Modulating the Structure and Function of an Aminoacyl-tRNA Synthetase Cofactor by Biotinylation*

Arc1p is a yeast-specific tRNA-binding protein that forms a ternary complex with glutamyl-tRNA synthetase (GluRSc) and methionyl-tRNA synthetase (MetRS) in the cytoplasm to regulate their catalytic activities and subcellular distributions. Despite Arc1p not being involved in any known biotin-dependent reaction, it is a natural target of biotin modification. Results presented herein show that biotin modification had no obvious effect on the growth-supporting activity, subcellular distribution, tRNA binding, or interactions of Arc1p with GluRSc and MetRS. Nevertheless, biotinylation of Arc1p was temperature dependent; raising the growth temperature from 30 to 37 °C drastically reduced its biotinylation level. As a result, Arc1p purified from a yeast culture that had been grown overnight at 37 °C was essentially biotin free. Non-biotinylated Arc1p was more heat stable, more flexible in structure, and more effective than its biotinylated counterpart in promoting glutamylation activity of the otherwise inactive GluRSc at 37 °C in vitro. Our study suggests that the structure and function of Arc1p can be modulated via biotinylation in response to temperature changes.

Aminoacyl-tRNA synthetases (aaRSs) 2 belong to an ancient family of enzymes. These enzymes are responsible for the attachment of an amino acid to its cognate tRNA. The resultant aminoacyl-tRNAs are then sent to ribosomes for mRNA decoding (1)(2)(3). Eukaryotes normally contain two distinct sets of aaRSs: one functioning in the cytoplasm and the other in mitochondria. In most cases, cytoplasmic and mitochondrial forms of an aaRS are encoded by two separate nuclear genes. In a few cases, both forms of an aaRS are encoded by a single nuclear gene through alternative transcription and translation (4 -8). Although the gene encoding yeast cytoplasmic glutamyl-tRNA synthetase (GluRS c ) is also dual functional, it specifies only a single protein form. GluRS c attaches glutamate to cytoplasmic tRNA Glu in the cytoplasm, but once imported into mitochondria, it attaches glutamate to mitochondrial tRNA Gln , forming the mischarged Glu-tRNA Gln (9).
The majority of yeast cytoplasmic aaRSs possess an N-or C-terminal appended domain, which is absent from their bacterial homologues (10). Many of these domains act in cis as an auxiliary tRNA-binding domain, examples of which include glutaminyl- (11), arginyl- (12), and valyl-tRNA synthetases (13). These domains enhance catalytic activities of the associated enzymes (14). In contrast, some appended domains are involved in specific protein-protein interactions, examples of include GluRS c , methionyl-(MetRS), and seryl-tRNA synthetases. GluRS c and MetRS form a ternary complex with the tRNA-binding protein Arc1p (15), whereas seryl-tRNA synthetase forms a binary complex with the peroxisome biogenesisrelated factor Pex21p (16). These interactions also enhance aminoacylation activities of the partner enzymes. In addition to acting as an aaRS cofactor, Arc1p also acts as a sorting platform to regulate subcellular distributions of GluRS c and MetRS upon a diauxic transition (17). Following dissociation from Arc1p, GluRS c and MetRS are, respectively, targeted to mitochondria and nuclei to coordinate OXPHOS gene expression.
Biotin, also known as vitamin H or B 7 , is an essential coenzyme synthesized by plants and many prokaryotes. It is required by all life forms and is covalently linked to a distinct set of carboxylases involved in fatty acid synthesis, branched-chain amino acid catabolism, and gluconeogenesis (18). These biotincontaining enzymes normally catalyze reactions involving the transfer of carbon dioxide (19). Biotin is covalently attached to the ⑀-amino group of specific lysine residues in carboxylases by the enzyme biotin protein ligase, for example, BirA in Escherichia coli, Bpl1p in yeast, and holocarboxylase synthetase in mammals (20). The biotinylated lysine residue is almost invariably positioned in the consensus sequence AMKM within carboxylases. As a result, biotin modification occurs across widely divergent species (21). Biotin modification has received copious attention recently due to its wide application in biotechnology. Biotin binds strongly to avidin and streptavidin (with a K d value of ϳ10 Ϫ15 M), and so this is one of the strongest ligand/protein interactions identified so far. Such a feature is often exploited in studies involving protein purification, immobilization, and localization.
Five biotin-containing carboxylases have been identified in the yeast Saccharomyces cerevisiae, including cytosolic and mitochondrial forms of acetyl-CoA carboxylase, cytosolic and mitochondrial forms of pyruvate carboxylase, and urea amidol-yase (22,23). Biotin is involved in the transfer of carbon dioxide and is thus essential for the activities of these enzymes. Despite the fact that Arc1p is not involved in any known carboxylation/ decarboxylation reaction and lacks the AMKM consensus sequence of biotin-binding domains, it is biotinylated in vivo (24). The biological significance of this modification has remained unclear. Our study shows that biotinylation of Arc1p is temperature sensitive, and increasing the growth temperature from 30 to 37°C significantly reduced its biotinylation level. Biotin-free Arc1p was more heat stable, more dynamic, and more effective than its biotinylated counterpart in activating GluRS c at high temperatures. This study highlights an unconventional role of biotinylation in modulating the structure and function of a non-carboxylase protein.

Results
Biotinylation of Yeast Arc1p-The N domain of Arc1p interacts with the N domains of GluRS c and MetRS, whereas its M plus C domains form a nonspecific tRNA-binding domain (Fig.  1A). Despite Arc1p lacking the AMKM consensus sequence of biotin-binding domains, biotinylation occurs at lysine 86 (Lys-86) within its N domain. Sequence alignment shows that sequences immediately surrounding Lys-86 considerably diverged among the Arc1p homologues of various yeast species. Nevertheless, a tetrapeptide, SSKD, was somewhat conserved among these homologues (Fig. 1B). To explore whether these Arc1p homologues are also biotinylated in vivo, Western blotting analysis using HRP-streptavidin as a probe was carried out. Note that all yeast Arc1p homologues tested here possessed a similar molecular mass of ϳ46 kDa. As shown in Fig. 1C, a protein band with a molecular mass of ϳ46 kDa was identified in S. cerevisiae, but not in its arc1 Ϫ allele, suggesting that the streptavidin-reactive protein is Arc1p. No protein bands with a similar size were identified in other yeast species tested, except for Vanderwaltozyma polyspora. Conceivably, biotinylation is not a common feature of Arc1p homologues.
Although Arc1p homologues in Candida albicans, Candida tropicalis, and Lodderomyces elongisporus also possessed a sequence similar to SSKD (TSKD, TSKD, and SSKE, respectively) at the corresponding position, they were not biotinylated in vivo. Conceivably, sequences surrounding SSKD also contribute to this specific modification. It is noteworthy that in addition to carboxylases, at least one protein (ϳ42 kDa) in L. elongisporus and two proteins (ϳ44 and ϳ60 kDa, respectively) in Pichia guilliermondii were also distinctly biotinylated in vivo (Fig. 1C). The identities of these streptavidin-reactive proteins remain to be resolved.
Interactions of Arc1p with GluRS c and MetRS-To investigate whether biotinylation is required for the interaction of Arc1p with GluRS c , we exploited a two-hybrid assay. The N domains of Arc1p and GluRS c were, respectively, cloned into a DNA-binding domain vector (pGBKT7) and a transcription activation domain vector (pGADT7), and the resultant constructs were co-transformed into a yeast reporter strain. As shown in Fig. 2A, the N domain of Arc1p strongly interacted with the N domain of GluRS c , which in turn turned on the reporter gene HIS3. The co-transformants robustly grew on selective medium lacking histidine at 30°C. The K86R mutation did not interrupt the interaction, suggesting that biotinylation is not required for the interaction of Arc1p with GluRS c . A similar scenario occurred at 20 and 37°C, but with poorer growth at 37°C for both the WT and K86R mutant. The mutant GluRS c T125R/R164A served here as a control, as mutations in these two amino acid residues had previously been shown to interrupt the interaction of GluRS c with Arc1p (25).
To provide additional evidence, we performed an in vitro Ni-NTA pulldown assay, where full-length GluRS c -His 6 and MetRS-His 6 were used as bait with GST, Arc1p WT -GST, or Arc1p K86R -GST as prey. (Note that the GST pulldown assay was not doable here because GluRS c itself contains a GST-like domain.) As expected, the K86R mutation impaired the biotinylation of Arc1p-GST (Fig. 2B, left panel). The Ni-NTA pulldown assay showed that GluRS c -His 6 and MetRS-His 6 , alone or in combination, effectively pulled down Arc1p K86R -GST as well as Arc1p WT -GST, but not GST (Fig. 2B, right panel). This result lends further support to the proposal that biotinylation is dispensable for the interaction of Arc1p with GluRS c and MetRS.
Effect of Biotinylation on the Rescue Activity, Localization, and Half-life in Vivo of Arc1p-Deletion of ARC1 is not lethal but reduces cellular growth, especially at low temperatures (15). To investigate whether biotinylation affects the ability of Arc1p to act as an aaRS cofactor in vivo, genes encoding the WT and K86R mutant of Arc1p were, respectively, cloned in pRS315, and the resultant constructs were tested in an arc1 Ϫ yeast strain at various temperatures. As shown in Fig. 3A, the knock-out strain showed a feeble growth phenotype on SD/-Leu plates at 20°C, but retained a near-normal growth phenotype at 30 and 37°C even in the absence of a functional ARC1 gene. This coldsensitive phenotype was effectively rescued by the WT and K86R mutant, suggesting that biotinylation is not required for the rescue activity of Arc1p. This finding is essentially consistent with an earlier report, in which the K86R mutant was shown to possess growth-supporting activity comparable with that of the WT (24).
We next tested whether biotinylation alters the subcellular localization of Arc1p. A GFP assay was carried out. As shown in Fig. 3B, both the WT and mutated Arc1p proteins were predominantly and evenly distributed in the cytoplasm at all temperatures tested (20, 30, and 37°C). However, due to resolution constraints of the imaging technology used, we could not rule out the possibility that a very minor portion of GFP fusion proteins was targeted to other cellular compartments.
To compare the relative protein stabilities (or half-lives) of biotinylated and non-biotinylated Arc1p proteins in vivo, a CHX-chase assay was carried out. Genes encoding the WT and mutated Arc1p-His 6 proteins were, respectively, cloned into the inducible vector pGAL1. The resultant constructs were transformed into a S. cerevisiae strain, and cultures of the resultant transformants were induced with galactose for 4 h at 30°C, followed by the addition of CHX to stop protein synthesis. CHX-treated cells were divided into groups, either maintained at 30°C or switched to 20 or 37°C, and harvested at various intervals (0ϳ16 h) post-induction. Protein extracts (40 g) were prepared for Western blotting analyses using an HRPconjugated anti-His 6 tag antibody. Fig. 3C shows that both the WT and K86R mutant were fairly stable at all temperatures tested; Ͻ15% of Arc1p proteins were degraded in the time period tested. Thus, biotin-free Arc1p retained a half-life comparable with that of biotinylated Arc1p in vivo.
Effect of Biotinylation on the tRNA Binding and Thermal Stability in Vitro of Arc1p-To explore whether biotinylation affects the tRNA-binding affinity of Arc1p in vitro, the WT and mutated Arc1p-His 6 proteins were purified through Ni-NTA affinity chromatography, and their tRNA-binding affinities were determined using polyacrylamide affinity coelectrophoresis (13). An aliquot of 32 P-labeled in vitro-transcribed yeast tRNA n Glu (nuclear-encoded tRNA Glu ) (ϳ1 nM) was loaded into each well of a 5% polyacrylamide gel, in which 2-fold dilutions of the purified Arc1p protein had been mixed into the gel, forming a protein gradient of 0.015-4 M. The far left lane contained no Arc1p protein and served as a control. Free tRNA moved faster than protein-bound tRNA in the gel. As shown in To explore whether the K86R mutation affects the thermal stability (secondary structure content) of Arc1p, purified Arc1p variants were subjected to CD spectroscopy. The secondary structure of a protein can be determined by CD spectroscopy in the far-UV region. An ␣-helix has negative bands at 222 and 208 nm, whereas a ␤-sheet has a negative band at 218 nm. Fig. 4B shows that Arc1p WT possessed a considerable level of secondary structures (with high molar ellipticity () values at 222 and 208 nm) at both 20 and 30°C, but it lost a significant portion of its secondary structure when the temperature reached 37°C. In contrast, Arc1p K86R was relatively unstable (with less secondary structure content), even at temperatures below 30°C (Fig. 4C). It was noteworthy that the 30°C CD spectrum of Acr1p K86R closely resembled the 37°C CD spectrum of Arc1p WT (Fig.  4B,C). No apparent aggregation was observed for either protein under the conditions used. Thus, the K86R mutation affects the biotinylation and thermal stability of Arc1p.
Effect of Biotinylation on the Cofactor Activity of Arc1p-To test whether biotinylation affects the cofactor activity of Arc1p, we carried out aminoacylation assays with GluRS c /Arc1p in a ratio of 1:1 (26). As shown in Fig. 5A, Arc1p K86R was as effective as Arc1p WT and promoted glutamylation activity of GluRS c (ϳ2-fold increase) at 20°C. A similar scenario occurred at 30°C (Fig. 5B). However, it should be noted that GluRS c alone was ϳ2-fold more active at 30°C than at 20°C. Most strikingly, the GluRS c alone possessed almost no aminoacylation activity at 37°C, suggesting that it is a heat-sensitive enzyme in vitro (Fig.  5C). Despite the fact that the addition of Arc1p WT or Arc1p K86R to the reaction mixture also promoted the glutamylation activity of the enzyme to a certain extent at 37°C, the overall activity was still insignificant (Fig. 5C).
Purification and Characterization of a Biotin-free Arc1p Variant-As the K86R mutation impairs the biotinylation and structural stability of Arc1p, Arc1p K86R might not truthfully represent a native biotin-free Arc1p. To find a better representative, we purified a WT Arc1p protein (designated herein as Arc1p BϪ ) from a yeast transformant that had been grown in a yeast-defined medium deficient in biotin (containing 0.2 g/liter of biotin). For comparison, we also purified a WT Arc1p protein (designated herein as Arc1p Bϩ ) from a yeast transformant that had been grown in a yeast-defined medium rich in biotin (containing 200 g/liter of biotin). (Note that normal yeast-defined medium contains ϳ2 g/liter of biotin.) To determine relative biotinylation levels of these Arc1p proteins, they were subjected to a streptavidin-based gel mobility shift assay.
As shown in Fig. 6A, Arc1p Bϩ possessed a biotinylation level (ϳ15%) comparable with that of Arc1p WT (a WT Arc1p protein purified from a yeast transformant that had been grown in normal yeast-defined medium), suggesting that the biotinylation level of Arc1p does not increase with higher levels of biotin in the growth medium. On the other hand, Arc1p BϪ possessed essentially no biotin modification (Fig. 6B). Hence, Arc1p BϪ can be regarded as a native biotin-free Arc1p. We next checked efficiencies of Arc1p WT , Arc1p Bϩ , and Arc1p BϪ as aaRS cofactors. As shown in Fig. 6C, Arc1p WT , Arc1p Bϩ , and Arc1p BϪ were almost equally effective in promoting GluRS c activity at 30°C (ϳ2-fold increase). However, Arc1p BϪ was much more effective than Arc1p WT and Arc1p Bϩ in promoting glutamylation activity of the otherwise inactive GluRS c at 37°C (Fig. 6D). Arc1p Bϩ and Arc1p WT only slightly promoted GluRS c activity at this temperature.
To compare the thermal stabilities of these two Arc1p variants, purified Arc1p proteins were subjected to CD spectroscopy. As shown in Fig. 7A, Arc1p Bϩ retained a CD spectrum that was almost indistinguishable from that of Arc1p WT (compare Figs. 4B and 7A). Arc1p Bϩ possessed ϳ30% of ␣ helixes and ϳ16% of ␤ sheets at 20°C. Increasing the test temperature from 20 to 30°C did not substantially alter its CD spectrum. However, like Arc1p WT , Arc1p Bϩ lost a significant portion of its secondary structure when the temperature reached 37°C. In contrast to the thermal instability of Arc1p Bϩ , Arc1p BϪ retained much of its secondary structure even when the temperature reached 37°C (Fig. 7B). Clearly, Arc1p BϪ is more heat stable than Arc1p Bϩ .
To obtain a more comprehensive picture, melting curves of Arc1p Bϩ and Arc1p BϪ were subsequently determined via CD spectroscopy at 222 nm and 10 -60°C. As shown in Fig. 7C, Arc1p Bϩ possessed a molar ellipticity value close to that of Arc1p BϪ at temperatures below 30°C. However, Arc1p Bϩ swiftly lost its secondary structure (namely the ␣ helix) at  temperatures above 30°C and had a melting temperature of ϳ37°C. In contrast, Arc1p BϪ possessed a relatively gentle melting curve. Arc1p BϪ slowly lost its secondary structure at temperatures above 30°C and had a melting temperature of ϳ47°C. This result suggests that biotinylation significantly reduced the thermal stability of Arc1p.
To examine whether biotinylation alters the structural flexibility of Arc1p, limited proteolysis with Arc1p/trypsin in a ratio of 1,000:1 was carried out at 30 and 37°C. This technique is often used to probe the structure and dynamics of proteins. Exposed regions such as loops and other flexible regions are more susceptible to the prolific protease. As shown in Fig. 7D, Arc1p Bϩ was much more resistant to the protease than was Arc1p BϪ at both temperatures. More than half of the Arc1p BϪ protein was hydrolyzed after 8 min of protease treatment at both temperatures, and no intact protein remained after 16 min of treatment. In contrast, almost no Arc1p Bϩ was hydrolyzed throughout the time period tested at either temperature. Thus, Arc1p BϪ was more flexible in structure than was Arc1p Bϩ . The higher structural flexibility, together with a higher thermal stability, might account for the higher cofactor activity of Arc1p BϪ at high temperatures (Fig. 6).
Temperature-dependent Biotinylation of Arc1p-The question arose as to whether temperature affects the biotinylation level of Arc1p, and to what extent. Pursuant to this objective, a yeast transformant harboring a plasmid-borne ARC1 gene (driven by its native promoter) was grown to an A 600 of 0.6. The culture was divided into groups and either maintained at 30°C or switched to 20 or 37°C for 3 or 12 h, and relative biotinylation levels of Arc1p were analyzed by Western blotting using HRP-streptavidin as a probe. As shown in Fig. 8A, biotinylation of Arc1p was sensitive to high temperatures. Raising the growth temperature from 30 to 37°C severely reduced the biotinylation level of Arc1p; almost no biotinylation was detected for Arc1p 12 h after switching the temperature to 37°C (Fig. 8A). On the other hand, no obvious difference in the biotinylation level of Arc1p was observed between 20 and 30°C. In line with this observation, Arc1p purified from a yeast transformant (carrying a plasmid-borne ARC1 gene driven by an ADH promoter) that had been grown overnight (24 h) at 20 or 30°C possessed ϳ15% biotinylation (Fig. 8B). In contrast, Arc1p purified from the same transformant that had been grown overnight at 37°C was essentially biotin free.
We next tested whether Arc1p 37 is more efficient than Arc1p 20 in stimulating the glutamylation activity of GluRS c in vitro. Arc1p 37 and Arc1p 20 , respectively, denote Arc1p variants purified from yeast transformants grown overnight (24 h) at 37 and 20°C. As expected, Arc1p 37 was as efficient as Arc1p 20 at 30°C, but was much more efficient than Arc1p 20 at 37°C in stimulating the glutamylation activity of GluRS c .

Discussion
Despite the fact that the occurrence of biotin-dependent enzymes is ubiquitous in nature, biotinylation is a relatively rare modification in cells (21). Both biotin-binding domains and biotin protein ligases are highly conserved throughout biology. Hence, biotin protein ligases can attach biotin to biotin acceptor proteins across widely divergent species. For example, the biotin protein ligases from Homo sapiens, S. cerevisiae, and Arabidopsis thaliana can effectively complement an E. coli birA mutant (27)(28)(29). Although Arc1p lacks a canonical biotinylation site and is not involved in any known biotin-dependent reaction, it is modified in a site-specific manner by the only biotin protein ligase in yeast, i.e. Bpl1p (Fig. 1) (24). Thus, SSKD may represent a secondary biotinylation site for yeast Bpl1p. This might explain why only ϳ15% of Arc1p was biotinylated even under biotin-rich conditions (Fig. 6). Additionally, this might explain why the E. coli BirA enzyme cannot modify yeast Arc1p (24). Because only two of nine yeast species tested possessed a biotinylable Arc1p homologue in vivo (Fig. 1C), such a modification may enable the aaRS cofactor, the translation machinery, and even the yeast species to more efficiently cope with stress conditions such as high temperatures.
Histones possess a globular C-terminal domain, which forms the nucleosome body, and a flexible N-terminal tail, which protrudes from the surface of the nucleosome body. The flexible tail carries many lysine, arginine, and serine residues, which are potential targets of post-translational modifications, such as acetylation, methylation, phosphorylation, ubiquitination, poly(ADP-ribosylation), and sumoylation (30). These modifications help mediate interactions of histones with DNA and maintain the structure of chromatin. Recent studies showed that biotin modifications also naturally occur in human H3 and H4 histones, albeit to a much lesser extent (31). Despite the fact that the Arc1p homologue of C. albicans is not biotinylated in vivo (Fig. 1), its H2A, H2B, and H4 histones are natural targets of biotin modification (32). Perhaps, biotinylation occurs more often than previously thought. In accordance with this view, at least one protein (ϳ42 kDa) in L. elongisporus and two proteins (ϳ44 and ϳ60 kDa, respectively) in P. guilliermondii were naturally biotinylated (Fig. 1).
Despite our expectations, biotinylation was not required for the rescue activity, tRNA binding, or interactions of Arc1p with GluRS c /MetRS under the conditions used (Figs. 2-4). Interestingly, biotinylation of Arc1p was temperature dependent in vivo. Increasing the growth temperature from 30 to 37°C significantly reduced its biotinylation level (Fig. 8). As a result, Arc1p purified from a yeast culture that had been grown overnight at 37°C (Arc1p 37 ) was practically biotin free. Biotin-free Arc1p (such as Arc1p BϪ ) was more heat tolerant, more dynamic, and more effective than its biotinylated counterpart as an aaRS cofactor at high temperatures (Figs. 6 and 7). These findings reinforce the hypothesis that Arc1p moonlights as a biotin reservoir under normal growth conditions (30°C), but it is unbiotinylated at high temperatures to maintain its structure and function as an effective aaRS cofactor. Given that GluRS c alone was quite vulnerable to heat (Fig. 6), this feature of Arc1p is particularly significant. In contrast to the dramatic effect of biotinylation on Arc1p, biotin modification appears to have little effect on the solution structure of the biotin-binding domain of E. coli acetyl-CoA carboxylase (33). Perhaps it is because biotin serves as a coenzyme in carboxylases, but as a structure modulator in Arc1p.
As only ϳ15% of Arc1p was biotinylated in Arc1p Bϩ (or the Arc1p WT ), it is elusive as to why there was such a dramatic difference between Arc1p Bϩ and Arc1p BϪ (Figs. 6 and 7). One likely possibility is that a biotinylated Arc1p molecule can somehow induce a conformational change of a non-biotinylated Arc1p molecule when mixed together, a scenario reminiscent of prions that cause scrapie in sheep and goats (34). Alternatively, biotinylation may enable Arc1p to escape from the ternary complex and be targeted to other cellular compartments for functioning. Due to resolution constraints of the imaging technology used, we could not rule out this possibility at the moment. Another possibility is that the biotinylated Arc1p variant gains a new function, enabling the protein to participate in a biochemical pathway other than protein translation. More efforts are underway to look into this issue. As for the K86R mutant of Arc1p, this mutation impaired the biotinylation and also the structural stability of Arc1p (Figs. 2 and 4). As a result, Arc1p K86R failed to act as an effective GluRS c cofactor at high temperatures (Fig. 5). Although Arc1p BϪ was also devoid of biotinylation, it was more heat stable and more effective than Arc1p WT in promoting glutamylation activity of GluRS c at 37°C (Figs. 6 and 7). Because Arc1p BϪ still retains a native protein sequence, it is probably a better representative of a biotin-free Arc1p. However, the most striking finding reported here is the discovery that biotin modification can be used to modulate the structure and function of a non-carboxylase protein, which might open new avenues for studying biotinylation.

Experimental Procedures
Plasmid Construction-Cloning of the wild-type (WT) ARC1 gene (Ϫ300 to ϩ1128 bp) into pRS315 (a low copy number yeast shuttle vector), pADH (a high copy number yeast shuttle vector with a constitutive ADH promoter), and pGAL1 (a high copy number yeast shuttle vector with an inducible GAL1 promoter) followed a previously described protocol (35). Note that a short sequence encoding a His 6 tag was inserted into the 3Ј end of the multiple cloning sites in these vectors. Cloning of the WT ARC1 gene into pADH and pGAL1 followed a similar protocol, except that only the open reading frame (ϩ1 to ϩ1128 bp) was polymerase chain reaction (PCR) amplified and cloned.
To make the K86R mutant, the WT ARC1 gene was cloned into pBluescript II KS(ϩ/Ϫ) (Agilent, Santa Clara, CA), and the resultant construct was used as a template for mutagenesis. Mutagenesis was carried out following standard protocols provided by the manufacturer (Stratagene, La Jolla, CA). To make fusion constructs Arc1p-GFP and Arc1p-GST, DNA sequences encoding the green fluorescent protein (GFP) and glutathione S-transferase (GST) were, respectively, amplified by a PCR as an XhoI-XhoI fragment and then inserted into the XhoI site at the 3Ј end of ARC1 in the appropriate constructs. Cloning of the N domains of Arc1p (N-terminal residues 1ϳ132) and GluRS c (N-terminal residues 1ϳ228) into the yeast two-hybrid vectors pGBKT7 and pGADT7 followed a similar protocol.
Ni-NTA Pulldown Assay-An interaction analysis of Arc1p, MetRS, and GluRS c was performed by Ni-NTA affinity chromatography (Qiagen, Hilden, Germany) using purified MetRS-His 6 and GluRS c -His 6 as bait with purified Arc1p-GST as prey. Ni-NTA beads were washed and equilibrated in equilibration buffer (20 mM HEPES at pH 7.4 and 150 mM NaCl). Afterward, Ni-NTA beads (20 l) were mixed with 20 g of bait protein (MetRS-His 6 or GluRS c -His 6 ) and 40 g of prey protein (Arc1p-GST) in 500 l of equilibration buffer (final volume) and incubated at 4°C for 1 h. Nonspecifically bound proteins were removed by washing three times each with 1 ml of equilibration buffer containing 20 mM imidazole, and target proteins were eluted with 100 l of equilibration buffer containing 200 mM imidazole. Protein contents of column eluates were analyzed by SDS-PAGE and Coomassie Brilliant Blue staining.
Circular Dichroism (CD) Spectroscopy-CD spectral measurements were carried out in a Jasco J-810 spectropolarimeter (Tokyo, Japan) in a buffer containing 50 mM potassium phosphate (pH 8.0). Spectra were recorded at 10 -60°C using a 1-mm path length cuvette for far-UV CD (200 -240 nm) measurements with a scan speed of 50 nm/min, a time constant of 1 s, and a bandwidth of 1 nm. The final spectrum was the average of three independent measurements. The final protein concentration of Arc1p used in the assay was 1 M.
Aminoacylation Assay-Aminoacylation reactions were carried out at temperatures as indicated in a buffer containing 50 mM HEPES (pH 7.5), 15 mM KCl, 6 mM MgCl 2 , 5 mM dithiothreitol, 10 mM ATP, 0.1 mg/ml of bovine serum albumin, 100 M unfractionated yeast tRNA (Roche Applied Science, Germany), and 20 M glutamate (2 M [ 3 H]glutamate; Moravek Biochemicals, Brea, CA). The final concentration of GluRS c used in the reactions was 200 nM. The specific activity of [ 3 H]glutamate used was 51.1 Ci/mmol. Reactions were quenched by spotting 10-l aliquots of the reaction mixture onto Whatman filters (Maidstone, UK) soaked in 5% trichloroacetic acid and 1 mM glutamate. Filters were washed three times for 15 min each in ice-cold 5% trichloroacetic acid before liquid scintillation counting. Data were obtained from three independent experiments and averaged. Error bars indicate Ϯ2 times the standard deviation.
Streptavidin-based Gel Mobility Shift Assay-Relative biotinylation levels of Arc1p variants were determined by a streptavidin-based gel mobility shift assay. Briefly, 0.25 g of purified Arc1p was mixed with 6ϫ SDS loading buffer and heated to 95°C for 3 min. Arc1p that had been boiled in SDS loading buffer was incubated with 1 g of core streptavidin (a tetramer) for 5 min at 25°C before loading into a 10% polyacrylamide gel with no further heating. Following electrophoresis, the resolved proteins were transferred using a semi-dry transfer device to a polyvinylidene fluoride membrane in buffer containing 30 mM glycine, 48 mM Tris base (pH 8.3), 0.037% SDS, and 20% methanol. The membrane was probed with a horseradish peroxidase (HRP) anti-His 6 tag antibody (Invitrogen) and then exposed to x-ray film following the addition of appropriate substrates.
Author Contributions-C. C. W. designed the study and wrote the paper. C. Y. C. and S. C. did CD assay. C. P. C. complete limited proteolysis assay. C. Y. C. performed other assays. S. W. W. designed the assays. T. K. T. analyzed the data. All authors analyzed the results and approved the final version of the manuscript.