The many applications of acid urea polyacrylamide gel electrophoresis to studies of tRNAs.

2.1. Preparation of total RNA under acidic conditions

2.2. Deacylation of aminoacylated tRNA and measurement of deacylation rates

Total RNA is isolated under acidic conditions as described above and resuspended in ice-cold sterile water.

• Preparation of bulk deacylated tRNA: A portion of the material is subjected to deacylation by addition of Tris-HCl pH 9.5 to a final concentration of 0.2 M. The deacylation reaction is usually performed at 37 °C for 30−120 min (most tRNAs will be deacylated after 30 min; however, certain tRNAs such as valine and isoleucine tRNAs require longer deacylation treatments [20]); deacylated tRNAs are re-precipitated with ethanol and stored in 10 mM sodium acetate pH 5.0; 10 mM Hepes-NaOH pH 7.4; or 10 mM Tris pH 7.5 at −80 °C.

• Measurement of deacylation rates: Deacylation is performed in 0.2 M Tris-HCl pH 9.2 at 37 °C. Aliquots corresponding to 0.1 A₂₆₀ are removed at appropriate time intervals, mixed with an equal volume of acid urea sample buffer (see section 2.3.1), and quick-frozen on dry-ice. Reaction products are subsequently analyzed by acid urea PAGE/Northern hybridization.

2.3. Acid urea polyacrylamide gel electrophoresis (acid urea PAGE)

tRNAs are separated by acid urea PAGE followed by identification of individual tRNAs using Northern hybridization [5].

2.4. Northern hybridization

3.1. Role of unique features of E. coli initiator tRNA^fMet in initiation of protein synthesis: Formylation

Initiator tRNAs are used exclusively for initiation of protein synthesis, whereas elongator tRNAs are used for the insertion of amino acids into internal peptide linkages. Translation initiation occurs universally with the amino acid methionine (Met) or its derivative formylmethionine (fMet). To fulfill their unique function in protein synthesis, initiator tRNAs possess a number of highly specific properties that direct them to the initiation pathway of translation [6-8]. These include (i) direct binding to the ribosomal P-site and (ii) exclusion from binding to elongation factor necessary for binding to the ribosomal A-site. In contrast to the protein synthesis systems of eukaryotes (cytoplasmic) and archaea, initiator tRNAs in eubacteria and eukaryotic organelles such as mitochondria and chloroplasts are unique in that the methionine attached to the initiator tRNA is formylated to formylmethionine (fMet∼tRNA^fMet), by methionyl-tRNA transformylase (MTF) (Figure 1A). The formyl group acts as a positive determinant for initiation factor IF-2 to select fMet∼tRNA^fMet from the pool of all cellular tRNAs and acts as a negative determinant for elongation factor EF-Tu to discriminate between fMet∼tRNA^fMet and elongator tRNAs [23-25,8]. In addition, the eubacterial fMet∼tRNA^fMet is insensitive to hydrolysis by peptidyl-tRNA hydrolase (PTH) [26], which cleaves the ester linkages between peptide and tRNAs dissociated from the ribosome and thereby allows tRNAs to recycle instead of accumulating as peptidyl∼tRNAs.

For in vivo studies of structure-function relationships of the E. coli initiator tRNA^fMet and to establish the critical role of formylation, various mutant tRNAs were generated and examined for their activity in initiation in vivo. Acid urea PAGE followed by Northern hybridization analysis was used to determine the extent of in vivo aminoacylation and formylation of these mutant tRNAs. Control experiments using markers generated by in vitro aminoacylation and formylation of tRNA^fMet show that the different forms of initiator tRNA – uncharged tRNA^fMet, aminoacylated Met∼tRNA^fMet, and formylaminoacylated fMet∼tRNA^fMet – can be separated clearly (Figure 1B). Results in Figure 1C show that formylation is strongly affected by mutations in the acceptor stem, in particular mutations at the first, second and third base pair. Studies like these (reviewed in [6-8]) proved to be critical for the elucidation of how tRNA^fMet is recognized by methionyl-tRNA synthetase (MetRS) and MTF, and how fMet∼tRNA^fMet is utilized in initiation in eubacteria.

As mentioned above, a well-established difference in translation initiation between E. coli and cytoplasmic protein synthesis of eukaryotes is the use of non-formylated $\mathrm{Met}\sim {\mathrm{tRNA}}_{\mathrm{i}}^{\mathrm{Met}}$ in the latter [27,28]. Ramesh et al. [14] showed that expression of E. coli MTF in S. cerevisiae led to formylation of the cytoplasmic initiator $\mathrm{Met}\sim {\mathrm{tRNA}}_{\mathrm{i}}^{\mathrm{Met}}$ (up to 70%) and possibly to initiation of protein synthesis with formylmethionine. The slow growth phenotype of yeast caused by expression of E. coli MTF could be rescued by co-expressing E. coli polypeptide deformylase (DEF), an enzyme that removes formyl groups from the N-terminus of proteins. Again, the in vivo state of formylation of the cytoplasmic yeast initiator ${\mathrm{tRNA}}_{\mathrm{i}}^{\mathrm{Met}}$ was monitored by acid urea PAGE/Northern hybridization linking the slow growth phenotype of yeast cells expressing MTF in the cytoplasm directly to formylation of the eukaryotic initiator ${\mathrm{tRNA}}_{\mathrm{i}}^{\mathrm{Met}}$ .

A striking example of the use of acid urea PAGE of tRNAs comes from studies of a mutant derived from the E. coli elongator tRNA (Mi:2 tRNA), which is active in initiation of protein synthesis in E. coli [29]. Activity of the mutant tRNA in initiation required the overproduction of E. coli MetRS. Acid urea PAGE/Northern hybridization explained the need for overproduction of MetRS. It was found that Mi:2 tRNA is normally aminoacylated in E. coli with lysine [30] and that Mi:2 tRNA aminoacylated with lysine is not a substrate for MTF [29] (Figure 2). Overproduction of MetRS leads to aminoacylation of a portion of the Mi:2 tRNA with methionine and to its consequent formylation by MTF (Figure 2B, lane 5). Besides highlighting the importance of the amino acid attached to the tRNA for its activity as a substrate for MTF, Figure 2B also shows quite clearly the separation by acid urea PAGE of four different forms of the same tRNA; tRNA, Lys∼tRNA, Met∼tRNA, and fMet∼tRNA.

3.2. Orthogonal suppressor tRNAs for site-specific insertion of unnatural amino acids

An important advance in protein engineering has been the site-specific insertion of unnatural amino acids into proteins in prokaryotes or in eukaryotes. The most common approach is based on the suppression of an amber termination codon (UAG) at a predetermined site in the mRNA by an amber suppressor tRNA aminoacylated with the desired unnatural amino acid [31]. The key requirement for this approach is that the suppressor tRNA is ‘orthogonal’ (Figure 3A), i.e. not recognized by any of the aminoacyl-tRNA synthetases present in the respective expression systems. Otherwise, once the aminoacylated suppressor tRNA has inserted the unnatural amino acid at the designated site in the target protein, it will be re-aminoacylated with a natural amino acid and insert the natural amino acid instead of the analogue, thus generating a heterogeneous pool of target protein molecules. Many such orthogonal suppressor tRNAs have been established over the past years, including suppressor tRNAs derived from the eubacterial tRNA^Tyr [31-36], archaeal tRNA^Tyr [37], yeast tRNA^Phe [38], bacterial tRNA^Trp [39], E. coli tRNA^Gln [15,18], and even human and E. coli initiator tRNAs [40,16,13]. The next requirement is the use of an orthogonal aminoacyl-tRNA synthetase that specifically recognizes the suppressor tRNA but no other tRNA in the cell (Figure 3A) (reviewed in [41-44]). The ultimate step is generation of mutants of the aminoacyl-tRNA synthetase, which use a specific unnatural amino acid instead of the natural amino acid and attach it to the orthogonal suppressor tRNA [37,35,45].

The first example of orthogonal suppressor tRNA/aminoacyl-tRNA synthetase pairs for use in mammalian cells consisted of an amber suppressor tRNA derived from E. coli tRNA^Gln and E. coli glutaminyl-tRNA synthetase (GlnRS) [15]. The suppressor tRNA was expressed in mammalian cells, and its activity as a suppressor tRNA in readthrough of amber codons in a reporter gene was shown to be strictly dependent upon co-expression of E. coli GlnRS. Based on this orthogonal suppressor tRNA/aminoacyl-tRNA synthetase pair, a complete set of orthogonal amber, ochre, and opal suppressor tRNA/E. coli GlnRS pairs was generated [18]. Acid urea PAGE/Northern hybridization was used to confirm their orthogonality in mammalian cells (Figure 3B). Introduction of additional mutations in the anticodon loop of the amber, ochre, and opal suppressor tRNAs led to much enhanced suppression efficiencies (observed increases up to 36, 156, and 200-fold, respectively) allowing the concomitant suppression of two or three termination codons in a mRNA [18].

Another amber suppressor tRNA, orthogonal in mammalian cells and yeast, was derived from the human initiator ${\mathrm{tRNA}}_{\mathrm{i}}^{\mathrm{Met}}$ [16,13]. Mutations in the anticodon (U35A36) and TψC-arm (U50G51:C63A64/U54/C60) of the initiator tRNA abolished its recognition by MetRS and by initiation factor eIF-2. The tRNA mutant containing a CUA anticodon (U35A36 mutation) was now a substrate for E. coli GlnRS, and acted as an amber suppressor in yeast cells expressing E. coli GlnRS [13] suggesting that the tRNA could now bind the elongation factor EF-1. In vivo aminoacylation with glutamine, estimated to be ∼50% (based on acid urea PAGE/Northern hybridization), allowed suppression of amber alleles in a yeast tester strain, S. cerevisiae HEY301−129 (met8−1am, trp1−1am, his4−580am) [32], indicated by growth on minimal media lacking methionine, tryptophan, or histidine (Figure 3C).

3.3. Formation of Cys∼tRNA^Cys in archaea

In most organisms, the synthesis of Cys∼tRNA^Cys is catalyzed by a canonical cysteinyl-tRNA synthetase (CysRS) that attaches cysteine to the corresponding tRNA^Cys. Surprisingly, certain methanogenic archaea (e.g. Methanocaldococcus jannaschii, Methanothermobacter thermoautotrophicus, and Methanopyrus kandleri) lack the gene encoding a canonical CysRS, yet genes encoding tRNA^Cys are present. This raised the intriguing question of how Cys∼tRNA^Cys, an essential component of protein synthesis, is synthesized in these organisms. A number of studies, using incorporation of radiolabeled amino acids into total tRNA, showed that in come cases prolyl-tRNA synthetase (ProRS) from methanogenic archaea could aminoacylate total tRNA with cysteine [46,47], leading to the hypothesis that ProRS attaches proline to tRNA^Pro and cysteine to tRNA^Cys. In other words, ProRS – hence called ProCysRS – could be responsible for the formation of Cys∼tRNA^Cys.

This intriguing hypothesis was put to rest (Figure 4A) [48,12] by aminoacylation of total M. jannaschii tRNA with nonradioactive proline or cysteine, followed by acid urea PAGE and Northern hybridization, using probes specific for tRNA^Pro and tRNA^Cys, to identify the tRNA species that had been aminoacylated [12]. The results showed clearly that purified recombinant ProRS – in addition to its canonical function to generate Pro∼tRNA^Pro – could also aminoacylate tRNA^Pro with cysteine, but was unable to aminoacylate tRNA^Cys with cysteine (Figure 4B). In addition, acid urea PAGE revealed a clear difference in mobility of Pro∼tRNA^Pro and Cys∼tRNA^Pro. These data showed that ProRS cannot be responsible for generation of Cys∼tRNA^Cys in vivo.

A subsequent study solved the mystery of how archaea lacking a canonical CysRS synthesize Cys∼tRNA^Cys [49]. A two-step pathway was discovered: first, tRNA^Cys is aminoacylated with O-phosphoserine (Sep) by O-phosphoseryl-tRNA synthetase (SepRS) forming the intermediate Sep∼tRNA^Cys; in a second step, Sep∼tRNA^Cys is converted to the final product Cys∼tRNA^Cys by Sep-tRNA:Cys-tRNA synthetase (SepCysS) (Figure 5A). Acid urea PAGE/Northern hybridization analysis of tRNA samples aminoacylated with a filtered M. jannaschii S-100 total cell extract (in the absence and presence of appropriate amino acid mixtures) was used to detect the formation of both Sep∼tRNA^Cys and Cys∼tRNA^Cys. Sep∼tRNA^Cys formation was further confirmed using purified recombinant M. jannaschii SepRS (Figure 5B). In requiring more that one step for the synthesis of Cys∼tRNA^Cys, the newly identified SepRS/SepCysS pathway in archaea is similar to the biosynthesis of Asn∼tRNA^Asn and Gln∼tRNA^Gln (in certain eubacteria and archaea), and of Sec∼tRNA^Sec [50].

3.4. Identification of the AUA-reading minor isoleucine tRNA in archaea

Annotation of the complete genome of the extreme halophilic archaeon Haloarcula marismortui does not include a tRNA for translation of AUA, the rare codon for isoleucine. This is a situation typical for most archaeal genomes sequenced to date. One hypothesis stated that the AUA-reading isoleucine tRNA is derived through alternative splicing of a transcript to produce both a tryptophan tRNA and the isoleucine tRNA [51]. However, using acid urea PAGE, we have shown that the minor AUA-decoding isoleucine tRNA in H. marismortui and other archaeal species is derived from a CAU anticodon-containing tRNA, currently annotated as methionine tRNA, in which C34 in the anticodon is post-transcriptionally modified (Figure 6A) [11]. In addition, we show that the post-transcriptional modification of C at position 34 in the anticodon of this tRNA, responsible for the switch in amino acid and decoding specificity, is different from those present at position 34 of isoleucine tRNA species in eubacteria and in eukaryotes [52,53].

To investigate whether either of the two CAU anticodon-containing elongator tRNAs, annotated as tRNA_12 and tRNA_34 (The Genomic tRNA Database at http://lowelab.ucsc.edu/GtRNAdb) in H. marismortui, is possibly aminoacylated with isoleucine in vivo, we measured the kinetics of chemical deacylation of aminoacyl-tRNAs corresponding to tRNA_12 and tRNA_34 (Figure 6A), exploiting the well-known fact that the stability of the ester link between tRNA and amino acid is determined by the nature of the amino acid attached to the tRNA [20]; with isoleucine and valine being the most stable and methionine being much less stable. Aminoacyl-tRNAs present in total RNA were isolated from H. marismortui under acidic conditions to retain the aminoacyl-ester linkage and were subjected to base-catalyzed deacylation. Samples were subsequently analyzed by acid urea PAGE and kinetics of deacylation followed by Northern hybridization using probes specific for tRNA_12 and tRNA_34, respectively (Figure 6B). Aminoacyl∼tRNA_12 is almost completely deacylated within 15 min, whereas deacylation of aminoacyl∼tRNA_34 is incomplete even after 120 min. The half-lives of deacylation of aminoacyl∼tRNA_12 is ∼5 min which is the same as that of Met∼ ${\mathrm{tRNA}}_{\mathrm{i}}^{\mathrm{Met}}$ ; whereas that of aminoacyl∼tRNA_34 is ∼35 min, which is the same as that of $\mathrm{Ile}\sim {\mathrm{tRNA}}_{\mathrm{GAU}}^{\mathrm{Ile}}$ [11]. These data, in combination with in vitro re-aminoacylation using purified recombinant H. marismortui MetRS and isoleucyl-tRNA synthetase (IleRS) , show that tRNA_12 is the elongator methionine ${\mathrm{tRNA}}_{\mathrm{CAU}}^{\mathrm{Met}}$ , and that tRNA_34 is the minor tRNA^Ile (renamed ${\mathrm{tRNA}}_{\mathrm{XAU}}^{\mathrm{Ile}}$ , X representing an unknown modified base).

The determination of deacylation kinetics has been used earlier to characterize the nature of the amino acid attached in vivo to a mutant human initiator tRNA initiating protein synthesis from non-AUG codons [17], and to establish the importance of the anticodon sequence for recognition of tRNAs by MetRS and valyl-tRNA synthetase in archaea [54].

3.5. Base modifications

Some modified nucleosides play critical roles in the structure and function of tRNAs; in particular, base modifications in the anticodon loop are thought to be important for recognition of tRNAs by their cognate aminoacyl-tRNA synthetases and/or for codon/anticodon interactions. Post-transcriptional modifications at the wobble position 34 and the purine 37, 3' adjacent to the anticodon, are known to be involved in stabilizing the geometry of the anticodon loop and its dynamics for optimum codon/anticodon interaction [55,56]. Early work on tRNA analysis used combinations of column, paper, and thin layer chromatography of partial and complete nuclease digests of tRNAs to identify and characterize modified nucleosides [2,3]. Recent analyses include separation of tRNA fragments or nucleoside mixtures by liquid chromatography followed by mass spectrometric analysis of individual peak fractions (LC-MS), pioneered by the laboratories of Nishimura, McCloskey and Crain [57,58].

One modified base of particular interest is lysidine, a modified cytosine with a lysine side-chain attached to the O² of cytidine. Found in the wobble position of eubacterial minor isoleucine tRNAs (${\mathrm{tRNA}}_{\mathrm{LAU}}^{\mathrm{Ile}}$ or ${\mathrm{tRNA}}_2^{\mathrm{Ile}}$), lysidine has a dual role (reviewed in [53]): (i) the presence of lysidine in the anticodon allows ${\mathrm{tRNA}}_{\mathrm{LAU}}^{\mathrm{Ile}}$ to translate specifically the minor isoleucine codon AUA; and (ii) modification of cytidine to lysidine is necessary for recognition of ${\mathrm{tRNA}}_{\mathrm{LAU}}^{\mathrm{Ile}}$ by IleRS and, at the same time, for blocking its aminoacylation by MetRS. The enzyme, tRNA^Ile-lysidine synthetase (TilS), responsible for modifying the minor isoleucine tRNA has been identified and characterized [59] (Figure 7A), and the crystal structure of the enzyme from Aquifex aeolicus has been solved [60]. For a recent study on archaeal minor isoleucine tRNA ([11]; see section 3.4), we have devised a simple method, involving the use of acid urea PAGE, to confirm the presence of lysidine in tRNAs. Total E. coli tRNA or a T7 transcript corresponding to E. coli ${\mathrm{tRNA}}_{\mathrm{LAU}}^{\mathrm{Ile}}$ is incubated with sulfo-NHS-LC-biotin [N-hydroxysuccinimide (NHS) ester of biotin], which reacts with the free α-NH₂ group of lysidine. Both, in vitro lysidinylation of the control T7 transcript using TilS and subsequent biotinylation of tRNAs can be monitored by acid urea PAGE followed by Northern blot analysis using a ${\mathrm{tRNA}}_{\mathrm{LAU}}^{\mathrm{Ile}}$-specific probe (Figure 7B). Introduction of lysidine into the T7 transcript causes a clear mobility shift (Figure 7B, compare lanes 1 and 3). This is expected since the α-NH₂ group of lysidine carries a positive charge under the acidic conditions of the gel electrophoresis. Biotinylation of the modified transcript causes an additional supershift (Figure 7B, lane 4), most likely caused by the added bulk from the biotin; as expected, a T7 transcript lacking lysidine does not show a shift (Figure 7B, lane 2). Reaction of total tRNA from E. coli with the NHS-activated biotin, also gave rise to a clear shift of ${\mathrm{tRNA}}_{\mathrm{LAU}}^{\mathrm{Ile}}$ confirming the presence of lysidine (Figure 7B, compare lanes 5 and 6); several minor shifts upon biotinylation were also observed, probably corresponding to partial reaction with other base modifications such as 3-(3-amino-3-carboxypropyl) uridine present in E. coli ${\mathrm{tRNA}}_{\mathrm{LAU}}^{\mathrm{Ile}}$ . In the course of the same study, we found that minor isoleucine tRNAs (${\mathrm{tRNA}}_{\mathrm{XAU}}^{\mathrm{Ile}}$) from H. marismortui (see section 3.4) and other archaea could not be biotinylated [11]. Thus, the base modification at C34 in the archaeal tRNA is not lysidine, which is consistent with the fact that, so far, no protein with significant homology to TilS could be identified in archaea. These results are intriguing from an evolutionary point of view; the overall pathway for supplying the cell with a minor isoleucine tRNA appears to be similar in eubacteria and most archaea and requires the post-transcriptional modification of a cytidine. Nevertheless, the base modification and the enzyme(s) responsible for this switch appear to be different between eubacteria and archaea.