A Novel Aminoacyl-tRNA Synthetase Appended Domain Can Supply the Core Synthetase with Its Amino Acid Substrate

The structural organization and functionality of aminoacyl-tRNA synthetases have been expanded through polypeptide additions to their core aminoacylation domain. We have identified a novel domain appended to the methionyl-tRNA synthetase (MetRS) of the intracellular pathogen Mycoplasma penetrans. Sequence analysis of this N-terminal region suggests the appended domain is an aminotransferase, which we demonstrate here. The aminotransferase domain of MpMetRS is capable of generating methionine from its α-keto acid analog, 2-keto-4-methylthiobutyrate (KMTB). The methionine thus produced can be subsequently attached to cognate tRNAMet in the MpMetRS aminoacylation domain. Genomic erosion in the Mycoplasma species has impaired many canonical biosynthetic pathways, causing them to rely on their host for numerous metabolites. It is still unclear if this bifunctional MetRS is a key part of pathogen life cycle or is a neutral consequence of the reductive evolution experienced by Mycoplasma species.


Introduction
Aminoacyl-tRNA synthetases (AARSs) are multi-domain proteins that catalyze attachment of a specific amino acid to its cognate tRNA, preparing the aminoacyl-tRNAs for template-directed protein synthesis at the ribosome. AARSs have been partitioned into two classes based on evolutionarily conserved structure and sequence similarities in their catalytic domains [1][2][3]. Over time, polypeptide domains and appendages have been added to this catalytic core, increasing the complexity and functionality of the AARS family [4]. Some additions improve the overall specificity and efficiency of the AARS, for example, the editing domains found in several class I enzymes [5]. In higher eukaryotes, accretion of novel domains has endowed AARSs with functions outside of the translational pathway [6]. In other cases, the appended domain serves to regulate the multimeric state of the AARS; for example, class I methionyl-tRNA synthetase (MetRS) is either a monomer or a dimer, depending on the presence of a C-terminal appended domain [7,8].
As more genomic sequences become available, additional examples of idiosyncratic acquired AARS domains surface. An extreme example is that of Mycoplasma penetrans, an obligate intracellular human pathogen first identified in HIV-infected patients [9]. Through reductive evolution, M. penetrans minimized its genome to 1.36 Mb, including 1038 protein encoding genes, and thereby lost enzymes not necessary for survival in its environmental niche of the human urogenital and respiratory tracts. The MYPE9380 open reading frame (ORF) was annotated as metS, encoding the M. penetrans MetRS homolog (MpMetRS) [10]. However, metS contains a large insertion upstream of the canonical MetRS-encoding gene, resulting in an apparent MpMetRS protein of 1087 amino acids which is approximately twice that of a typical bacterial MetRS. BLAST analysis revealed that the far N-terminal region, consisting of 166 amino acids, does not share significant sequence similarity with any reported protein. Overall, the novel acquired domain is 46% similar to the Anabaena sp. alanine-glyoxylate aminotransferase (AGAT), accession number WP015216524. The Anabaena sp. AGAT is a member of the aspartate aminotransferase superfamily of proteins and contains a highly conserved pyridoxal 5'-phosphate (PLP) binding domain [11]. The C-terminal half of the MpMetRS ORF shares 45% sequence similarity with Escherichia coli MetRS (EcMetRS). A truncated version of MpMetRS encoding only the canonical MetRS core retains function and specificity similar to that of other characterized MetRSs [12,13].
It is of particular interest that one of the acquired domains exhibits sequence similarity with an aminotransferase, as aminotransferases are essential to the methionine salvage pathway (MSP). In this route, the enzyme would catalyze conversion of 2-keto-4-methylthiobutyrate (KMTB) to methionine, using an amino acid as the amine donor [14]. The MSP serves to salvage sulfur from 5'-methylthioadenosine, which is a by-product of polyamine (for example, spermine) and ethylene synthesis. Transaminase enzymes exhibiting activity towards KMTB are mainly aromatic transaminases or aspartate transaminase homologues [15]. Genome analysis has revealed that M. penetrans has lost canonical pathways for methionine biosynthesis, leaving the MSP as the only means by which it can regulate cellular methionine concentration beyond what is available from the host [16].
We sought to determine if the fusion of AGAT and MetRS genes into a single ORF arose because the AGAT is able to function cooperatively with the MetRS to supply methionine in M. penetrans, and, in doing so, it facilitates efficient formation of Met-tRNA Met . This could be accomplished by the aminotransferase domain acting upon a KMTB-tRNA Met intermediate to form Met-tRNA Met , similar to the strategy employed in glutamine-tRNA Gln formation from Glu-tRNA Gln [17]. Alternatively, the transferase could act directly on KMTB to form methionine, which is then a substrate for the aminoacylation domain to synthesize Met-tRNA Met . Here we present a characterization of the MpMetRS aminotransferase domain, its capacity to generate methionine from the keto acid precursor, and the ability of the full MpMetRS protein to catalyze Met-tRNA Met synthesis from KMTB and tRNA Met substrates, using a general amine donor.

Materials
DNA oligonucleotides were synthesized by Integrated DNA Technologies (IDT; Coralville, IA, USA). HisTrap nickel columns were from Cytiva (Marlborough, MA, USA). α-32 P ATP was purchased from Perkin-Elmer (Waltham, MA, USA). Restriction enzymes were from New England Biolabs (Ipswitch, MA, USA). Expression vector pQE70 was from Qiagen. Pfu Ultra DNA polymerase, XL-10 Gold cells and QuikChange multi-site directed mutagenesis kit were from Agilent (Santa Clara, CA, USA). SuperSignal West Pico Rabbit IgG Detection Kit was from ThermoFisher Scientific (Waltham, MA, USA). M. penetrans protein lysate was a gift from Dr. M. Balish of Miami University of Ohio. Salts, buffers, and growth media were from Fisher Scientific (Waltham, MA, USA).

Cloning, Mutagenesis, and Expression
The metS gene was cloned into the pQE70 vector, and the MpMetRS protein was expressed as previously described [12]. Variants Met-568→Ala, Lys-386→Ala, Asp-616→Ala and Trp-1005→Ala were generated by QuikChange mutagenesis, following the manufacturer's protocol, with primers obtained from IDT. The M568A substitution was made to remove an internal start site within the MetRS gene and was used as the functional wild-type MpMetRS in all experiments. Three other alanine variants (K386A, D616A, and W1005A) were generated to produce catalytically inactive aminotransferase and synthetase domains, as described below. Following protein purification, all MpMetRS variants were separated by SDS-PAGE to determine homogeneity (Supplementary Materials Figure S3).

Western Blot Analysis
MpMetRS samples were separated on a 10% SDS-PAGE, using 40 µg M. penetrans lysate, 0.1 µg recombinant MpMetRS, and 40 µg XL-Gold E. coli lysate containing empty vector. Proteins were transferred to a nitrocellulose membrane at 100 V for 1 h. The membrane was blocked overnight at 4 • C with 4% milk in Tris-buffered saline (pH 7.5)/0.1% Tween-20. The membrane was incubated with rabbit α-MpMetRS (ThermoScientific; 1:10,000) and rocked overnight, at 4 • C. The membrane was washed six times, incubated with goat α-rabbit-HRP (1:4000) at 4 • C for 1.5 h, and again washed six times. The remainder of the procedure was performed by the manufacturer's guidelines for the SuperSignal West Pico Rabbit IgG Detection Kit and visualized on an Amersham Imager 600 (Cytiva, Marlborough, MA, USA).

In Vitro tRNA Synthesis
M. penetrans tRNA Met (Mp-tRNA Met ) was synthesized in vitro from a DNA template generated by extending overlapping primers (Supplementary Materials Table S2) with the Klenow fragment as previously described [18]. The duplex tDNA template (250 µL of Klenow reaction) was combined with 0.1 mg/mL bovine serum albumin (BSA), 2 mM spermidine, 250 mM HEPES·NaOH (pH 7.5), 30 mM MgCl 2 , 5 mM NTPs, 40 mM DTT, 10 U yeast inorganic pyrophosphatase, and 100 µL T7 RNA polymerase. The transcription reaction (2 mL total volume) proceeded for 12-16 h, in a shaking incubator, at 37 • C, prior to treatment with DNase I for 2 h and ethanol precipitation. The transcript was purified on a 10% polyacrylamide (19:1 acrylamide:bisacrylamide) gel with 8 M urea in 1X TBE. The tRNA band was identified by UV shadowing, excised from the gel, and eluted in 500 mM NH 4 OAc (pH 5.2) and 1 mM EDTA, overnight, in a shaking incubator, at 37 • C. The extracted tRNA was passed through a 0.2 µm syringe filter, ethanol precipitated, and resuspended in 20 mM HEPES·NaOH (pH 7.5).

Liquid Chromatography-Mass Spectrometry Detection of Transamination
The capacity of the MpMetRS aminotransferase domain to catalyze transamination was assayed in 100 mM ammonium bicarbonate buffer (pH 7.4), 5 mM phenylalanine, 1-40 mM KMTB, and 50 nM MpMetRS over one minute for steady-state kinetic analysis of KMTB; 100 mM ammonium bicarbonate buffer (pH 7.4), 10 mM KMTB, 0.5-16 mM phenylalanine, and 50 nM MpMetRS over one minute for steady-state kinetic analysis of phenylalanine. Screening of transamination substrates was performed, using 1 mM amino acid and 1 mM α-keto acid with 1 µM MpMetRS, over a period of 40 min. MpMetRS alanine variants were assayed at 100 nM with 1 mM KMTB and 5 mM phenylalanine with a time course of 40 min. All reactions were quenched (1:10) in Optima H 2 O/0.1% formic acid. A ZORBAX SB-C18 column (5 µm, 4.6 × 150 mm) was equilibrated for 30 min on an amaZon LC/MS with 50% Optima H 2 O/0.1% formic acid (solvent A): 50% MeOH/0.1% formic acid (solvent B) at a flow rate of 0.400 mL/min prior to sample analysis. After equilibration, a 10 µL sample was analyzed at the same flow rate, under the following LC/MS gradient: increase solvent B from 50% to 65% over 10 min, 65% to 100% over 2 min, isocratic for 5 min, 100% to 50% over 2 min, and equilibration for 10 min.

The MpMetRS metS Gene
The complete genome of the obligate intracellular parasite Mycoplasma penetrans HF-2 was sequenced in 2002 by Sasaki et al. [10]. By comparative genomics, it was clear that the annotated M. penetrans metS gene (3264 nt) was unusually long compared to the E. coli homolog (2034 nt) [12]. Sequence alignment revealed that the 3 -half of the M. penetrans metS gene encodes a canonical methionyl-tRNA synthetase with HIGH and KMSKS sequence motifs and 45% similarity to E. coli MetRS. The 5 -half of the gene is also an open reading frame with 46% similarity to the Anabaena sp. alanine glyoxylate aminotransferase (AGAT). We previously cloned the M. penetrans metS gene into a pCR2.1-TOPO vector, mutated the 12 UGA codons (used as tryptophan in mycoplasma but stop in E. coli) to UGG tryptophan codons, and cloned the resulting gene into a pQE70 expression vector [12]. The Anabaena AGAT crystal structure (PDB 1VJO) is co-crystalized with the pyridoxal 5 -phosphate (PLP) cofactor [21]. Sequence alignment of the Anabaena AGAT with the N-terminal domain of MpMetRS reveals that several key PLP binding residues are conserved. A spectroscopic method adapted from Wada et al. was used to determine that recombinant MpMetRS co-purifies with PLP at 72% occupancy (Supplementary Materials Table S1) [22]. Initial expression of the recombinant MpMetRS resulted in the generation of two bands on a denaturing SDS-PAGE, one at 126.4 kDa and the other at 61.4 kDa (Supplementary Materials Figure S3). The lower band was subjected to Edman sequencing analysis and corresponds to the C-terminal core synthetase domain (MpMetRS∆). It has previously been suggested that recombinant mycoplasma genes may contain internal ribosomal binding sites, producing more than one product following heterologous expression [23]. To test whether the smaller band was due to proteolytic cleavage or an internal ribosomal binding site, we generated a Met-568→Ala variant, as Met-568 is at the beginning of the region that aligns with the canonical bacterial MetRS. The resulting overexpressed protein migrated as a single band at 126.4 kDa, suggesting that the smaller fragment was indeed the result of an internal translational start and that the M568A variant corresponds to full-length MpMetRS.

Evidence for Full-Length MpMetRS In Vivo
Given the presence of two products following recombinant expression in E. coli, we wanted to identify whether MpMetRS exists in vivo as the proposed aminotransferase-synthetase fusion protein.
We therefore performed immunoblots on both M. penetrans cell lysate and recombinant proteins. The band present in each sample just below the 130 kDa marker is consistent with the 126.4 kDa size of full-length MpMetRS ( Figure 1). An additional smaller band in the recombinant MpMetRS corresponds to MpMetRS∆ (61.4 kDa); this smaller fragment is absent in the MpMetRS M568A variant. A similarly sized smaller band is also present in the lysate sample; we hypothesize that M. penetrans uses an internal start codon to produce the isolate aminoacylation domain. The ratio of full-length to fragment seems variable, and ongoing experiments seek to identify the variability and significance of the alternate translational forms of MpMetRS.
As the Anabaena sp. aminotransferase exists as a dimer, we probed the oligomeric state of MpMetRS. Size exclusion chromatography-multiangle light scattering (SEC-MALS) was performed to assess the molecular weight and polydispersity of the enzyme [24,25]. This analysis determined that full-length MpMetRS exists in solution as a dimer with an apparent molecular weight of 253 kDa and a polydispersity of one, while MpMetRS∆ is a monomer of 77 kDa and a polydispersity of one (Supplementary Materials Figure S1).
As the Anabaena sp. aminotransferase exists as a dimer, we probed the oligomeric state of MpMetRS. Size exclusion chromatography-multiangle light scattering (SEC-MALS) was performed to assess the molecular weight and polydispersity of the enzyme [24,25]. This analysis determined that full-length MpMetRS exists in solution as a dimer with an apparent molecular weight of 253 kDa and a polydispersity of one, while MpMetRSΔ is a monomer of 77 kDa and a polydispersity of one (Supplementary Materials Figure S1).

MpMetRS Is a Bifunctional Enzyme
We have previously established that MpMetRSΔ alone is capable of catalyzing Met-tRNA Met synthesis [12]. When the full-length MpMetRS was compared to the aminoacylation capability of MpMetRS∆, both enzymes aminoacylated tRNA Met ; the role of the MpMetRS appended domain was unclear. Consistent with our earlier findings, aminoacylation conducted in this work, using fulllength MpMetRS (M568A), remained robust for its tRNA substrate ( Figure 2). While MpMetRS∆ is able to aminoacylate tRNA Met , the activity is low and requires an elevated enzyme concentration. Subsequent analyses of MpMetRS in this work use the M568A variant as the wild-type, full-length enzyme.  Figure S2) [26]. We therefore set out to establish whether MpMetRS can catalyze the interconversion of α-keto acids and amino acids. The enzyme was tested with several different amine group donors and acceptors, using LC-MS to monitor reaction progress. Amine group transfer to the methionine precursor KMTB was tested with three different amino acids: phenylalanine, aspartate, and glutamate. The three amine group donors

MpMetRS Is a Bifunctional Enzyme
We have previously established that MpMetRS∆ alone is capable of catalyzing Met-tRNA Met synthesis [12]. When the full-length MpMetRS was compared to the aminoacylation capability of MpMetRS∆, both enzymes aminoacylated tRNA Met ; the role of the MpMetRS appended domain was unclear. Consistent with our earlier findings, aminoacylation conducted in this work, using full-length MpMetRS (M568A), remained robust for its tRNA substrate ( Figure 2). While MpMetRS∆ is able to aminoacylate tRNA Met , the activity is low and requires an elevated enzyme concentration. Subsequent analyses of MpMetRS in this work use the M568A variant as the wild-type, full-length enzyme.
Genes 2020, 11, x FOR PEER REVIEW 5 of 12 As the Anabaena sp. aminotransferase exists as a dimer, we probed the oligomeric state of MpMetRS. Size exclusion chromatography-multiangle light scattering (SEC-MALS) was performed to assess the molecular weight and polydispersity of the enzyme [24,25]. This analysis determined that full-length MpMetRS exists in solution as a dimer with an apparent molecular weight of 253 kDa and a polydispersity of one, while MpMetRSΔ is a monomer of 77 kDa and a polydispersity of one (Supplementary Materials Figure S1).

MpMetRS Is a Bifunctional Enzyme
We have previously established that MpMetRSΔ alone is capable of catalyzing Met-tRNA Met synthesis [12]. When the full-length MpMetRS was compared to the aminoacylation capability of MpMetRS∆, both enzymes aminoacylated tRNA Met ; the role of the MpMetRS appended domain was unclear. Consistent with our earlier findings, aminoacylation conducted in this work, using fulllength MpMetRS (M568A), remained robust for its tRNA substrate (Figure 2). While MpMetRS∆ is able to aminoacylate tRNA Met , the activity is low and requires an elevated enzyme concentration. Subsequent analyses of MpMetRS in this work use the M568A variant as the wild-type, full-length enzyme.  Figure S2) [26]. We therefore set out to establish whether MpMetRS can catalyze the interconversion of α-keto acids and amino acids. The enzyme was tested with several different amine group donors and acceptors, using LC-MS to monitor reaction progress. Amine group transfer to the methionine precursor KMTB was tested with three different amino acids: phenylalanine, aspartate, and glutamate. The three amine group donors  Figure S2) [26]. We therefore set out to establish whether MpMetRS can catalyze the interconversion of α-keto acids and amino acids. The enzyme was tested with several different amine group donors and acceptors, using LC-MS to monitor reaction progress. Amine group transfer to the methionine precursor KMTB was tested with three different amino acids: phenylalanine, aspartate, and glutamate. The three amine group donors varied in their capacity as substrates, with phenylalanine being the best in vitro donor among those tested ( Figure 3A). varied in their capacity as substrates, with phenylalanine being the best in vitro donor among those tested ( Figure 3A). To test the catalytic capacity of MpMetRS to use different amine group acceptors, the enzyme was supplied with either KMTB or oxaloacetate, and phenylalanine was the amine donor for production of methionine or aspartate, respectively ( Figure 3B). The conversion of oxaloacetate to aspartate was quantified by measuring the depletion of phenylalanine rather than the accumulation of aspartate. In control experiments, the LC-MS signal for aspartate was not sensitive enough for accurate quantification, for reasons that remain unclear.

The Domains of MpMetRS Are Functionally Independent
While we have shown that MpMetRSΔ was capable of aminoacylation in the absence of its Nterminal domain, we wanted to determine if the two domains function independently in the context of the full-length enzyme. Site-specific substitutions in each catalytic domain were generated at essential amino acid positions. The first substitution was a Lys-386→Ala (K386A) in the aminotransferase domain; this conserved lysine aligns with the Lys-209 residue of Anabaena variabilis alanine-glyoxylate aminotransferase, which forms a Schiff base intermediate with PLP during amine group transfer [21]. Two individual variants were generated in the aminoacylation domain, Asp-616→Ala (D616A) and Trp-1005→Ala (W1005A). These residues align with Asp-52 and Trp-461 of E. coli MetRS, where the acidic aspartate orients ATP during the formation of the methionine adenylate, and the tryptophan stacks with C34 of the tRNA Met anticodon [21,27,28]. To ensure there were no drastic structural change as a result of these alanine substitutions at conserved positions, we performed circular dichroism spectroscopy (Supplementary Materials Figure S4), thermal melting studies, and PLP occupancy tests (Supplementary Materials Table S1). The alanine variants were then compared to wild-type MpMetRS activity in both an LC-MS assay to monitor methionine synthesis and a TLC assay to quantify Met-tRNA Met formation. For each variant, only the activity of the domain containing the substitution was inhibited, while the activity of the adjacent domain was comparable to the wild-type enzyme (Figure 4). To test the catalytic capacity of MpMetRS to use different amine group acceptors, the enzyme was supplied with either KMTB or oxaloacetate, and phenylalanine was the amine donor for production of methionine or aspartate, respectively ( Figure 3B). The conversion of oxaloacetate to aspartate was quantified by measuring the depletion of phenylalanine rather than the accumulation of aspartate. In control experiments, the LC-MS signal for aspartate was not sensitive enough for accurate quantification, for reasons that remain unclear.

The Domains of MpMetRS Are Functionally Independent
While we have shown that MpMetRS∆ was capable of aminoacylation in the absence of its N-terminal domain, we wanted to determine if the two domains function independently in the context of the full-length enzyme. Site-specific substitutions in each catalytic domain were generated at essential amino acid positions. The first substitution was a Lys-386→Ala (K386A) in the aminotransferase domain; this conserved lysine aligns with the Lys-209 residue of Anabaena variabilis alanine-glyoxylate aminotransferase, which forms a Schiff base intermediate with PLP during amine group transfer [21]. Two individual variants were generated in the aminoacylation domain, Asp-616→Ala (D616A) and Trp-1005→Ala (W1005A). These residues align with Asp-52 and Trp-461 of E. coli MetRS, where the acidic aspartate orients ATP during the formation of the methionine adenylate, and the tryptophan stacks with C34 of the tRNA Met anticodon [21,27,28]. To ensure there were no drastic structural change as a result of these alanine substitutions at conserved positions, we performed circular dichroism spectroscopy (Supplementary Materials Figure S4), thermal melting studies, and PLP occupancy tests (Supplementary Materials Table S1). The alanine variants were then compared to wild-type MpMetRS activity in both an LC-MS assay to monitor methionine synthesis and a TLC assay to quantify Met-tRNA Met formation. For each variant, only the activity of the domain containing the substitution was inhibited, while the activity of the adjacent domain was comparable to the wild-type enzyme (Figure 4).

MpMetRS Aminotransferase Domain Can Supply tRNA Met With Methionine
The ability of MpMetRS to use KMTB as a transaminase substrate and the presence of fused aminotransferase and tRNA synthetase active sites led us to investigate whether the product of one domain could provide the substrate for the other. To test this hypothesis, we performed an aminoacylation assay in the presence of KMTB and phenylalanine with 32 P-labeled tRNA Met . When MpMetRS is supplied with methionine directly, Met-tRNA Met is formed, as previously shown. When the aminotransferase is supplied with phenylalanine and KMTB in the absence of methionine, Met-tRNA Met is synthesized, albeit at reduced efficiency ( Figure 5). Formation of Met-tRNA Met when the enzyme is supplied Met directly is 2.2 pmol/min, which is only ~2-fold greater than the rate when Met is generated in the aminotransferase domain from KMTB and Phe (1.2 pmol/min). Aminoacylation from KMTB alone is negligible (0.04 pmol/min). The reaction kinetics of each domain were determined through either LC-MS or aminoacylation assays and are summarized in Table 1.

MpMetRS Aminotransferase Domain Can Supply tRNA Met with Methionine
The ability of MpMetRS to use KMTB as a transaminase substrate and the presence of fused aminotransferase and tRNA synthetase active sites led us to investigate whether the product of one domain could provide the substrate for the other. To test this hypothesis, we performed an aminoacylation assay in the presence of KMTB and phenylalanine with 32 P-labeled tRNA Met . When MpMetRS is supplied with methionine directly, Met-tRNA Met is formed, as previously shown. When the aminotransferase is supplied with phenylalanine and KMTB in the absence of methionine, Met-tRNA Met is synthesized, albeit at reduced efficiency ( Figure 5). Formation of Met-tRNA Met when the enzyme is supplied Met directly is 2.2 pmol/min, which is only~2-fold greater than the rate when Met is generated in the aminotransferase domain from KMTB and Phe (1.2 pmol/min). Aminoacylation from KMTB alone is negligible (0.04 pmol/min). The reaction kinetics of each domain were determined through either LC-MS or aminoacylation assays and are summarized in Table 1.

MpMetRS Aminotransferase Domain Can Supply tRNA Met With Methionine
The ability of MpMetRS to use KMTB as a transaminase substrate and the presence of fused aminotransferase and tRNA synthetase active sites led us to investigate whether the product of one domain could provide the substrate for the other. To test this hypothesis, we performed an aminoacylation assay in the presence of KMTB and phenylalanine with 32 P-labeled tRNA Met . When MpMetRS is supplied with methionine directly, Met-tRNA Met is formed, as previously shown. When the aminotransferase is supplied with phenylalanine and KMTB in the absence of methionine, Met-tRNA Met is synthesized, albeit at reduced efficiency ( Figure 5). Formation of Met-tRNA Met when the enzyme is supplied Met directly is 2.2 pmol/min, which is only ~2-fold greater than the rate when Met is generated in the aminotransferase domain from KMTB and Phe (1.2 pmol/min). Aminoacylation from KMTB alone is negligible (0.04 pmol/min). The reaction kinetics of each domain were determined through either LC-MS or aminoacylation assays and are summarized in Table 1.   Note: MpMetRS∆ kinetic parameters were previously reported by Jones et al. [12].

Discussion
While many AARSs contain appended domains that expand their functionality beyond tRNA aminoacylation [6], MpMetRS is the first known occurrence of an AARS exhibiting aminotransferase activity. This fusion protein is not observed in any of the other well-characterized members of the Mycoplasma spp. and is to our knowledge unique to M. penetrans MetRS.
Immunoblotting of M. penetrans cell lysates confirmed that the full-length protein does exist in vivo. The presence of both full-length and core MpMetRS suggests that both forms are biologically relevant, although the larger polypeptide is more abundant in the cell lysates analyzed here. It remains to be seen whether the ratio of full-length to core proteins varies with cellular conditions. The likelihood that an internal ribosomal start site is responsible for the two alternate forms upon overexpression in E. coli is strengthened by observation of a single polypeptide product in the M568A variant corresponding to full-length protein (Figure 1).
When three different versions of wild-type MpMetRS (MpMetRS, MpMetRS∆, and M568A) were used in an aminoacylation reaction, MpMetRS and M568A aminoacylate Mp-tRNA Met efficiently, while MpMetRS∆ is active but less so than the full-length enzymes ( Figure 2).
Expression of the core aminoacylation domain alone (MpMetRS∆) corresponds to a structural switch from dimer to monomer. The SEC-MALS analysis for MpMetRS∆ suggests a molecular weight of 77 kDa, somewhat higher than the calculated 61.4 kDa. The MpMetRS∆ sample did display greater scatter in generated molecular weights (Supplementary Materials Figure S1), suggesting that the aminoacylation domain alone may have lower structural stability. This observation is consistent with the decreased catalytic efficiency of MpMetRS∆. While the two enzymes have similar K M values for the Mp tRNA Met substrate, there is a notable difference in the chemical step of aminoacylation, represented by k cat (Table 1). While each domain can catalyze its reaction independently, as demonstrated by single substitutions at critical domain-specific positions (Figure 4), structural integrity may be necessary for full catalytic efficiency.
The aminotransferase domain of MpMetRS was shown to utilize more than one set of amine group donor and acceptor. While some aminotransferase enzymes are specific for a given amino acid-keto acid pair, others are more promiscuous in their substrate use [29][30][31][32]. Our observation that phenylalanine was the best in vitro amine group donor ( Figure 3A) was consistent with work of Berger and coworkers, who noted that aromatic and branched chain amino acids are often amine donors in methionine regeneration [33,34]. It was surprising to note that oxaloacetate was a better keto acid substrate for the aminotransferase activity ( Figure 3B); the optimal in vivo substrate remains unknown. It may be that the MpMetRS transaminase domain has been co-opted to catalyze Met synthesis in the context of the fused enzyme (keeping in mind the lack of substrate stringency for many aminotransferases) but has a primary function elsewhere as an aspartate transaminase.
We have shown here that the fused aminotransferase domain is capable of producing methionine that can be used by the core synthetase domain to synthesize Met-tRNA Met . Over the course of two minutes, the initial rate for Met-tRNA Met synthesis beginning with KMTB and Phe was half that observed with methionine added directly to the reaction ( Figure 5). For the net aminoacylation reaction, phenylalanine and KMTB were preincubated with enzyme for three minutes prior to tRNA addition to allow for methionine accumulation. Without this pre-incubation, there was a 15 s lag in product formation, although the rate following the lag period was not significantly different from the linear initial rate shown here.
Given the two distinct active sites on this novel enzyme, and the capacity for the synthetase domain to use methionine produced by the aminotransferase domain, the question of methionine transport arises. Crystal structures exist for isolated proteins homologous to these domains: E. coli MetRS (PDB 1QQT) and Anabaena sp. AGAT (PDB 1VJO) [21,34]. However, the orientation of the domains relative to one another and the arrangement of the monomers within the dimer are both unknown. We have considered whether methionine could migrate through substrate channeling in MpMetRS. Substrate channeling between distal enzyme active sites is best exemplified by amidotransferases such as glutamine phosphoribosylpyrophosphate (PRPP) amidotransferase (GPAT). This enzyme hydrolyzes the glutamine side chain amide and shields the NH 3 group from aqueous environment through a hydrophobic tunnel to the PRPP active site 16 Å away [35]. A rationale for the presence of substrate channeling is either to protect a labile substrate from hydrolysis or to prevent its use by other enzymes [36]. One evolutionary advantage to the fused MpMetRS could be the direct delivery of methionine for attachment to tRNA Met , while preventing enzymes like S-adenosylmethionine synthase from competing for its common substrate. The MpMetRS fusion protein could allow direct transfer of methionine from the aminotransferase domain to the aminoacylation domain, either within one monomer (intramolecular channeling) or across the dimer interface (intermolecular channeling). Alternatively, methionine may simply be released from the aminotransferase active site for use by a neighboring enzyme. While the existence of a bifunctional enzyme does not necessitate substrate channeling, there is precedent for monofunctional enzymes that exist as a fusion protein in another organism. One clear example is the thymidylate synthase (TS) and dihydrofolate reductase (DHFR) enzyme pair; in humans, they are isolated monofunctional enzymes, while in some protozoan species, they are bifunctional fusion proteins [37]. Substrate channeling has been demonstrated for some of the protozoan bifunctional TS-DHFR proteins. While we do not know if MpMetRS exhibits channeling, it has been suggested that synchronized reaction rates would facilitate channeling [38]. Our kinetic studies show that the reaction rates in the two domains are similar (k cat~6 s −1 ). Additionally, there was little benefit to providing the enzyme time to produce methionine prior to tRNA addition, suggesting that simple diffusion may not explain how the MetRS sequesters its methionine substrate. Future work includes structural investigations of the individual catalytic domains and the full protein fusion to understand whether substrate channeling is possible.
The question remains as to why M. penetrans would retain such a gene fusion in the context of a condensed genome lacking many key metabolic enzymes [39,40]. As these bacteria are obligate intracellular parasites, they obtain most molecular building blocks from their host rather than through canonical biosynthetic pathways; indeed, M. penetrans lacks methionine biosynthesis enzymes. Use of an aminotransferase would enable M. penetrans to increase the local concentration of methionine, using KMTB and an amine-donating amino acid. Synthesis of methionine from KMTB is the last step in the methionine salvage pathway; our data suggest that the N-terminal domain of MpMetRS is a transaminase for methionine production.
Recombination events during genomic erosion of mycoplasma species may have led to unique protein fusions like the one seen with MpMetRS. The essentiality of the MpMetRS appended domain remains unknown, despite the demonstration here that it possesses aminotransferase activity in vitro. Until recently, the extreme N-terminal 166 amino acids of MpMetRS did not align with any known proteins; however, a PHYRE2 structure-based search of the MpMetRS appended domain suggested homology to another fusion protein, phosphonate-specific cytidylyltransferase, and 2-aminoethylphosphonate transaminase (PDB 6PD1), from the bacterium Treponema denticola [41]. The 2-aminoethylphosphanate aminotransferase (AEPT) is characterized as a Class V aminotransferase, consistent with our earlier BLAST results aligning MpMetRS with the Anabaena spp. AGAT. It remains to be determined if the MpMetRS aminotransferase has an affinity for 2-aminoethylphosphonates or whether it possesses nucleotidyltransferase capabilities suggested by this new alignment. Future work will focus on structure determination of this novel fusion protein and its possible role in the metabolic life cycle of the human pathogen M. penetrans.

Conclusions
In conclusion, the results presented here demonstrate the first example of an AARS for which the canonical aminoacylation activity is augmented by aminotransferase function. The capacity for MpMetRS to generate its own methionine substrate from a metabolic precursor suggests an intriguing adaptation to environmental conditions. It remains to be determined whether this catalytic duality contributes to the M. penetrans life cycle and whether the acquisition of the aminotransferase domain is unique among AARS enzymes or may yet be identified in additional species.