Role of Pseudouridine Formation by Deg1 for Functionality of Two Glutamine Isoacceptor tRNAs

Loss of Deg1/Pus3 and concomitant elimination of pseudouridine in tRNA at positions 38 and 39 (ψ38/39) was shown to specifically impair the function of tRNAGlnUUG under conditions of temperature-induced down-regulation of wobble uridine thiolation in budding yeast and is linked to intellectual disability in humans. To further characterize the differential importance of the frequent ψ38/39 modification for tRNAs in yeast, we analyzed the in vivo function of non-sense suppressor tRNAs SUP4 and sup70-65 in the absence of the modifier. In the tRNATyrGψA variant SUP4, UAA read-through is enabled due to an anticodon mutation (UψA), whereas sup70-65 is a mutant form of tRNAGlnCUG (SUP70) that mediates UAG decoding due to a mutation of the anticodon-loop closing base pair (G31:C39 to A31:C39). While SUP4 function is unaltered in deg1/pus3 mutants, the ability of sup70-65 to mediate non-sense suppression and to complement a genomic deletion of the essential SUP70 gene is severely compromised. These results and the differential suppression of growth defects in deg1 mutants by multi-copy SUP70 or tQ(UUG) are consistent with the interpretation that ψ38 is most important for tRNAGlnUUG function under heat stress but becomes crucial for tRNAGlnCUG as well when the anticodon loop is destabilized by the sup70-65 mutation. Thus, ψ38/39 may protect the anticodon loop configuration from disturbances by loss of other modifications or base changes.


Introduction
In addition to standard nucleosides, tRNA is known to contain a large variety of modified nucleosides, formed by the addition of chemical groups or isomerization. Pseudouridine (ψ) is an isomer of uridine (5-ribosyl-uracil) and represents the initially discovered and most abundant modified nucleoside present in all three domains of life [1,2]. The model organism Saccharomyces cerevisiae has ten characterized pseudouridine synthases (Pus1-9 and Cbf5) that modify cytoplasmic and mitochondrial tRNAs and/or other types of RNA, including rRNA, small nuclear RNA (snRNA) and mRNA [2]. Despite its ubiquitous presence in tRNA, ψ appears to affect tRNA function rather subtly, since it is not generally required for cell viability, as demonstrated by the non-essential nature of all of the PUS genes in the yeast system [3]. However, the absence of PUS3/DEG1 in yeast uniquely causes slow growth, in particular at elevated temperature [4]. The general effects of ψ on tRNA function are likely mediated by increasing the rigidity of the sugar phosphate backbone and base stacking [2,5]. In humans, homozygous mutation of the DEG1 orthologue PUS3 is correlated with intellectual disability, suggesting a significant contribution to tRNA functioning not only in yeast but also in humans [6].
In yeast, Deg1 dependent formation of ψ occurs in a majority of tRNAs either at position 38 or 39 [4,7]. Slow growth of yeast mutants lacking Deg1 is strongly aggravated if parts of the wobble uridine modification 5-methoxy-carbonylmethyl-2-thiouridine (mcm 5 s 2 U) are simultaneously absent [8,9]. The mcm 5 component of the mcm 5 s 2 U modification is formed by a pathway requiring the Elongator complex and a variety of accessory and potentially regulatory proteins [10][11][12][13][14][15][16][17][18][19]. The separate Urm1 pathway mediates the transfer of sulfur to the carbon 2 of the uracil base. This requires the sulfur carrier and ubiquitin-like modifier protein Urm1 and a number of additional proteins operating upstream (Nfs1, Tum1, Uba4) and downstream (Ncs2, Ncs6) of Urm1 resulting in sulfur flow from cysteine to mcm 5 U [11,[20][21][22][23][24][25]. The absence of either part alone (mcm 5 U or s 2 U) induces shared pleiotropic phenotypes, including growth defects at elevated temperature and sensitivity to a number of exogenous stressors including the target of rapamycin complex 1 (TORC1)-inhibiting drug rapamycin [26,27]. On a functional level, partial loss of mcm 5 s 2 U reduces binding of tRNA to the ribosomal A-site and in turn causes reading frame slippage [28,29]. Combined loss of mcm 5 and s 2 U increases the severity of these phenotypes and causes a defect in protein homeostasis resulting in the accumulation of protein aggregates [30][31][32]. Similarly, combination of Elongator or Urm1 pathway mutations with a deg1 deletion aggravates growth defects and induces protein aggregation, which is further linked to severe problems in cytoskeleton organization, cell polarity and nuclear segregation [9].
Pleiotropic phenotypes of mutants lacking ψ38/39 together with mcm 5 s 2 U can be specifically rescued by overexpressing tRNA Gln UUG , which is reasonable since this tRNA alone carries both modifications [8,9]. In addition to tRNA Gln UUG , there is an isoacceptor tRNA with the anticodon sequence CUG that is thought to decode the G-ending codon for Gln. Genetic data from yeast suggests that the U34-containing isoacceptor mainly decodes the A-ending codon and only inefficiently reads the G-ending codon [33]. Because of this, the single tQ(CUG) gene has been found to be essential for yeast cell viability [33]. Except for the wobble nucleoside and a variable base (A/G) in position 42, tRNA Gln CUG contains an anticodon stem loop (ASL) identical with that of tRNA Gln UUG and is also modified by Deg1 at position U38 [34]. Interestingly, however, growth defects of deg1 single mutants were found to be suppressible by overexpression of tRNA Gln UUG , but not tRNA Gln CUG . This raises the possibility of differential functional dependencies of the two Gln isoacceptor tRNAs on the presence of ψ38. In this study, we investigated potential reasons for such divergent effects of one and the same modification on two highly similar tRNA species.

Phenotypic Rescue of deg1 Mutants by tRNA Overexpression
tRNA Gln UUG functionality has been shown to depend on ψ38, in particular under conditions of heat stress [8]. Since the tRNA Gln CUG isoacceptor harbours an ASL almost identical to that of tRNA Gln UUG , we first investigated whether there are indeed differential dependencies of the two Gln isoacceptors on the presence of ψ38. The thermosensitive growth phenotype of deg1 mutants was shown to be specifically suppressible by overexpression of tRNA Gln UUG but not tRNA Gln CUG or any other tRNA carrying U38 or U39, which is modified by Deg1 to ψ [8]. We confirm this result and find a clear suppression of deg1-induced thermosensitivity by multicopy tRNA Gln UUG but not tRNA Gln CUG or other tester tRNAs ( Figure 1A). However, rescue by multicopy tRNA Gln UUG is clearly incomplete as growth at 39 • C is not fully restored to wild type levels ( Figure 1A). Therefore, we checked whether co-overexpression of the isoacceptor tRNA Gln CUG together with tRNA Gln UUG improves growth at elevated temperature (39 • C) compared to overexpression of tRNA Gln UUG alone. We find that under these conditions, growth of deg1 mutants already overexpressing tRNA Gln UUG is not further improved by additional overexpression of tRNA Gln CUG ( Figure 1B). These results suggest that under heat stress conditions, growth defect of yeast cells lacking ψ38 results almost exclusively from malfunction of tRNA Gln UUG rather than tRNA Gln CUG , despite containing almost identical ASLs.  It should also be taken into account that copy numbers of the two Gln isoacceptors in yeast differ significantly due to the number of the respective tRNA genes (nine copies of tQ(UUG) versus one copy of tQ(CUG) (Saccharomyces cerevisiae genome database)). This correlates with a biased codon usage and may impact on the efficiency of phenotypic suppression by tRNA overexpression. Based on this consideration, phenotypic effects of the DEG1 deletion may be rescuable by overexpression of tRNA Gln UUG, since this is the major Gln isoacceptor and consequently more important for overall translation as compared to the minor tRNA Gln CUG. However, the single tQ(CUG) gene is in fact known to be essential for yeast viability [33], indicating that this C34-containing tRNA is indeed indispensable for translation, despite its lower frequency of utilization compared to the one containing mcm 5 s 2 U34. Hence, the differential dependency of both isoacceptors on ψ38 as indicated by differential phenotypic suppression of the deg1 phenotype is It should also be taken into account that copy numbers of the two Gln isoacceptors in yeast differ significantly due to the number of the respective tRNA genes (nine copies of tQ(UUG) versus one copy of tQ(CUG) (Saccharomyces cerevisiae genome database)). This correlates with a biased codon usage and may impact on the efficiency of phenotypic suppression by tRNA overexpression. Based on this consideration, phenotypic effects of the DEG1 deletion may be rescuable by overexpression of tRNA Gln UUG , since this is the major Gln isoacceptor and consequently more important for overall translation as compared to the minor tRNA Gln CUG . However, the single tQ(CUG) gene is in fact known to be essential for yeast viability [33], indicating that this C34-containing tRNA is indeed indispensable for translation, despite its lower frequency of utilization compared to the one containing mcm 5 s 2 U34. Hence, the differential dependency of both isoacceptors on ψ38 as indicated by differential phenotypic suppression of the deg1 phenotype is presumably, at least in part, due to the difference at the wobble base. Interestingly, tRNA Gln UUG is known to harbour a hypomodified wobble uridine under mild heat stress conditions, which results from a destabilization of the URM1 pathway and in turn suppresses the tRNA thiolation reaction that converts mcm 5 U to mcm 5 s 2 U [8,25,[35][36][37]. Thus, an explanation why tRNA Gln UUG depends more than tRNA Gln CUG on ψ38 under heat stress conditions might be the presence of an additional functional impairment of the former tRNA due to wobble uridine hypomodification. tRNA Gln CUG carries a different unmodified wobble base (C34) and therefore remains unaffected by the temperature effect on wobble uridine thiolation.

Genetic Interactions of DEG1 under TORC1 Inhibition
In support of a specific requirement for ψ38 in tRNA Gln UUG in the event of wobble base hypomodification, strong synthetic growth defects were observed upon combining the deg1 mutation with mutations in Elongator or Urm1 pathway genes, indicative of functional crosstalk between the modifications at position 34 and 38 in tRNA Gln UUG [8,9,38]. Elongator and Urm1 pathway mutants are further known to exhibit shared phenotypes, including translational inaccuracy and sensitivity to TORC1 inhibiting agents such as caffeine or rapamycin and these are synergistically increased in double mutants lacking both modification activities [9,[30][31][32]. To check whether negative genetic interactions between DEG1, Elongator and URM1 can be extended to conditions of TORC1 inhibition, we scored rapamycin sensitivity of deg1, urm1 and elp3 single and all possible double mutant combinations. As shown in Figure 2, there is a clear rapamycin sensitivity in all three single mutants and this sensitivity is strongly aggravated in all the double mutants, with stronger effects seen in elp3 deg1 and urm1 deg1, where U34 and U38/39 are hypomodified. Thus, there is an overlapping phenotype in deg1, elp3 and urm1 mutants and all affected modifications (mcm 5 /s 2 U, ψ38/39) appear to mediate rapamycin tolerance independently of each other, possibly by maintaining a related aspect of tRNA function. In particular, the observed rapamycin sensitivity of deg1 single mutants in the absence of heat stress suggests that ψ38 alone becomes important for tRNA function under at least one stress condition that is different from heat. Since there is strong negative genetic interaction on rapamycin between DEG1 and URM1/Elongator, the latter of which affect tRNA Gln UUG but not tRNA Gln CUG , we assume that these effects are caused by specific malfunction of tRNA Gln UUG . presumably, at least in part, due to the difference at the wobble base. Interestingly, tRNA Gln UUG is known to harbour a hypomodified wobble uridine under mild heat stress conditions, which results from a destabilization of the URM1 pathway and in turn suppresses the tRNA thiolation reaction that converts mcm 5 U to mcm 5 s 2 U [8,25,[35][36][37]. Thus, an explanation why tRNA Gln UUG depends more than tRNA Gln CUG on ψ38 under heat stress conditions might be the presence of an additional functional impairment of the former tRNA due to wobble uridine hypomodification. tRNA Gln CUG carries a different unmodified wobble base (C34) and therefore remains unaffected by the temperature effect on wobble uridine thiolation.

Genetic Interactions of DEG1 under TORC1 Inhibition
In support of a specific requirement for ψ38 in tRNA Gln UUG in the event of wobble base hypomodification, strong synthetic growth defects were observed upon combining the deg1 mutation with mutations in Elongator or Urm1 pathway genes, indicative of functional crosstalk between the modifications at position 34 and 38 in tRNA Gln UUG [8,9,38]. Elongator and Urm1 pathway mutants are further known to exhibit shared phenotypes, including translational inaccuracy and sensitivity to TORC1 inhibiting agents such as caffeine or rapamycin and these are synergistically increased in double mutants lacking both modification activities [9,[30][31][32]. To check whether negative genetic interactions between DEG1, Elongator and URM1 can be extended to conditions of TORC1 inhibition, we scored rapamycin sensitivity of deg1, urm1 and elp3 single and all possible double mutant combinations. As shown in Figure 2, there is a clear rapamycin sensitivity in all three single mutants and this sensitivity is strongly aggravated in all the double mutants, with stronger effects seen in elp3 deg1 and urm1 deg1, where U34 and U38/39 are hypomodified. Thus, there is an overlapping phenotype in deg1, elp3 and urm1 mutants and all affected modifications (mcm 5 /s 2 U, ψ38/39) appear to mediate rapamycin tolerance independently of each other, possibly by maintaining a related aspect of tRNA function. In particular, the observed rapamycin sensitivity of deg1 single mutants in the absence of heat stress suggests that ψ38 alone becomes important for tRNA function under at least one stress condition that is different from heat. Since there is strong negative genetic interaction on rapamycin between DEG1 and URM1/Elongator, the latter of which affect tRNA Gln UUG but not tRNA Gln CUG, we assume that these effects are caused by specific malfunction of tRNA Gln UUG.

Figure 2.
Interactions between DEG1 and genes coding for Elongator (ELP3) or Urm1 (URM1) pathway components under TORC1 inhibition. Indicated single and double mutants were spotted on drug-free yeast peptone dextrose (YPD) plates which were incubated at 30 °C or 37 °C or YPD medium containing 5 nM of the TORC1 inhibiting drug rapamycin.

Figure 2.
Interactions between DEG1 and genes coding for Elongator (ELP3) or Urm1 (URM1) pathway components under TORC1 inhibition. Indicated single and double mutants were spotted on drug-free yeast peptone dextrose (YPD) plates which were incubated at 30 • C or 37 • C or YPD medium containing 5 nM of the TORC1 inhibiting drug rapamycin.

Replacement of tQ(CUG) by the sup70-65 Allele in Haploid Yeast
The finding that tRNA Gln UUG lacking ψ38 becomes vulnerable to disturbances of the natural anticodon loop configuration (loss of U34 modification) suggested that tRNA Gln CUG devoid of ψ38 might also be sensitized to destabilization of the ASL. However, as C34 is unmodified and no modification except ψ38 is known for the ASL of tRNA Gln CUG [7,34], we utilized a variant of tRNA Gln CUG with a disturbance of the ASL due to a sequence change (G31A). This mutation, known as the tQ(CUG) allele sup70-65, replaces the Watson-Crick base pair G31:C39 normally closing the anticodon loop of tRNA Gln CUG by a mismatch (A31:C39) and reduces stability and charging efficiency of the mutant tRNA [39][40][41]. Despite these strong effects on the tRNA product of the essential single-copy gene tQ(CUG), diploid yeast cells carrying the homozygous sup70-65 mutation are viable [39][40][41] and may provide an opportunity to analyze the effects of ψ38 removal on tRNA Gln CUG in the presence of a destabilized ASL. We utilized a haploid strain (S288C) lacking the single genomic copy of tQ(CUG) and carrying the wild type tQ(CUG) gene on a plasmid that is counter-selectable by 5-fluoro-orotate (5-FOA) [31,40]. This strain was transformed with a second centromeric plasmid carrying the sup70-65 allele of the tQ(CUG) gene. Subsequently, the former plasmid was removed by growing cells on 5-FOA medium, resulting in a strain carrying sup70-65 as the sole genetic source of tQ(CUG) ( Figure 3A). The viability of this strain suggests that even in haploid S288C cells, the mutant form of tQ(CUG) is active to maintain sufficient levels of translation. However, we find that in comparison to wild type, a strong thermosensitivity is caused by the mutation, while no significant growth defect was observed at 30 • C ( Figure 3B). This finding possibly indicates that the destabilized form of tRNA Gln CUG (sup70-65) is sufficiently active in translation at normal, but not at elevated temperatures. The absence of defects in growth or CAG decoding in diploid sup70-65 cells was already documented [39]. However, more recent work demonstrated a reduced expression of reporter genes with tandem CAG codons and several natural CAG-codon-containing mRNAs [40,41]. Hence, a translational defect is already present in sup70-65 cells at 30 • C and this may be aggravated at elevated temperatures, as evidenced by the observed growth defect ( Figure 3B). It is noteworthy that tRNA Gln CUG can be functionally replaced by the inefficient CAG decoder tRNA Gln UUG if the latter is present at elevated levels and carries the full mcm 5 s 2 U modification [33]. In such strain, CAG codons are exclusively decoded by tRNA Gln UUG , involving U-G wobble and the mcm 5 s 2 U modified wobble base. In particular, the absence of s 2 U in tRNA Gln UUG entirely prevents the suppression of lethal effects of a tQ(CUG) deletion by multi-copy tQ(UUG) [33]. Thus, the strong growth defect of haploid sup70-65 cells at elevated temperatures may also indicate that tRNA Gln UUG contributes, to some extent, to CAG (Gln) decoding in this mutant, since tRNA Gln CUG is functionally impaired. If so, elevated temperature would repress thiolation and in turn prevent the engagement of tRNA Gln UUG in decoding of the G-ending Gln codon (CAG).
anticodon loop configuration (loss of U34 modification) suggested that tRNA Gln CUG devoid of ψ38 might also be sensitized to destabilization of the ASL. However, as C34 is unmodified and no modification except ψ38 is known for the ASL of tRNA Gln CUG [7,34], we utilized a variant of tRNA Gln CUG with a disturbance of the ASL due to a sequence change (G31A). This mutation, known as the tQ(CUG) allele sup70-65, replaces the Watson-Crick base pair G31:C39 normally closing the anticodon loop of tRNA Gln CUG by a mismatch (A31:C39) and reduces stability and charging efficiency of the mutant tRNA [39][40][41]. Despite these strong effects on the tRNA product of the essential single-copy gene tQ(CUG), diploid yeast cells carrying the homozygous sup70-65 mutation are viable [39][40][41] and may provide an opportunity to analyze the effects of ψ38 removal on tRNA Gln CUG in the presence of a destabilized ASL. We utilized a haploid strain (S288C) lacking the single genomic copy of tQ(CUG) and carrying the wild type tQ(CUG) gene on a plasmid that is counter-selectable by 5-fluoro-orotate (5-FOA) [31,40]. This strain was transformed with a second centromeric plasmid carrying the sup70-65 allele of the tQ(CUG) gene. Subsequently, the former plasmid was removed by growing cells on 5-FOA medium, resulting in a strain carrying sup70-65 as the sole genetic source of tQ(CUG) ( Figure 3A). The viability of this strain suggests that even in haploid S288C cells, the mutant form of tQ(CUG) is active to maintain sufficient levels of translation. However, we find that in comparison to wild type, a strong thermosensitivity is caused by the mutation, while no significant growth defect was observed at 30 °C ( Figure 3B). This finding possibly indicates that the destabilized form of tRNA Gln CUG (sup70-65) is sufficiently active in translation at normal, but not at elevated temperatures. The absence of defects in growth or CAG decoding in diploid sup70-65 cells was already documented [39]. However, more recent work demonstrated a reduced expression of reporter genes with tandem CAG codons and several natural CAG-codon-containing mRNAs [40,41]. Hence, a translational defect is already present in sup70-65 cells at 30 °C and this may be aggravated at elevated temperatures, as evidenced by the observed growth defect ( Figure 3B). It is noteworthy that tRNA Gln CUG can be functionally replaced by the inefficient CAG decoder tRNA Gln UUG if the latter is present at elevated levels and carries the full mcm 5 s 2 U modification [33]. In such strain, CAG codons are exclusively decoded by tRNA Gln UUG, involving U-G wobble and the mcm 5 s 2 U modified wobble base. In particular, the absence of s 2 U in tRNA Gln UUG entirely prevents the suppression of lethal effects of a tQ(CUG) deletion by multi-copy tQ(UUG) [33]. Thus, the strong growth defect of haploid sup70-65 cells at elevated temperatures may also indicate that tRNA Gln UUG contributes, to some extent, to CAG (Gln) decoding in this mutant, since tRNA Gln CUG is functionally impaired. If so, elevated temperature would repress thiolation and in turn prevent the engagement of tRNA Gln UUG in decoding of the G-ending Gln codon (CAG). To test whether sup70-65 is still able to replace wild type tQ(CUG) when ψ38 is removed, we deleted the DEG1 gene in a strain carrying a genomic deletion of tQ(CUG) and containing the wild type tQ(CUG) gene on the counter-selectable plasmid. Then, the ability to lose the tQ(CUG) plasmid in the presence of the sup70-65 plasmid was assayed in comparison to the strain lacking the deg1 mutation. We find that colony formation on 5-FOA medium, which counter-selects the tQ(CUG) wild type allele, is severely reduced in the strain lacking DEG1 ( Figure 3C). This result supports the interpretation that ψ38 becomes critical for function of tRNA Gln CUG in the presence of a destabilized ASL.

Comparison of Non-Sense Suppression by sup70-65 and SUP4 in the Presence and Absence of ψ38/39
Since sup70-65 is known as an amber suppressor [39,40], we analyzed how loss of ψ38 influences the efficiency of UAG stop codon read-through by sup70-65. If ψ38 is indeed crucial for tRNA Gln CUG function in the presence of a destabilized ASL, then stop codon read-through by sup70-65 could be impaired by the deg1 mutation. As a comparison, we scored the efficiency of SUP4, a distinct non-sense suppressor tRNA, derived from tRNA Tyr GψA [10]. SUP4 can read-through ochre (UAA) stop codons due to a base change at the wobble position and was already shown to crucially depend on the presence of the Elongator-mediated mcm 5 U modification (mcm 5 UψA) [10]. Since tRNA Tyr GψA harbours a DEG1 dependent ψ in position 39 [7], SUP4 could in principle also be affected by the deg1 mutation. To test this, we utilized a W303-1B-derived strain carrying SUP4 and ade2-1, an allele of ADE2 causing adenine auxotrophy due to an ochre mutation [10]. Non-sense suppression of the ade2-1 transcript is known to cause adenine prototrophy and change of colony color from red (no read-through) to white (read-through) [10]. In this background, we deleted the DEG1 gene and scored adenine auxotrophy as well as color colony in direct comparison with an elp3 mutant that is known to affect SUP4-mediated read-through. While the presence of SUP4 causes the expected change in color and adenine prototrophy, we find that this indeed depends on a functional ELP3 gene but not on DEG1 ( Figure 4A). Since this assay might be too insensitive to detect smaller reductions in SUP4-mediated read-through and to confirm the obtained result with a premature ochre codon in another reporter gene, we utilized a quantitative assay relying on a lacZ reporter with an engineered ochre codon [33,42]. Again, read-through in this reporter was strongly enhanced by the presence of SUP4 and suppressed by the elp3 mutation, yet not by deg1 ( Figure 4B). Further, even application of mild temperature stress (37 • C) to cells carrying the reporter constructs did not significantly reduce read-through levels of the deg1 mutant as compared to the wild type SUP4 strain under the same conditions ( Figure 4B). Thus, Deg1-mediated ψ39 in itself does not detectably impact on the function of SUP4 in non-sense suppression in two different read-through contexts. This result is consistent with the absence of suppression of temperature sensitive growth of the deg1 mutant by multi-copy tRNA Tyr GψA ( Figure 1A) and supports the conclusion that ψs formed by Deg1 in distinct tRNA species may have different functional consequences.
For sup70-65-mediated non-sense suppression, we utilized strain W303-1B, which carries the trp1-1 allele that contains a premature amber (UAG) codon and is efficiently suppressible by sup70-65 when the mutant tRNA gene is present on a multi-copy plasmid [40]. We deleted the DEG1 gene in W303-1B and introduced sup70-65 on a multi-copy construct. As a control, we used the parental DEG1 wild type strain and introduced sup70-65 on both, single and multi-copy vectors [40]. While there is no major difference in growth on tryptophan-supplemented minimal media, the presence of the sup70-65 gene in both single and multi-copy conferred an ability to grow on tryptophan-free media specifically to the DEG1 wild type. As expected, multi-copy expression of sup70-65 was required for induction of efficient growth in the absence of tryptophan, due to relatively inefficient amber suppression by sup70-65 via first base wobble (U1:G36/A2:U35/G3:C34) [40]. In the absence of the suppressor tRNA, no growth on tryptophan-free medium was observed and, most importantly, growth was also absent in the deg1 mutant with the multi-copy sup70-65 construct causing efficient read-through in the DEG1 wild type strain. Thus, indeed sup70-65-mediated non-sense suppression critically depends on the presence of ψ38 formation by Deg1.
tryptophan-free media specifically to the DEG1 wild type. As expected, multi-copy expression of sup70-65 was required for induction of efficient growth in the absence of tryptophan, due to relatively inefficient amber suppression by sup70-65 via first base wobble (U1:G36/A2:U35/G3:C34) [40]. In the absence of the suppressor tRNA, no growth on tryptophan-free medium was observed and, most importantly, growth was also absent in the deg1 mutant with the multi-copy sup70-65 construct causing efficient read-through in the DEG1 wild type strain. Thus, indeed sup70-65-mediated non-sense suppression critically depends on the presence of ψ38 formation by Deg1.  strain UMY2893, carrying SUP4; SUP4 deg1: deletion of DEG1 in UMY2893; SUP4 elp3: deletion of ELP3 in UMY2893. Serial dilutions were spotted on adenine-free (-Ade) or adenine-containing medium (+Ade); (B) Quantitative SUP4-mediated lacZ read-through assay. Strains as described in (A) were transformed with pUKC815 and pUKC817 and relative read-through levels measured as described in Materials and Methods. 30 • C and 37 • C indicate growth temperatures. Averages of at least three biological replicates are shown along with standard deviation; (C) sup70-65-mediated read-through of trp1-1. Strain W303-1B or W303-1B deg1 was transformed with empty vector (vector), single-copy (s.c.) or high-copy (h.c.) constructs with the sup70-65 allele of tQ(CUG). The ability of strains to grow on tryptophan-free (-Trp) medium by read-through of the trp1-1 marker was scored by drop dilution assays. +Trp: tryptophan-containing medium condition.

Strains, Plasmids and General Methods
Yeast strains used are listed in Table 1. Standard methods were used in this study for yeast growth [43]. Yeast was routinely grown in complex yeast peptone dextrose (YPD) medium. To select transformants or maintain plasmids, either synthetic complete (SC) medium lacking specific nutrients or YPD medium containing 200 µg·mL −1 G418 was used. Deletion of DEG1 involved PCR-mediated synthesis of deletion cassettes using previously described oligonucleotides [9] and template plasmids pUG6 and pUG27 [44]. Transformation was done using the PEG/lithium acetate method [45]. Correct genomic replacement of the DEG1 gene was verified by PCR using oligonucleotides described [9]. Multicopy plasmids for overexpression of tRNA Glu UUC , tRNA Lys UUU , tRNA Arg UCU , tRNA Leu UAA , tRNA Tyr GUA and tRNA Gly GCC were described earlier [13] as well as the construct for tRNA Gln UUG [46]. Wild type tRNA Gln CUG overexpression used 2 µ plasmid pSUP70 [40]. pAK01 is a single copy (CEN/ARS) plasmid carrying the wild type tQ(CUG) gene as well as URA3 as a counter-selectable marker [40]. Counter-selection was done using synthetic complete medium containing uracil and 5-FOA at 1 mg·mL −1 . Plasmids pSUP70-65-2 µ and pSUP70-65-CEN carry the sup70-65 allele of tQ(CUG) and either the 2 µ origin or CEN-ARS, respectively [40].

Phenotypic Assays
Yeast strains were either grown on YPD or appropriate SC solid medium to select for the presence of plasmids for 24 to 36 h. Cells were recovered from the plates and resuspended in sterile water. Cell densities were measured using absorbance at 600 nm and dilutions prepared from this suspension with final OD 600nm values of 0.15, 0.015, 0.0015 and 0.00015. These were spotted on either drug-free solid YPD plates or YPD plates supplemented with sterile rapamycin at a final concentration of 5 nM. Subsequently, plates were incubated at different temperatures for 36-72 h and photographed. The longer incubation time (72 h) was used for phenotypic assay with deg1 elp3 double mutants. For qualitative read-through of ade2-1 and trp1-1 alleles, serial dilution were spotted on SC lacking adenine or tryptophan respectively and incubated for 48 to 60 h at 30 • C.

Quantitative Read-through Assay
Cells carrying pUKC815 (wild type lacZ) or pUKC817 (UAA ochre insertion) [42] were grown at 30 • C or 37 • C in liquid SC medium lacking uracil to OD 600 nm of 2-3 and harvested by centrifugation. Cells were washed and resuspended in Z-buffer (60 mM Na 2 HPO 4 , 40 mM NaH 2 PO 4 , 10 mM KCl, 50 mM 2-mercaptoethanol, pH 7). Cell density was measured by absorbance at 600 nm. To 500 µL aliquots, two drops of 0.01% sodium dodecyl sulfate (SDS) solution and chloroform were added and samples mixed on a vortex for 30 s each. Following incubation at 37 • C for 5 min, 100 µL of 4 mg·mL −1 ortho-nitrophenyl-β-galactoside dissolved in Z-buffer was added. Reactions were stopped by the addition of 250 µL 1 M Na 2 CO 3 and absorbance at 420 nm measured. Activity units were calculated by employing Miller's formula [48]. Relative read-through efficiency (%) was calculated by dividing the beta galactosidase activity measured with the pUKC817 construct by the one measured with the pUKC815 construct [33,42]. For each strain, at least three independent cultures were measured with both constructs.

Conclusions
The emerging picture from the presented work and previous studies places Deg1-mediated ψ38 in a role to maintain the function of both Gln isoacceptor tRNAs. However, the presence of ψ38 becomes critical for tRNA function only in the presence of an additional disturbance of the ASL structure. In one case (tRNA Gln UUG ), a strong growth defect indicative of a significant loss of function is observed specifically when deg1 mutants are shifted to conditions known to downregulate the formation of a distinct modification (mcm 5 s 2 U) (Figures 1 and 2). In another case (tRNA Gln CUG ) where this modification is absent per se, the introduction of a destabilization of the ASL can substitute for a hypomodified wobble base and sensitize this tRNA as well for the negative effects of the deg1 mutation (Figures 3 and 4). It appears possible that other tRNA species such as tRNA Tyr GψA do not lose function in the absence of ψ38/39 due to the existence of additional modifications that may stabilize the ASL. For example, tRNA Tyr GψA -derived suppressor tRNAs are known to carry N 6 -isopentenyl-adenosine (i 6 A) in position 37 [7]. The presence of i 6 A has been shown to support efficient ochre read-through [49] and therefore may protect the ASL from destabilization. We assume that multiple modifications of the ASL provide multiple layers of structural protection, which routinely are not ruptured in single tRNA modification mutants and this functional redundancy may explain the paradox situation of missing phenotypes for many of the evolutionary conserved tRNA modification genes.