Life without post-transcriptional addition of G-1: two alternatives for tRNAHis identity in Eukarya.

The identity of tRNA(His) is strongly associated with the presence of an additional 5'-guanosine residue (G-1) in all three domains of life. The critical nature of the G-1 residue is underscored by the fact that two entirely distinct mechanisms for its acquisition are observed, with cotranscriptional incorporation observed in Bacteria, while post-transcriptional addition of G-1 occurs in Eukarya. Here, through our investigation of eukaryotes that lack obvious homologs of the post-transcriptional G-1-addition enzyme Thg1, we identify alternative pathways to tRNA(His) identity that controvert these well-established rules. We demonstrate that Trypanosoma brucei, like Acanthamoeba castellanii, lacks the G-1 identity element on tRNA(His) and utilizes a noncanonical G-1-independent histidyl-tRNA synthetase (HisRS). Purified HisRS enzymes from A. castellanii and T. brucei exhibit a mechanism of tRNA(His) recognition that is distinct from canonical G-1-dependent synthetases. Moreover, noncanonical HisRS enzymes genetically complement the loss of THG1 in Saccharomyces cerevisiae, demonstrating the biological relevance of the G-1-independent aminoacylation activity. In contrast, in Caenorhabditis elegans, which is another Thg1-independent eukaryote, the G-1 residue is maintained, but here its acquisition is noncanonical. In this case, the G-1 is encoded and apparently retained after 5' end processing, which has so far only been observed in Bacteria and organelles. Collectively, these observations unearth a widespread and previously unappreciated diversity in eukaryotic tRNA(His) identity mechanisms.

INTRODUCTION

Recognition and aminoacylation of a specific tRNA by its cognate aminoacyl-tRNA synthetase (aaRS) is quantitatively important for error-free protein synthesis. tRNAs and aaRS utilize various identity elements that facilitate discrimination between substrate and nonsubstrate tRNA at a biochemical level, and are often necessary and sufficient for aminoacylation in vitro and in vivo (McClain 1993; Giege et al. 1998; Ardell 2010). Identity elements are commonly present in the primary sequence of a tRNA, but some are acquired post-transcriptionally in the form of extra nucleotides or modifications of existing nucleotides.

A critical identity element for histidyl-tRNA (tRNAHis) is a single additional guanosine residue at the −1 position (G−1) that is present in all three domains of life. Biochemical analyses of histidyl-tRNA synthetase (HisRS) from Escherichia coli and Saccharomyces cerevisiae revealed that G−1, and specifically the 5′-monophosphate on this additional residue, contributes substantially toward discrimination of tRNAHis from nonsubstrate tRNAs, although the discriminator nucleotide at position 73 and nucleotides of the GUG anticodon can further aid recognition (Himeno et al. 1989; Nameki et al. 1995; Rudinger et al. 1997; Rosen and Musier-Forsyth 2004; Rosen et al. 2006).

Two distinct biochemical pathways have been identified that lead to acquisition of G−1. In Bacteria and many Archaea, the G−1 residue is encoded in the tRNAHis gene sequence (Jühling et al. 2009). The presence of genomically encoded G−1 opposite the bacterial C73 leads to formation of an atypical 8-bp acceptor stem in pre-tRNAHis, causing alternative cleavage of the 5′-leader by Ribonuclease P (RNase P) and retention of the encoded G−1 in mature tRNAHis (Orellana et al. 1986). Alternatively, in Eukarya and other Archaea, the G−1 residue is not genomically encoded and instead is added after tRNA 5′ end processing by a unique 3′–5′ nucleotide addition reaction catalyzed by the tRNAHis guanylyltransferase (Thg1) (Cooley et al. 1982; Gu et al. 2003).

The occurrence of the G−1 residue is highly conserved in Bacteria, Archaea, and Eukarya, with two notable exceptions: (1) the α-proteobacterium Caulobacter crescentus, where the isolated tRNAHis lacks a G−1 and in turn the corresponding HisRS enzymes uses alternative identity elements for tRNAHis aminoacylation (Ardell and Andersson 2006; Wang et al. 2007; Yuan et al. 2011), and (2) the protist Acanthamoeba castellanii, where it was recently shown that the HisRS (AcaHisRS) readily aminoacylates a G−1-lacking tRNA in vitro (Rao et al. 2013). Despite the latter example representing the only such occurrence in a eukaryote described to date, neither the mechanistic basis for altered substrate specificity, nor the ability of the noncanonical A. castellanii synthetase to function independently of G−1 in vivo has been demonstrated.

Here we show that the alternative tRNAHis identity mechanism exhibited by A. castellanii is not a unique instance of noncanonical recognition among Eukarya. In Trypanosoma brucei, like A. castellanii, mature tRNAHis lacks the G−1 element, and the corresponding HisRS from T. brucei also catalyzes aminoacylation of tRNAHis independent of G−1. Through biochemical analysis, we demonstrate that the noncanonical tRNAHis recognition mechanisms exhibited by T. brucei and A. castellanii HisRS result from a substantial dependence on the tRNAHis anticodon for substrate recognition and do not require the 5′-monophosphate that is critical for identity in canonical G−1-dependent HisRS enzymes. Moreover, we identify a third completely distinct eukaryotic pathway for tRNAHis identity in Caenorhabditis elegans, wherein the tRNAHis retains G−1, but in this case the identity element is acquired by a non-Thg1-dependent pathway that predictably results from noncanonical 5′ end processing mechanisms. Unlike the α-proteobacterial case, which seems to have arisen from a common ancestral trait, the scattered distribution of organisms that possess noncanonical tRNAHis/HisRS suggests a polyphyletic origin for these activities, implying that these mechanisms may have arisen multiple independent times throughout evolution and that reliance on G−1 among eukaryotes is not as ubiquitous as previously accepted. Taking advantage of these newly described non-G−1-dependent HisRS enzymes, we also describe the construction of an S. cerevisiae genetic system that exhibits robust growth in the absence of the G−1-addition enzyme THG1, thus permitting future investigation into alternate functions of Thg1-catalyzed 3′–5′ polymerase activities in biology.

RESULTS

Mature tRNAHis in T. brucei lacks the G−1 nucleotide

Despite the widespread distribution of Thg1 family members throughout eukaryotes, a number of species do not encode an identifiable THG1 gene (Fig. 1; Jackman et al. 2012). Previous studies focused on A. castellanii, which lacks a bona fide Thg1 capable of adding G−1 to tRNAHis, but encodes two related family members, which are Thg1-like proteins (TLPs) that apparently function in other processes beyond tRNAHis metabolism (Abad et al. 2011; Rao et al. 2011). However, a number of other eukaryotes do not encode any identifiable gene with sequence similarity to either Thg1 or TLP, and thus seem to lack any enzyme capable of catalyzing 3′–5′ addition (highlighted in red in Fig. 1). We first chose to investigate the disease-causing parasite Trypanosoma brucei as a representative organism. To evaluate the G−1 status of T. brucei tRNAHis, we used a previously described approach to sequence the 5′ end of tRNAHis by circularization of isolated tRNA from T. brucei and amplification of the junction between the 5′- and 3′ ends of tRNAHis (Rao et al. 2013). Nine clones were obtained for otherwise fully mature tRNAs (containing the G+1 nucleotide ligated directly to A76 of the intact 3′-CCA end) (Fig. 2A, clones 1–9). The other seven clones contained additional nucleotides at the junction between G+1 and A76 that appear to correspond to aberrant processing intermediates (Fig. 2A, clones 10–16). The five clones with additional A, C, or CCA sequences (Clones 10–14) likely correspond to additional nucleotides added to the 3′ end by the CCA-adding enzyme. However, the possibility that the single additional C nucleotide observed in clone 13 is derived from incomplete processing at the 5′ end cannot be excluded, since the tRNAHis gene contains an encoded C−1 immediately adjacent to G+1. Interestingly, two clones (15 and 16) contained an additional U residue between the G+1 nucleotide and the mature 3′-CCA end, despite the fact that U is not derived from any annotated 5′-leader sequence or would not predictably result from aberrant activity of the CCA-adding enzyme. The lack of G−1 in any obtained clone (from either mature or alternatively processed tRNA) clearly indicates that T. brucei, like A. castellanii, lacks the highly conserved G−1 identity element on tRNAHis.

FIGURE 1.

Thg1 is not ubiquitous among eukaryotes. A phylogenetic tree to demonstrate Thg1 diversity in eukaryotes from the indicated supergroups was generated using PhyloT and visualized using the Interactive Tree of Life (iTOL) tool. Organism names indicated in color represent organisms that lack an identifiable Thg1 gene; species shown in blue lack any identifiable Thg1 protein, but encode Thg1-like proteins (TLPs) that catalyze RNA repair reactions; species shown in red lack any gene with identifiable similarity to Thg1 (either Thg1 or TLPs).

FIGURE 2.

Recognition of tRNAHis in T. brucei does not require the G−1 identity element. (A) tRNA 5′- and 3′-end sequences of 16 independent clones obtained by circularization and RT-PCR using primers targeting the T. brucei tRNAHis shown on the diagram. Additional nucleotides observed between A76 and G+1 in some clones (presumably derived from alternative processing events) are shown in gray italics. (B) In vitro aminoacylation assays with purified HisRS enzymes from S. cerevisiae (Sce), A. castellanii (Aca), and T. brucei (Tbr), enzymes were used in serial dilutions from ∼125 to 5 ng/µL. Reactions contained S. cerevisiae tRNAHis substrates either lacking (Δ) or containing (G−1) the G−1 residue as indicated. Lanes indicated by – are no enzyme controls. Positions of radiolabeled 5′-32P-A76 derived from unreacted substrate (p*A) and 5′-32P-A76 with the ester-linked amino acid (p*A-His), derived from products, are as indicated.

Noncanonical eukaryotic HisRS enzymes do not depend on G−1 for aminoacylation either in vitro or in vivo

We tested whether purified T. brucei (TbrHisRS) requires G−1 to recognize and aminoacylate its tRNAHis substrates in vitro by performing aminoacylation assays with tRNAHis transcripts containing (G−1) or lacking (Δ) the G−1 nucleotide (Wolfson et al. 1998). Purified TbrHisRS displayed a clear lack of dependence on the presence of G−1 for tRNAHis aminoacylation (Fig. 2B). The ability of TbrHisRS to aminoacylate tRNAHis independent of G−1 in vitro is consistent with the lack of G−1 observed on isolated tRNAHis from T. brucei, and therefore another noncanonical HisRS/tRNAHis pair, in addition to the previously described pair in A. castellanii, is identified in Eukarya.

Although the observed in vitro activities of TbrHisRS and AcaHisRS are consistent with the lack of G−1 on isolated tRNAHis from each organism, these altered tRNAHis recognition properties had not been demonstrated in a biological system with fully modified tRNA. In S. cerevisiae Thg1 is essential for normal growth, due to the strong dependence of SceHisRS in the presence of the G−1 residue for tRNAHis recognition. The requirement for Thg1 for viability can only be bypassed by overexpression of both SceHisRS and its tRNAHis substrate, although these strains still exhibit a growth defect compared with THG1-containing yeast strains (Preston and Phizicky 2010). This suggests an epistasis between Thg1 function and tRNAHis aminoacylation, and that the dependence on Thg1 could be bypassed by expression of a G−1-independent HisRS in S. cerevisiae. To test this, we transformed plasmids for expression of AcaHisRS or TbrHisRS into the previously described S. cerevisiae thg1Δ strain in which viability is maintained by the presence of wild-type THG1 on a URA3 plasmid (Gu et al. 2003; Abad et al. 2010). Growth on media containing 5-fluoroorotic acid (5-FOA) causes selective loss of the THG1-expressing URA3 plasmid and in the absence of another source of THG1 (such as in the vector control strain), cells selected on 5-FOA are not viable (Fig. 3A).

FIGURE 3.

Genetic complementation of thg1Δ in S. cerevisiae. (A) The thg1Δ yeast strain JJY20 (see Materials and Methods for relevant genotype) was transformed with either empty CEN LEU2 plasmid (Vector) or the same plasmid containing the indicated genes (S. cerevisiae THG1 or HisRS enzymes from S. cerevisiae [SceHisRS], A. castellanii [AcaHisRS], T. brucei [TbrHisRS], or C. elegans [CelHisRS]) under control of a galactose-inducible promoter. Independent transformants (A and B) from each strain were tested by replica plating on the indicated media; pictures represent growth after 4 d at 30°C. (B) The thg1Δ strain JJY20 was transformed with the same CEN LEU2 plasmids (empty vector or TbrHisRS) used in A and with a 2µ HIS3 plasmid containing tRNAHis genes (+200-bp flanking sequence) from either yeast (Sce) or T. brucei (Tbr), as indicated. Independent transformants (A and B) were tested by replica plating on the indicated media; pictures represent growth after 5 d at 30°C. (C) Analysis of 5′ end status of tRNAHis by primer extension of total RNA isolated from Thg1-complemented or AcaHisRS-complemented cells. Reactions used labeled primer (indicated by starred line on tRNA diagram) corresponding to the D-loop and stem of S. cerevisiae tRNAHis. Position of the unextended primer is shown in the lane with no added RNA (–); positions of the primer extension stop corresponding to tRNA with G−1 and without G−1 (5′ end at G+1) are indicated with arrows. (D) Western blot to test for expression of the indicated enzymes in JJY20 (thg1Δ) yeast transformed with CEN LEU2 plasmids expressing either S. cerevisiae Thg1 (Thg1) or various HisRS enzymes from A. castellanii (Aca), T. brucei (Tbr), and C. elegans (Cel). Antibodies targeted N-terminal Flag epitopes incorporated into each construct and were used to probe blots from crude lysates prepared from each strain. Expected sizes for the indicated enzymes are as follows: Thg1 (29 kDa); SceHisRS (59 kDa); AcaHisRS (55 kDa); TbrHisRS (54 kDa); CelHisRS (60 kDa). The reason for the second (lower) apparent molecular weight band observed in strains expressing CelHisRS is not known. Positions of molecular weight standards are as shown on the left of the image, determined with commercial dye-labeled standards.

As hypothesized, S. cerevisiae thg1Δ strains expressing AcaHisRS grow similarly on 5-FOA-containing media to the SceThg1-expressing control strain, thus indicating that AcaHisRS genetically complements SceThg1 function in vivo (Fig. 3A). To ensure that the observed growth of AcaHisRS-complemented yeast is not due to the presence of residual G−1-containing tRNAHis, we isolated total RNA from the AcaHisRS-complemented strain and analyzed the 5′ end of tRNAHis by primer extension, which revealed a lack of any detectable G−1 on the tRNA (Fig. 3C). Thus, AcaHisRS aminoacylates tRNAHis in the absence of G−1, both in vitro and in vivo in yeast. Interestingly, TbrHisRS failed to complement yeast growth in the presence of 5-FOA, suggesting that these cells still require Thg1 for viability (Fig. 3A). The observed phenotype was not due to lack of TbrHisRS expression; indeed AcaHisRS and TbrHisRS were expressed to similar levels in the complemented strains as judged by Western blots targeting a Flag epitope added to each protein (Fig. 3D).

To determine whether there was a biochemical explanation in terms of catalytic efficiency for the difference in ability of AcaHisRS and TbrHisRS to complement the thg1Δ phenotype, we compared steady-state kinetic parameters for aminoacylation of ScetRNAHis. Steady-state assays were performed at saturating concentrations of histidine and ATP with 32P-labeled A76 to visualize the extent of aminoacylation. Since previous kinetic characterization of SceHisRS utilized a different (charcoal filter) assay with radiolabeled amino acid, the steady-state kinetic parameters for SceHisRS were also determined under the same buffer conditions reported previously for G−1-containing tRNAHis (Rosen et al. 2006). The KM,tRNA for SceHisRS observed with this assay was <0.3 µM (reflecting the lowest concentration of tRNA that could be tested in the assay, see Supplemental Fig. S1) compared with 0.14 µM measured earlier. The measured kcat is ∼60-fold lower, as expected since 5′-triphosphorylated tRNAHis transcripts were used here and SceHisRS displays a preference for 5′-monophosphorylated tRNAHis (Table 1; Rosen et al. 2006). However, since neither AcaHisRS nor TbrHisRS exhibit a similar phosphate preference (see below), these assays utilized the 5′-triphosphorylated form of each tRNA that is readily produced by in vitro transcription. The measured catalytic efficiency (as reflected by kcat/KM) of aminoacylation of ScetRNAHis by AcaHisRS and TbrHisRS was very similar to that observed with SceHisRS and its optimal (G−1-containing) substrate, regardless of the presence of the G−1 residue, and AcaHisRS even displays an approximately fivefold preference for the ΔG−1 tRNAHis (Table 1; Supplemental Fig. S1).

TABLE 1.

Steady-state kinetic parameters for aminoacylation by HisRS enzymes

Despite similar overall efficiencies of aminoacylation (as judged by kcat/KM), TbrHisRS exhibited a substantially (∼20–50-fold) higher KM than either SceHisRS or AcaHisRS for S. cerevisiae tRNAHis (Table 1; Supplemental Fig. S1). We reasoned that the inability of TbrHisRS to complement the yeast thg1Δ phenotype could be attributed to a preference for features of its native T. brucei tRNAHis. Indeed, we observed that kcat/KM of TbrHisRS for TbrtRNAHis was ∼84-fold higher as compared with that observed with ScetRNAHis, and importantly, that the KM,tRNA was substantially decreased (Table 1; Supplemental Fig. S1). Thus, we hypothesized that overexpression of this preferred tRNA in yeast might confer the ability to complement the yeast thg1Δ phenotype. We transformed the same yeast strain used for the previous complementation tests with a 2µ (multiple copy) vector that encodes either ScetRNAHis or TbrtRNAHis, and again tested growth on media containing 5-FOA. Interestingly, TbrHisRS supported growth of the yeast thg1Δ strain in transformants that overexpressed either ScetRNAHis or TbrtRNAHis (Fig. 3B). Thus, consistent with the elevated KM,tRNA values (Table 1), TbrHisRS may require higher levels of tRNAHis for efficient aminoacylation, but again expression of the noncanonical HisRS can complement the loss of THG1.

C. elegans contains a genomically encoded G−1 on tRNAHis

Although numerous examples of eukaryotes that lack Thg1 are found among lower eukaryotic genomes (Jackman et al. 2012), among well-sequenced metazoa there is only a single organism identified so far that lacks an identifiable Thg1 enzyme in its genome, which is Caenorhabditis elegans (Fig. 1). Given the previous observations with A. castellanii and T. brucei, we tested whether the lack of a Thg1 enzyme in C. elegans could be accommodated by a similar mechanism that does not require the presence of the G−1 identity element on tRNAHis. However, C. elegans HisRS (CelHisRS) did not support growth in the absence of SceThg1 (Fig. 3A), and moreover, purified CelHisRS exhibits a detectable dependence on G−1, although unlike SceHisRS, CelHisRS exhibits a small amount of activity even with the ΔG−1 substrate (Fig. 4A).

FIGURE 4.

C. elegans utilizes an encoded G−1 identity element for tRNAHis recognition. (A) In vitro aminoacylation assays with purified HisRS enzymes from S. cerevisiae (Sce) and C. elegans (Cel); enzymes were used in serial dilutions from ∼125 to 5 ng/µL. Reactions contained S. cerevisiae tRNAHis substrates either lacking (Δ) or containing (G−1) the G−1 residue as indicated. Lanes indicated by – are no enzyme controls. (B) tRNA 5′- and 3′-end sequences of 16 independent clones obtained by circularization and RT-PCR using primers targeting the C. elegans tRNAHis shown on the diagram. The additional G−1 residue observed in 13 clones is underlined; nucleotides corresponding to identifiable 3′ end trailer sequences in some clones are shown in gray italics. The N7–N66 base pair that distinguishes different tRNAHis genes is also highlighted for each sequence.

To directly address the nature of tRNAHis in C. elegans, 5′ end sequencing of circularized C. elegans tRNAHis was performed, revealing that the G−1 nucleotide is indeed present, since 12/12 clones derived from mature tRNA contained a G−1 (Fig. 4B, clones 1–12). Among these 12 clones, multiple tRNAHis genes are represented, since sequences were obtained that contain either G7–C66 base pairs (encoded by 11 of the 17 C. elegans tRNAHis genes) or U7–A66 base pairs (encoded by the remaining six genes). Thus, the sample of sequences obtained here appears to be representative of the overall tRNA pool in C. elegans and not just derived from a small subset of G−1-containing tRNAHis. These data are consistent with widespread occurrence of G−1 on tRNAHis in C. elegans, which would be recognized by a canonical G−1-dependent HisRS.

Interestingly, among the 17 annotated tRNAHis genes in C. elegans, we noted that 16 contain a genomically encoded guanosine at −1 position in the leader sequence of tRNAHis, thus suggesting that a G−1 residue could be incorporated during transcription of the pre-tRNA. No prior examples of a similar genomically encoded G−1 have been documented in any nucleus-encoded eukaryotic tRNAHis gene, and the lack of the typical Thg1-dependent mechanism for acquisition of G−1 appears to have been compensated for by encoding the G−1 nucleotide in the precursor transcript, which is unprecedented in Eukarya.

Four sequences were obtained from immature tRNAHis species, all of which correspond to tRNA with matured 5′ ends but unprocessed 3′ ends (Fig. 4B). One sequence corresponds to the product of tRNA gene Y40H7A.t1 (Clone 13), which is clearly identifiable based on its unique 3′-trailer sequence (ACUAUUUU). As with the tRNAs described above, this gene also encodes a G−1 nucleotide that is detected in the immature tRNA. The other three immature tRNA sequences (Clones 14–16) were all clearly attributable to another unique gene (B0513.t1), which is the only one of the 17 tRNAHis genes that does not encode G−1. The observation of clones derived from this gene confirms that the tRNA is transcribed, but the fact that no fully mature sequences (after 3′ end processing) were obtained raises questions about whether this single tRNA species that apparently lacks G−1 is functional. From the limited number of clones obtained for this analysis, we cannot rule out the possibility that there is also a pool of mature tRNAHis derived from this gene (and presumably lacking G−1), but the abundance of other mature clones that contain the G−1 nucleotide suggests that these would be minor species in the cell.

Distinct mechanism of tRNAHis recognition by AcaHisRS and TbrHisRS

Since AcaHisRS and TbrHisRS clearly lack a dependence on the canonical G−1 element, we investigated the biochemical basis for tRNAHis recognition. Since tRNA identity is often associated with the anticodon, we tested the role of the GUG anticodon using tRNAHis variants in which the GUG anticodon was replaced with the GAA sequence of tRNAPhe (Jackman and Phizicky 2006a). Both AcaHisRS and TbrHisRS are strongly dependent on the tRNAHis anticodon, GUG (Fig. 5A,B). Interestingly, transplantation of the GUG anticodon into an otherwise wild-type tRNAPhe supported some, albeit reduced, acquisition of aminoacylation activity with AcaHisRS, but a similar effect was not observed for TbrHisRS.

FIGURE 5.

G−1-independent HisRS from A. castallanii and T. brucei depend on the presence of the tRNAHis anticodon but do not require a tRNA 5′-monophosphate. In vitro aminoacylation assays were performed with serial dilutions (∼125–5 ng/µL) of AcaHisRS (A) or TbrHisRS (B) with the indicated S. cerevisiae tRNA variants. Lanes indicated with – are no enzyme controls; p*A and p*A-His represent substrate and product of the reactions, as described above. C demonstrates the classical 5′-monophosphate dependence observed with SceHisRS (serial dilutions from 125 to 5 ng/µL) and S. cerevisiae tRNAHis substrate containing the indicated 5′-phosphorylation status; ppp, 5′-triphosphate; p, 5′-monophsophate; -, 5′-hydroxyl. (D) In vitro aminoacylation of the same 5′-phosphate variants analyzed in C with AcaHisRS (serial dilutions from 125 to 0.125 ng/µL).

We noted that native TbrtRNAHis contains an unusual U73 discriminator nucleotide, which is not found in other annotated tRNAHis species (Fig. 2A). However, alteration of the wild-type A73 discriminator to U73 in the context of ScetRNAHis resulted in only modest, if any, improvement in kcat/KM for TbrHisRS aminoacylation, and the KM remained high for this ScetRNAHis variant (Table 1). Although an effect of the U73 discriminator in conjunction with other elements of TbrtRNAHis cannot be ruled out, U73 alone is not a sufficient identity element for recognition by TbrHisRS, and additional elements remain to be identified.

The 5′-monophosphate on G−1, more so than the guanine base itself, was shown to be critical for tRNAHis recognition through previous kinetic analysis of canonical HisRS enzymes (Rosen and Musier-Forsyth 2004; Rosen et al. 2006). Therefore, an alternate interaction that similarly positions the 5′-monophosphate in the HisRS active site, even in the absence of G−1, could explain the altered G−1 dependence exhibited by noncanonical AcaHisRS and TbrHisRS. To test this, we compared aminoacylation activity of the various HisRS enzymes with the in vitro transcribed ScetRNAHis substrate, which naturally contains a 5′-triphosphate (ppp), to the activities observed with the same tRNA treated with either tobacco acid pyrophosphatase (TAP), which cleaves the α-β phosphate linkage to yield a 5′-monophosphate (p), or calf intestinal phosphatase (CIP), which removes all terminal phosphates and yields a 5′-OH end (-). Control activity assays with SceHisRS recapitulated the previous observation of optimal activity with 5′-monophosphorylated tRNAHis, followed by reduced activity with the ppp-tRNA and little or no detectable aminoacylation of the OH-tRNA under these conditions (Fig. 5C).

However, with both AcaHisRS (Fig. 5D) and TbrHisRS (Supplemental Fig. S2), aminoacylation assays revealed a strikingly different pattern of activity, with neither enzyme dependent on the number, or even the presence of any phosphates on the 5′ end of tRNAHis. The observed dependence of the amount of aminoacylated product on the concentration of each enzyme included in the reactions is nearly identical, suggesting that the overall efficiency of aminoacylation is similar for the three tRNAs. These observations are in stark contrast to the role of 5′-terminal phosphorylation of tRNAHis for E. coli and SceHisRS and underscore the distinct nature of the molecular mechanism used to interact with tRNAHis by the noncanonical AcaHisRS and TbrHisRS.

A. castellanii encodes both canonical and noncanonical HisRS enzymes

Eukaryotes can encode separate tRNAHis genes in nuclear and organellar genomes, raising the possibility of different mechanisms of 5′-processing to yield G−1 in the two compartments, as is observed in S. cerevisiae where G−1 is encoded in the mitochondrial tRNAHis and added by Thg1 to the nucleus-encoded tRNA (Burkard and Söll 1988; L'Abbé et al. 1990; Marechal-Drouard et al. 1996b). Although T. brucei does not encode any mitochondrial tRNAs (mt-tRNAs) and therefore utilizes nucleus-encoded tRNA for mitochondrial translation, A. castellanii also encodes a tRNAHis gene in its mitochondrial genome. Interestingly, despite the lack of G−1 in the nucleus-encoded tRNAHis, A. castellanii mt-tRNAHis encodes a G−1 opposite a C73 discriminator nucleotide, and thus predictably retains the G−1 residue on the mt-tRNA during 5′ end processing. Accordingly, the A. castellanii nuclear genome encodes two distinct HisRS, only one of which has an identifiable mitochondrial targeting sequence and hence is predicted to localize to the mitochondria. Apart from the typical highly conserved catalytic residues, sequence comparison of the cytosolic and mitochondrial HisRSs from A. castellanii reveals relatively low similarity, indicating that they are most likely not products of gene duplication (Fig. 6).

FIGURE 6.

Multiple sequence alignment of HisRS enzymes. The HisRS sequences from E. coli (E.co), C. crescentus (C.cr), A. castellanii predicted mitochondrial HisRS- (A.ca(m)), T. brucei (T.br), A. castellanii predicted cytosolic HisRS (A.ca(c)), S. cerevisiae (S.ce), H. sapiens (H.sa), D. discoideum (D.di), and C. elegans (C.el) were aligned using Clustal Omega (Sievers et al. 2011; Sievers and Higgins 2014). Coloring of organism labels indicates whether the HisRS is canonical G−1-dependent (blue) or noncanonical G−1-independent (red). The black line separates HisRS enzymes derived from bacteria/organelles (above the line) and eukaryotic cytosol (below the line). Conserved motifs I and II and HisA and HisB boxes are indicated by blue lines above the alignment; the motif II amino acids that interact with G−1 in bacterial HisRS are colored in red on the alignment. The numbers in parenthesis after each organism label indicate the number of N-terminal amino acids for each enzyme that are not shown on this alignment. Sequence alignment image was prepared using GeneDoc.

To investigate the possibility that two distinct HisRS enzymes with different tRNAHis specificity are expressed in the same organism, we tested the biochemical dependence of the purified A. castellanii mitochondrial HisRS (Acamt-HisRS) on the G−1 identity element. For these assays, we used a transcript based on mitochondrial tRNAHis (either cloned from A. castellanii [Acamt-tRNAHis] or S. cerevisiae [Scemt-tRNAHis]), since Acamt-HisRS was not active with cytosolic ScetRNAHis transcripts (Fig. 7A). Indeed, purified Acamt-HisRS exhibited a preference for G−1-containing tRNAHis with either mt-tRNA substrate (Fig. 7B,C). As reported previously, SceHisRS also readily aminoacylated the mitochondrial tRNAHis even in the absence of G−1 (Fig. 7C; Su et al. 2011).

FIGURE 7.

Role of G−1 as tRNAHis identity element for mitochondrial HisRS. All in vitro aminoacylation assays were performed with A76 32P-labeled tRNAHis substrates which either lacked (Δ) or contained (G−1) a G−1. (A) Aminoacylation of yeast cytosolic tRNAHis was tested with A. castellanii mitochondrial HisRS (Acamt-HisRS) using yeast HisRS (Sce) as control. Effect of G−1 on aminoacylation of mitochondrial tRNAHis substrates by S. cerevisae HisRS (Sce) and A. castellanii mitochondrial HisRS (Acamt-HisRS) was tested using mitochondrial tRNAHis from (B) A. castellanii (Acamt-RNAHis) and (C) S. cerevisiae (Scemt-RNAHis). Assays contained 125–5 and 250–10 ng/µL of Sce and Acamt-HisRS, respectively.

DISCUSSION

The critical nature of the G−1 residue for recognition of tRNAHis and the highly conserved nature of the domain-specific pathways that lead to incorporation of this identity element are widely acknowledged (Rudinger et al. 1997; Giege et al. 1998; Heinemann et al. 2012). However, a deeper investigation of the eukaryotic family tree revealed multiple diverse organisms in which these rules for specifying tRNAHis identity are not universally followed (Fig. 1). On the basis of G−1 occurrence and biochemical properties of HisRS enzymes, we propose that tRNAHis/HisRS pairs can now be broadly divided into at least three classes. These include (1) the well-characterized scenario where G−1 is present and there is a canonical HisRS (e.g., E. coli and S. cerevisiae), (2) the scenario where G−1 is absent and a noncanonical HisRS ensures tRNAHis identity (e.g., C. crescentus, A. castellanii, and T. brucei), and finally, (3) the case where G−1 is present, but it is obtained by an atypical mechanism (e.g., C. elegans). The practically ubiquitous presence of an encoded G−1 residue throughout tRNAHis gene sequences in Bacteria suggests that the first scenario remains the overwhelming rule in this domain. Although only a limited number of reports detail the actual G−1 status of tRNAHis from bacterial species, the ∼20 α-proteobacteria from the RCS clade that lack this encoded residue remain the only exceptions to this rule identified to date (Wang et al. 2007).

However, in Archaea and Eukarya, the data described here suggest the possibility of significantly more variability. A number of archaeal species from both euryarchaeal and crenarchaeal clades contain an encoded G−1 and C73 that would predictably lead to cotranscriptional G−1 incorporation as in Bacteria, but a roughly equivalent number of available archaeal genomes lack this encoded G−1. Thus, if present, G−1 must be acquired by an alternative pathway in these organisms. Interestingly, members of the Thg1 enzyme family (Thg1-like proteins, or TLPs) are found in Archaea that lack an encoded G−1, and could thus function analogously to Thg1 in tRNAHis metabolism (Abad et al. 2010). However, archaeal TLPs are also found in species that contain the encoded G−1, raising questions about their function in these species, and moreover, a substantial biochemical preference for catalyzing tRNA 5′ end repair has now been associated with TLPs, including several from Archaea (Rao et al. 2011).

In Eukarya, these newly identified mechanisms for specifying tRNAHis identity indicate a more diverse landscape than could have been previously appreciated from the study of a relatively limited number of model organisms. Most notably, in contrast to the single isolated lineage of G−1-independent HisRS enzymes observed in the RCS clade of α-proteobacteria, noncanonical HisRS enzymes are found sporadically throughout the eukaryotic family tree, and in distantly related eukaryotes, suggesting that differences in tRNAHis recognition may have arisen multiple times throughout evolution (Fig. 1). While a fully quantitative picture of the relative proportions of each type of tRNAHis/HisRS pair in Eukarya awaits further investigation, examples of species that lack an identifiable Thg1 homolog are readily identified among all taxa. Therefore, it is reasonable to assume that additional eukaryotic species exhibit alternative mechanisms for tRNAHis recognition that may be similar to, or even different from, the mechanisms identified here.

These data also constitute the first biochemical investigation of the basis for tRNAHis recognition by a G−1-independent eukaryotic HisRS. Although in the absence of G−1, the tRNAHis anticodon clearly contributes to tRNAHis recognition, additional identity elements remain to be identified, since little to no aminoacylation was observed by either Tbr or AcaHisRS with the tRNAPhe chimera containing the tRNAHis anticodon, GUG (Fig. 5). Interestingly, while T. brucei tRNAHis contains an unusual U73 nucleotide at the discriminator position, this relatively conspicuous feature does not uniquely contribute to recognition by noncanonical TbrHisRS (Table 1). Moreover, the currently accepted model for tRNAHis binding to HisRS enzymes is based almost entirely on the need to correctly position the tRNA 5′-monophosphate in the active site; indeed the identity of the actual nucleotide base found at the −1 position of tRNAHis contributes quantitatively less to tRNA recognition than does the phosphorylation status at the tRNA 5′ end (Hawko and Francklyn 2001; Rosen and Musier-Forsyth 2004; Rosen et al. 2006). Moreover, the fact that this critical role for the 5′-monophosphate is shared by E. coli and S. cerevisiae HisRS led to the idea that this is an evolutionarily conserved pathway, despite local differences in G−1:N73 recognition between enzymes from different domains of life. Again, the lack of dependence on the 5′-phosphorylation status observed here reveals that a completely different manner of tRNA recognition occurs with G−1-independent HisRS enzymes, and that the basis for tRNA selectivity remains to be fully determined.

Although the noncanonical HisRS enzymes investigated in this work share some biochemical similarities with the α-proteobacterial HisRS from C. crescentus (CcrHisRS), there are some important differences that distinguish these two types of HisRS (Ardell and Andersson 2006; Yuan et al. 2011). As with AcaHisRS and TbrHisRS, the nucleotides of the anticodon contribute greatly to recognition by CcrHisRS, and single alterations to these residues decrease kcat/KM for CcrHisRS by 100–1000-fold (Yuan et al. 2011). However, CcrHisRS also strongly prefers the presence of a G+1–U72 wobble base pair, and conversion to a Watson–Crick +1–72 pair (as is found in eukaryotic tRNAHis, including in A. castellanii and T. brucei) causes a substantial decrease in kcat/KM for CcrHisRS. Furthermore, the presence of a G−1 residue is actually somewhat detrimental for the efficiency of aminoacylation by CcrHisRS, whereas aminoacylation by AcaHisRS and TbrHisRS occurs with similar catalytic efficiency in the absence or presence of G−1. Moreover, AcaHisRS and TbrHisRS enzymes differ from CcrHisRS in the motif (motif II) that is associated with recognition of G−1, and specifically do not have an analogous residue to the G118 residue of CcrHisRS demonstrated to play an important role in recognition of tRNAHis without G−1 (Fig. 6; Ardell and Andersson 2006; Ardell 2010). Thus, although a reduced dependence on G−1 superficially appears to be a common theme for substrate recognition between AcaHisRS, TbrHisRS, and CcrHisRS, this detailed comparison of their respective protein sequences and biochemical preferences for tRNAHis substrates suggests that these are distinct and likely independently evolved mechanisms. Notably, A. castellanii and T. brucei are both human pathogens for which treatment has proven to be challenging. A better understanding of the mechanistic features that distinguish these enzymes may lead to the development of novel compounds that could selectively target HisRS from these pathogens.

The identification of the non-Thg1-dependent pathway for acquisition of G−1 raises interesting questions about the nature of tRNAHis processing in C. elegans, and presumably other Caenorhabditis species, which similarly lack Thg1. To date, the only mechanism that allows retention of a genomically encoded G−1 residue is the bacterial pathway that depends on the presence of a C73 discriminator nucleotide capable of base-pairing with the encoded G−1, thus creating an alternative cleavage site for the 5′-maturation enzyme RNase P (Burkard et al. 1988). However, C. elegans tRNAHis retains the universal eukaryotic A73 discriminator nucleotide and thus G−1 cannot participate in a similar Watson–Crick interaction (Fig. 4B). If C. elegans G−1 is retained after 5′ end processing, this would require an alternative pathway in which either RNase P or another 5′-nuclease exhibits an unusual cleavage specificity that accommodates a non-Watson–Crick G−1:A73 base pair at the cleavage site. Although C. elegans is the only eukaryotic system in which a Thg1-independent mechanism for G−1 acquisition has been observed to date, other eukaryotes (such as Bigelowiella natans) similarly lack a Thg1 enzyme and yet encode a G residue in the −1 position of their nuclear tRNAHis genes (Chan and Lowe 2009). Therefore, as with the noncanonical HisRS/tRNAHis pairs, this pathway is likely to account for tRNAHis identity in a number of eukaryotes.

Based on the widespread observation of an encoded G−1 (and bacteria-like C73) in organellar tRNAHis genes, the G−1 residue was predicted to serve as an identity element not only in the cytosol, but also for organellar-encoded tRNAHis. In A. castellanii, we have identified the first example of both modes of tRNAHis recognition operating in the same organism, with two distinct HisRS enzymes that exhibit different G−1-dependence encoded in the nuclear genome. These observations indicate a division of labor in A. castellanii, wherein two independent and coexisting mechanisms for tRNAHis identity have evolved that are strictly compartmentalized. Inconsistencies in the occurrence of even the encoded G−1 on tRNAHis have been previously reported in plant mitochondria and chloroplast (Marechal-Drouard et al. 1996a,b; Placido et al. 2005, 2010).

The yeast genetic system in which the requirement for the tRNAHis-related function of Thg1 has been bypassed by expression of a G−1-independent AcaHisRS provides an exciting opportunity to study other potential functions for Thg1 enzymes in eukaryotes. The original observation of the Watson–Crick templated 3′–5′ polymerase activity catalyzed by SceThg1 was at odds with the fact that Thg1 in yeast (and other eukaryotes) was apparently only required to add a single G−1 nucleotide opposite A73 in tRNAHis (Jackman and Phizicky 2006b). The reason for eukaryotic Thg1 enzymes to have even retained the ability to recognize a Watson–Crick base pair was not understood. Previous studies that bypassed the essential requirement for Thg1 by overexpression of HisRS and tRNAHis were complicated by the fact that there was still a strong growth defect evident in this strain, and whether this was due to incomplete aminoacylation or other defects associated with loss of Thg1 could not be completely decoupled (Preston and Phizicky 2010). However, the AcaHisRS-complemented thg1Δ strain grows similarly to the THG1-expressing strain under these conditions, and therefore will allow investigation of alternative function(s) of Thg1 enzymes, including those that may involve 3′–5′ polymerase activity. Indeed, several unexplained cell cycle defects have been associated with depletion of Thg1 in S. cerevisiae and humans, and a biochemical interaction between SceThg1 and the Orc2 component of the origin-recognition complex is not understood (Guo et al. 2004; Rice et al. 2005). It is intriguing to speculate that these effects may not be completely explained by a simple role for Thg1 solely in tRNAHis maturation.

MATERIALS AND METHODS

tRNAHis 5′ end sequencing

5′-End sequencing of tRNAHis was performed by circularization of total RNA isolated from T. brucei and C. elegans, using T4 RNA ligase, followed by reverse transcription and PCR-amplification of the ligated junction, as described previously (Rao et al. 2013).

Cloning and purification of HisRS and tRNAHis

HisRS from T. brucei (TbrHisRS) was cloned by RT-PCR from total T. brucei RNA (generously provided by Juan Alfonzo, Ohio State University). S. cerevisiae HisRS (HTS1; SceHisRS) was PCR amplified from a clone generously provided by Eric Phizicky, University of Rochester. Genes for A. castellanii mitochondrial HisRS (Acamt-HisRS) and C. elegans HisRS (CelHisRS) were PCR amplified from synthetic constructs (Genewiz, PA). PCR products were all cloned using ligation-independent cloning in to the previously described AVA421 for expression as N-terminal His6-tagged proteins (Quartley et al. 2009). All HisRS constructs were overexpressed at 18°C in E. coli BL21(DE3) pLysS and purified by metal-ion affinity chromatography using the protocol described in Abad et al. (2011). Purified proteins were dialyzed against buffer containing 50% glycerol, 20 mM Tris (pH 7.5), 4 mM MgCl2, 0.05 M NaCl, 1 µM ethylenediamine tetraacetic acid (EDTA) and 1 mM dithiothreitol (DTT).

tRNAHis genes (T. brucei tRNAHis and A. castellanii/S. cerevisiae mitochondrial tRNAHis were cloned by either RT-PCR from total RNA (T. brucei and A. castellanii) or PCR from genomic DNA (S. cerevisiae) for in vitro transcription by T7 RNA polymerase. Constructs with and without G−1 were generated by the use of alternate 5′-end primers during amplification. In vitro transcription reactions were performed with linearized DNA plasmids digested for run-off transcription and transcripts were purified on denaturing acrylamide gels, followed by phenol extraction and ethanol precipitation.

Yeast strains, media, and complementation

HisRS genes (as described above) were PCR-amplified and cloned into a vector for expression under control of a galactose-inducible promoter with an N-terminal Flag epitope [CEN LEU2 P] and transformed into yeast strain JJY20 (relevant genotype: thg1Δ::kanMX his3-1 leu2Δ ura3Δ [CEN URA3 P]). The plasmid for expression of T. brucei tRNAHis in S. cerevisiae was generated by PCR amplification of the T. brucei tRNAHis gene (with 100-bp upstream and downstream sequences) from T. brucei genomic DNA and restriction cloning into the yeast 2µ YEpLAC vector [2µ HIS3 TbrtRNAHis]. YEpLAC and yeast tRNAHis-expressing [2µ HIS3 ScetRNAHis] vectors were generously provided by Eric Phizicky. Growth was tested by replica plating and in liquid at 30°C in media containing either 2% dextrose or galactose and 0.1% 5-fluoroorotic acid (FOA), if indicated.

tRNA labeling and aminoacylation assays

tRNAHis substrates were radiolabeled with 32P at the 5′-phosphate of A76 by the activity of tRNA nucleotidyltransferase as described (Wolfson et al. 1998). Aminoacylation assays were performed and reaction products were resolved as described previously (Rao et al. 2013). Steady-state parameters were determined using a minimum of fivefold excess of [tRNAHis]/[HisRS], in the presence of saturating concentrations of histidine (40 µM) and ATP (2.5 mM). Linear initial rates were determined at varied concentrations of tRNAHis (0.3–30 µM), plotted as a function of [tRNAHis] and fit to the Michaelis–Menten equation using Kaleidagraph (Synergy software). To generate 5′-monophosphorylated and 5′-hydroxyl tRNAHis substrates, A7632P-labeled tRNAHis was incubated with either tobacco acid pyrophosphatase (10 units; Epicentre) or calf intestinal alkaline phosphatase (20 units; NEB), respectively, according to the manufacturer's instructions and subsequently purified by phenol extraction and ethanol precipitation.

SUPPLEMENTAL MATERIAL

Supplemental material is available for this article.