Duplication of leucyl-tRNA synthetase in an archaeal extremophile may play a role in adaptation to variable environmental conditions

Aminoacyl-tRNA synthetases (aaRSs) are ancient enzymes that play a fundamental role in protein synthesis. They catalyze the esterification of specific amino acids to the 3′-end of their cognate tRNAs and therefore play a pivotal role in protein synthesis. Although previous studies suggest that aaRS-dependent errors in protein synthesis can be beneficial to some microbial species, evidence that reduced aaRS fidelity can be adaptive is limited. Using bioinformatics analyses, we identified two distinct leucyl-tRNA synthetase (LeuRS) genes within all genomes of the archaeal family Sulfolobaceae. Remarkably, one copy, designated LeuRS-I, had key amino acid substitutions within its editing domain that would be expected to disrupt hydrolytic editing of mischarged tRNALeu and to result in variation within the proteome of these extremophiles. We found that another copy, LeuRS-F, contains canonical active sites for aminoacylation and editing. Biochemical and genetic analyses of the paralogs within Sulfolobus islandicus supported the hypothesis that LeuRS-F, but not LeuRS-I, functions as an essential tRNA synthetase that accurately charges leucine to tRNALeu for protein translation. Although LeuRS-I was not essential, its expression clearly supported optimal S. islandicus growth. We conclude that LeuRS-I may have evolved to confer a selective advantage under the extreme and fluctuating environmental conditions characteristic of the volcanic hot springs in which these archaeal extremophiles reside.

Unexpectedly, three unique cases of Mycoplasma LeuRSs with degenerated or missing CP1 (connective polypeptide) editing domains have also been identified (19). In addition to an editing defective LeuRS, Mycoplasma mobile was also found to encode a phenylalanyl-tRNA synthetase (PheRS) containing a defunct editing domain (19). By analyzing mass spectral data previously obtained from M. mobile samples (20), errors in protein translation as a direct consequence of editing-defective LeuRS and PheRS were identified that resulted in amino acid substitutions within the proteome of this parasitic bacterium (19). However, rather than being lethal, it was suggested that purposeful mistranslation in Mycoplasma could be adaptive, leading to antigen diversity and allowing this parasite to modulate its interactions with host defense systems (19).
Herein, using bioinformatic approaches, we searched for other examples, particularly for LeuRS, that might have adapted similarly, compared with Mycoplasma, in compromising mechanisms of fidelity. Significantly, we discovered that each member of the archaeal family Sulfolobaceae, within the phylum Crenarchaeota, contains a LeuRS (LeuRS-I) that has a disrupted CP1 editing domain (Fig. 1). Surprisingly, LeuRS-I appears to be a duplication of a second LeuRS (LeuRS-F) that possesses canonical aminoacylation and CP1 editing domains.

The Sulfolobaceae uniformly contain a LeuRS duplication with an altered editing domain
The CP1 domains of LeuRS were searched to identify unusual sequences that might compromise or eliminate its hydrolytic editing activity. Bioinformatic analysis of 39 unique strains within the Sulfolobaceae revealed two distinct isoforms that occupy and cluster within distinct branches of a phylogenetic tree (Fig. 1). Pairs of the full-length LeuRS enzymes within a given organism maintain ϳ40% amino acid sequence identity. However, depending upon the particular genera and species, pairwise alignments among one of the isoforms (Fig. 1, denoted LeuRS-F in orange) reveal conservation of 59 -69% amino acid sequence identity, whereas the second isoform ( Fig. 1, denoted LeuRS-I in blue) preserves 48 -53% identity.
One of the LeuRS variants, LeuRS-F, contained canonical sequences within the CP1 editing domain that would be expected to confer hydrolytic activity to maintain fidelity. In contrast, the second isoform, LeuRS-I, has critical substitutions of key amino acids that define the CP1 editing active site (Fig. 2). Within each species investigated, the universal aspartic acid essential to post-transfer editing (9) is substituted in LeuRS-I by a histidine (His-320). We hypothesized that this would render the Sulfolobaceae LeuRS-I editing deficient.
Each of the Sulfolobus islandicus and Sulfolobus solfataricus LeuRS-I's are also missing two additional residues important for post-transfer editing. A highly conserved threonine-rich sequence within the LeuRS CP1 domain forms a binding pocket for amino acids mischarged onto tRNA Leu (21,22). Mutation of the conserved threonine that forms the bottom of this binding pocket (Escherichia coli Thr-252) to alanine relaxes specificity of the CP1 editing site, allowing hydrolysis of correctly charged Leu-tRNA Leu (21). In the case of LeuRS-I, the position of this conserved threonine residue is occupied by a bulkier leucine (Leu-227, Fig. 2A). When bulkier residues such as leucine, phenylalanine, or tyrosine replace this threonine, mischarged amino acids are blocked from the editing site and tRNA Leu is stably aminoacylated with isoleucine and valine (23,24).
Moreover, it was hypothesized, and later shown, that Val-338 from E. coli LeuRS confers specificity to the CP1 domain (9,25). Mutating this residue to a bulkier phenylalanine disrupted E. coli LeuRS editing of Ile-tRNA Leu (25). In S. islandicus and S. solfataricus LeuRS-I, E. coli Val-388 corresponds to a tyrosine (Tyr-313) ( Fig. 2A). In concert with the substitution of the universally conserved aspartic acid, we hypothesize that these two mutations will abolish the editing capabilities of LeuRS-I.
The sequences of the Sulfolobaceae LeuRS isoforms were compared with Pyrococcus horikoshii LeuRS, which is the closest relative that has a solved crystal structure (26,27). There were no gross deletions, insertions, or rearrangements of domains suggesting that the Sulfolobaceae LeuRSs resemble their archaeal/eukaryal counterparts (Fig. 2B). However, the LeuRS C-terminal domains are significantly diverged with sequence identities that approached 20%. For example, whereas the C-terminal domain of S. islandicus LeuRS-F (residues 781-944) has a predicted pI of 8.6, its counterpart in LeuRS-I(772-934) is very acidic with a pI of 5.0. This results in an overall predicted pI for LeuRS-F and LeuRS-I of 7.1 and 5.6, respectively (Fig. 2B).

Histidine substitution of the universal aspartic acid inactivates post-transfer editing
Based on previous experiments that substituted the universally conserved aspartic acid in the CP1 editing domain of LeuRS, we hypothesized that a natural histidine substitution in Sulfolobaceae LeuRS-I would disable its mechanism to clear mischarged tRNA Leu . Using E. coli LeuRS as a model, we mutated the universally conserved aspartic acid to histidine to test if editing activity was inactivated. The mutant E. coli LeuRS-D345H was recombinantly expressed and affinity-purified via its N-terminal His 6 tag. The purified mutant enzyme aminoacylated E. coli tRNA Leu to similar levels as WT E. coli LeuRS (data not shown).
We generated mischarged E. coli Ile-tRNA Leu using editingdefective E. coli LeuRS-Y330A/D342A/D345A (28). The mischarged tRNA was isolated and used to test if editing activity was inactivated with E. coli LeuRS-D345H. Compared with WT E. coli LeuRS, the mutant E. coli LeuRS-D345H's deacylation activity was significantly decreased (Fig. 3). As would be expected, E. coli LeuRS-D345H also produced mischarged Ile-tRNA Leu (data not shown). Together, these results support that LeuRS-I, which has the same amino acid substitution, is deficient in editing.

S. islandicus LeuRS-I fails to aminoacylate in vitro transcribed tRNA Leu
We hypothesized that LeuRS-I generated statistical mutations similar to Mycoplasma LeuRS that is missing its CP1 domain (19). The genes for S. islandicus LeuRS-F and LeuRS-I

LeuRS-I is vital for optimal growth of S. islandicus
were cloned and expressed as fusions with N-terminal His 6 tags. Preliminary experiments determined that whereas LeuRS-F possessed minimal aminoacylation activity at ambient temperatures below 40°C, LeuRS-I lacked any measurable activity.
Because Sulfolobus species reside in fluctuating acidic environments at extreme temperatures, we screened for conditions to maximize activity using LeuRS-F. LeuRS-F aminoacylated in vitro transcribed S. islandicus tRNA Leu optimally at 55°C and pH 6.5. Interestingly, the intracellular pH of the related Sulfolobus acidocaldairus is 6.5 (29). Although LeuRS-F exhibited robust charging activity at elevated temperature and under acidic conditions, LeuRS-I still failed to aminoacylate this S. islandicus in vitro transcribed tRNA Leu substrate (Fig. 4A).
We further investigated whether this LeuRS paralog could bind tRNA Leu . Quenching of LeuRS intrinsic fluorescence was measured with increasing concentrations of in vitro transcribed tRNA Leu -UAG. LeuRS-F was found to bind in vitro transcribed tRNA Leu with an apparent affinity of 27.1 Ϯ 5.1 M (Fig. 4B). At 75 M tRNA, LeuRS-F is ϳ85% bound. Surprisingly, LeuRS-I also bound in vitro transcribed tRNA Leu with an apparent affinity of 31.4 Ϯ 2.5 M (Fig. 4B) and was 74% bound in the presence of 75 M tRNA Leu -UAG. These binding affinities are up to 10-fold lower than those reported for the archaeal LeuRS enzymes in the haloalkaliphile Natrialba magadii (30). It is possible the differences are a result of the technique utilized to obtain them, or that these unusual S. islandicus aaRSs require leucine and/or ATP to achieve optimal tRNA binding. Although the measured binding affinities are lower than expected for synthetase-tRNA interactions, the affinities for LeuRS-F and LeuRS-I are similar, suggesting LeuRS-I does not

LeuRS-I is vital for optimal growth of S. islandicus
have a tRNA-binding defect. Because LeuRS-I retains tRNAbinding activity, it is possible that LeuRS-I aminoacylates a novel tRNA or even a protein. It is also possible that the tRNAbound LeuRS-I functions outside of translation.
Key motifs within the Class I synthetase core of LeuRS-I appear to be intact with both HIGH and KMSKS signature sequences being maintained for amino acid activation. Using pyrophosphate exchange assays, we tested whether the recombinant LeuRS isoforms from S. islandicus could activate leucine. Using the optimal conditions of 55°C and pH 6.5 that were established for aminoacylation by LeuRS-F, we determined that LeuRS-I activates leucine, albeit at a lower efficiency than LeuRS-F (Fig. 4C). Interestingly, activation of noncognate amino acids were similar for LeuRS-F and LeuRS-I (data not shown). LeuRS-I generated the adenylates of both isoleucine and norvaline, albeit to a much lesser extent than leucine. Therefore, even at the conditions where LeuRS-I failed to aminoacylate S. islandicus in vitro transcribed tRNA Leu , it is clear that this LeuRS-like protein maintains the capacity to be enzymatically active. Given the enzyme's strong preference for leucine compared with other amino acids, it is possible that leucine activation is required for a noncanonical activity, to aminoacylate a novel tRNA or protein.

LeuRS-F is essential for S. islandicus viability
Because LeuRS-I failed to aminoacylate S. islandicus in vitro transcribed tRNA Leu , we wondered if its gene was defunct and even expressed in S. islandicus. Therefore, as a first step, we probed whether the duplicated leuRS genes are actively transcribed into mRNA. S. islandicus M.16.4 (31) was cultured at 78°C and harvested at mid-log phase. Total RNA was isolated and reverse transcribed into a cDNA library. Using gene-specific primers and the cDNA as a template, PCR yielded products of the expected size confirming that both leuRS-F and leuRS-I are actively transcribed (Fig. 5A).
We then asked whether one or both of these LeuRS paralogs were essential to the viability of S. islandicus. To address this question, we sought to generate knockout strains of both leuRS-F and leuRS-I. Because S. islandicus serves as a model organism for DNA replication, repair, and recombination within the Crenarchaeota, many genetic systems for in vivo analysis have been developed (see reviews in Refs. 32 and 33). Until recently, the methodologies for generating knockouts within Sulfolobus typically relied on uracil prototrophic selection and 5-fluoroorotic acid (5-FOA) counterselection using a pyrEF cassette (32)(33)(34)(35)(36)(37)(38)(39) or on the simvastatin selection marker 3-hydroxy-3-methylglutaryl coenzyme A (CoA) reductase gene (hmgA) (40,41). Because both of these markers pose significant

LeuRS-I is vital for optimal growth of S. islandicus
experimental challenges, a third, more robust selection was developed based on agmatine prototrophy (42,43) and applied to S. islandicus (44). Arginine decarboxylase, the protein product of argD, generates agmatine from arginine. Without this critical metabolite, cells cannot synthesize polyamines and are therefore not viable. Thus, cells deficient in this enzyme can be selected for in agmatine-supplemented media. We attempted to make the leuRS knockouts utilizing a vector, pRJW8 (44), harboring argD, pyrEF, and lacS markers and the previously developed plasmid integration and segregation (PIS) (Fig. 6) (34) and marker insertion and unmarked gene deletion (MID) methodologies ( Fig. S1) (39).
Both PIS and MID knockout methods rely on two discrete rounds of recombination (34,39). The first round occurs upon transformation of S. islandicus with a knockout plasmid. Upon transforming cells with a PIS plasmid, it integrates into its target site via single crossover recombination ( Fig. 6 and Fig. S1A), whereas for a linearized MID plasmid, a double crossover event will lead to integration (Fig. S1B). As the pRJW8 plasmid backbone contains genes for argD, pyrEF, and lacS, RJW004 (⌬argD ⌬lacS ⌬pyrEF) S. islandicus cells (44) that have integrated the plasmid into their genome are no longer ⌬argD ⌬lacS ⌬pyrEF, and are therefore prototrophic for agmatine and uracil. Thus, integrants can be propagated on medium no longer supplemented with agmatine and uracil. The gene lacS also allows for convenient blue/white screening of transformants. After colony purification, pyrEF ϩ integrants can be forced to undergo a second round of recombination when exposed to 5-FOA. This counterselection will yield either internal, in-frame leuRS deletions, WT revertants, or spontaneous pyrEF Ϫ integrants for the PIS method ( Fig. 6 and Fig. S1A) or internal, in-frame leuRS deletions or spontaneous pyrEF Ϫ integrants for the more stringent MID approach (Fig. S1B).
Approximately 20 and 25 colonies were obtained from cellular transformations that involved RJW004 with LeuRS-F and LeuRS-I PIS knockout plasmids, respectively. A subset of these colonies was treated with 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal), and as expected, stained blue supporting that the plasmid was stably integrated. Intermediates were subjected to counterselection with 5-FOA, resulting in numerous 5-FOA-resistant colonies. Several of these colonies were treated with X-gal and failed to stain blue indicating that the second round of recombination was successful. At least 10 colonies from the LeuRS-F and LeuRS-I counterselections were screened for the presence or absence of the gene targeted for deletion. Significantly, colonies representing leuRS-I knockout strains were readily obtained (Figs. 5B and 7A), suggesting that this LeuRS-like protein is not essential under standard laboratory conditions. As might be expected then, each of the screened leuRS-F counterselection colonies maintained the gene for this canonical LeuRS that contains an intact editing domain (Fig. 7B), suggesting that this protein is essential for S. islandicus viability. Even a second, more stringent, MID knockout method failed to yield the targeted leuRS-F deletion, further supporting the essentiality of leuRS-F. Therefore, we propose that LeuRS-F is essential to protein translation and functions primarily in tRNA Leu aminoacylation.

leuRS-I is required for S. islandicus wildtype growth
We compared the growth of our in-frame deletion strain of leuRS-I to its isogenic WT strain to probe for altered phenotypes. Sulfolobaceae thrive in challenging environments, which can fluctuate dramatically (45). Therefore, we monitored the growth of these two strains as a function of temperature. A small difference in growth between WT RJW004 and the deletion strain was apparent at the optimal growth temperature of 78°C. Although both strains reached similar levels of saturation at similar times, there was a slight, but consistent, reduction in lag and/or early log phase growth (Fig. 8A). This lag in growth was also observed at the higher temperature of 85°C (Fig. 8B). Interestingly, dropping the temperature to 60°C amplified the delayed growth phenotype (Fig. 8C). Moreover, at this temperature, deletion of leuRS-I results in a higher plateau suggesting greater proliferation compared with the WT strain. Alternatively, it is possible that the knockout strain aggregates, or that a metabolic shift could confer cells that are larger than WT cells, and therefore, scatter more light. Overall, whereas LeuRS-I is not essential for S. islandicus, it is clear that the gene is required for optimal growth, particularly at temperatures that are outliers for the extremophile.

Discussion
A long-standing paradigm is that within a given organism, a single aaRS gene would be maintained for each of the standard amino acids (46), except when a second copy is destined for the mitochondria or chloroplast in eukaryotes (47). However, aaRS duplications have occurred, generating paralogs that have adapted for alternate functions (48 -53). For example, the first identified synthetase duplication was E. coli lysyl-tRNA synthetase (LysRS) (51,52). Interestingly, the cellular function of this duplication is not entirely clear, although the second copy's (LysU) ability to generate Ap 4 A under conditions of cellular stress appears to be a dominant function (54,55). In addition, Actinomycetes, including Mycobacterium tuberculosis, produce mycothiol as a mechanism to protect against oxidative stress and electrophilic toxins. MshC, the enzyme responsible for catalyzing the penultimate step in the synthesis of this compound, has a tertiary-fold similar to that of cysteinyl-tRNA synthetase (53). Although the vast majority remain

LeuRS-I is vital for optimal growth of S. islandicus
uncharacterized, it is expected that aaRS paralogs are providing opportunities for the acquisition, or evolution, of new functions (48).
Duplication of LeuRS genes within the Crenarchaeota, specifically the Sulfolobaceae is striking. Because each species within this family contains the LeuRS duplication, acquisition of the second copy likely occurred prior to the divergence of the Sulfolobus, Metallosphaera, and Acidianus genera. Both LeuRS isoforms are rooted in Archaea, but are evolving on unique trajectories. Our work shows that LeuRS-F, with its intact CP1 domain, is essential with an evolutionary commitment to maintain fidelity.
The precise role of LeuRS-I remains puzzling given the absence of tRNA aminoacylation activity while maintaining capacity to bind tRNA Leu and also produce leucyl-adenylates. We hypothesize that its diverged acidic C-terminal domain plays a key role in defining the cellular function of LeuRS-I.
It is well-established that the canonical CTD for LeuRS is crucial for tRNA binding by the archaeal enzyme. Several resi- Figure 6. LeuRS-I is not essential for S. islandicus. In-frame deletion of leuS-I utilizing a PIS recombination method. Plasmid pRJW8-siI_PIS_Com contains a hybrid marker, lacS-pyrEF-argD (green, yellow, and cyan, respectively), in addition to homologous DNA sequences flanking leuS-I (indicated as UP and DOWN, respectively). S. islandicus M.16.4 RJW04 (WT-⌬lacS ⌬pyrEF ⌬argD) was transformed with this plasmid. Upon cellular uptake, the plasmid was integrated into the host chromosome via a single crossover at either UP (i) or DOWN (ii), resulting in prototrophic intermediates selected on uracil-and agmatine-free medium. In the presence of 5-FOA, integrants will undergo a second round of recombination at either UP or DOWN resulting in an in-frame internal deletion of leuS-I (X) or a RJW04 revertant (Y). Genes (gray) above the solid horizontal line are coded on the 3Ј-5Ј DNA strand, whereas leuS-I (orange) and other genes (gray) below this line are coded on the 5Ј-3Ј DNA strand.

LeuRS-I is vital for optimal growth of S. islandicus
dues within this domain recognize the conserved A47c and G47d nucleotides in the extra-long variable loop of archaeal tRNA Leu (26,30,56). In the absence of its CTD, archaeal LeuRS either fails to aminoacylate, or aminoacylation of tRNA Leu is inefficient (27,30). In contrast, the CTD domain can be dispensed in some cases for archaeal LeuRS post-transfer editing. For example, whereas Natrialba magadii (Nm) LeuRS-⌬CTD could not deacylate norvaline-NmtRNA Leu (30), PhLeuRS-⌬CTD retains this editing activity toward mis-charged PhtRNA Leu (27). Significantly, this domain is also important in preventing PhLeuRS from misediting correctly charged Ile-tRNA Ile (57).
Although LeuRS-I clearly binds to in vitro transcribed tRNA Leu -UAG, its CTD has potential to confer functional interactions with a specific tRNA Leu isoacceptor, a noncanonical tRNA, or even a unique tRNA-like molecular partner (58). If LeuRS-I's functional substrate is a particular tRNA Leu isoacceptor, this charged species could be utilized for a noncanonical function outside of protein synthesis (59). Alternatively, the intact catalytic core of LeuRS-I, which maintains the HIGH and KMSKS signature sequences required for aminoacylation (60 -65), could be dependent not only on a specific tRNA, but one that contains a particular modification. It is also possible that the bound tRNA Leu -LeuRS-I complex functions as a ribonucleoprotein for signaling.
LeuRS-I may also utilize the leucyl-adenylate for post-translational modifications of proteins rather than aminoacylating a tRNA, although these two scenarios are not mutually exclusive. Indeed, it was recently shown that all mammalian aaRSs can serve as amino acid sensors by functioning as aminoacyl transferases, modifying the ⑀-amine of lysine residues within their substrate proteins (66). In this case it could activate or suppress alternate metabolic pathways that affect cell responses to fluctuating environmental conditions. This could be consistent with the impact on optimal cell growth upon deletion of the LeuRS-I gene.
Although the cellular function of LeuRS-I remains unclear, a growing body of evidence in the literature supports a general role for the aaRSs, which occurs under stressful circumstances, to assist the organism in responding or adapting by promoting mistranslation through mis-aminoacylation of cognate, and/or noncognate tRNAs (67,68). For example, methionyl-tRNA synthetase (MetRS) is prone to making errors under oxidative stress (69 -71). Reactive oxygen species lead to increased MetRS phosphorylation with a concomitant decrease in tRNA specificity. Thus, increased methionylation of nonmethionyl tRNAs translates to increased methionine incorporation within the proteome. The methionine statistical mutations are proposed to act as efficient oxidant scavengers protecting residues that are important for structure or activity (72).
It is also possible that a temperature or pH shift within the Sulfolobaceae environment could alter tRNA specificities as in the case of MetRS from the hyperthermophilic crenarchaeon Aeropyrum pernix. At the lower end of its 70 to 100°C temperature range, MetRS fidelity for tRNA Met decreases, yielding a temperature-dependent shift in charging preference for tRNA Leu , generating Met-tRNA Leu (73). The leucine to methionine mistranslation was proficient in increasing the activity of A. pernix proteins at low temperatures (73). Intriguingly, the growth defect of S. islandicus ⌬leuRS-I at low temperatures (60°C) is markedly pronounced compared with their isogenic WT strain suggesting that an A. pernix MetRS-like mechanism is at least plausible for LeuRS-I and the Sulfolobaceae.

LeuRS-I is vital for optimal growth of S. islandicus
sential, plasmid-born copy (AgnB2). This organism generates TM 84, which inhibits LeuRS in the plant-tumor causing pathogen Agrobacterium tumefaciens, preventing the pathogen's growth (76). Interestingly, TM 84 also targets the essential genomic form of LeuRS within A. radiobacter, whereas the nonessential AgnB2 is resistant, thus preventing A. radiobacter cellular suicide (76). Intriguingly, in each identified instance (74 -76), a biologically active compound targets the aaRS in question and a second aaRS acts as a defense mechanism, imparting resistance by binding the molecule, thereby allowing the other copy to function normally.
Although members of the Sulfolobaceae can dominate local populations in low pH springs, S. islandicus species are rare in all environments (77). Furthermore, these species are in constant contact with several major genera of archaea (77,78). Therefore, it is possible that LeuRS-I's function has evolved as a protective measure against other acidothermophilic organisms.

Bioinformatics analysis of duplicated LeuRSs within the Sulfolobaceae
Seventy-eight LeuRS sequences representing 39 unique strains within the Sulfolobaceae were retrieved from the NCBI (SCR_003257) and Uniprot (SCR_002380) databases. Sequences were aligned using Clustal Omega (79). Alignment outputs were imported into MEGA X for phylogenetic analysis (80), using the minimum evolution algorithm and the Poisson substitution model. Bootstrap analysis was performed with 1000 replicates and those having values less than 25 are not shown. Selected sequences were also aligned against the E. coli LeuRS CP1 domain. Alignments were imported and manipulated within Jalview 2 (81).

Construction of LeuRS overexpression vectors
The coding sequence for leuRS-F was amplified from S. islandicus genomic DNA (gDNA) using Pfu DNA polymerase (Stratagene) and primer pair LRS1_YNf and LRS1_YNr containing XhoI and BamHI sites, respectively. The amplified DNA product was subcloned between the XhoI and BamHI sites of pET-14b (Novagen) generating the expression vector p14b_F. A SpeI site was introduced into pET-14b using primer pair p14M-4F and p14M-4R and a QuikChange site-directed mutagenesis kit (Stratagene), effectively eliminating the BamHI site generating the plasmid p14Bam-Spe. As done with leuRS-F, the coding sequence for leuRS-I was amplified with primer pair LRS2_YNf and LRSI_YNr, which include XhoI and SpeI sites, respectively. The amplified DNA product was cloned between the XhoI and SpeI sites of p14Bam-Spe generating the expression vector p14B-S_I. E. coli LeuRS-D345H was generated using plasmid p15EC3-1 (21,82), mutagenic primer pair D345Hf and D345Hr, and a QuikChange site-directed mutagenesis kit (Stratagene). All recombinant clones were verified via DNA sequencing. All primer sequences can be found in Table S1.

Protein overexpression and purification
Proteins were expressed using the T7 system (83). For S. islandicus proteins, a 1-liter culture of BL21(DE3) cells con-taining an expression plasmid was grown at 37°C in Luria-Bertani broth with 100 g/ml of ampicillin to an optical density (600 nm) of 0.4 to 0.7 before protein expression was induced with the addition of isopropyl ␤-D-thiogalactopyranoside to a final concentration of 1 mM. Induced cells were shifted to room temperature and allowed to grow overnight with continued aeration prior to harvesting. Cell pellets were resuspended in 15 ml of buffer 5 (10 mM Tris, 50 mM sodium phosphates, 300 mM NaCl, 10 mM imidazole, and 10% glycerol, pH 8.0). Cell suspensions were subjected to one round of a freeze-thaw cycle and then supplemented with 1 mM phenylmethylsulfonyl fluoride and 1 mg/ml of lysozyme. Cell lysates were then allowed to incubate on ice (45 min) before sonication (Fisher Scientific model 120 Sonic Dismembrator). After sonication, 10 units of RNase-free DNase (New England Biolabs) was added to the lysates prior to passing them twice through an 18-gauge needle. The suspensions were incubated on ice for 45 min and then centrifuged (30 min, 12,000 rpm, 4°C) to remove cellular debris. To remove additional E. coli contaminating proteins, the cleared lysates were collected and incubated at 60°C for 30 min with occasional mixing. The precipitated proteins were pelleted via centrifugation as before. The resulting supernatant was collected and loaded onto a 2-ml gravity nickel-nitrilotriacetic acid metal affinity column (Qiagen) pre-equilibrated with buffer 5. The column was then washed with 20 ml of buffer 5, 10 ml of buffer 6, 6 ml of buffer 7, 6 ml of buffer 8, and 6 ml of elution buffer (buffers 6, 7, and 8, and the elution buffer resemble buffer 5, but have increasing concentrations of imidazole: 25, 50, 75, and 250 mM, respectively). SDS-PAGE established that the majority of the desired proteins were within buffer 8 and elution buffer fractions. These fractions were collected and thoroughly dialyzed against storage buffer (50 mM potassium phosphate, 5 mM MgCl 2 , 25 mM KCl, 0.1 mM EDTA, 5 mM ␤-mercaptoethanol, 5 mM DTT, and 50% glycerol, pH 7.6) and placed at Ϫ20°C. For E. coli LeuRS and LeuRS-D345H, the following modifications were made. Cell pellets were resuspended in 7 ml of HA-1 buffer (10 mM Tris, 20 mM sodium phosphates, 300 mM NaCl, 5 mM imidazole, and 5% glycerol, pH 8.0) and then sonicated. Following sonication, the lysate was cleared via centrifugation at 12,000 rpm for 30 min at 4°C. The supernatant was collected and loaded onto a 2-ml gravity nickel-nitrilotriacetic acid metal affinity column pre-equilibrated with HA-1 buffer. The resin was then washed with a total of 100 ml of HA-2 buffer (10 mM Tris, 20 mM sodium phosphates, 500 mM NaCl, 5 mM imidazole, and 5% glycerol, pH 8.0), and the protein was eluted with 10 ml of HA-1 buffer at 100 mM imidazole. The collected protein solution was extensively dialyzed against storage buffer (100 mM potassium phosphates, 10 mM ␤-mercaptoethanol, and 50% glycerol, pH 6.8) and stored at Ϫ20°C. All protein concentrations were determined at 280 nm using extinction coefficients generated by the ProtParam tool (84) on the ExPASy Bioinformatics Resource Portal (SCR_012880) and via Bio-Rad protein assays (Bio-Rad) based on a modified Bradford assay (85). The results from these two assays were averaged.

LeuRS-I is vital for optimal growth of S. islandicus tRNA construction and RNA preparation
The gene for S. islandicus tRNA Leu (UAG) was constructed as previously described (86) with the following modifications. The gene having an upstream T7 RNA polymerase promoter, a 5Ј EcoRI site, a 3Ј BstNI site, and a terminal 3Ј PstI site was constructed from 14 overlapping synthetic DNA oligonucleotides (IDT). Oligonucleotides were phosphorylated in reactions that contained 1ϫ T4 DNA ligase buffer (New England Biolabs) containing 1 mM ATP, 4 M oligonucleotides, and 20 units of T4 polynucleotide kinase. Phosphorylated oligonucleotides were mixed in equimolar ratios, heated to 90°C, and then slow cooled to room temperature to allow for their annealing. The annealed product was added to EcoRI-and PstI-digested pUC18 (5:1 ratio) in a ligation reaction containing 1ϫ T4 DNA ligase buffer and 400 units of T4 DNA ligase, generating plasmid pUC-18_SiUAG 1-2. After sequence verification, this plasmid was used to generate template DNA for in vitro transcription reactions. Briefly, Vent DNA polymerase (New England Biolabs) and primer pair M13 Forward (Ϫ21)/Seq3A were used to amplify the gene and its flanking DNA sequences. The amplified product was cleaned using a QIAquick PCR purification kit (Qiagen) and subsequently digested using BstNI. After cleanup and concentrating, the DNA was used in transcription reactions at a final concentration of 18 ng/l. In vitro transcription reactions also contained the following: 40 mM Tris, pH 8.3, 1 mM spermidine, 0.05 mg/ml of bovine serum albumin (BSA), 5 mM DTT, 7.5 mM each nucleotide triphosphate, 32 mM MgCl 2 , 0.5 units/l of RNasin (Promega), 1 units/ml of yeast inorganic pyrophosphatase (PPiase, New England Biolabs), and 50 ng/l of WT T7 RNA polymerase. Reactions proceeded at 37°C for 4.5 h. Product tRNA was gauged greater than 95% pure by electrophoresis using an analytical 8% acrylamide, 4 M urea gel. Reactions were precipitated using ethanol and ammonium acetate at a final concentration of 2.5 M. Precipitated and cleaned tRNA was resuspended in 10 mM HEPES, 0.1 mM EDTA, pH 7.5, and frozen at Ϫ20°C until further use. Concentrations were determined using absorbance values at 260 nm and an extinction coefficient of 932.81 mM Ϫ1 cm Ϫ1 .

Aminoacylation and deacylation reactions
To fold S. islandicus tRNA Leu (UAG), a stock aliquot was first incubated in a 50°C water bath for 3 min. Samples of this preheated tRNA were used in folding reactions that contained 10 mM MES, pH 6.5, 5 mM MgCl 2 , and 80 M tRNA. The reactions were heated in an 85°C water bath for 3 min. The water bath/ tRNA folding reaction was then transferred to an ice bath and allowed to slow cool to 25°C. Aminoacylation reactions contained 60 mM MES, pH 6.5, 15 mM MgCl 2 , 2 mM DTT, 20 M [ 3 H]leucine (5 Ci/mmol), 20 M folded tRNA, and 2 mM ATP. The reactions were initiated with protein at a final concentration of 100 nM. At selected time intervals, aliquots (2 l) were quenched on Grade 3 Whatman filters (GE LifeSciences) presoaked in 10% TCA containing 10 mM leucine. After thorough washing, first in 10% TCA, then in 5% TCA, and finally in 95% ethanol, the samples were analyzed using a TRI-CARB 4910TR scintillation counter (PerkinElmer Life Sciences).
E. coli tRNA Leu -UAA was prepared using T7 in vitro transcription, and mischarged by [ 3 H]isoleucine using an E. coli LeuRS-Y330A/D342A/D345A triple mutant previously described (28). The hydrolytic activity of WT E. coli LeuRS and E. coli LeuRS-D345H toward mis-aminoacylated tRNA Leu was measured at 25°C in a reaction mixture containing 60 mM Tris-HCl, pH 7.5, 10 mM MgCl 2 , and 6.5 M [ 3 H]Ile-tRNA Leu . The reaction was initiated by the addition of 100 nM enzyme. At selected time intervals, 5-l reaction aliquots were quenched on pads that had been prepared as described above and processed similarly.

PPi (PP i ) exchange reactions
Reaction mixtures consisted of 100 mM MES, pH 6.5, 10 mM MgCl 2 , 2 mM DTT, 1 mM [ 32 P]PP i (375 Ci/ml), 1 mM leucine, 100 nM enzyme and were initiated by the addition of ATP to a final concentration of 1 mM. Reactions were run at 55°C. Aliquots (2 l) were collected at selected time points and quenched in 4 l of quenching solution (400 mM sodium acetate, 0.1% SDS). Quenched samples (2 l) were spotted on polyethylenimine TLC plates (Macherey-Nagel) that had been prerun in water. TLC plates were developed in 750 mM potassium phosphates, pH 3.5, at room temperature. After drying, separated radiolabeled bands were detected by phosphorimaging using a FUJIX BAS Cassette 2040 phosphorscreen (Fujifilm Medical Systems USA, Inc.) and a Typhoon (GE LifeSciences). Bands were quantified via densitometry using ImageJ (87).

tRNA-binding reactions
Although we sought to utilize microscale thermophoresis (88,89) to interrogate LeuRS-tRNA interactions, we consistently observed a systematic decrease in intrinsic fluorescence of both LeuRS-F and LeuRS-I when these proteins were titrated with increasing concentrations of tRNA, nullifying thermophoresis as a viable means of detecting these interactions. However, SDS-denaturation (SD) tests (NanoTemper Technologies recommendation) revealed that the observed fluorescence quenching was due to synthetase-tRNA interactions and not from ligand-induced sample aggregation, precipitation, or adsorption to capillaries. Therefore, fluorescence quenching was used to monitor LeuRS-F and LeuRS-I interactions with in vitro transcribed tRNA Leu -UAG.
The total volume of the binding reactions was 15 l and contained 60 mM MES, pH 6.5, 15 mM MgCl 2 , 10 M ZnSO 4 , 2 mM DTT, 25 mM NaCl, 12.5% glycerol, 0.1% pluronic F-127, in vitro transcribed tRNA Leu -UAG, and 500 nM of either LeuRS-F or LeuRS-I that had been centrifuged at 13,000 rpm for 10 min immediately prior to use. Folded tRNA Leu -UAG concentrations covered 3 orders of magnitude, ranging from 75 (high concentration) to 0.59 M (low concentration).
Initial experiments revealed that time was a key factor in obtaining consistent data. Therefore, all reactions were incubated at room temperature for at least 1 h prior to loading into standard-treated capillaries (NanoTemper Technologies). Measurements were made using the Monolith NT.Label Free (NanoTemper Technologies). Raw fluorescence scans were loaded into the NT Analysis software (NanoTemper Technologies) and normalized to the fraction of complexed molecules.

LeuRS-I is vital for optimal growth of S. islandicus
Data for three biological replicates were then imported into KaleidaGraph (Synergy Software), averaged, and fit to the Hill equation for determining apparent binding affinities. Reported errors are standard deviations of the binding affinities determined from fitting each of the three biological replicates individually for a given paralog.
SD-tests were performed to rule out ligand-induced protein loss as an explanation for the observed tRNA-dependent fluorescence changes. They were performed as suggested (Nano-Temper Technologies). Because capillaries 1 (75 M tRNA Leu ) and 8 (0.59 M tRNA Leu ) displayed the greatest difference in fluorescence intensities, their remaining reaction volumes were collected and centrifuged at 14,000 rpm for 11 min. A portion of this centrifuged reaction volume was removed and combined with an equal volume of 4% SDS, 40 mM DTT, followed by incubation at 95°C for 5 min. The solution was loaded into standard-treated capillaries and rescanned using the Monolith NT.Label Free. In the case of tRNA-induced quenching of LeuRS-F and LeuRS-I, the fluorescence of these denatured protein samples should be identical for both capillaries. If, however, a difference in fluorescence intensities between capillaries 1 and 8 is observed, protein was lost either by aggregation and the subsequent centrifugation or by unspecific adsorption to capillary walls. Because both capillaries had fluorescence counts around 2500 and 2000 for LeuRS-F and LeuRS-I experiments, respectively, the observed decrease in fluorescence was due to tRNA-induced quenching.

Construction of leuRS knockout plasmids
A PIS method (34) utilizing a second generation PIS vector encoding a selectable marker for agmatine prototrophy was employed to create markerless, internal, in-frame deletions of both leuRS-F and leuRS-I within S. islandicus M.16.4 ⌬argD ⌬lacS ⌬pyrEF (RJW004) (44). Both genes are encoded on the 5Ј to 3Ј Watson strand. For leuRS-F, 961-base pair (bp) upstream and 940-bp downstream fragments flanking the gene were amplified using S. islandicus gDNA and primer pairs CW-M1_Sen/CW-M2_Ant and CW-M3_Sen/CW-M4_Ant. The amplified products contain the last 39 and first 24 bases of leuRS-F, respectively. The two products were digested with EagI/BamHI and BamHI/SphI, respectively. The digested fragments were inserted, by triple ligation, into pRJW8 that had been digested with EagI and SphI. This resulted in recombinant plasmid pRJW8-siF_PIS_Com. Similarly, 906-bp upstream and 1000-bp downstream fragments flanking leuRS-I were amplified using primer pairs CW-M5_Sen/CW-M6_Ant and CW-M7_Sen/CW-M8_Ant. The amplified products contained the last 39 and first 45 bases of leuRS-I, respectively. The upstream and downstream fragments were digested with EagI/ BamHI and BamHI/SphI, respectively, and inserted via a triple ligation into EagI/SphI-digested pRJW8. This produced recombinant plasmid pRJW8-siI_PIS_Com. If successful, the internal deletion of leuRS-F would generate a protein product of 20 amino acids from the disrupted gene (8 N-terminal residues fused to 12 C-terminal residues), whereas for leuRS-I, a protein product of 27 amino acids (15 N-terminal residues fused to 12 C-terminal residues) would originate from the disrupted ORF.
As the PIS knockout strategy failed to generate deletions of leuRS-F (see "Results"), a MID strategy was also employed in an attempt to disrupt this gene (39). Upstream and downstream fragments flanking leuRS-F of sizes 940 and 922 bp, respectively, were amplified using S. islandicus gDNA and primer pairs CW-M32_Sen/CW-M33_Ant and CW-M30_Sen/CW-M31_Ant. The amplified products contain the last 18 and first 6 bases of leuRS-F, respectively. These fragments were digested with BamHI/SphI and EagI/BamHI, respectively, and inserted into pRJW8 via a triple ligation between the SphI and EagI sites. This yielded recombinant plasmid pRJW8-siF_UD_MID. A 700-bp amplification product corresponding to the 3Ј-end of leuRS-F (Tg-arm) was amplified using primer pair CW-M28_ Sen/CW-M29_Ant. This fragment was digested with KpnI and SalI and inserted into KpnI/SalI-digested pRJW8-siF_ UD_MID, yielding recombinant plasmid pRJW8-siF_MID-_Com. An internal deletion of leuRS-F using this plasmid would result in the generation of a 7-residue peptide (2 N-terminal residues fused to 5 C-terminal residues) originating from the ORF. All PIS and MID recombinant clones were sequenced verified.

Generating S. islandicus leuRS markerless, in-frame deletions
Cells used for transformations were prepared essentially as described (90) with the following modifications. Prior to use, cells were passaged into fresh DT-A/U medium at least two times. For the second passage, the culture was diluted to an OD 600 nm of 0.008. This culture was allowed to grow to an OD 600 nm of 0.2-0.3 prior to harvesting via centrifugation (15 min, 8,000 rpm, 4°C). Pelleted cells were resuspended in 20 ml of ice-cold 20 mM sucrose, and thereafter kept on ice. The resuspended cells were pelleted and washed once more with 10 ml of ice-cold sucrose. After a final pelleting step, cells were resuspended to an OD 600 nm of 10 to 15. Electrocompetent RJW004 was transformed with circular PIS plasmids or linearized MID plasmids (1-1.5 g) using a Bio-Rad Gene Pulser TM with input parameters of 1.2 kV, 25 F, and 600 ⍀ in 0.1-cm gap LeuRS-I is vital for optimal growth of S. islandicus electrocuvettes (Bio-Rad). Immediately following electroporation, 800 l of 78°C recovery media (1% (w/v) sucrose, 20 mM ␤-alanine, 1.5 mM malic acid, 10 mM MgSO 4 , pH 4.5) was added to the cells. This suspension was incubated at 78°C for 30 min, standing. As the pRJW8 plasmid backbone encodes argD and pyrEF, the entire transformation volume was added to 10 ml of 78°C DT plate medium supplemented with 0.4% (w/v) Gelrite and poured onto pre-heated 25-ml DT plates previously solidified with 0.7% (w/v) Gelrite (no agmatine or uracil supplementation needed). Transformants were allowed to grow within this overlay ("overlay cultivation") for 10 -15 days at 78°C double-bagged and sealed.
Colonies resulting from transformations were placed into 2 ml of DT media and grown at 78°C until cell growth was apparent. Small samples of these cultures (100 l) were assayed for LacS activity by the addition of X-gal to a final concentration of 2.5 mg/ml. The assay solution was allowed to incubate at 78°C for 1 h. A blue precipitate was indication of successful plasmid integration into the chromosome of RJW004. Cultures that stained when treated with X-gal were passaged into 20-ml volumes of DT media. When an OD 600 nm of 0.2-0.3 was reached, these cultures were again passaged into 20-ml volumes of DT media at an initial OD 600 nm of 0.008. At an OD 600 nm of 0.1-0.2, 10-fold serial dilutions of these cultures were plated as described above. The resulting purified colonies were isolated, grown as described, and counterselected on DT-A/U plates supplemented with 5-FOA. Knockouts were identified using X-gal staining and PCR screens. Established knockouts were verified via DNA sequencing.

Growth curves
To compare the growth of RJW004 and RJW004 ⌬leuRS-I, cells were passaged at least two times (see above) and subsequently harvested by centrifugation (15 min, 8,000 rpm, 4°C). The cell pellets were rinsed 2 times in 20 mM sucrose. After an additional pelleting step, the cells were resuspended in DT-A/U medium. These cells were used to inoculate 45 ml of DT-A/U medium at an initial OD 600 nm of 0.008. OD 600 nm measurements were obtained at the indicated time points. Each data point represents the average OD 600 nm of three independent cultures, and error bars represent the mean Ϯ S.D. of these measurements.

RT-PCR analysis
Samples (10 ml) of RJW004 and RJW004 ⌬leuRS-I were collected at an OD 600 nm of 0.25. Total RNA from these samples was isolated using a mirVana TM miRNA Isolation kit according to the manufacturer's protocol. Total RNA was digested with 1000 units of RNase-free DNase I (New England Biolabs) in the presence of 4000 units of RNasin (Promega). After digestion and cleanup, total RNA was used to make cDNA using an iScript cDNA Synthesis kit (Bio-Rad). To detect the presence of contaminating gDNA, control reactions were assembled containing no iScript reverse transcriptase. cDNA corresponding to leuRS-F, leuRS-I, and gapDH (glyceraldehyde-3-phosphate dehydrogenase) were targeted in PCR with primer pairs CSW49_Sen/CSW50_Ant, CSW35_Sen/CSW36_Ant, and CSW53_Sen/ CSW54_Ant, respectively. Control reactions were also carried out using template from the no iScript control RT reactions.