ability of CaLeuRS for hctRNA Leu , we further explored the capacity of ScLeuRS to aminoacylate bacterial and human tRNA Leu s as well as yeast tRNA. ScLeuRS aminoacylated yeast total tRNA with K m and k cat values of 0.332 ± 0.037 mM and 0.188 ± 0.025 s À1 , respectively. However, its k cat values for over-expressed EctRNA Leu or hctRNA Leu increased >10-fold (2.09 ± 0.16 and 2.19 ± 0.22 s À1 , respectively), although the K m values differed from each other remarkably (2.19 ± 0.47 mM for over-expressed EctRNA Leu and 0.111 ± 0.027 mM for over-expressed hctRNA Leu ). These data demonstrated that over-expressed hctRNA Leu was the best aminoacylation substrate for ScLeuRS (catalytic efficiency 19 729.73 s À1 mM À1 ) and furthermore suggested that base modification was important during recognition or catalysis. ScLeuRS recognized transcribed EctRNA Leu with a similar K m (1.71 ± 0.22 mM) but a sharply decreased k cat (0.134 ± 0.011 s À1 ) compared with the values of over-expressed EctRNA Leu . It also recognized transcribed hctRNA Leu with an increased K m (0.926 ± 0.170 mM) and a decreased k cat (0.887 ± 0.114 s À1 ) compared with the values for over-expressed hctRNA Leu (Table 2).
Overall, both CaLeuRS and ScLeuRS recognized bacterial, yeast and human tRNA Leu s. Interestingly, recognition of CatRNA Leu by hcLeuRS was negligible (Supplementary Figure S1A). Futhermore, EcLeuRS failed to acylate CatRNA Leu (Supplementary Figure  S1B). These results were unexpected because it is widely accepted that aaRSs from higher organisms are able to aminoacylate tRNAs from lower organisms.

Amino acid activation capacity of CaLeuRS
Various LeuRSs have been shown to mis-activate a series of non-cognate amino acids. To investigate mis-activation of non-cognate amino acids by CaLeuRS, we included ABA, Nva, Val, Ile, Met, Ser in the ATP-PPi exchange reaction. The data clearly showed that CaLeuRS significantly mis-activated Nva; furthermore, ABA was also misactivated to an obvious level compared with the control reaction conducted in the absence of amino acids. In contrast, mis-activation of Val, Ile, Met and Ser was comparable with that of the control reaction conducted in the absence of amino acids (Figure 2). To further define the quantitative discrimination capacity of the aminoacylation active site of CaLeuRS, we measured the activation kinetics for cognate Leu and non-cognate Nva and ABA of CaLeuRS. CaLeuRS gave much higher K m values for Nva (5487 ± 645 mM) and ABA (120387 ± 1698 mM); however, the k cat values were comparable with that for Leu, equating to discriminator factors for Nva and ABA of 220 and 3462, respectively (Table 3). These results indicated that Nva is a real challenge for CaLeuRS and that removal of Nva-AMP and/or Nva-tRNA Leu is required to maintain the translational quality control. However, the discrimination against ABA was below the proposed threshold of 1/3000, indicating that editing of ABA may not be necessary.
CaLeuRS exhibited little tRNA-dependent pre-transfer editing for Nva The hydrolysis of Nva-AMP or Nva-tRNA Leu may be separately or simultaneously catalyzed by CaLeuRS. Editing leads to the net consumption of ATP (yielding AMP) due to repetitive cycles of synthesis-hydrolysis of the non-cognate products. This is the basis of the TLCbased AMP formation methodology, in which the editing capacity is measured by monitoring the quantity of AMP produced (9)(10)(11)36,37). In the presence of tRNA and non-cognate amino acid, the TLC assay measures the global editing activity, including the tRNA-independent and tRNA-dependent pre-transfer editing in addition to the post-transfer editing. In the absence of tRNA, but with non-cognate amino acid, AMP is produced only from tRNA-independent pre-transfer editing activity (10). We initially assayed Nva-included AMP formation by CaLeuRS with or without the CatRNA Leu transcript. CaLeuRS showed similar observed rate constants (k obs ) with (0.28 ± 0.04 s À1 ) or without (0.25 ± 0.03 s À1 ) CatRNA Leu , indicating that CaLeuRS possesses little tRNA-dependent editing capability (Table 4). To reveal any effect of the residue at position 919, we also determined the editing capacity of CaLeuRS-Leu 919 and obtained comparable k obs values in the absence (0.22 ± 0.02 s À1 ) and presence (0.25 ± 0.05 s À1 ) of the CatRNA Leu transcript. These data suggested that CaLeuRS possesses negligible CatRNA Leu -dependent editing capability. It is also possible that modified bases of tRNA Leu play a crucial role in Nva-editing. However, the lack of availability of over-expressed CatRNA Leu impeded exploration of the potential role of modified bases in editing. Therefore, we performed AMP formation assays with Nva in the presence of transcribed or overproduced EctRNA Leu in E. coli, which could be leucylated by CaLeuRS. In accordance with our findings, the k obs values with unmodified or modified EctRNA Leu were 0.40 ± 0.06 or 0.51 ± 0.04 s À1 , respectively. Similarly, transcribed or over-produced hctRNA Leu in E. coli, both of which were effectively aminoacylated by CaLeuRS, stimulated Nva-editing of CaLeuRS with k obs values of 0.33 ± 0.02 or 0.38 ± 0.05 s À1 , respectively (Table 4). These data showed that the modified bases of tRNA Leu had little effect on the tRNA-dependent editing of CaLeuRS. Based on data from various transcripts or the tRNA Leu with modified bases, we concluded that CaLeuRS has little tRNA-dependent editing activity for Nva. Whether it was deficient in post-transfer editing would be explored later in the text. By comparing the k obs values with or without tRNAs, we also observed that post-transfer editing, if it occurred, contributed little to the total editing, and that the observed k obs with tRNAs was almost a reflection of the tRNA-independent pre-transfer editing.

CaLeuRS exhibited obvious and efficient post-transfer editing to prevent synthesis of Nva-tRNA Leu
The absence of significant stimulation of editing of Nva by CaLeuRS with various tRNA Leu s prompted us to investigate its post-transfer editing capability. Usually, the posttransfer editing ability of various LeuRSs is monitored by hydrolysis of Ile-or Met-tRNA Leu , which are easily obtained by mis-charging tRNA Leu with commercially available radioactive Ile or Met using a LeuRS mutant without post-transfer editing capability. Because we focused on the Nva-editing properties of CaLeuRS and Nva labeled with radioactive isotope was not commercially available, the 3 0 end of CatRNA Leu was first labeled with [a-32 P]ATP by E. coli CCase, and then Nva-[ 32 P]CatRNA Leu was prepared by editing-deficient   Discrimination factor corresponds to the loss of catalytic efficiency relative to Leu.
ScLeuRS-D419A (13,18). Hydrolytic analysis clearly showed that CaLeuRS edited Nva-[ 32 P]CatRNA Leu when compared with the control experiment conducted in the absence of the enzyme ( Figure 3A and B). To confirm the post-transfer editing reaction catalyzed by CaLeuRS, we mutated the conserved and post-transfer editing-essential Asp 422 to generate CaLeuRS-D422A. Asp 422 corresponds to Asp 373 , Asp 419 , Asp 444 and Asp 399 of AaLeuRS, ScLeuRS, GlLeuRS and hcLeuRS, respectively, which are crucial to post-transfer editing by these LeuRSs (9,10,18,19,27). Indeed, CaLeuRS-D422A did not hydrolyze Nva-[ 32 P]CatRNA Leu and was deficient in posttransfer editing, indicating that this mutation inactivated the CP1 domain of CaLeuRS ( Figure 3A and B). Further mis-aminoacylation of [ 32 P]CatRNA Leu with Nva showed that a significant amount of Nva-[ 32 P]CatRNA Leu was formed by CaLeuRS-D422A; however, a negligible amount of mis-charged CatRNA Leu was formed by CaLeuRS ( Figure 3C and D). Overall, these data showed that CaLeuRS harbored an obvious and efficient capability for post-transfer editing of Nva-[ 32 P]CatRNA Leu , the loss of which caused accumulation of mis-charged tRNA Leu . No further accumulation of AMP after the addition of CatRNA Leu in the AMP formation assay (Table 4) suggested that post-transfer editing of mis-charged CatRNA Leu contributed little to the total editing. To further explore the absence of tRNA-dependent pre-transfer editing of Nva by CaLeuRS, we tested the AMP formation of CaLeuRS-D422A in the presence of Nva with or without over-produced hctRNA Leu . The data showed that over-produced hctRNA Leu with modified bases did not stimulate further AMP production after abolishing post-transfer editing (k obs 0.22 ± 0.06 versus 0.25 ± 0.04 s À1 ), confirming the lack of tRNA-dependent pre-transfer editing of Nva by CaLeuRS (Table 4).
CaLeuRS inhibited synthesis of Nva-tRNA Ser by non tRNA species-specific post-transfer editing The ability of CaLeuRS to efficiently mis-activate noncognate Nva and recognize non-cognate CatRNA Ser raises the interesting question of how to prevent the formation of Nva-CatRNA Ser . To test for the presence of post-transfer editing activity that hydrolyzes potentially synthesized Nva-CatRNA Ser , mis-aminoacylation of [ 32 P]CatRNA Ser with Nva by CaLeuRS was compared with that of the post-transfer editing-deficient CaLeuRS-D422A. The data clearly showed that, mutation of Asp 422 resulted in significant synthesis of Nva-[ 32 P]CatRNA Ser by the mutant, in contrast to wild-type enzyme, which generated negligible amounts of Nva-[ 32 P]CatRNA Ser , indicating that CaLeuRS used post-transfer editing to prevent Nva-CatRNA Ser synthesis ( Figure 4A and B). We then prepared Nva-CatRNA Ser for use in hydrolysis assays to more directly monitor the post-transfer editing activity. Obvious hydrolysis of Nva-[ 32 P]CatRNA Ser was mediated by CaLeuRS but not CaLeuRS-D422A ( Figure 4C and D). Above all, these data showed that the post-transfer editing by CaLeuRS was not only CatRNA Leu specific but also efficient for CatRNA Ser to inhibit synthesis of both Nva-CatRNA Leu and Nva-CatRNA Ser .
CaLeuRS possessed weak tRNA-dependent pre-transfer editing capacity for ABA ABA was selected to test whether CaLeuRS possessed any tRNA-dependent pre-transfer editing of other noncognate amino acids because it was obviously activated by CaLeuRS. CatRNA Leu transcript, transcribed or over-produced EctRNA Leu and hctRNA Leu were used to trigger editing of ABA by CaLeuRS ( Table 5). The data showed that over-produced hctRNA Leu obviously stimulated editing by increasing the k obs 5-fold [(23.19 ± 3.62) Â 10 À3 s À1 ] compared with that in the absence of tRNA [(4.69 ± 0.51) Â 10 À3 s À1 ]. Overexpressed EctRNA Leu led only to an $3-fold increase in k obs [(14.58 ± 2.14) Â 10 À3 s À1 ]. However, CatRNA Leu , EctRNA Leu and hctRNA Leu transcripts had little effect on the rate of ABA-editing (Table 5). These data implied that editing of ABA was tRNA modification-dependent.
As over-produced hctRNA Leu was the most efficient stimulator of ABA-editing, we measured AMP formation by the editing-deficient CaLeuRS-D422A mutant in the presence of ABA with over-produced hctRNA Leu . Mutation of Asp 422 , which abolished post-transfer editing, apparently decreased the rate of AMP formation with a k obs of (12.16 ± 1.98) Â 10 À3 s À1 . Therefore, with over-produced hctRNA Leu , post-transfer editing of ABA by CaLeuRS accounted for 47.6% of the total editing [(23.19 À 12.16)/23.19], whereas tRNA-independent and   (Figure 3C and D). These data implied that editing of ABA by CaLeuRS was not sufficient to prevent the synthesis of ABA-[ 32 P]CatRNA Leu in the presence of saturating ABA concentrations. This paradox between ABA mis-aminoacylation and charging accuracy may be solved by fine discrimination against ABA at the aminoacylation active site (Table 3).
These results revealed that CaLeuRS exhibits a weak level of tRNA-dependent pre-transfer editing activity for ABA. In addition, the total ABA-editing capacity is not sufficient to avoid the formation of mis-charged tRNA Leu , which is different from Nva-editing capacity.
ScLeuRS, like CaLeuRS, also exhibited little tRNA-dependent pre-transfer editing capacity for Nva The natural deficiency in tRNA-dependent pre-transfer editing of Nva by CaLeuRS prompted us to investigate whether it is a common characteristic of other yeast LeuRS. Therefore, we assayed the Nva-included AMP formation catalyzed by ScLeuRS in the absence or presence various tRNA Leu s. The data showed that overproduced hctRNA Leu obviously stimulated editing (k obs of 0.64 ± 0.04 s À1 ) compared with that observed in the absence of tRNA (0.10 ± 0.02 s À1 ). However, transcribed CatRNA Leu , EctRNA Leu , hctRNA Leu and over-expressed EctRNA Leu failed to trigger further editing by ScLeuRS ( Table 6). The formation of AMP stimulated by tRNA Leu should be derived from tRNA-dependent pre-transfer editing and/or post-transfer editing. To distinguish between these two pathways, the conserved Asp 419 was mutated to generate ScLeuRS-D419A, which has been shown to be defective in post-transfer editing and is homologous with Asp 422 of CaLeuRS (13,18). Assay of the Nva-included AMP formation by ScLeuRS-D419A showed that the k obs with over-produced hctRNA Leu was only slightly greater (0.100 ± 0.010 s À1 ) than that observed in the absence of tRNA (0.094 ± 0.001 s À1 ), indicating that inactivation of post-transfer editing totally abolished the triggering of AMP formation by tRNA, and that the increase in AMP production by over-produced hctRNA Leu was due to post-transfer editing. Therefore, like CaLeuRS, ScLeuRS did not significantly catalyze tRNA-dependent pre-transfer editing for Nva.
Eukaryotic, but not archaeal or bacterial LeuRSs, recognized CatRNA Ser It is interesting that LeuRSs from some Candida species recognize a uniquely evolved tRNA Ser to introduce ambiguity at CUG codons (28). Unfortunately, no elements of CaLeuRS have been identified to account for the interaction with anti-codon stem and/or loop of CatRNA Ser . To test whether other eukaryotic, archaeal, bacterial or mitochondrial LeuRSs could potentially recognize CatRNA Ser , we performed aminoacylation of [ 32 P]CatRNA Ser with Leu by CamtLeuRS, EcLeuRS, ScLeuRS, hcLeuRS and PhLeuRS. The data showed that only eukaryotic LeuRSs (including CaLeuRS, ScLeuRS and hcLeuRS) could aminoacylate CatRNA Ser with Leu; however, other LeuRSs, including bacterial EcLeuRS, mitochondrial CamtLeuRS and archaeal PhLeuRS, failed to charge it ( Figure 5). Strikingly, under the same conditions, ScLeuRS and hcLeuRS mediated more efficient aminoacylation of CatRNA Ser .

Insertion of Ser or Leu at CUG codons might not be incidental
In C. albicans and most other CUG clade species, a mutant tRNA Ser (CAG) has evolved to decode the Leu CUG codon both as Ser and Leu (28,29). This peculiarity is derived from its combined tRNA Leu and tRNA Ser identity elements (38). This tRNA is mainly aminoacylated by SerRS and charged by LeuRS to a small extent (29). Both biochemical and structural data have revealed that ambiguity at the single CUG codon of SerRS induces local structural rearrangement, leading to a slightly increased activity (27%) of CaSerRS-Leu 197 compared with the wild-type CaSerRS-Ser 197 (30). Furthermore, genetic studies showed that increased Leu incorporation across all the CUG codons of C. albicans had no visible effect on the growth phenotype but had an impressive impact on cell morphology (39). Therefore, it was proposed that CUG decoding ambiguity has a potential regulatory role in protein structure and/or function (30). CaLeuRS is another crucial player in this genetic code alteration and also contains only one CUG codon at position 919. This site is located at the C-terminal domain of LeuRS, which has been shown to be responsible for binding the variable loop of tRNA Leu and involved in the aminoacylation activity; however, this domain is not strictly conserved among archaeal/eukaryotic LeuRSs ( Figure 1A and B). Here, we revealed that both CaLeuRS-Leu 919 and CaLeuRS-Ser 919 catalyzed Leu activation and aminoacylation, but the former was more active (30%) than the latter, indicating that the conformation of the 919-containing a29 helix might be finely controlled by the introduction of either Ser or Leu. This phenomenon was also observed in another crucial player in the CUG decoding alteration pathway, CaSerRS (30). We suggested that insertion of either Ser or Leu at the CUG codon was not random and incidental. The relative amounts of CaLeuRS-Ser 919 /CaLeuRS-Leu 919 should be strictly regulated by an unidentified but precise molecular mechanism in vivo. Whether the fine balance of CaLeuRS-Ser 919 /CaLeuRS-Leu 919 is critical for decoding other Leu codons and correlates with the ratio of CaSerRS-Ser 197 /CaSerRS-Leu 197 requires further investigation.

Yeast LeuRS exhibited a relaxed tRNA recognition capacity
In tRNA aminoacylation, species-specific charging, where a tRNA from one taxonomic domain is not aminoacylated by an aaRS from another, is widespread. This may be as a result of the co-evolution of synthetase/tRNA pairs by the addition of species-specific elements. For instance, human tyrosyl-tRNA synthetase does not recognize bacterial tRNA Tyr , and E. coli tyrosyl-tRNA synthetase is unable to charge eukaryotic tRNA Tyr (40), and there is no cross-recognition of E. coli and human tRNA Gly by the respective glycyl-tRNA synthetases (41). Similarly, E. coli isoleucyl-tRNA synthetase is unable to charge eukaryotic tRNA Ile (42). Yeast ArgRS charges E. coli tRNA Arg ; however, E. coli ArgRS acylates only its cognate E. coli tRNA (43). Human cysteinyl-tRNA synthetase charges bacterial tRNA Cys , but E. coli cysteinyl-tRNA synthetase is non-functional in aminoacylating human tRNA Tyr (44).
Here, we showed that both hcLeuRS and EcLeuRS failed to aminoacylate CatRNA Leu ; however, both CaLeuRS and ScLeuRS readily aminoacylated bacterial, yeast and even human tRNA Leu s. These results showed that yeast LeuRSs exhibit a more relaxed recognition specificity compared with other LeuRSs. Indeed, CatRNA Ser itself harbors only tRNA Leu recognition elements in the anticodon loop with other parts being crucial for SerRS recognition. In addition, G33 is also unfavorable for LeuRS; even in this adverse state, CaLeuRS aminoacylates it in vivo (29). Comparison between transcribed and overexpressed tRNA Leu s showed that base modification of tRNA Leu plays an important role in both binding and catalysis. CaLeuRS was deficient in tRNA-dependent pre-transfer editing but exhibited efficient post-transfer editing for Nva Nva is inherently mis-activated by various LeuRSs to a significant level that requires editing for translational accuracy (9,(18)(19)(20). With an elevated ratio of Nva to Leu, Nva can escape the safeguarding of EcLeuRS and replace Leu in proteins rich in Leu codons, indicating that Nva-tRNA Leu can escape further checking by the ribosome and pose a direct threat to the accuracy of newly synthesized proteins (24). From the viewpoint of editing, some LeuRSs with degenerated (e.g. hmtLeuRS) or deleted CP1 (e.g. MmLeuRS) domains are exceptional examples, which use alternative pathways (efficient discrimination at the active site) for translational quality control (hmtLeuRS) (26) or do not edit misaminoacylation product to produce proteome ambiguity (MmLeuRS) (20). However, all LeuRSs with functional CP1 domains studied so far display tRNA-independent, tRNA-dependent pre-transfer editing and post-transfer editing for Nva. Through inactivation of CP1 or utilization of LeuRS inhibitors, post-transfer editing has been successfully isolated (9,10,18,19,27). Similarly, by mutating a crucial Tyr residue to Asp in EcLeuRS and AaLeuRS, both tRNA-dependent pre-transfer and posttransfer editing are inactivated (10,33). Interestingly, this study identified that CaLeuRS itself is naturally defective in tRNA-dependent pre-transfer editing for Nva. With CatRNA Leu , no tRNA-dependent pre-transfer editing was identified. In contrast, weak tRNA-dependent pretransfer editing for ABA in the presence of specific tRNA Leu was observed, despite the indication that ABAediting might not be necessary in vivo based on fine discrimination at the active site. Similarly, ScLeuRS did not mediate tRNA-dependent pre-transfer editing. These results indicate that the capacity for tRNA-dependent pre-transfer editing for Nva has been lost by CaLeuRS (also ScLeuRS), and that ABA is also rarely induced. The reason for this deficiency in tRNA-dependent pretransfer editing and the pathway by which this deficiency was introduced remains to be elucidated. Post-transfer editing contributed little or negligibly to the total Nva-editing since addition of any tRNAs in the AMP formation assays did not significantly induce additional AMP. Thus, the produced AMP was mainly derived from tRNA-independent pre-transfer editing. However, the energy-saving post-transfer editing pathway critically controls the accuracy of aminoacylation. Mutation at the conserved Asp 422 of CaLeuRS led to a LeuRS with abolished post-transfer editing capacity; consequently, Nva-tRNA Leu was synthesized. Similarly, ScLeuRS did not synthesize Ile-tRNA Leu ; however, ScLeuRS-D419A readily generated significant amounts of Ile-tRNA Leu (13,18). Using these unique CaLeuRS and ScLeuRS models devoid of tRNA-dependent pre-transfer editing capacity, we concluded that the post-transfer editing pathway is the most economic but efficient editing mechanism for LeuRS. Consistent with other LeuRS models and even other aaRS systems, once post-transfer editing is impaired, the mis-charged tRNA is unavoidably accumulated (9)(10)(11)18,19,27,45,46).
Our results also revealed that post-transfer editing by CaLeuRS is not tRNA-species specific, as Nva-CatRNA Ser was also a substrate. Indeed, based on the poor discrimination against Nva in the active site, Nva-tRNA Ser is possibly synthesized but should be removed. Otherwise, the CUG codon might be decoded as Ser, Leu and Nva in vivo. It has been proposed that the acceptor end of the tRNA switches from a hairpin conformation to a helical conformation for editing by class I aaRSs, whereas the reverse change in conformation occurs at the acceptor end of the tRNA for editing by class II aaRSs (47). Notably, CatRNA Ser corresponds to a class II SerRS; however, results here showed that CatRNA Ser could switch from a hairpin to a helical conformation for editing by a class I LeuRS.

Eukaryotic LeuRSs recognized CatRNA Ser
In addition, we revealed that other eukaryotic LeuRSs could efficiently aminoacylate CatRNA Ser , implying that the introduction or evolution of this type of tRNA in other eukaryotic systems would reprogram or discombobulate the genetic code, leading to proteome chaos. In other words, a specific eukaryotic genetic code could be artificially reprogramed by expression of this tRNA Ser . Indeed, CatRNA Ser has been shown to be efficiently produced, processed and aminoacylated in S. cerevisiae, with its expression triggering a stress response, blocking mating and re-defining the gene expression model of S. cerevisiae (48). Notably, archaeal LeuRS is in the same group with eukaryotic LeuRS according to primary or higher structure (14) and only differs at the C-terminal tRNA binding domain, indicating that this domain in eukaryotic LeuRSs is a key element for recognition of CatRNA Ser . This observation is consistent with the structural and biochemical results showing that the C-terminal domain of archaeal LeuRS specifically contacts the variable loop but not the anti-codon loop of archaeal tRNA Leu (16,49). However, the anti-codon loop, which is a key recognition element in both CatRNA Ser (28,29,38) and yeast tRNA Leu (49), is likely to be bound by the C-terminal domain of eukaryotic LeuRS. This proposal requires confirmation from eukaryotic LeuRS-tRNA Leu /tRNA Ser structures.

Concluding remarks
Translational machinery of human pathogen C. albicans is of particular interest because its CUG codon in the genome is decoded as both Ser and Leu by a unique CatRNA Ser , leading to proteome ambiguity (28,29). One of the most crucial components in this decoding process is CaLeuRS, which catalyzes two successive steps, aminoacylation and editing reactions, which together are essential for ensuring high specificity of tRNA charging. In aminoacylation, we showed that Leu isoform was more active than Ser isoform of CaLeuRS in charging CatRNA Leu , implying the existence of an in vivo mechanism regulated by balance of CaLeuRS-Leu 919 and CaLeuRS-Ser 919 . In addition, as a yeast LeuRS model, CaLeuRS recognized tRNA Leu s from bacteria, yeast and higher eukaryote. In editing, CaLeuRS efficiently misactivated non-cognate Nva. One of the most interesting findings was that CaLeuRS lacked tRNA-dependent pretransfer editing for Nva, which has been well investigated for bacterial and human LeuRSs (10,19). Instead, CaLeuRS prevented insertion of Nva into proteome mainly via post-transfer editing no matter whether Nva has been loaded onto either CatRNA Leu or CatRNA Ser . Collectively, we further improved our understanding of mechanism and significance of genetic code ambiguity in C. albicans and revealed interesting properties of both aminoacylation and editing reactions by CaLeuRS. Furthermore, the capacity of eukaryotic LeuRSs at aminoacylating CatRNA Ser suggests the possibility of reconstructing proteome of other eukaryotes by simply introducing this unique tRNA Ser .