Use of a Yeast tRNase Killer Toxin to Diagnose Kti12 Motifs Required for tRNA Modification by Elongator

Saccharomyces cerevisiae cells are killed by zymocin, a tRNase ribotoxin complex from Kluyveromyces lactis, which cleaves anticodons and inhibits protein synthesis. Zymocin’s action requires specific chemical modification of uridine bases in the anticodon wobble position (U34) by the Elongator complex (Elp1-Elp6). Hence, loss of anticodon modification in mutants lacking Elongator or related KTI (K. lactis Toxin Insensitive) genes protects against tRNA cleavage and confers resistance to the toxin. Here, we show that zymocin can be used as a tool to genetically analyse KTI12, a gene previously shown to code for an Elongator partner protein. From a kti12 mutant pool of zymocin survivors, we identify motifs in Kti12 that are functionally directly coupled to Elongator activity. In addition, shared requirement of U34 modifications for nonsense and missense tRNA suppression (SUP4; SOE1) strongly suggests that Kti12 and Elongator cooperate to assure proper tRNA functioning. We show that the Kti12 motifs are conserved in plant ortholog DRL1/ELO4 from Arabidopsis thaliana and seem to be involved in binding of cofactors (e.g., nucleotides, calmodulin). Elongator interaction defects triggered by mutations in these motifs correlate with phenotypes typical for loss of U34 modification. Thus, tRNA modification by Elongator appears to require physical contact with Kti12, and our preliminary data suggest that metabolic signals may affect proper communication between them.

Elevated tRNA Glu levels also protect against zymocin, and this effect is suppressed by overexpression of the class II gene KTI1/TRM9 [15,16]. Trm9 is the catalytic subunit of a methyl-transferase (Trm9•Trm112), which methylates the preceding cm 5 (s 2 )U34 modification generated by Elongator and also contributes to zymocin sensitivity [16][17][18]. Consistently, loss of U34 methylation in trm9 and trm112 mutants confers resistance to zymocin and γ-toxin tRNase [18,19]. Thus, growth inhibition of S. cerevisiae by zymocin apparently involves restriction of U34-modified tRNAs affecting protein biosynthesis. In support of this strategy, which is shared by other microbial anticodon nucleases [20,21], zymocin depletes tRNA Glu in vivo, and overexpression of tRNA ligase suppresses the tRNA attack showing that tRNA repair is able to 'heal' the damage in the anticodon loop [22]. Use of the γ-toxin tRNase as a tool diagnostic for Elongator activity identified additional class II gene products with related roles in tRNA modification [2,5,19,[23][24][25]. Among them, a casein kinase I isozyme (Hrr25/Kti14), a protein phosphatase (Sit4) and an Elongator partner protein (Kti12) were found to constitute a complicated network that regulates the phosphorylation status of Elongator's largest subunit (Elp1) [26][27][28]. This suggested that phosphorylation affects Elongator activity, a notion supported by data showing that Hrr25 phosphorylation sites on Elp1 are critical for U34 modification [29] and that Hrr25 association with Elongator requires Kti12 [28].
Sequence alignments between Kti12 and its plant ortholog DRL1/ELO4 suggested that both share a P-loop motif characteristic for nucleoside triphosphate (NTP) binders [30] and a putative calmodulin (CaM) binding domain (CBD) [31]. With ELO4 known to bind CaM in vitro and associate with plant Elongator in vivo [32,33], the presence of cofactor binding motifs points towards conserved roles for Kti12 and ELO4. To further analyze Kti12 requirements for Elongator activity, we studied a pool of zymocin resistant kti12 mutants previously used [5] to identify the KTI12 locus in yeast. We map mutations in the CBD and P-loop motifs and other regions conserved between Kti12 and ELO4 that, based on genetic and biochemical assays, trigger traits typical of Elongator-linked loss of U34 modification and defects in Elongator interaction. In summary, our data reinforce the view that Kti12 relates in function to the yeast Elongator complex and stimulates its tRNA modification activity.

Results and Discussion
Based on its similarity to yeast Kti12, the Arabidopsis protein ELO4 was previously proposed to play an Elongator-related role in plants [24,32]. This was confirmed when ELO4 was shown to associate with TAP-tagged plant Elongator and ELO4 mutations found to cause phenotypes typical of plant Elongator mutants [33]. In addition, alignments between Kti12 and ELO4 proteins from fungal and plant species ( Figure S1) revealed several regions of high similarity. Among these are P-loop motifs, putative calmodulin (CaM) binding domains (CBDs) [24,31,32] and other homology regions ( Figure S1). While P-loops may be indicative of NTP binding [30], the CBDs likely mediate the demonstrated in vitro binding of ELO4 to CaM [31,32]. In comparison to strong CBDs (type 1-5-10) used, e.g., in tandem affinity purification (TAP) protocols [34], the CBDs in Kti12 or ELO4 (type 1-8-14) show weaker CaM affinity [35]. In summary, similarity in sequence and domain organisation may indicate that the motifs found in Kti12 and ELO4 are evolutionarily conserved because they functionally interact with Elongator.
Toxins 2017, 9,272 4 of 12 the kti12-4 and kti12-8 alleles ( Figure S2), although to a lesser extent than elp1∆. In summary, an Elongator-like phenotypic signature of KTI12 loss-of-function alleles documents the usefulness of zymocin as a tool diagnostic for Kti12 protein motifs that are important for Elongator activity.

Conserved Kti12 Motifs Support Elongator-Dependent tRNA Suppression
Physiologically, Elongator-dependent tRNA modification fine-tunes tRNA decoding [40][41][42][43] and maintains ribosomal reading frame accuracy during translation [44][45][46][47]. In support, stop codon read-through by U34 carrying tRNA suppressors (SUP4; sup9e) requires Elongator function [8,16,39]. Inspired by previous data showing that some class II kti mutants also have tRNA modification defects [8,19], we compared SUP4 read-through between members of the kti12 mutant collection and an elp3Δ knock-out in reporter strain UMY2893. The latter maintains ochre (UAA) alleles of ADE2 (ade2-1) and CAN1 (can1-100), which upon suppression by SUP4, confer adenine prototrophy (Ade + ) and canavanine sensitivity (Can S ), respectively [8,16]. In agreement with previous reports, SUP4 readthrough of ade2-1 and can1-100 was abolished by ELP3 gene deletion yielding adenine auxotrophic (Ade − ) and canavanine resistant (Can R ) elp3∆ cells ( Figure 2A). Notably, both traits (Ade − , Can R ) were copied by all the kti12 mutant alleles (kti12-1, kti12-3, kti12-4, kti12-6, kti12-7 and kti12-8) we tested ( Figure 2A) and complemented after reintroducing a KTI12 wild-type gene into kti12∆ cells ( Figure  2A). The latter finding supports our notion that recessive KTI12 loss-of-function alleles trigger Elongator-linked tRNA modification defects and phenotypes in yeast (Figures 1 and 2). Consistent with our previous observation that Elongator dependent processes (including e.g., zymocicity) can be suppressed by overexpression of KTI12 [5,26], multi-copy KTI12 also countered tRNA suppression of can1-100 by SUP4 ( Figure 2A). As for ade2-1, however, multi-copy KTI12, allowed weak growth in medium lacking adenine, a trait in between the phenotypes of wild-type (Ade + ) or elp3/kti12 mutant (Ade − ) cells ( Figure 2A). The latter read-out may indicate that the two ochre reporters differ in their level of sensitivity level to Elongator defects. Presumably, a weaker Ade + trait could be interpreted in a way that increased intracellular Kti12 levels also decrease the tRNA modification activity of Elongator ( Figure 2A). It will be interesting to address this scenario in greater detail in future studies using quantitative lacZ ochre read-through assay or direct LC-MS/MS measurements [29,48,49]. These techniques will allow comparison of the relative abundance of Elongator dependent U34 modification levels between mutants lacking Kti12 and cells expressing Kti12 at elevated levels from either single-copy or multi-copy plasmids. Consistent with our previous observation that Elongator dependent processes (including e.g., zymocicity) can be suppressed by overexpression of KTI12 [5,26], multi-copy KTI12 also countered tRNA suppression of can1-100 by SUP4 ( Figure 2A). As for ade2-1, however, multi-copy KTI12, allowed weak growth in medium lacking adenine, a trait in between the phenotypes of wild-type (Ade + ) or elp3/kti12 mutant (Ade − ) cells (Figure 2A). The latter read-out may indicate that the two ochre reporters differ in their level of sensitivity level to Elongator defects. Presumably, a weaker Ade + trait could be interpreted in a way that increased intracellular Kti12 levels also decrease the tRNA modification activity of Elongator (Figure 2A). It will be interesting to address this scenario in greater detail in future studies using quantitative lacZ ochre read-through assay or direct LC-MS/MS measurements [29,48,49]. These techniques will allow comparison of the relative abundance of Elongator dependent U34 modification levels between mutants lacking Kti12 and cells expressing Kti12 at elevated levels from either single-copy or multi-copy plasmids.
In addition, we observed that similar to kti12∆ and elp1∆ cells, kti12-4 and kti12-8 alleles also antagonized tRNA missense suppression of cdc8-1 ts by SOE1 ( Figure 2B), a mutant tRNA Glu that reads Lys codons in a fashion dependent on U34 modification by Elongator [39,50]. Hence, proper Elongator and Kti12 activity are required for SOE1 to rectify Glu to Lys substitutions in cdc8-1 ts cells and thus suppress their thermosensitivity at 36 • C ( Figure 2B). Together with previous data showing that missense (SOE1) and nonsense (SUP4; sup9e) suppressor tRNAs depend on Elongator [8,16,39], Kti12 can thus be considered to promote suppressor tRNA functionality, most likely through its regulatory role in Elongator's tRNA modification reaction.

Kti12 Expression and Interaction Profiles
To examine whether and how the mutated Kti12 proteins are expressed, we subjected kti12-2, kti12-3, kti12-4 and kti12-8 alleles to genomic c-Myc-tagging. Following Western blot analyses, all epitope-tagged Kti12 variants were found to be produced at the expected molecular weights ( Figure 3A); Kti12-4 and Kti12-8 are full-length (~40 kDa) while Kti12-2 (~26 kDa) and Kti12-3 (~35 kDa) represent C-terminal truncations ( Figure 3A). In addition, while Kti12-3 and Kti12-4 levels were reduced, Kti12-2 and Kti12-8 expression compared favourably with Kti12 wild-type levels ( Figure 3A). To avoid abundance related secondary effects, we conducted subsequent analyses with the latter two mutants. Since Kti12 and Elongator associate with each other [6,24,28,36], we examined whether inappropriate interaction between both may underlie the loss-of-function traits observed in kti12-2 and kti12-8 cells (Figures 1 and 2). Therefore, c-Myc-tagged KTI12, kti12-2 and kti12-8 alleles ( Figure 3A) were each co-expressed with an HA-tagged copy of the ELP2 gene for co-immune precipitation (IP) assays ( Figure 3B). Unlike the strong interaction observed between wild-type Kti12 and Elp2 (Figure 3B), the Elongator subunit failed to co-precipitate Kti12-2 and Kti12-8 ( Figure 3B) and both mutants hardly co-purified Elp2 in reciprocal IPs (data not shown). This indicates that a proper interaction with Elongator requires both the structural integrity of CBD2 and conserved Kti12 residues further downstream (truncated in kti12-2 cells) and an intact CBD1 motif (mutated in kti12-8 cells) ( Figure 1A). Whether the G92D mutation ( Figure 1A) affects Elongator interaction in kti12-8 cells through changes in local hydrophobicity ( Figure S4) is an option currently under investigation. In summary, loss of Elongator binding due to mutations in specific and putative functional Kti12 domains correlates with Elongator-minus traits (Figures 1 and 2). These observations again support our previously presented working hypothesis that direct physical contact with Kti12 is required for Elongator's ability to modify tRNA anticodons [6,28,29]. In addition, we observed that similar to kti12∆ and elp1∆ cells, kti12-4 and kti12-8 alleles also antagonized tRNA missense suppression of cdc8-1 ts by SOE1 ( Figure 2B), a mutant tRNA Glu that reads Lys codons in a fashion dependent on U34 modification by Elongator [39,50]. Hence, proper Elongator and Kti12 activity are required for SOE1 to rectify Glu to Lys substitutions in cdc8-1 ts cells and thus suppress their thermosensitivity at 36 °C ( Figure 2B). Together with previous data showing that missense (SOE1) and nonsense (SUP4; sup9e) suppressor tRNAs depend on Elongator [8,16,39], Kti12 can thus be considered to promote suppressor tRNA functionality, most likely through its regulatory role in Elongator's tRNA modification reaction.

Kti12 Expression and Interaction Profiles
To examine whether and how the mutated Kti12 proteins are expressed, we subjected kti12-2, kti12-3, kti12-4 and kti12-8 alleles to genomic c-Myc-tagging. Following Western blot analyses, all epitope-tagged Kti12 variants were found to be produced at the expected molecular weights ( Figure  3A); Kti12-4 and Kti12-8 are full-length (~40 kDa) while Kti12-2 (~26 kDa) and Kti12-3 (~35 kDa) represent C-terminal truncations ( Figure 3A). In addition, while Kti12-3 and Kti12-4 levels were reduced, Kti12-2 and Kti12-8 expression compared favourably with Kti12 wild-type levels ( Figure  3A). To avoid abundance related secondary effects, we conducted subsequent analyses with the latter two mutants. Since Kti12 and Elongator associate with each other [6,24,28,36], we examined whether inappropriate interaction between both may underlie the loss-of-function traits observed in kti12-2 and kti12-8 cells (Figures 1 and 2). Therefore, c-Myc-tagged KTI12, kti12-2 and kti12-8 alleles ( Figure  3A) were each co-expressed with an HA-tagged copy of the ELP2 gene for co-immune precipitation (IP) assays ( Figure 3B). Unlike the strong interaction observed between wild-type Kti12 and Elp2 ( Figure 3B), the Elongator subunit failed to co-precipitate Kti12-2 and Kti12-8 ( Figure 3B) and both mutants hardly co-purified Elp2 in reciprocal IPs (data not shown). This indicates that a proper interaction with Elongator requires both the structural integrity of CBD2 and conserved Kti12 residues further downstream (truncated in kti12-2 cells) and an intact CBD1 motif (mutated in kti12-8 cells) ( Figure 1A). Whether the G92D mutation ( Figure 1A) affects Elongator interaction in in kti12-8 cells through changes in local hydrophobicity ( Figure S4) is an option currently under investigation. In summary, loss of Elongator binding due to mutations in specific and putative functional Kti12 domains correlates with Elongator-minus traits (Figures 1 and 2). These observations again support our previously presented working hypothesis that direct physical contact with Kti12 is required for Elongator's ability to modify tRNA anticodons [6,28,29].

Affinity of Kti12 for CaM
As the putative CBDs may be involved in Kti12 function (Figures 2 and 3) [24,31,32], we examined next whether yeast Kti12 binds to CaM. In detail, we subjected total yeast extracts containing c-Myc-tagged Kti12 or Kti12-8 to CaM Sepharose affinity chromatography. Extracts containing Kti13-c-Myc, a protein without any CBD motif, or Kti12-TAP with a high-affinity CaM-tag were used as negative and positive controls, respectively (Figure 4). Protein bound to the CaM matrix was eluted with EGTA to chelate calcium and analyzed in Western blots. Both wild-type Kti12-c-Myc and Kti12-8-c-Myc could be detected in the eluate, whereas the negative control Kti13-c-Myc could not (Figure 4). The flow-through fractions indicate that, as expected, the affinity of Kti12-c-Myc is lower than that of Kti12-TAP, which carries an additional CBD derived from the TAP tag [34]. Whether the second CBD ( Figure S1) might contribute to residual binding of the kti12-8 mutant to CaM is not known. Collectively, our data indicate that Kti12 and Kti12-8 bind CaM in a calcium sensitive manner. Although they cannot unambiguously provide evidence that the Gly-92 residue in Kti12 is part of a bona fide CBD, they show that Kti12 has CaM affinity, a finding consistent with demonstrated in vitro binding of plant ELO4 to CaM [32]. Whether the ability of Kti12 to bind CaM contributes to Elongator function has to await further investigation.

Affinity of Kti12 for CaM
As the putative CBDs may be involved in Kti12 function (Figures 2 and 3) [24,31,32], we examined next whether yeast Kti12 binds to CaM. In detail, we subjected total yeast extracts containing c-Myc-tagged Kti12 or Kti12-8 to CaM Sepharose affinity chromatography. Extracts containing Kti13-c-Myc, a protein without any CBD motif, or Kti12-TAP with a high-affinity CaMtag were used as negative and positive controls, respectively (Figure 4). Protein bound to the CaM matrix was eluted with EGTA to chelate calcium and analyzed in Western blots. Both wild-type Kti12-c-Myc and Kti12-8-c-Myc could be detected in the eluate, whereas the negative control Kti13c-Myc could not (Figure 4). The flow-through fractions indicate that, as expected, the affinity of Kti12c-Myc is lower than that of Kti12-TAP, which carries an additional CBD derived from the TAP tag [34]. Whether the second CBD ( Figure S1) might contribute to residual binding of the kti12-8 mutant to CaM is not known. Collectively, our data indicate that Kti12 and Kti12-8 bind CaM in a calcium sensitive manner. Although they cannot unambiguously provide evidence that the Gly-92 residue in Kti12 is part of a bona fide CBD, they show that Kti12 has CaM affinity, a finding consistent with demonstrated in vitro binding of plant ELO4 to CaM [32]. Whether the ability of Kti12 to bind CaM contributes to Elongator function has to await further investigation.

Cross-Complementation Analysis between KTI12 and ELO4
Based on several lines of evidence including sequence similarities (27.9% identity; 45.3% similarity, Figure S1), elongator-related functions in vivo and CaM binding in vitro (Figure 4) [31][32][33], Kti12 and ELO4 may be functional counterparts. To study structure and functional conservation between subdomains ( Figure 5A), we engineered hybrid alleles of KTI12 encoding motif-swapped proteins in which the P-loop or CBD1 motif from Kti12 is replaced with the respective ELO4 sequence ( Figure 5B). After introducing these into a kti12∆ reporter strain, both hybrids were able to complement the loss-of function phenotype of the mutant; sensitivity to growth inhibition by the γtoxin tRNase subunit of zymocin ( Figure 5C) and ochre codon read-through of ade2-1 by SUP4 ( Figure  5D) were restored. In contrast, the plant ELO4 gene leaves γ-toxin resistance and adenine auxotrophy (Ade − ) of kti12∆ cells unchanged ( Figure 5C,D). Hence, the Arabidopsis homologue of Kti12 is unable to replace Kti12 function in yeast, whereas the hybrid proteins are functional. We conclude that sequence differences in the highly conserved motifs are unlikely to be responsible for the specificity of Kti12. Notably, the KTI12 gene from K. lactis, which is more closely related to that of S. cerevisiae, is also unable to complement the kti12∆ mutation (data not shown). To address the question of whether Elongator interactions are retained in the motif-swapped Kti12-c-Myc variants, they were co-expressed with Elp1-HA, and IPs were performed with anti-HA antibodies. Elp1-HA coprecipitated native Kti12 and both hybrids ( Figure 5E). Based on protein expression in the extracts washes) and eluted (EGTA elution) protein fractions were subjected to anti-TAP (top panel) and anti c-Myc Western blots (other panels); (B) killer eclipse assays between the same S. cerevisiae strains used for CaM affinity chromatography and a K. lactis zymocin producer (killer strain). 'S' or 'R' denote zymocin sensitivity or resistance.

Cross-Complementation Analysis between KTI12 and ELO4
Based on several lines of evidence including sequence similarities (27.9% identity; 45.3% similarity, Figure S1), elongator-related functions in vivo and CaM binding in vitro (Figure 4) [31][32][33], Kti12 and ELO4 may be functional counterparts. To study structure and functional conservation between subdomains ( Figure 5A), we engineered hybrid alleles of KTI12 encoding motif-swapped proteins in which the P-loop or CBD1 motif from Kti12 is replaced with the respective ELO4 sequence ( Figure 5B). After introducing these into a kti12∆ reporter strain, both hybrids were able to complement the loss-of function phenotype of the mutant; sensitivity to growth inhibition by the γ-toxin tRNase subunit of zymocin ( Figure 5C) and ochre codon read-through of ade2-1 by SUP4 ( Figure 5D) were restored. In contrast, the plant ELO4 gene leaves γ-toxin resistance and adenine auxotrophy (Ade − ) of kti12∆ cells unchanged (Figure 5C,D). Hence, the Arabidopsis homologue of Kti12 is unable to replace Kti12 function in yeast, whereas the hybrid proteins are functional. We conclude that sequence differences in the highly conserved motifs are unlikely to be responsible for the specificity of Kti12. Notably, the KTI12 gene from K. lactis, which is more closely related to that of S. cerevisiae, is also unable to complement the kti12∆ mutation (data not shown). To address the question of whether Elongator interactions are retained in the motif-swapped Kti12-c-Myc variants, they were co-expressed with Elp1-HA, and IPs were performed with anti-HA antibodies. Elp1-HA co-precipitated native Kti12 and both hybrids ( Figure 5E). Based on protein expression in the extracts prior to IP that were compared to wild-type Kti12 levels, the CBD1-swapped hybrid may be slightly reduced in Elp1 interaction. In summary, we conclude that while alleles coding for hybrids, in which conserved cofactor binding motifs of ELO4 (P-loop, CBD1) have been engineered into yeast Kti12, are functionally exchangeable, full-length plant ELO4 is not. Cross-complementation was also absent with plant ELO4 homologs from rice (OsELO4) and moss (PpELO4) ( Figure S3). The incompatibility of ELO4 with Elongator function in S. cerevisiae contrasts with previous findings showing that elp1∆ mutants can be partially rescued by ELO2, the plant homolog of yeast ELP1 [51]. Strikingly, the most highly conserved Elongator subunit gene (ELP3) gene cannot be replaced in yeast by the Arabidopsis ELP3 homolog even though the plant subunit assembled into holo-Elongator complexes [9]. However, functional complementation in yeast was achieved when the two plant Elongator genes (ELO2, ELP3) were co-expressed in the genetic background of an elp1∆ elp3∆ double mutant [9]. Hence, non-conserved plant-specific sequences in ELO4 may mediate a selective physical or functional interaction with individual Elongator subunits, Hrr24 or Sit4 to explain the failure of cross-complementation in our gene shuffle experiments ( Figure 5C,D). Regions of species-specific sequence variation may thus provide first insights into the evolutionary differences between plant ELO4 and yeast Kti12. prior to IP that were compared to wild-type Kti12 levels, the CBD1-swapped hybrid may be slightly reduced in Elp1 interaction. In summary, we conclude that while alleles coding for hybrids, in which conserved cofactor binding motifs of ELO4 (P-loop, CBD1) have been engineered into yeast Kti12, are functionally exchangeable, full-length plant ELO4 is not. Cross-complementation was also absent with plant ELO4 homologs from rice (OsELO4) and moss (PpELO4) ( Figure S3). The incompatibility of ELO4 with Elongator function in S. cerevisiae contrasts with previous findings showing that elp1∆ mutants can be partially rescued by ELO2, the plant homolog of yeast ELP1 [51]. Strikingly, the most highly conserved Elongator subunit gene (ELP3) gene cannot be replaced in yeast by the Arabidopsis ELP3 homolog even though the plant subunit assembled into holo-Elongator complexes [9]. However, functional complementation in yeast was achieved when the two plant Elongator genes (ELO2, ELP3) were co-expressed in the genetic background of an elp1∆ elp3∆ double mutant [9]. Hence, non-conserved plant-specific sequences in ELO4 may mediate a selective physical or functional interaction with individual Elongator subunits, Hrr24 or Sit4 to explain the failure of crosscomplementation in our gene shuffle experiments ( Figure 5C,D). Regions of species-specific sequence variation may thus provide first insights into the evolutionary differences between plant ELO4 and yeast Kti12.

Conclusions
To conclude, the zymocin tRNase ribotoxin can be exploited as a molecular diagnostic tool to assign function to motifs conserved in two Elongator partner proteins, Kti12 and plant ELO4. These motifs are potentially involved in cofactor binding (nucleotide, calmodulin) and when mutated, abolish Elongator interaction and trigger phenotypes typical of tRNA modification defects in known Elongator mutants. Thus, tRNA modification by Elongator appears to require physical contact and proper communication with Kti12, a notion further supported by our findings that Kti12 and Elongator cooperate in proper performance of nonsense and missense tRNA suppressors (SUP4, SOE1). Despite their structural and functional similarities, Kti12 and ELO4 cannot be functionally exchanged between yeast and plant cells. This suggests the occurrence of species-specific barriers or evolutionary diversification of Elongator and its associated regulatory proteins between lower and higher eukaryotes. In summary, we provide further evidence for a dynamic and complicated network underlying potential regulation of the U34 modification pathway. It constitutes regulatory proteins including Kti11/Dph3, Kti12, Kti13/Ats1 and Kti14/Hrr25 [28,29,36,39,[52][53][54][55] whose associations with the Elongator complex appear to be sensitive to certain metabolic signals. In support of this notion are recent findings showing that loss of U34 modifications cause altered metabolic profiles in Elongator mutants [56].

KTI12 Expression Analyses and Elongator or CaM Interaction Studies
Immune detection of epitope-tagged Kti12 and Elongator subunits and immune precipitations (IPs) from total yeast extracts used anti-c-Myc (9E10) and anti-HA (3F10) antibodies (Roche) as previously described [6,24]. Protein loading was checked with anti-Cdc19 serum (a kind donation by Dr J. Thorner, University of California, Berkley, CA, USA), which recognizes pyruvate kinase. CaM binding assays in vitro involved TAP-tagged or c-Myc-marked Kti12 and Kti13 proteins from total yeast extracts (~2 mg) adjusted to 1 mM CaCl 2 . The material (2 mL) was mixed with CaM-Sepharose for 30 min and equilibrated with 2 mL CaM binding buffer (CBB) as described [32]. The slurry was packed onto a column, the buffer drained, and the column rinsed twice with 2.5 volumes of CBB. Elution of bound proteins was in EGTA buffer (4 mM Tris-HCl, pH 7.5, 200 mM NaCl, 1 mM MgCl 2 , 2 mM EGTA, and 0.1 mM DTT). Equal proportions of load, flow-through, wash, and elution fractions were separated by 15% SDS-PAGE, and Kti12 material was detected using anti-TAP and anti-c-Myc Western blots.

Kti12 Secondary Structure Predictions and Alignments with ELO4/DRL1
Secondary structure predictions for Kti12 and ELO4 proteins were obtained using the semi-automated homolog-based protein fold recognition server Phyre2.0 [60]. Best matching records, in both cases based on the crystal structure of archaeal O-phosphoseryl-tRNA(Sec) kinase (PSTK, PDB: 3A4M) [61], were further used to visualize Kti12 domain architecture and illustrate a motif-swap experiment ( Figure 5B). All structures were rendered using PyMOL Molecular Graphics System (http://www.pymol.org/) (Version 1.8 Schrödinger, LLC, New York, NY, USA). Sequences of Kti12 or its homologs from indicated organisms were obtained from NCBI. Multiple alignments were generated using MUSCLE available at EMBL-EBI web page (http://www.ebi.ac.uk/Tools/msa/muscle/). MSA files were handled using JalView software [62]. Homolog-based model and multiple alignment were used for sequence conservation illustration with the CONSURF server (http://consurf.tau.ac.il/) [63].

Conflicts of Interest:
The authors declare no conflict of interest.