Proofreading and Aminoacylation of tRNAs Before Export from the Nucleus

See allHide authors and affiliations

Science  11 Dec 1998:
Vol. 282, Issue 5396, pp. 2082-2085
DOI: 10.1126/science.282.5396.2082


After synthesis and processing in the nucleus, mature transfer RNAs (tRNAs) are exported to the cytoplasm in a Ran·guanosine triphosphate–dependent manner. Export of defective or immature tRNAs is avoided by monitoring both structure and function of tRNAs in the nucleus, and only tRNAs with mature 5′ and 3′ ends are exported. All tRNAs examined can be aminoacylated in nuclei of Xenopusoocytes, thereby providing a possible mechanism for functional proofreading of newly made tRNAs. Inhibition of aminoacylation of a specific tRNA retards its appearance in the cytoplasm, indicating that nuclear aminoacylation promotes efficient export.

In eukaryotic cells, most RNAs are processed in the nucleus. Nuclear maturation of tRNAs involves base modification, processing of the 5′ and 3′ ends, and in some cases splicing of tRNA precursors (pre-tRNAs), with enzymes such as ribonuclease P (RNase P), tRNA nucleotidyl transferase, and tRNA splicing endonuclease and ligase (1–5). In addition, low amounts of factors that normally interact with tRNAs during protein synthesis in the cytoplasm have been identified in the nucleus, including aminoacyl-tRNA synthetases (5) and translation elongation factor EF-1α (6).

A complex involved in tRNA export contains tRNA, the tRNA export receptor exportin-t, and the guanosine 5′-triphosphate (GTP)–bound form of Ran (Ran·GTP) (7), a GTPase required for nucleocytoplasmic transport of both RNAs and proteins (8, 9), but no cargo-specific adapter protein. In X. laevis oocytes, a single pathway appears to function for export of all tRNAs (7,10); in Saccharomyces cerevisiae there may be more than one pathway, because cells lacking the homologous tRNA export receptor (Los1p) are viable (11, 12).

Only mature tRNAs are found in the cytoplasm (1,2), and defective tRNAs, which might disrupt translation, appear to be excluded. We show here that in X. laevis oocytes the absence of nonfunctional tRNAs from the cytoplasm can be explained by kinetic differences in maturation steps and by monitoring of mature tRNAs for their abilities to be aminoacylated, before export from the nucleus.

To determine if processing enzymes contribute to the proofreading of tRNA transcripts, precursors of X. laevistRNATyr (13) containing an intron (intervening sequence, IVS) and extra sequences at the 5′ and 3′ ends (5′, 3′ ext.) were injected into nuclei of X. laevis oocytes (14). Wild-type transcripts were rapidly matured to tRNA (Fig. 1A, lanes 2 and 3), with the transient accumulation in the nucleus of a heterogeneous population of processing intermediates that were spliced (ΔIVS) but not end-matured (5′, 3′ ext.) (15). In contrast, injection of high, nonphysiological amounts of the same precursor led to generation of a different processing intermediate (Fig. 1B, lanes 2 and 3) that was fully matured at its 5′ and 3′ ends but not spliced (+IVS) (15). A comparable dose-dependent switch in processing pathways was observed when a DNA template for tRNATyr was injected (compare lanes 1 of Fig. 1, C and D); low to moderate amounts of tRNATyr genes produced exclusively spliced intermediates (Fig. 1C), whereas high amounts led to accumulation of the intron-containing intermediate (Fig. 1D). The change in the types of intermediates that accumulated in response to different amounts of pre-tRNA shows that excess substrate saturates splicing more easily than it saturates end maturation.

Figure 1

Structural requirements for nuclear maturation and export of tRNATyr. (A) Disruption of conserved tertiary interactions within tRNATyrinterferes with both processing and export. Low amounts of32P-labeled RNAs corresponding to wild-type (WT) or mutant (T55C and G57C) precursors ofX. laevis tRNATyr were injected into nuclei ofXenopus oocytes (14), and 1 (lanes 2, 4, 7, 9, 12, and 14) and 2 hours (lanes 3, 5, 8, 10, 13, and 15) later the intracellular distributions of the injected primary pre-tRNA transcripts (lanes 1, 6, and 11), the spliced (ΔIVS) but not end- matured (5′, 3′ ext.) processing intermediates, and the mature tRNA (tRNATyr), were determined by analysis of total nuclear (N) and cytoplasmic (C) RNAs. The small amounts of injected precursor RNAs in the cytoplasm probably resulted from inefficient retention (22) shortly after injection. (B) Saturation of the tRNA splicing machinery leads to nuclear accumulation and export of nonspliced, end-matured tRNATyr. High amounts of wild-type primary transcript were injected into oocyte nuclei, and export of unspliced (+IVS) and mature tRNATyr was monitored at 0.75 (lanes 2 and 5), 1.5 (lanes 3 and 6), and 4 hours (lanes 4 and 7) after injection. (C) Disruption of the RanGTPase system blocks export but not processing of tRNATyr. Low amounts of the gene for X. laevis tRNATyr (14) were injected into control oocytes (lanes 1 and 2) or oocytes preinjected with RanT24N, and export and processing of the newly made tRNATyr transcripts were monitored 4.5 hours later. The processing intermediates are fully spliced but end-immature (18), as observed in (A) upon injection of low amounts of WT pre-tRNATyr. (D) Nonspliced tRNATyris exported via the tRNA pathway. High amounts of the gene for X. laevis tRNATyr (14) were injected with [α-32P]GTP into nuclei of control oocytes (lanes 1 and 2) or oocytes preinjected with anti-nucleoporin (mAb 414, lanes 3 and 4) or wheat germ agglutinin (lanes 5 and 6), and export and processing of the newly made tRNATyr transcripts were analyzed 20 hours later.

In contrast to the intron-containing intermediate (Fig. 1, B and D), the spliced but end-immature intermediates were not exported (Fig. 1, A and C), indicating that end maturation but not splicing is a prerequisite for export. When lower, more physiological amounts of precursor are present in oocyte nuclei, splicing is an early event that occurs before end maturation, thereby ensuring that processing intermediates are not normally exported. Previous studies with very high amounts of tRNATyr template DNA reported accumulation and export of the unspliced intermediate, leading to the inappropriate conclusion that splicing is normally the last step in the processing pathway (16).

The unspliced tRNA intermediate appears to be exported by the pathway used for mature tRNA, because both types of molecules responded similarly to export inhibitors. Export was resistant to an antibody to nucleoporins [monoclonal antibody (mAb) 414] (17) (Fig. 1D, lanes 3 and 4), but was competed by coinjected tRNAPhe (18) and was sensitive to injection of the lectin wheat germ agglutinin (17) (lanes 5 and 6). Exportin-t–mediated export of an unspliced intermediate has recently been observed by others (10).

Although depletion of nuclear Ran·GTP by injection of the dominant-negative mutant RanT24N or the RanGTPase-activating protein RanGAP (9, 19) blocked export of the mature tRNA, as shown by others (9), it had no effect on splicing or end maturation of pre-tRNATyr (Fig. 1C, compare lanes 1 and 3) (15, 18). In yeast cells disruption of the RanGTPase system leads to rapid accumulation of unspliced intermediates (20), perhaps reflecting tighter coupling between export and tRNA splicing in those cells.

The influence of tertiary structure on maturation and export of tRNATyr was assessed by injection of precursors having mutations that would disrupt conserved interactions between loops I and III (the D- and T-ψ-C-G loops, respectively). The precursor with a T55C mutation or with either C or T at position 57 (normally G in tRNATyrC) (13) was processed much more slowly than wild-type (Fig. 1A, lanes 7, 8, 12, and 13), and the unprocessed mutant precursors were unstable, providing evidence for proofreading during tRNA maturation (18, 21). Those few mutant tRNAs that were processed to the mature form were exported poorly (lanes 7 to 10) (21). Also, the binding affinity of exportin-t in vitro is reported to be higher for wild-type tRNA than for end-immature or mutated tRNAs (7, 10), indicating a role of exportin-t in export cargo selection.

The requirement in tRNA export for maturation of both 5′ and 3′ ends was demonstrated by the export behavior of several variants of the human initiator tRNA, tRNAi Met (14). tRNAi Met with three extra nucleotides at its 5′ end and the mature 3′ end appeared in the cytoplasm only after processing of the 5′ end (Fig. 2A). tRNAi Met lacking the 3′-terminal adenosine was processed rapidly at the 5′ end but was not exported until the 3′ end had also been matured (Fig. 2B); likewise, when generation of the 3′ terminal CCA by tRNA–nucleotidyl transferase was not possible (because of the lack of unpaired nucleotides at the 3′ end), the transcript remained in the nucleus (Fig. 2C) (22). As expected, a transcript having three extra nucleotides at its 5′ end but lacking three (CCA) at the 3′ end (23) was processed at both ends before export, yielding a molecule with the same length and mobility as the precursor (Fig. 2D).

Figure 2

Requirements for both 5′ and 3′ end maturation before tRNA export. 32P-labeled RNAs corresponding to human tRNAi Met with the indicated 5′ and 3′ terminal sequences (14) were injected into oocyte nuclei, and the intracellular distributions of the RNAs were analyzed at 45, 100, and 220 min (A and B) and at 30, 60, and 120 min (C and D). The tRNAi Met injected in (A) and (D) were primary transcripts synthesized in vitro, whereas the RNAs injected in (B) and (C) were generated by periodate oxidation and β-elimination (14) of the RNAs used in (A) and (D), respectively. In all cases, the major cytoplasmic forms correspond to fully mature tRNAi Met, and the small amount of this form in (C) (lanes 5 to 7) is due to heterogeneity of the 3′ ends of the injected RNA (18).

The requirements for both correct folding and mature ends of tRNAs raised the possibility that nuclear tRNAs might be subject to “functional proofreading” by aminoacylation before export. To test for aminoacylation of nuclear tRNAs, potential aminoacyl-tRNA linkages were stabilized by isolation and electrophoresis of tRNAi Met under acidic conditions (24); brief incubation at pH 9 (which causes deacylation) permitted comparison with markers of acylated and deacylated forms of cytoplasmic tRNAi Met (Fig. 3A, lanes 4 and 5). Within 20 min of injection a new form of nuclear tRNAi Met accumulated that comigrated with cytoplasmic aminoacylated tRNAi Met (lane 2). Like the acylated cytoplasmic tRNAi Met, the nuclear RNA was resistant to periodate oxidation unless first deacylated at pH 9 (18), thereby confirming the presence of a blocking group at the 3′ end, presumably an amino acid.

Figure 3

Aminoacylation of tRNA within the nucleus. (A) Aminoacylation of tRNAi Metoccurs in the nucleus.32P-labeled pre-tRNAi Met (Fig. 2A) was injected into oocyte nuclei, and the intracellular RNA distributions were monitored 5 and 20 min later (upper panels). The state of aminoacylation (lower panels) was assessed by isolation and analysis of RNAs under acidic conditions (14, 24) without (−) or with (+) prior deacylation. (B) Nuclear aminoacylation of tRNAi Met occurs independently of export. Export (upper panels) and the extents of aminoacylation (lower panel) of tRNAi Met were monitored 1.5 hours after nuclear injection into oocytes that had been depleted of Ran·GTP (14) by prior intranuclear injection of RanGAP protein (+) or control oocytes (−). (C) Nuclear aminoacylation of tRNAi Met is specific. Mixtures of unlabeled human tRNAi Met (lane 1) or yeast tRNAPhe (lane 2) (4 to 5 pmol/oocyte) and 32P-labeled control RNAs were injected into nuclei of oocytes depleted of Ran·GTP and labeled with35S-methionine (14). Five hours later, nuclear RNAs were isolated and analyzed under acidic conditions. (D) tRNATyr also is aminoacylated in the nucleus. Newly made nuclear and cytoplasmic tRNATyrtranscripts (Fig. 1D, lanes 1 and 2) were isolated under acidic conditions and analyzed in both neutral (lanes 1 and 2) and acidic gels (lanes 3 to 6). Aminoacylated forms of tRNATyr and end-mature but unspliced pre-tRNATyr (lanes 3 and 5) are indicated by lower and upper brackets, respectively, whereas nonacylated forms (lanes 4 and 6) are indicated by dots. Differences in the extents of aminoacylation were confirmed by assaying for resistance to periodate oxidation (18).

To test if the aminoacylated tRNAi Met in the nuclear fraction represented molecules that had been exported but remained associated with the outside of the nucleus, transport was blocked by depletion of the nuclear pool of Ran• GTP (compare with Fig. 1D) (9, 19). When tRNA export was inhibited in this manner, essentially all of the nuclear tRNAi Metmigrated as the aminoacylated form (Fig. 3B, lanes 1 and 2), demonstrating that it was aminoacylated within the nucleus.

The specificity of nuclear aminoacylation was examined by injection of unlabeled tRNAi Met or tRNAPhe into oocyte nuclei containing 35S-methionine, under conditions where tRNA export was blocked. Only the tRNA of nuclei receiving tRNAi Met was labeled by the 35S-methionine (Fig. 3C), and this label was released from the tRNAi Met by incubation under deacylation conditions (18). Thus, nuclear tRNAi Met was aminoacylated with its cognate amino acid.

Other tRNAs also can be aminoacylated in the nucleus. Both the mature and the unspliced forms of tRNATyr were aminoacylated in the nucleus, although aminoacylation was more extensive for the mature tRNA (Fig. 3D, lanes 3 and 4) (25); in the cytoplasm, both forms were almost fully charged (lane 5). Injection of pxt210 DNA (26) resulted in the nuclear accumulation ofX. laevis tRNAs specific for alanine, asparagine, leucine, lysine, methionine, tyrosine, and phenylalanine. By both the deacylation and periodate sensitivity assays, all seven nuclear tRNAs appeared to be aminoacylated (18). Thus it is very likely that all tRNAs can be aminoacylated with their cognate amino acids before export from the nucleus.

Confirmation that the mobility shift of the nuclear tRNATyr is due to aminoacylation was obtained through the use of tyrosyl sulfamoyl adenosine (Tyr-AMS), a strong competitive inhibitor of the requisite intermediate, tyrosyl-AMP (Tyr-AMP) (27). Tyr-AMS blocked aminoacylation of tRNATyr (both nuclear and cytoplasmic) but not tRNAi Met, as assayed by the absence or presence of mobility shifts upon incubation under deacylation conditions (Fig. 4A, left panel; compare lanes 3 and 4, and lanes 7 and 8; right panel, lanes 2 and 4). As expected, protein synthesis was strongly inhibited in oocytes treated with Tyr-AMS or the asparaginyl-AMP analog Asn-AMS, reflecting the absence of specific charged tRNAs (18).

Figure 4

Effect of blockage of aminoacylation on tRNA export. (A) Tyr-AMS blocks both nuclear and cytoplasmic aminoacylation of tRNATyr. pre-tRNATyr (left) or pre-tRNAi Met (right) was injected into nuclei of oocytes that had been preinjected in the cytoplasm with Tyr-AMS (+) or treated with cycloheximide (−) (14), and 50 and 100 min later the extent of aminoacylation of the tRNAs in the nuclear (N) and cytoplasmic (C) fractions were assayed. Differences in gel mobilities of the nuclear nonacylated tRNAs may represent differences in modifications. (B) Tyr-AMS specifically inhibits export of tRNATyr. Oocytes were injected as in (A), and RNA export was monitored with time. The intracellular distributions of tRNATyr, tRNAi Met, and U1Sm at 50 min (lanes 2, 4, 6, and 8) and 100 min (lanes 3, 5, 7, and 9) after RNA injection are shown; the injected RNAs are shown in lane 1 (Inj).

The effect of nuclear aminoacylation on tRNATyr export was monitored with Tyr-AMS. In the absence of aminoacylation, export of tRNATyr was significantly retarded (Fig. 4B, top panel, compare lanes 2 to 5 with lanes 6 to 9), but export of tRNAi Met (middle panel) and U1Sm RNA (bottom panel) was not affected; the control oocytes received cycloheximide, to account for nonspecific effects due to inhibition of protein synthesis. In a similar series of experiments, Asn-AMS specifically blocked aminoacylation of nuclear and cytoplasmic Asn-tRNAAsn and interfered with the export of tRNAAsn, but had no effect on aminoacylation and export of tRNATyr (18). Thus, nuclear aminoacylation affects the rate of export of several, and perhaps all, tRNAs (28).

Although aminoacylation is important for tRNA export, it is not essential under the conditions used here, because uncharged tRNAs can be exported, albeit more slowly. Even with the lowest amounts of tRNAs injected, the concentration in the nucleus far exceeds that of endogenous nuclear tRNAs, possibly driving formation of tRNA-containing complexes that normally might not be stable. Aminoacylation greatly strengthens the binding of tRNAs to the translation factor EF-1α (29), and charging could serve a similar function in promoting the association of nuclear tRNAs with exportin-t (plus Ran·GTP); artificially high levels of nuclear tRNA would diminish but not abolish the importance of this effect. Similarly, the high amounts of intron-containing processing intermediates that accumulate when splicing is saturated (Fig. 1B) favor their inappropriate export.

Proofreading of transcripts before export prevents precursors from being prematurely separated from nuclear processing enzymes. In oocyte nuclei, splicing of pre-tRNAs normally occurs faster than end maturation (Fig. 1A), thereby ensuring that unspliced intermediates are not exported. Moreover, export of any end-mature but unspliced tRNA is reduced because export is coupled to nuclear aminoacylation (Fig. 4) and unspliced tRNAs are inefficiently charged (Fig. 3D) (1). Such coupling may account for the absence of unspliced tRNAs from the cytoplasm in yeast, where splicing and end maturation can occur in either order (30).

Unspliced mRNAs are generally retained in the nucleus, and it is likely that other RNAs are also subject to proofreading before export, by mechanisms that are yet to be defined. The presence of aminoacylated tRNA in the nucleus supports proposals that some mRNAs may undergo functional monitoring before or during export from the nucleus, as a step in nonsense-mediated decay (31); similarly, newly made ribosomes may be subjected to a functional quality control check before export.

  • * To whom correspondence should be addressed. E-mail: dahlberg{at}


View Abstract

Navigate This Article