Only a couple of years after Crick proposed his adaptor hypothesis, experimenters discovered both the adaptor molecules and the enzymes involved in attaching the amino acids to the adaptors. The adaptors turned out to be contained in the soluble RNA fraction as named by their discoverer Mahlon Hoagland, working in Paul Zamecnik’s laboratory in 1958; they were later termed transfer RNAs or tRNAs. Hoagland also recognized the existence, in crude cellular extract, of amino acyl-tRNA synthetases that specifically attach an amino acid to its respective cognate tRNA.
tRNA molecules fold into four-arm cloverleaf structures
Each cell, whether bacterial, archaeal, or eukaryotic, contains a set of tRNA molecules, each of which mediates incorporation of one of the 20 canonical amino acids present in proteins into the polypeptide chain during translation. We know that there are 61 meaningful codons that specify the 20 amino acids. Recall from Chapter 7 that because of the redundancy of the genetic code, some amino acids are encoded by more than one codon. Does this mean that there are 61 different types of tRNA in each cell? The answer is no, because some tRNAs can recognize more than one codon; the molecular basis for this is defined by the wobble hypothesis proposed by Crick in 1966. In Escherichia coli, for example, there are 40 tRNA types to accommodate the 20 amino acids and the 61 meaningful or sense codons. tRNA molecules containing different anticodons that specify the same amino acid are termed isoacceptor tRNA molecules. The specificity of a tRNA in terms of the amino acid it carries is designated by a superscript, for example, tRNAThr. The three remaining codons usually serve as stop codons, but in a few situations they can act as codons for unusual amino acids.
tRNAs are relatively short RNA molecules, usually 73–74 nucleotides long, whose sequence allows the formation of four different arms by intramolecular base pairing. The cloverleaf structure representations of two-dimensional (2D) tRNA structures are useful for visualizing the general patterns of hydrogen bonding and for denoting the individual arms (Figure 1 and Figure 2). These figures also show that the three-dimensional (3D) folding of tRNA is more complex than the 2D patterns might suggest. Two of the arms are named to reflect their function: the anticodon and acceptor arms. The other two, the TψC arm and the D arm, are given their respective names because they contain a substantial fraction of invariant positions and modified bases in which tRNAs are especially rich. The chemical structures of some typical modified nucleosides in tRNA are given in Figure 3.
Fig1. Generalized cloverleaf secondary structure and three-dimensional (3D) folding of tRNA. Distinct regions in (A) the cloverleaf model and (B) the 3D structure are color-coded in the same scheme. Some arms derive their names from specifically modified residues that occur frequently in these parts of the polynucleotide: D, dihydrouridine; T, ribothymidine; ψ, pseudouridylate. The amino acid attaches through its –COOH group to the 3′- or the 2′-OH groups of the ribose of the terminal adenine, A, residue in the universal CCA triplet at the acceptor stem’s terminus. The anticodon arm contains the triplet anticodon, which specifies the amino acid that is to be incorporated into the growing peptide chain by interacting with the codon of the mRNA.
Fig2. Primary, secondary, and tertiary structure of tRNAPhe from yeast. (A) Nucleotide sequence and cloverleaf secondary structure, with two-dimensional base-pairing depicted by red dots. The bases in blue are invariantly present in all tRNAs. (B) The crystal structure reveals the three-dimensional organization of the helices, with the individual domains represented in different colors. The structure is stabilized by non-Watson–Crick base-pairing between the TψC loop and the D loop; stacking interactions align the TψC arm with the acceptor arm and the D arm with the anticodon arm. Note the exposed bases that form the anticodon; this organization helps the interaction with the codon on the mRNA [Adapted, courtesy of Yakrazuul, Wikimedia.]
Fig3. Some typical modified nucleosides in tRNA.
The anticodon arm carries the anticodon in the loop of the arm; the three bases that form the anticodon are exposed to the solution (see Figure 2), which facilitates their interaction with the mRNA codon during protein synthesis. It is important to understand that when the anticodon pairs with the codon, a very short stretch of double-stranded A-form RNA helix is formed, following the rules of complementarity and anti-parallelism that govern double-helical nucleic acid structure. This means that when writing or depicting codon–anticodon interactions, care has to be exercised to reflect these rules. For example, in Figure 4, the anticodon in the tRNA is given in the usual 5′ → 3′ direction, but the codon in the mRNA is written in the 3′ → 5′ direction, UCG; the correct reading of the codon, 5′ → 3′, in this case is, of course, GCU.
Fig4. Wobble rules illustrated by the example of yeast tRNAAla. (A) Wobble rules as proposed by Crick in 1966. These rules have been extended on the basis of experimental data to include differently modified bases in the anticodon region. (B) The partial sequence presented is that of the anticodon arm, in blue letters. The wobble base hypoxanthine, which is the base portion of the modified nucleoside inosine (see Figure 3) in the anticodon, shown in red, can pair with U, C, or A in the codon, according to the rules in part A.
As mentioned above, there are fewer tRNAs than sense codons, meaning that several anticodons, and thus several tRNAs, must be capable of recognizing the same codon during translation. It turns out, as proposed by Crick and later proven experimentally, that the 5′-base of the anticodon can wobble, making alternative, non Watson–Crick hydrogen bonds with several different bases at the 3′-position of the codon. The general wobble rules are given in Figure 4, which also illustrates the wobbling of the modified inosine in the 5′-position of the anticodon of yeast alanine tRNA.
The acceptor arm at the top or 3′-end of the cloverleaf structure contains the universal triplet CCA as a single-strand extension to the arm. Recall from Chapter 14 that this triplet is added post-transcriptionally to all tRNAs with the help of a specific enzyme, tRNA nucleotidyltransferase, and that no template sequence is needed for this addition. The 2′- or 3′-OH group of the terminal adenosine, A, serves as an acceptor for the amino acid during the aminoacylation reaction.
The other arms play a role in the 3D folding and stabilization of the molecule that give rise to its characteristic L-shape (see Figure 1); both non-Watson–Crick base pairing and stacking interactions participate.
tRNAs are aminoacylated by a set of specific enzymes, aminoacyl-tRNA synthetases
In order to serve as an adaptor molecule between mRNA and an amino acid, tRNA has to carry the cognate amino acid at its acceptor arm and interact with the respective codon on the mRNA via its anticodon arm. How is the appropriate amino acid attached? How is the fidelity of the genetic code ensured? In other words, how is the amino acid selected so that its carrier tRNA has the anticodon that specifies this, and no other, amino acid? The answer is, by the specificity of action of a group of enzymes called aminoacyl-tRNA synthetases.
As pointed out by Christian de Duve, the scientist who shared the 1974 Nobel prize with Albert Claude and George E. Palade “for their discoveries concerning the structural and functional organization of the cell,” the accuracy of translation of the genetic code depends on the precision of two successive independent matchings: first, that of amino acids with tRNAs, and second, that of charged tRNAs with ribosome-linked messenger RNA. The second recognition event is rather straightfor ward, as it involves only complementary interactions between nucleotide sequences, the codon and anticodon triplets. The matching between amino acids and tRNA is quite different: it is indirect and bilingual. The aminoacyl-tRNA synthetase enzymes that do the matching use recognition between structural features of the enzyme itself and structural features of the tRNA that they charge. De Duve suggested the existence of a second genetic code imprinted into the structure of aminoacyl-tRNA that matches the amino acid with the cognate tRNA; he even introduced a term for these features, paracodons. He found the term convenient for defining a code, different from the codon–anticodon interactions, that involves protein–tRNA recognition.
Aminoacylation of tRNA is a two-step process
Before delving into the recognition problem, we first consider the chemistry of the tRNA aminoacylation reaction. This reaction takes place in two steps (Figure 5): (1) activation of the amino acid and (2) its subsequent addition to the CCA end of the tRNA. Both steps take place on the aminoacyl-tRNA synthetase and can occur on either the 2′- or 3′-hydroxyl of the A residue. The choice between amino acid attachment to a 3′- or a 2′-OH is determined by the specific structural class to which a given aminoacyl-tRNA synthetase belongs. Class I synthetases contain a characteristic Rossmann fold catalytic domain, act as monomers, and couple the aminoacyl group to the 2′-OH of the tRNA. Class II synthetases share an anti-parallel β-sheet fold flanked by α-helices, are mostly dimeric or multimeric, and prefer the 3′-OH group of the tRNA. If the amino acid is attached initially to the 2′-OH group, it is then shifted to the 3′-position in an additional step, as only 3′-attached amino acids can serve as substrates in translation. Despite the considerable structural differences between the two enzymatic classes, tRNA binding involves an α-helical structure that is conserved between them.
Fig5. Aminoacylation of tRNA: The chemistry. (Step 1) The nucleophilic carboxyl ate group of the amino acid attacks the γ-phosphorus of adenosine triphosphate, ATP. The pyrophosphate that is released during this step is immediately hydrolyzed by an abundant enzyme, pyrophosphatase, to two molecules of inorganic phosphate. This prevents the amino acid activation reaction from going into reverse. (Step 2) The 3′-OH or 2′-OH group of the terminal A in the universal CCA tail of tRNA attacks the carboxylate of the activated amino acid, releasing adenosine monophosphate, AMP, and creating an ester link age between the amino acid and the ribose. This is the form in which the amino acids are brought to the ribosome for incorporation into the nascent polypeptide chain. Both steps occur on the aminoacyl-tRNA synthetase.
Aminoacyl-tRNA synthetases possess a catalytically active site, where ATP binds to activate the amino acid, and two separate sites for binding of amino acid and tRNA (Figure 6). The crystal structures of complexes comprising synthetases and their cognate tRNA are available; an example is presented in Figure 7. The majority of the contacts are with the acceptor and the anticodon stem, but other interactions are also important for the recognition of specific tRNAs through the identity elements in tRNAs (Figure 8). Identity elements are defined by their ability to present the correct tRNA to the correct enzyme in an in vitro reaction. A major step forward was the identification, by Paul Schimmel’s laboratory in 1988, of a single base pair, G3U70, in the acceptor stem of E. coli tRNAAla as the identity element critical for binding to a cognate aminoacyl-tRNA synthetase. Moreover, this specificity could be conferred onto two other tRNAs by introduction of this base pair into their acceptor stems. As shown by the tRNAAla example in Figure 15.9, other base pairs in the acceptor stem also participate in the identification process. There was a major surprise concerning identity elements: although the anticodons are indeed used as such elements in many cases, some tRNAs can happily recognize their cognate enzyme without the participation of their unique anticodon triplets.
Fig6. Schematic of the overall structure of tRNA synthetase. The catalytic ATP binding site and binding sites for tRNA and amino acid are depicted in the upper part. ATP and the amino acid bind first, followed by tRNA; the blue line depicts the single-stranded universal CCA end. The lower part illustrates the structure at the end of catalysis, before release of the aminoacyl-tRNA. Often, a single aminoacyl-tRNA synthetase recognizes several isoacceptor tRNAs.
Fig7. E. coli tRNAGln bound to glutaminyl-tRNA synthetase. The enzyme is in the ribbon representation, and the tRNA is shown as a stick model. Note that the enzyme interacts with both the tRNA acceptor stem and the anticodon. Molecular dynamics simulations indicate that, upon binding, the tRNA induces conformational changes throughout the protein-tRNA interface and within the catalytic site. The charging reaction is stimulated by interactions between the enzyme and the tRNA identity elements, which, for the most part, are located far away from the tRNA acceptor stem. Thus, there must be an allosteric signal passed between the anticodon and the distant catalytic site. The amino acid residues and tRNA nucleotides involved in these allosteric interactions are evolutionarily conserved.
Fig8. Diversity of the major identity elements in tRNAs that identify the tRNA to the cognate tRNA synthetase. (A) Three different tRNAs—tRNAAla, tRNAGln, and tRNATyr—in E. coli illustrate the diverse locations and chemical nature of the respective identity elements. The identity elements are depicted as red dots. Note that the antico don nucleotides, shown as blue letters, are not always involved in the recognition of some cognate tRNAs, for example, tRNAAla. Generally, nucleotides in the acceptor stem confer a major portion of the specificity of recognition. (B) Identity elements in tRNATyr molecules from Methanocaldococcus jannaschii, an archaeon; E. coli, a bacterium; and Saccharomyces cerevisiae, a eukaryote. Note that evolutionarily conserved elements are present in all species.
Fortified with the concept of identity elements, we now return to the structures and dynamic networks in tRNA synthetase complexes (see Figure 7). It is known that the charging reaction is stimulated by specific interactions between the enzyme and the tRNA identity elements, but it is not exactly clear how this occurs, especially when the sites involved are far apart. Theoretical analysis has solved the coupling problem: allosteric signals pass between the anticodon and the distant catalytic site, with many possible paths of transferring the signal. The amino acid residues and nucleotides involved in these communication pathways are evolutionarily conserved.
Quality control or proofreading occurs during the aminoacylation reaction
Molecular recognition allows the possibility that an amino acid structurally similar to the cognate one can be erroneously bound by the wrong tRNA. Why can mischarging occur? The aminoacylation enzymes have to deal with two separate recognition issues: they need to correctly recognize both the cognate amino acid and its tRNA partner. The choice of tRNA poses no major problem, because the aminoacyl-tRNA synthetases make extensive contacts with the tRNA over a large surface area, between 2500 and 5700 Å2, allowing for numerous specific interactions. The real challenge for the enzyme is to recognize the cognate amino acid. Amino acids are small and very similar in structure and chemical reactivity, differing only in their side chains. The problematic amino acids are those whose side chains differ only slightly, for example, by just one methylene or OH group. These similar amino acids are usually described as near-cognate amino acids, to distinguish them from the noncognate amino acids that are significantly different from the cognate ones. The spontaneous rate of misincorporation of near-cognate amino acids is expected to be rather high. The error rate in vivo is, however, at least several orders of magnitude lower than that expected. What is the secret?
Biochemical and structural studies led to the discovery, in some aminoacyl-tRNA synthetases, of a second catalytic center that can perform hydrolytic release of a noncognate amino acid even after it has been linked to the tRNA. This center is thus responsible for the proofreading activity, or editing activity, of the enzymes.
In totality, quality control during the aminoacylation of tRNA involves a rather com plex set of processes: some acting after activation of the amino acid but before its attachment to the tRNA, known as pre-transfer mechanisms, and others acting after formation of the aminoacyl-tRNA adduct, known as post-transfer mechanisms.
Figure 9 also provides a background schematic illustrating the sequence of events that occur during the normal process of aminoacylation and its connection to the downstream events in protein synthesis. To complicate things even more, the post-transfer mechanisms can act in cis or in trans, depending on whether the editing involves the same enzyme molecule that performed the original aminoacylation reaction, acting in cis, or whether another aminoacyl-tRNA synthetase molecule is used to edit the mischarged tRNA following its release from the original enzyme, acting in trans (Figure 10). The process involving a second enzyme is known as resampling and is used in other steps of protein synthesis. Finally, the trans pathway of post-transfer editing may involve freestanding editing factors, that is, proteins other than the aminoacyl-tRNA synthetases (see Figure 10).
Fig9. Quality control during the formation of an aminoacyl-tRNA occurs in two stages. (A) Desirable outcome of the tRNA charging reaction that results in the forma tion of a tRNA molecule carrying its cognate amino acid. (B) Quality control that prevents tRNA charging with a noncognate amino acid acts at either of two different stages, pre- or post transfer, that is, before or after attachment of the activated amino acid to the tRNA. First, the synthetic site excludes amino acids that are larger than the cognate one or that cannot establish proper specific contacts with the enzyme. Another pre-transfer mechanism may act by sim ply hydrolyzing the labile mischarged aminoacyl (aa) adenylate. The hydrolysis can occur at the editing site of aminoacyl-tRNA synthetase or spontaneously in solution after the noncognate aa-AMP is expelled from the enzyme. Second, if a noncognate amino acid becomes bound to the tRNA, then there is a movement of the amino acid from the synthetic site into the editing site, where the RNA–amino acid ester linkage is hydrolyzed. This movement is possible because of the flexibility of the CCA end of the acceptor stem of the tRNA. [Adapted from Cochella L & Green R (2005) Curr Biol 15: R536–R540. With permission from Elsevier.]
Fig10. Discrimination between cis and trans post-transfer editing following mischarging. The cis pathway involves the action of an editing site on the same aminoacyl-tRNA synthetase molecule that catalyzed the original aminoacylation reaction. In trans editing, there are two possible scenarios, both of which occur following release of the mischarged aa-tRNA from the original synthetase: (1) the mischarged tRNA binds to another synthetase molecule, directly using its editing site, or (2) freestanding editing factors perform the hydrolysis reaction.
Interestingly, removal of d-stereoisomers of amino acids from tRNA is performed by a special class of such factors, called d-aminoacyl-tRNA deacylases. Although the intrinsic editing sites in aminoacyl-tRNA synthetases are highly conserved among living things, the trans-acting factors are much more diverse and may have arisen more than once during evolution. Thus, although d-aminoacyl-tRNA deacylases are universally required for cell viability, three unrelated classes have been found: one in bacteria and some eukaryotes, another in archaea and plants, and a third in cyanobacteria.
The existence and proper functioning of proofreading during charging of tRNAs with their cognate amino acids is of extreme importance to the fidelity of translation. It is one of only two major steps that exert control over the fidelity of translation; the second step occurs during codon–anticodon recognition at the elongation step of translation. If for some reason this step is not working properly and mischarging occurs, the consequences for the organism could be drastic.
Insertion of noncanonical amino acids into polypeptide chains is guided by stop codons
An important noncanonical amino acid is the formylmethionine that is used for initiating new polypeptide chains in bacteria, mitochondria, and chloroplasts. The insertion of this amino acid does not require an additional amino acyl-tRNA synthetase because addition of the formyl group occurs after tRNAiMet is charged with methionine by the regular methionyl-tRNA synthetase.
For a long time it was believed that, except for N-formylmethionine, only the 20 canonical amino acids could be inserted into proteins. It is now known that at least two additional amino acids, selenocysteine and pyrrolysine (Figure 11), can be incorporated internally into polypeptide chains. Selenocysteine is a cysteine analog in which the sulfur is replaced with its immediate downstairs neighbor in the periodic table, selenium. Selenocysteine is found in several essential enzymes, and the activities of these enzymes depend on its presence. Selenoproteins have been described in all three domains of life, with 25 representatives in humans. Pyrrolysine-containing proteins are less broadly distributed, being restricted to just a handful of archaea and bacteria.
Fig11. Co-translational insertion of unusual amino acids into the nascent polypeptide chain. (A) Selenocysteine, Sec; (B) pyrrolysine, Pyl. In both cases, the incorporation is guided by stop codons, which are followed by specific sequences in the mRNA that form secondary structures. Similar mechanisms have been implicated in the insertion of these two amino acids. First, a special cognate tRNA carrying the anticodon that recognizes the stop codon is mis-acylated by a normal aminoacyl-tRNA synthetase; this is a mis-acylation, for the enzyme should not normally recognize this anticodon. Then the noncognate amino acid bound to the cognate tRNA is converted, with the help of specific enzymes, into the cognate selenocysteine or pyrrolysine, respectively. The binding of these amino acids to the A site of the ribosome requires specific elongation factors or EFs, because the canonical EF, EF-Tu, would not recognize these structures. The binding of the specific EFs is aided by nearby insertion elements. Bacteria differ from archaea and eukarya in the exact location of the selenocysteine insertion element or SECIS; in bacteria, SECIS is located adjacent to the UGA, whereas in archaea and eukaryotes, it is in the 3′-untranslated region.
In both cases, insertion of these two extra amino acids into the polypeptide chain is guided by stop codons in a specific mRNA sequence context and is preceded by a complex sequence of biochemical reactions that attach the respective amino acid to their cognate tRNAs. These designated tRNAs contain anticodons that would recognize the context-dependent stop codons during translation (see Figure 11). In addition, once formed, these specific aminoacylated tRNAs need specific elongation factors to bring them to the A site of the ribosome during translation elongation.
Experimental technologies are now so advanced that it is possible to introduce even unnatural amino acids into proteins. This capability is just starting to affect a number of areas of protein research, bioengineering, and other areas. Such modification is expected to become a major source of designed proteins that have highly useful properties.
