|Home | About | Journals | Submit | Contact Us | Français|
The fidelity of tRNA aminoacylation is dependent in part on amino acid editing mechanisms. A hydrolytic activity that clears mischarged tRNAs typically resides in an active site on the tRNA synthetase that is distinct from its synthetic aminoacylation active site. A second pre-transfer editing pathway that hydrolyzes the tRNA synthetase aminoacyl adenylate intermediate can also be activated. Pre- and post-transfer editing activities can co-exist within a single tRNA synthetase resulting in a redundancy of fidelity mechanisms. However, in most cases one pathway appears to dominate, but when compromised, the secondary pathway can be activated to suppress tRNA synthetase infidelities.
The aminoacyl-tRNA synthetases (aaRSs) are responsible for committing amino acid–tRNA pairs in the first step of protein synthesis. Once the amino acid is linked to its cognate tRNA isoacceptor, it is passed from the aaRS to an elongation factor and then ultimately to the ribosome for incorporation into the nascent polypeptide chain. Decades of experimental and theoretical studies have emphasized that the synthetases and other tRNA partners are exquisitely adapted to ensure fidelity. Indeed, compromised fidelity results in amino acid toxicities that cause cell death in microbes  and neurological disease in mammals .
Even before the tRNA aminoacylation reaction was discovered in 1956 , Linus Pauling predicted that proteins would lack the discriminatory power to fully distinguish isosteric substrates such as amino acids that differed by a single methyl group ( i.e. ala-nine and glycine). Yet early in vivo studies using chick ovalbumin extractions indicated that the fidelity of protein synthesis was quite high . Subsequently, Alan Fersht proposed a double sieve model for the aaRSs to account for this high fidelity . In essence, he hypothesized that when one enzyme active site cannot adequately discriminate between pairs of structurally related amino acids, two active sites with two different strategies for amino acid recognition could increase fidelity for protein synthesis to the threshold levels that are required by the physiology of the cell.
Since the double sieve model was first proposed  biochemistry and structural biology investigations have revealed that about half of the aaRSs contain a wholly separate domain with a hydrolytic active site for amino acid editing . Thus, only those aaRSs where amino acid discrimination is sufficiently threatened have evolved to meet the fidelity demands of the cell. In addition, independent hydrolytic tRNA deacylases aid in clearing mischarged tRNAs, and in some cases provide a third auxiliary sieve to ensure fidelity [8,9]. These editing active sites of the aaRS and the tRNA deacylases must be crafted to eliminate binding or translocation of the activated cognate amino acid, so that it can bypass the hydrolysis pathway and allows correctly charged tRNA to be efficiently released to EF-Tu for protein synthesis. Uncoupling discrimination of the cognate amino acid from the editing active site would result in an ATP-consumptive futile cycle of aminoacylation and editing, such as in the case of the Escherichia coli leucyl-tRNA synthetase (LeuRS) T252A mutant that confers deacylation of correctly charged Leu-tRNALeu (; Fig. 1).
Amino acid fidelity relies on a number of checkpoints as the activated amino acid moves through the two-step aminoacylation reaction of the aaRS and onto the elongation factor, which is responsible for binding all of the charged tRNAs and shuttling them to the ribosome. Post-transfer editing by the aaRS targets the mischarged tRNA for hydrolysis to cleave the incorrect amino acid (; Fig. 2) and clear its mistakes before they are incorporated into the proteome as statistical mutations. EF-Tu can also take full advantage of the aaRS post-transfer editing activity by recycling a prematurely released mischarged tRNA back to the aaRS . Post-transfer editing activity by the aaRS or an independent tRNA deacylase can be readily investigated by monitoring the deacylation activity of the editing enzyme in the presence of mischarged tRNA. In addition, a number of X-ray crystal structures have clearly defined the hydrolytic active site in the multi-domain editing aaRS that clips the amino acid from the mischarged tRNA .
Pre-transfer editing hydrolytically clears the misactivated aminoacyl adenylate that is produced after the first step of the aminoacylation reaction ([13–15]; Fig. 2). Because of the transient nature of the adenylate intermediate and its instability in aqueous environments, the pre-transfer amino acid editing pathway has proven difficult to isolate and characterize. Thus, it has long been controversial since it was first proposed by Berg and coworkers [13,15] to explain the fidelity mechanism of isoleucyl-tRNA synthetase (IleRS).
Based on rapid quench kinetic approaches by Fersht, two fidelity models emerged with IleRS  and valyl-tRNA synthetase (; ValRS) in which they respectively relied upon pre- and post-transfer editing to clear their mistakes and achieve fidelity of protein synthesis (Fig. 3A). In addition, work from Friedrich Cramer's laboratory suggested that the fidelity strategies for LeuRSs from different origins (yeast cytoplasmic versus E. coli) could partition between pre- and post-transfer editing mechanisms to ensure accuracy (; Fig. 3A). This would suggest then that distinct sets of molecular determinants dictate pre- and post-transfer editing activities.
Significantly, although IleRS and yeast cytoplasmic LeuRS were proposed to maintain fidelity via a pre-transfer editing mechanism, both are also quite capable of clearing mischarged tRNAs through their homologous CP1 editing domains [11,18]. This suggests a redundancy of fidelity mechanisms and raises questions about when pre- versus post-transfer editing is activated. It is possible that one activity is simply more efficient and predominates as the amino acid clearance mechanism. Moreover, it is again likely that sets of molecular determinants drive the balance of editing activities between the pre- and post-transfer editing pathways. E. coli wild-type LeuRS has been reported to maintain amino acid fidelity exclusively by a post-transfer editing mechanism .
However, three different sets of mutations in this enzyme have unmasked an inherent pre-transfer editing activity. One single mutation A293D that is located on the surface of the CP1 editing domain  reduced the levels of mischarged tRNALeu when introduced into a post-transfer editing-inactivated LeuRS (; Fig. 3B). It is noteworthy that this aspartic acid substitution in E. coli LeuRS is actually conserved in the primary sequence alignment of many other LeuRSs and could suggest that this site is part of a mechanistic switch point to tip the balance between pre- and post-transfer editing.
A second mutation on the surface of the canonical aminoacylation core at Lys 186 of a post-transfer editing-inactivated LeuRS also enhanced fidelity (; Fig. 3C). The Lys 186-based surface peptide on the aminoacylation domain is in good proximity to the Ala/Asp 293 – based surface peptide on the CP1 domain and could implicate domain–domain interactions within the synthetase that are important to pre-transfer editing. It is also significant that the lysine residue in LeuRS is found at a homologous site in IleRS. Interestingly, in IleRS this conserved lysine has been suggested to serve as a “hinge” that is critical to the enzyme's fidelity mechanism . Thus, it is possible that this fidelity mechanism that appears to be based at least in part on this surface lysine may universally influence whether pre-transfer editing is dominant in the homologous LeuRS, IleRS, and ValRS enzymes.
The CP1 editing domain has also been completely deleted in E. coli and yeast mitochondrial LeuRS . As would be expected, post-transfer editing was abolished, but surprisingly both enzymes maintained amino acid fidelity by activation of a pre-transfer editing mechanism (Fig. 3D). These mutational examples in LeuRS demonstrate that pre- and post-transfer editing activities can co-exist within a single enzyme, albeit one pathway can be masked because the other pathway operates as the dominant mechanism to maintain amino acid fidelity. In addition, the partition that sets the aaRS's dependence on pre- versus post-transfer editing can be shifted and in some cases the shift appears to occur quite readily. At least in the examples presented above, the shift is dependent on a set of protein molecular determinants that would mechanistically block and/or activate one of the pathways. However, it is also possible that the shift between pre- and post-transfer editing (or vice versa) could be triggered by the tRNA or environmental cues within the cell. For those aaRSs that activate multiple non-cognate amino acids, the shift could even be dictated by the identity of the non-cognate amino acid. This would be reminiscent of the ProRS that requires a triple sieve to clear misactivated alanine and cysteine .
Coexistence of pre- and post-transfer editing activities within an aaRS raises questions about the co-localization or separation of active sites. Crystal structures for all the editing aaRSs, as well as extensive biochemical experiments, clearly demonstrate that the post-transfer editing active site resides in a discrete domain that is completely separated from the aminoacylation active site . This is consistent with the original predictions of the double sieve model of aaRS fidelity that was proposed by Fersht and Ding-wall .
The location of pre-transfer editing activity has been more perplexing. X-ray crystal structures of editing tRNA synthetases that are bound to pre- and post-transfer substrate analogs have suggested that these two hydrolytic editing active sites could physically overlap [18,23,24], In LeuRS, the pre- and post-transfer substrate analogs shared common amino acid and adenine binding pockets in the editing active site . To accommodate the different geometries, the ribose ring and phosphate linkages contort in relation to each other. In IleRS, sub-binding sites within the hydrolytic active site of the editing domain were identified based on the crystal structure . Likewise, molecular determinants for pre- and post-transfer editing activity in IleRS have been mutationally dissected within these overlapping active sites .
Pre-transfer editing within the CP1 domain would require that the labile misactivated adenylate is translocated over 30 Å from the aminoacylation site to the editing active site. While it remains unclear how this translocation mechanism might operate, the first X-ray crystal structure for the tRNA-bound complex of IleRS suggested a putative tRNA-induced channel for translocation of the small adenylate molecule . When considering adenylate translocation to a remote editing domain, it is also important to consider whether correctly activated adenylate is discriminated by the translocation mechanism. At least in IleRS, fluorescence-tagged ATP substrates have been used to show that the evacuation of the adenylate binding site is coupled with amino acid editing and suggests that translocation is restricted to misactivated amino acid .
As an alternate, the pre-transfer editing pathway has also been proposed to be controlled by the aminoacylation active site. The best characterized example is methionyl-tRNA synthetase (MetRS), which uses its aminoacylation active site to aid an intramolecular cyclization of bound homocysteinyl adenylate to form a thiolactone . Lactone production during editing has also been reported for yeast cytoplasmic LeuRS . ProRS enzymatically hydrolyzes a misactivated adenylate intermediate and also edits by selectively releasing the labile intermediate to the aqueous milieu of the cell, albeit this latter mechanism appears to be the minor pathway [29,30]. Similar to MetRS, both of these ProRS pre-transfer editing pathways are tRNA-independent. In contrast, a non-editing aaRS, glutaminyl-tRNA synthetase (GlnRS) was shown via kinetic measurements to enzymatically hydrolyze the cognate glutaminyl-adenylate in a tRNA-dependent mechanism . In yet another divergent example of pre-transfer editing for bacterial and mitochondrial LeuRS, deletion of the entire CP1 editing domain activated tRNA-dependent pre-transfer editing activity (Fig. 3D) suggesting that the activity is associated with the canonical aminoacylation core . This would contrast biochemical and structural evidence for LeuRS , but it remains possible that the enzyme's aminoacylation core is attempting to translocate misactivated adenylate intermediate to a CP1 domain that is simply not there in the deletion mutant.
Some might speculate that it is troublesome to encounter such a diversity of potential editing active sites and mechanisms to ensure aminoacylation fidelity as well as a system that ultimately lacks uniform mechanistic guidelines. However, it is important to consider the ancient nature of the aaRSs and their long evolutionary history that has likely marched through different eras in protein synthesis as the genetic code expanded and the complexity of the cell increased. The power of evolution has long found different and contrasting mechanisms where the end justified the means. Indeed, a perfect example of this occurs with the two completely unrelated classes of aaRSs [32,33], where nature found two entirely different structural and mechanistic means to achieve efficient aminoacylation of tRNA for protein synthesis. In one case, the Rossmann fold-based class I aaRS active site is found ubiquitously throughout many diverse protein families that bind ATP derivatives. In contrast, with just a few exceptions the canonical aminoacylation core of the class II aaRSs is quite unique. The single example of cross-over between the two classes for lysyl-tRNA synthetase (LysRS)  further emphasizes how evolution can take seemingly unique pathways to meet the operational goals of the aaRS family. The class I and class II LysRSs even co-exist naturally in some species .
In the case of amino acid editing, the diversity of fidelity mechanisms among the aaRSs is likely an imprint of the evolutionary path of protein synthesis. It is noteworthy that these idiosyncratic fidelity mechanisms can extend beyond the aaRSs to other tRNA binding proteins too. For example, in cells that rely on Glu-tRNAGln or Asp-tRNAAsn and an amidotransferase to convert these mischarged tRNAs to the correctly charged tRNA for protein synthesis , EF-Tu can be more evolved to discriminate between the mischarged and correctly charged tRNA , rather than introduce statistical mutations into the proteome. Recent work by the Ibba, Musier-Forsyth, and Frederick labs have shown that EF-Tu can recycle mischarged tRNAs that have escaped an editing aaRS . Thus, the path to protein synthesis is marked with a number of fidelity checkpoints and notably in some cases, these checkpoints are idiosyncratic.
The origins of post-transfer editing in aaRSs that clear mischarged tRNA appear to be linked to the incorporation of a discrete hydrolytic domain into an early aaRS polypeptide chain. Free-standing AlaX and ProX tRNA deacylases exist that are homologues of the editing modules of alanyl-tRNA synthetase (AlaRS) and ProRS . It is possible that these free-standing hydrolytic enzymes preceded their incorporation into the contemporary class II editing aaRSs, but this is unclear. In the case of the related CP1 editing domains that are found in the class I LeuRS, ValRS, and IleRS, an independent protein that might represent their evolutionary origins has so far escaped identification.
Although it has been proposed that the emergence of post-transfer editing pre-dates pre-transfer editing , examples of pre-transfer editing activities that are localized to the canonical aminoacylation core, suggest that this mode of adenylate hydrolysis could have preceded post-transfer editing in an early period for protein synthesis. Indeed, in some cases such as MetRS, pre-transfer editing of homocysteine remains sufficient for modern protein synthesis machinery and the needs of bacteria, lower eukaryotes and mammalian cells [28,40,41]. However, for other aaRS ancestors, their primitive fidelity mechanisms were inadequate for the evolving cell. Based on the E. coli LeuRS ΔCP1 mutant that retains fidelity , it is quite possible that the addition of the CP1 editing domain provided a more efficient fidelity mechanism that might have been commensurate with greater demands on protein synthesis by the developing cell.
The addition of a separate domain for amino acid editing would undoubtedly have enlisted the tRNA as an active player to bridge the protein–protein interactions and mediate the activity between the separated active sites. In LeuRS, the tRNA end binds near the editing site in an exit/entrance complex [42,43] and is proposed to sweep through this hydrolytic active site enroute to the aminoacylation active site . Remote molecular determinants that influence tRNA translocation have been identified in the CP1 domain  and emphasize the RNA's intimate involvement in cross-communication between the aminoacylation and editing domains. This extends to pre-transfer editing too, where the tRNA has long been proposed to be a co-factor [14,45] or if mischarged, could be used to prime the mechanism for clearing misactivated adenylates . The tRNA may even be replaced by a DNA aptamer that acts as a co-factor to trigger pre-transfer editing in IleRS .
In summary, evolution has driven the family of aaRSs and protein synthesis toward high fidelity with many checks and balances. It is tempting to speculate though if the introduction of mechanisms that would lower fidelity for amino acid selection and would potentially present statistical mutations in the proteome would have been beneficial to the cell at certain times. This might be akin to DNA mutations that allow the cell to adapt. However, statistical mutations in the proteome would only provide a transient advantage to the cell. It is possible though that that a short-term mechanism to introduce diversity into the proteome could be coupled to a permanent DNA mutation, such as in the case of a ValRS editing defect that allowed E. coli to adapt for subsistence on a non-standard amino acid . Significantly, some mitochondrial aaRSs have naturally lost their editing abilities [48–50]. Although in some cases, amino acid discrimination in the aminoacylation active site is increased, it is probable that this potential decrease in aaRS fidelity was driven by poorly understood physiology requirements of the mitochondria and its evolutionary path. A compromise in amino acid fidelity by the aaRSs could likewise have been an important mechanism to enable expansion of the genetic code as well as to introduce a broader and more diverse complement of amino acids into the proteome.
This work was support by a grant from the National Institutes of Health (GM63789).