All tRNAs have a similar structure

The smallest tRNAs are only 74 nucleotides in length, and the largest are rarely more than 90 nucleotides. Because of their small size, and because it is possible to purify individual tRNAs, they were among the first nucleic acids to be sequenced, way back in 1965 by Robert Holley's group at Cornell University, New York. The sequences revealed one unexpected feature, that as well as the standard RNA nucleotides (A, C, G and U), tRNAs contain a number of modified nucleotides, 5–10 in any particular tRNA, with over 50 different modifications known altogether (Section 10.3).

Examination of the first tRNA sequence, for tRNA^Ala of Saccharomyces cerevisiae, showed that the molecule could adopt various base-paired secondary structures. After more tRNAs had been sequenced, it became clear that one particular structure could be taken up by all of them. This is the cloverleaf (Figure 11.2), and has the following features:

• The acceptor arm is formed by seven base pairs between the 5′ and 3′ ends of the molecule. The amino acid is attached to the extreme 3′ end of the tRNA, to the adenosine of the invariant CCA terminal sequence (Section 10.2.2).
• The D arm, named after the modified nucleoside dihydrouridine (see Table 10.5), which is always present in this structure.
• The anticodon arm contains the triplet of nucleotides called the anticodon which base-pair with the mRNA during translation.
• The V loop contains 3–5 nucleotides in Class 1 tRNAs or 13–21 nucleotides in Class 2 tRNAs.
• The TΨC arm, named after the sequence thymidine-pseudouridine-cytosine, which is always present.

Figure 11.2

The cloverleaf structure of a tRNA. The tRNA is drawn in the conventional cloverleaf structure, with the different components labeled. Invariant nucleotides (A, C, G, T, U, Ψ, where Ψ = pseudouridine) and semi-invariant nucleotides (abbreviations: (more...)

The cloverleaf structure can be formed by virtually all tRNAs, the main exceptions being the tRNAs used in vertebrate mitochondria, which are coded by the mitochondrial genome and which sometimes lack parts of the structure. An example is the human mitochondrial tRNA^Ser, which has no D arm. As well as the conserved secondary structure, the identities of nucleotides at some positions are completely invariant (always the same nucleotide) or semi-invariant (always a purine or always a pyrimidine), and the positions of the modified nucleotides are almost always the same.

Many of the invariant nucleotide positions are important in the tertiary structure of tRNA. X-ray crystallography studies have shown that nucleotides in the D and TΨC loops form base pairs that fold the tRNA into a compact L-shaped structure (Figure 11.3; Clark, 2001). Each arm of the L-shape is approximately 7 nm long and 2 nm in diameter, with the amino acid binding site at the end of one arm and the anticodon at the end of the other. The additional base-pairing means that the base-stacking (see page 14) is almost continuous from one end of the tRNA to the other, providing stability to the structure.

Figure 11.3

The three-dimensional structure of a tRNA. Additional base pairs, shown in black and mainly between the D and TΨC loops, fold the cloverleaf structure shown in Figure 11.2 into this L-shaped configuration. Depending on its sequence, the V loop (more...)

Aminoacyl-tRNA synthetases attach amino acids to tRNAs

The attachment of amino acids to tRNAs - ‘charging’ in molecular biology jargon - is the function of the group of enzymes called aminoacyl-tRNA synthetases. The chemical reaction that results in aminoacylation occurs in two steps. An activated amino acid intermediate is first formed by reaction between the amino acid and ATP, and then the amino acid is transferred to the 3′ end of the tRNA, the link being formed between the -COOH group of the amino acid and the -OH group attached to either the 2′ or 3′ carbon on the sugar of the last nucleotide, which is always an A (Figure 11.4).

Figure 11.4

Aminoacylation of a tRNA. The result of aminoacylation by a Class II aminoacyl-tRNA synthetase is shown, the amino acid being attached via its -COOH group to the 3′-OH of the terminal nucleotide of the tRNA. A Class I aminoacyl-tRNA synthetase (more...)

With a few exceptions, organisms have 20 aminoacyl-tRNA synthetases, one for each amino acid. This means that groups of isoaccepting tRNAs are aminoacylated by a single enzyme. Although the basic chemical reaction is the same for each amino acid, the 20 aminoacyl-tRNA synthetases fall into two distinct groups, Class I and Class II, with several important differences between them (Table 11.1). In particular, Class I enzymes attach the amino acid to the 2′-OH group of the terminal nucleotide of the tRNA, whereas Class II enzymes attach the amino acid to the 3′-OH group (Ibba et al., 2000).

Table 11.1

Features of aminoacyl-tRNA synthetases.

Aminoacylation must be carried out accurately: the correct amino acid must be attached to the correct tRNA if the rules of the genetic code are to be followed during protein synthesis. It appears that an aminoacyl-tRNA synthetase has high fidelity for its tRNA, the result of an extensive interaction between the two, covering some 25 nm² of surface area and involving the acceptor arm and anticodon loop of the tRNA, as well as individual nucleotides in the D and TΨC arms. The interaction between enzyme and amino acid is, of necessity, less extensive, amino acids being much smaller than tRNAs, and presents greater problems with regard to specificity because several pairs of amino acids are structurally similar. Errors do therefore occur, at a very low rate for most amino acids but possibly as frequently as one aminoacylation in 80 for difficult pairs such as isoleucine and valine. Most errors are corrected by the aminoacyl-tRNA synthetase itself, by an editing process that is distinct from aminoacylation, involving different contacts with the tRNA (Hale et al., 1997; Silvian et al., 1999).

In most organisms, aminoacylation is carried out by the process just described, but a few unusual events have been documented. These include a number of instances where the aminoacyl-tRNA synthetase attaches the incorrect amino acid to a tRNA, this amino acid subsequently being transformed into the correct one by a second, separate chemical reaction. This was first discovered in the bacterium Bacillus megaterium for synthesis of glutamine-tRNA^Gln (i.e. glutamine attached to its tRNA). This aminoacylation is carried out by the enzyme responsible for synthesis of glutamic acid-tRNA^Glu, and initially results in attachment of a glutamic acid to the tRNA^Gln (Figure 11.5A). This glutamic acid is then converted to glutamine by transamidation catalyzed by a second enzyme. The same process is used by various other bacteria (although not Escherichia coli) and by the archaea. Some archaea also use transamidation to synthesize asparagine-tRNA^Asn from aspartic acid-tRNA^Asn (Ibba et al., 2000). In both of these cases, the amino acid that is synthesized by the modification process is one of the 20 that are specified by the genetic code. There are also two examples where the modification results in an unusual amino acid. The first example is the conversion of methionine to N-formylmethionine (Figure 11.5B), producing the special aminoacyl-tRNA used in initiation of bacterial translation (Section 11.2.2). The second example occurs in both prokaryotes and eukaryotes and results in synthesis of selenocysteine, which is specified in a context-dependent manner by some 5′-UGA-3′ codons (Section 3.3.2). These codons are recognized by a special tRNA^SeCys, but there is no aminoacyl-tRNA synthetase that is able to attach selenocysteine to this tRNA. Instead, the tRNA is aminoacylated with a serine by the seryl-tRNA synthetase, and then modified by replacement of the -OH group of the serine with an -SeH, to give selenocysteine (Figure 11.5C; Low and Berry, 1996).

Figure 11.5

Unusual types of aminoacylation. (A) In some bacteria, tRNA^Gln is aminoacylated with glutamic acid, which is then converted to glutamine by transamidation. (B) The special tRNA used in initiation of translation in bacteria is aminoacylated with methionine, (more...)

Ultracentrifugation was used to measure the sizes of ribosomes and their components

The initial progress in understanding the detailed structure of the ribosome came not from observing them with the electron microscope but by analyzing their components by ultracentrifugation (Technical Note 2.2). Intact ribosomes have sedimentation coefficients of 80S for eukaryotes and 70S for bacteria, and each can be broken down into smaller components (Figure 11.10):

• Each ribosome comprises two subunits. In eukaryotes these subunits are 60S and 40S; in bacteria they are 50S and 30S. Note that sedimentation coefficients are not additive because they depend on shape as well as mass; it is perfectly acceptable for the intact ribosome to have an S value less than the sum of its two subunits.
• The large subunit contains three rRNAs in eukaryotes (the 28S, 5.8S and 5S rRNAs) but only two in bacteria (23S and 5S rRNAs). In bacteria the equivalent of the eukaryotic 5.8S rRNA is contained within the 23S rRNA.
• The small subunit contains a single rRNA in both types of organism: an 18S rRNA in eukaryotes and a 16S rRNA in bacteria.
• Both subunits contain a variety of ribosomal proteins, the numbers detailed in Figure 11.10. The ribosomal proteins of the small subunit are called S1, S2, etc.; those of the large subunit are L1, L2, etc. There is just one of each protein per ribosome, except for L7 and L12, which are present as dimers.

Figure 11.10

The composition of eukaryotic and bacterial ribosomes. The details refer to a ‘typical’ eukaryotic ribosome and the Escherichia coli ribosome. Variations between different species mainly concern the numbers of ribosomal proteins.

Probing the fine structure of the ribosome

Once the basic composition of eukaryotic and bacterial ribosomes had been worked out, attention was focused on the way in which the various rRNAs and proteins fit together. Important information was provided by the first rRNA sequences, comparisons between these identifying conserved regions that can base-pair to form complex two-dimensional structures (Figure 11.11). This suggested that the rRNAs provide a scaffolding within the ribosome, to which the proteins are attached, an interpretation that under-emphasizes the active role that rRNAs play in protein synthesis but which nonetheless was a useful foundation on which to base subsequent research.

Figure 11.11

The base-paired structure of the Escherichia coli 16S rRNA. In this representation, standard base pairs (G-C, A-U) are shown as bars; non-standard base pairs (e.g. G-U) are shown as dots.

Much of that subsequent research has concentrated on the bacterial ribosome, which is smaller than the eukaryotic version and available in large amounts from extracts of cells grown to high density in liquid cultures. A number of technical approaches have been used to study the bacterial ribosome:

• Nuclease protection studies (Section 2.2.1) enable contacts between rRNAs and proteins to be identified.
• Protein-protein crosslinking identifies pairs or groups of proteins that are located close to one another in the ribosome.
• Electron microscopy has gradually become more sophisticated, enabling the overall structure of the ribosome to be resolved in greater detail. For example, innovations such as immunoelectron microscopy, in which ribosomes are labeled with antibodies specific for individual ribosomal proteins before examination, have been used to locate the positions of these proteins on the surface of the ribosome.
• Site-directed hydroxyl radical probing makes use of the ability of Fe(II) ions to generate hydroxyl radicals that cleave RNA phosphodiester bonds located within 1 nm of the site of radical production. This technique was used to determine the exact positioning of ribosomal protein S5 in the E. coli ribosome. Different amino acids within S5 were labeled with Fe(II) and hydroxyl radicals induced in reconstituted ribosomes. The positions at which the 16S rRNA was cleaved were then used to infer the topology of the rRNA in the vicinity of S5 protein (Figure 11.12; Heilek and Noller, 1996).

Figure 11.12

Positions within the Escherichia coli 16S rRNA that form contacts with ribosomal protein S5. The distribution of the contact positions (shown in red) for this single ribosomal protein emphasizes the extent to which the base-paired secondary structure (more...)

In recent years these techniques have been increasingly supplemented by X-ray crystallography (Section 9.1.3), which has been responsible for the most exciting insights into ribosome structure. Analyzing the massive amounts of X-ray diffraction data that are produced by crystals of an object as large as a ribosome is a huge task, particularly at the level needed to obtain a structure that is detailed enough to be informative about the way in which the ribosome works (Pennisi, 1999). This challenge has been met, and structures have been deduced for ribosomal proteins bound to their segments of rRNA (Conn et al., 1999; Agalarov et al., 2000), for the large and small subunits (Ban et al., 2000; Wimberly et al., 2000), and for the entire bacterial ribosome attached to mRNA and tRNAs (Yusupov et al., 2001). As well as revealing the structure of the ribosome (Figure 11.13), this recent explosion of information has had an important impact on our understanding of the translation process, as we will see in the next section.

Figure 11.13

The bacterial ribosome. The picture shows the ribosome of the bacterium Thermus thermophilus. The small subunit is at the top, with the 16S rRNA in light blue and the small subunit ribosomal proteins in dark blue. The large subunit rRNAs are in grey and (more...)

Initiation in bacteria requires an internal ribosome binding site

The main difference between initiation of translation in bacteria and eukaryotes is that in bacteria the translation initiation complex is built up directly at the initiation codon, the point at which protein synthesis will begin, whereas eukaryotes use a more indirect process for locating the initiation point, as we will see in the next section.

When not actively participating in protein synthesis, ribosomes dissociate into their subunits, which remain in the cytoplasm waiting to be used for a new round of translation. In bacteria, the process initiates when a small subunit, in conjunction with the translation initiation factor IF-3 (Table 11.2), attaches to the ribosome binding site (also called the Shine-Dalgarno sequence). This is a short target site, consensus 5′-AGGAGGU-3′ in E. coli (Table 11.3), located about 3–10 nucleotides upstream of the initiation codon, the point at which translation will begin (Figure 11.14). The ribosome binding site is complementary to a region at the 3′ end of the 16S rRNA, the one present in the small subunit, and it is thought that base-pairing between the two is involved in the attachment of the small subunit to the mRNA.

Table 11.2

Functions of the bacterial translation factors.

Table 11.3

Examples of ribosome binding sequences in Escherichia coli.

Figure 11.14

The ribosome binding site for bacterial translation. In Escherichia coli, the ribosome binding site has the consensus sequence 5′-AGGAGGU-3′ and is located between 3 and 10 nucleotides upstream of the initiation codon.

Attachment to the ribosome binding site positions the small subunit of the ribosome over the initiation codon (Figure 11.15). This codon is usually 5′-AUG-3′, which codes for methionine, although 5′-GUG-3′ and 5′-UUG-3′ are sometimes used. All three codons can be recognized by the same initiator tRNA, the last two by wobble. This initiator tRNA is the one that was aminoacylated with methionine and subsequently modified by conversion of the methionine to N-formylmethionine (see Figure 11.5B). The modification attaches a formyl group (-COH) to the amino group, which means that only the carboxyl group of the initiator methionine is free to participate in peptide bond formation. This ensures that polypeptide synthesis can take place only in the N→C direction. The initiator tRNA_i ^Met is brought to the small subunit of the ribosome by a second initiation factor, IF-2, along with a molecule of GTP, the latter acting as a source of energy for the final step of initiation. Note that the tRNA_i ^Met is only able to decode the initiation codon; it cannot enter the complete ribosome during the elongation phase of translation during which internal 5′-AUG-3′ codons are recognized by a different tRNA^Met carrying an unmodified methionine.

Figure 11.15

Initiation of translation in Escherichia coli. See the text for details. Note that the different components of the initiation complex are not drawn to scale. Abbreviation: fM, N-formylmethionine.

Completion of the initiation phase occurs when IF-1 binds to the initiation complex. The precise role of IF-1 is unclear (see Table 11.2), but it may induce a conformational change in the initiation complex, enabling the large subunit of the ribosome to attach. Attachment of the large subunit requires energy, which is generated by hydrolysis of the bound GTP, and results in release of the initiation factors.

Initiation in eukaryotes is mediated by the cap structure and poly(A) tail

Only a small number of eukaryotic mRNAs have internal ribosome binding sites (see the next section). Instead, with most mRNAs the small subunit of the ribosome makes its initial attachment at the 5′-end of the molecule and then scans along the sequence until it locates the initiation codon. The process requires a plethora of initiation factors and there is still some confusion over the functions of all of these (Table 11.4). The details are as follows (Figure 11.16; Dever, 1999).

Table 11.4

Eukaryotic translation factors.

Figure 11.16

Initiation of translation in eukaryotes. (A) Assembly of the pre-initiation complex and its attachment to the mRNA. See the text for details. For clarity, several proteins whose precise roles are not understood have been omitted. The overall configuration (more...)

The first step involves assembly of the pre-initiation complex. This structure comprises the 40S subunit of the ribosome, a ‘ternary complex’ made up of the initiation factor eIF-2 bound to the initiator tRNA^Met and a molecule of GTP, and three additional initiation factors, eIF-1, eIF-1A, and eIF-3. As in bacteria, the initiator tRNA is distinct from the normal tRNA^Met that recognizes internal 5′-AUG-3′ codons but, unlike bacteria, it is aminoacylated with normal methionine, not the formylated version.

After assembly, the pre-initiation complex associates with the 5′ end of the mRNA. This step requires the cap binding complex (sometimes called eIF-4F), which comprises the initiation factors eIF-4A, eIF-4E and eIF-4G. The contact with the cap might be made by eIF-4E alone (as shown in Figure 11.16) or might involve a more general interaction with the cap binding complex (Pestova and Hellen, 1999). The factor eIF-4G acts as a bridge between eIF-4E, bound to the cap, and eIF-3, attached to the pre-initiation complex (Hentze, 1997). The result is that the pre-initiation complex becomes attached to the 5′ region of the mRNA. Attachment of the pre-initiation complex to the mRNA is also influenced by the poly(A) tail, at the distant 3′ end of the mRNA. This interaction is thought to be mediated by the polyadenylate-binding protein (PADP), which is attached to the poly(A) tail (Section 10.1.2). In yeast and plants it has been shown that PADP can form an association with eIF-4G, this association requiring that the mRNA bends back on itself. With artificially uncapped mRNAs, the PADP interaction is sufficient to load the pre-initiation complex onto the 5′ end of the mRNA, but under normal circumstances the cap structure and poly(A) tail probably work together (Preiss and Hentze, 1998). The poly(A) tail could have an important regulatory role, as the length of the tail appears to be correlated with the extent of initiation that occurs with a particular mRNA.

After becoming attached to the 5′ end of the mRNA, the initiation complex, as it is now called, has to scan along the molecule and find the initiation codon. The leader regions of eukaryotic mRNAs can be several tens, or even hundreds, of nucleotides in length and often contain regions that form hairpins and other base-paired structures. These are probably removed by a combination of eIF-4A and eIF-4B. eIF-4A, and possibly also eIF-4B, has a helicase activity and so is able to break intramolecular base pairs in the mRNA, freeing the passage for the initiation complex (Figure 11.16B). The initiation codon, which is usually 5′-AUG-3′ in eukaryotes, is recognizable because it is contained in a short consensus sequence, 5′-ACCAUGG-3′, referred to as the Kozak consensus.

Once the initiation complex is positioned over the initiation codon, the large subunit of the ribosome attaches. As in bacteria, this requires hydrolysis of GTP and leads to release of the initiation factors. Two final initiation factors are involved at this stage: eIF-5, which aids release of the other factors, and eIF-6, which is associated with the unbound large subunit and prevents it from attaching to a small subunit in the cytoplasm.

Initiation of eukaryotic translation without scanning

The scanning system for initiation of translation does not apply to every eukaryotic mRNA. This was first recognized with the picornaviruses, a group of viruses with RNA genomes which includes the human poliovirus and rhinovirus, the latter being responsible for the common cold. Transcripts from these viruses are not capped but instead have an internal ribosome entry site (IRES) which is similar in function to the ribosome binding site of bacteria, although the sequences of IRESs and their positions relative to the initiation codon are more variable than the bacterial versions (Mountford and Smith, 1995). The presence of IRESs on their transcripts means that picornaviruses can block protein synthesis in the host cell by inactivating the cap binding complex, without affecting translation of their own transcripts, although this is not a normal part of the infection strategy of all picornaviruses.

Remarkably, no virus proteins are required for recognition of an IRES by a host ribosome. In other words, the normal eukaryotic cell possesses proteins and/or other factors that enable it to initiate translation by the IRES method (Holcik et al., 2000). Because of their variability, IRESs are difficult to identify by inspection of DNA sequences, but it is becoming clear that a few nuclear gene transcripts possess them and that these are translated, at least under some circumstances, via their IRES rather than by scanning. Examples are the mRNAs for the mammalian immunoglobulin heavy-chain binding protein and the Drosophila Antennapedia protein (Section 12.3.3). IRESs are also found on several mRNAs whose protein products are translated when the cell is put under stress, for example by exposure to heat, irradiation, or low oxygen conditions. Under these circumstances, cap-dependent translation is globally suppressed (as described in the next section). The presence of IRESs on the ‘survival’ mRNAs therefore enables these to undergo preferential translation at the time when their products are needed.

Regulation of translation initiation

The initiation of translation is an important control point in protein synthesis, at which two different types of regulation can be exerted. The first of these is global regulation, which involves a general alteration in the amount of protein synthesis occurring, with all mRNAs translated by the cap mechanism being affected to a similar extent. In eukaryotes this is commonly achieved by phosphorylation of eIF-2, which results in repression of translation initiation by preventing eIF-2 from binding the molecule of GTP that it needs before it can transport the initiator tRNA to the small subunit of the ribosome. Phosphorylation of eIF-2 occurs during stresses such as heat shock, when the overall level of protein synthesis is decreased and IRES-mediated translation takes over.

Transcript-specific regulation involves mechanisms that act on a single transcript or a small group of transcripts coding for related proteins. The most frequently cited example of transcript-specific regulation involves the operons for the ribosomal protein genes of E. coli (Figure 11.17A). The leader region of the mRNA transcribed from each operon contains a sequence that acts as a binding site for one of the proteins coded by the operon. When this protein is synthesized it can either attach to its position on the ribosomal RNA, or bind to the leader region of the mRNA. The rRNA attachment is favored and occurs if there are free rRNAs in the cell. Once all the free rRNAs have been assembled into ribosomes, the ribosomal protein binds to its mRNA, blocking translation initiation and hence switching off further synthesis of the ribosomal proteins coded by that particular mRNA. Similar events involving other mRNAs ensure that synthesis of each ribosomal protein is coordinated with the amount of free rRNA in the cell.

Figure 11.17

Transcript-specific regulation of translation initiation. (A) Regulation of ribosomal protein synthesis in bacteria. The L11 operon of Escherichia coli is transcribed into an mRNA carrying copies of the genes for the L11 and L1 ribosomal proteins. When (more...)

A second example of transcript-specific regulation, one occurring in mammals, involves the mRNA for ferritin, an iron-storage protein (Figure 11.17B). In the absence of iron, ferritin synthesis is inhibited by proteins that bind to sequences called iron-response elements located in the leader region of the ferritin mRNA. The bound proteins block the ribosome as it attempts to scan along the mRNA in search of the initiation codon. When iron is present, the binding proteins detach and the mRNA is translated. Interestingly, the mRNA for a related protein - the transferrin receptor involved in the uptake of iron - also has iron-response elements, but in this case detachment of the binding proteins in the presence of iron results not in translation of the mRNA but in its degradation. This is logical because when iron is present in the cell, there is less requirement for transferrin receptor activity because there is less need to import iron from outside.

Elongation in bacteria and eukaryotes

Attachment of the large subunit results in two sites at which aminoacyl-tRNAs can bind. The first of these, the P or peptidyl site, is already occupied by the initiator tRNA_i ^Met, charged with N-formylmethionine or methionine, and base-paired with the initiation codon. The second site, the A or aminoacyl site, covers the second codon in the open reading frame (Figure 11.18). The structures revealed by X-ray crystallography show that these sites are located in the cavity between the large and small subunits of the ribosome, the codon-anticodon interaction being associated with the small subunit and the aminoacyl end of the tRNA with the large subunit (Figure 11.19; Yusupov et al., 2001).

Figure 11.18

Elongation of translation. The diagram shows the events occurring during a single elongation cycle in Escherichia coli. See the text for details regarding eukaryotic translation. Abbreviations: fM, N-formylmethionine; T, threonine.

Figure 11.19

The important sites in the ribosome. The structure on the left is the large subunit of the Thermus thermophilus ribosome; that on the right is the small subunit. The views look down onto the two surfaces that contact one another when the subunits are (more...)

The A site becomes filled with the appropriate aminoacyl-tRNA, which in E. coli is brought into position by the elongation factor EF-Tu, which ensures that only tRNAs that carry the correct amino acid are able to enter the ribosome, mischarged tRNAs being rejected at this point (Ibba, 2001). EF-Tu is an example of a G protein, meaning that it binds a molecule of GTP which it can hydrolyze to release energy. In eukaryotes the equivalent factor is called eEF-1, which is a complex of four subunits: eEF-1a, eEF-1b, eEF-1d and eEF-1g (see Table 11.4). The first of these exists in at least two forms, eEF-1a1 and eEF-1a2, which are highly similar proteins that probably have equivalent functions in different tissues (Hafezparast and Fisher, 1998). Specific contacts between the tRNA, mRNA and the 16S rRNA within the A site ensure that only the correct tRNA is accepted. These contacts are able to discriminate between a codon-anticodon interaction in which all three base pairs have formed, and one in which one or more mis-pairs are present, the latter signaling that the wrong tRNA is present (Yoshizawa et al., 1999). This is probably just one part of a series of safeguards that ensure the accuracy of the translation process (Rodnina and Wintermeyer, 2001a; 2001b).

When the aminoacyl-tRNA has entered the A site, a peptide bond is formed between the two amino acids. This involves a peptidyl transferase enzyme, which releases the amino acid from the initiator tRNA_i ^Met and then forms a peptide bond between this amino acid and the one attached to the second tRNA. In bacteria, the peptidyl transferase activity resides in the 23S rRNA of the large subunit, and so is an example of a ribozyme (Section 10.2.3; see Research Briefing 11.1). The reaction is energy dependent and requires hydrolysis of the GTP attached to EF-Tu (eEF-1 in eukaryotes). This inactivates EF-Tu, which is ejected from the ribosome and regenerated by EF-Ts. A eukaryotic equivalent of EF-Ts has not been identified, and it is possible that one of the subunits of eEF-1 has the regenerative activity.

Box 11.1

Peptidyl transferase is a ribozyme. A ribosome-associated protein that has the peptidyl transferase activity needed to synthesize peptide bonds during translation has never been isolated. The reason for this lack of success is now known: the enzyme activity (more...)

Now the dipeptide corresponding to the first two codons in the open reading frame is attached to the tRNA in the A site. The next step is translocation, during which three things happen at once (see Figure 11.18):

• The ribosome moves along three nucleotides, so the next codon enters the A site.
• The dipeptide-tRNA in the A site moves to the P site.
• The deacylated tRNA in the P site moves to a third position, the E or exit site, in bacteria or, in eukaryotes, is simply ejected from the ribosome.

Translocation requires hydrolysis of a molecule of GTP and is mediated by EF-G in bacteria and by eEF-2 in eukaryotes. Electron microscopy of ribosomes at different intermediate stages in translocation suggests that the two subunits rotate slightly in opposite directions, opening up the space between them and enabling the ribosome to slide along the mRNA (Frank and Agrawal, 2000). Translocation results in the A site becoming vacant, allowing a new aminoacyl-tRNA to enter. The elongation cycle is now repeated, and continues until the end of the open reading frame is reached.

Frameshifting and other unusual events during elongation

The straightforward codon-by-codon translation of an mRNA is looked upon as the standard way in which proteins are synthesized. But an increasing number of unusual elongation events are being discovered. One of these is frameshifting, which occurs when a ribosome pauses in the middle of an mRNA, moves back one nucleotide or, less frequently, forward one nucleotide, and then continues translation (Farabaugh, 1996). The result is that the codons that are read after the pause are not contiguous with the preceding set of codons: they lie in a different reading frame (Figure 11.20A).

Figure 11.20

Three unusual translation elongation events occurring in Escherichia coli. (A) Programmed frameshifting during translation of the dnaX mRNA. During synthesis of the γ subunit the ribosome shifts back one nucleotide, immediately after a series (more...)

Spontaneous frameshifts occur randomly and are deleterious because the polypeptide synthesized after the frameshift has the incorrect amino acid sequence. But not all frameshifts are spontaneous: a few mRNAs utilize programmed frameshifting to induce the ribosome to change frame at a specific point within the transcript. Programmed frameshifting occurs in all types of organism, from bacteria through to humans, as well as during expression of a number of viral genomes. An example occurs during synthesis of DNA polymerase III in E. coli, the main enzyme involved in replication of DNA (Section 13.2.2). Two of the DNA polymerase III subunits, γ and τ, are coded by a single gene, dnaX. Subunit τ is the full-length translation product of the dnaX mRNA, and subunit γ is a shortened version. Synthesis of γ involves a frameshift in the middle of the dnaX mRNA, the ribosome encountering a termination codon immediately after the frameshift and so producing the truncated γ version of the translation product. It is thought that the frameshift is induced by three features of the dnaX mRNA:

• A hairpin loop, located immediately after the frameshift position, which stalls the ribosome;
• A sequence similar to a ribosome binding site immediately upstream of the frameshift position, which is thought to base-pair with the 16S rRNA (as does an authentic ribosome binding site), again causing the ribosome to stall;
• The codon 5′-AAG-3′ at the frameshift position. The presence of a modified nucleotide at the wobble position of the tRNA^Lys that decodes 5′-AAG-3′ means that the codon-anticodon interaction is relatively weak at this position, enabling the frameshift to occur.

A similar phenomenon - translational slippage - enables a single ribosome to translate an mRNA that contains copies of two or more genes (Figure 11.20B). This means that, for example, a single ribosome can synthesize each of the five proteins coded by the mRNA transcribed from the tryptophan operon of E. coli (see Figure 2.20B). When the ribosome reaches the end of one series of codons it releases the protein it has just made, slips to the next initiation codon, and begins synthesizing the next protein. A more extreme form of slippage is translational bypassing (Herr et al., 2000) in which a larger part of the transcript, possibly a few tens of base pairs, is skipped, and elongation of the original protein continues after the bypassing event (Figure 11.20C). The bypass starts and ends either at two identical codons or at two codons that can be translated by the same tRNA by wobble. This suggests that the jump is controlled by the tRNA attached to the growing polypeptide, which scans the mRNA as the ribosome tracks along, and halts the bypass when a new codon to which it can base-pair is reached. Translational bypassing of 44 nucleotides occurs in E. coli during translation of the mRNA for gene 60 of T4 bacteriophage, which codes for a DNA topoisomerase subunit. Similar events have also been identified in a variety of other bacteria. Bypassing could result in two different proteins being synthesized from one mRNA - one protein from normal translation and one from bypassing - but whether this is its general function is not yet known.

Not all proteins fold spontaneously in the test tube

The notion that the amino acid sequence contains all the information needed to fold the polypeptide into its correct tertiary structure derives from experiments carried out with ribonuclease in the 1960s (Anfinsen, 1973). Ribonuclease is a small protein, just 124 amino acids in length, containing four disulfide bridges and with a tertiary structure that is made up predominantly of β-sheet, with very little α-helix. Studies of its folding were carried out with ribonuclease that had been purified from cow pancreas and resuspended in buffer. Addition of urea, a compound that disrupts hydrogen bonding, resulted in a decrease in the activity of the enzyme (measured by testing its ability to cut RNA) and an increase in the viscosity of the solution (Figure 11.24), indicating that the protein was being denatured by unfolding into an unstructured polypeptide chain. The critical observation was that when the urea was removed by dialysis, the viscosity decreased and the enzyme activity reappeared. The conclusion is that the protein refolds spontaneously when the denaturant (in this case, urea) is removed. In these initial experiments the four disulfide bonds remained intact because they were not disrupted by urea, but the same result occurred when the urea treatment was combined with addition of a reducing agent to break the disulfide bonds: the activity was still regained on renaturation. This shows that the disulfide bonds are not critical to the protein's ability to refold, they merely stabilize the tertiary structure once it has been adopted.

Figure 11.24

Denaturation and spontaneous renaturation of a small protein. As the urea concentration increases to 8M, the protein becomes denatured by unfolding: its activity decreases and the viscosity of the solution increases. When the urea is removed by dialysis, (more...)

More detailed study of the spontaneous folding pathways for ribonuclease and other small proteins has led to the following general two-step description of the process (Hartl, 1996):

1.

The secondary structural motifs along the polypeptide chain form within a few milliseconds of the denaturant being removed. This step is accompanied by the protein collapsing into a compact, but not folded, organization, with its hydrophobic groups on the inside, shielded from water.

2.

During the next few seconds or minutes, the secondary structural motifs interact with one another and the tertiary structure gradually takes shape, often via a series of intermediate conformations. In other words, the protein follows a folding pathway. There may, however, be more than one possible pathway that a protein can follow to reach its correctly folded structure (Radford, 2000). The pathways may also have side-branches into which the protein can be diverted, leading to an incorrect structure. If an incorrect structure is sufficiently unstable then partial or complete unfolding may occur, allowing the protein a second opportunity to pursue a productive route towards its correct conformation (Figure 11.25).

Figure 11.25

An incorrectly folded protein might be able to refold into its correct conformation. The blue arrow represents the correct folding pathway, leading from the unfolded protein on the left to the active protein on the right. The red arrow leads to an incorrectly (more...)

For several years it was more or less assumed that all proteins would fold spontaneously in the test tube, but experiments have shown that only smaller proteins with less complex structures possess this ability. Two factors seem to prevent larger proteins from folding spontaneously. The first of these is their tendency to form insoluble aggregates when the denaturant is removed: the polypeptides may collapse into interlocked networks when they attempt to protect their hydrophobic groups from water in step 1 of the general folding pathway. Experimentally, this can be avoided by using a low dilution of the protein, but this is not an option that the cell can take to prevent its unfolded proteins from aggregating. The second factor that prevents folding is that a large protein tends to get stuck in non-productive side branches of its folding pathway, taking on an intermediate form that is incorrectly folded but which is too stable to unfold to any significant extent. Concerns have also been raised about the relevance of in vitro folding, as studied with ribonuclease, to the folding of proteins in the cell, because a cellular protein might begin to fold before it has been fully synthesized. If the initial folding occurs when only part of the polypeptide is available, then there might be an increased possibility of incorrect branches of the folding pathway being followed. These various considerations prompted research into folding in living cells.

In cells, folding is aided by molecular chaperones

Most of our current understanding of protein folding in the cell is founded on the discovery of proteins that help other proteins to fold. These are called molecular chaperones and have been studied in most detail in E. coli. It is clear that both eukaryotes and archaea possess equivalent proteins, although some of the details of the way they work are different (Hartl, 1996; Slavotinek and Biesecker, 2001).

The molecular chaperones in E. coli can be divided into two groups:

• The Hsp70 chaperones, which include the proteins called Hsp70 (coded by the dnaK gene and sometimes called DnaK protein), Hsp40 (coded by dnaJ) and GrpE;
• The chaperonins, the main version of which in E. coli is the GroEL/GroES complex.

Molecular chaperones do not specify the tertiary structure of a protein, they merely help the protein find that correct structure. The two types of chaperone do this in different ways. The Hsp70 family bind to hydrophobic regions of proteins, including proteins that are still being translated (Figure 11.26A). They prevent protein aggregation by holding the protein in an open conformation until it is completely synthesized and ready to fold. The Hsp70 chaperones are also involved in other processes that require shielding of hydrophobic regions in proteins, such as transport through membranes and disaggregation of proteins that have been damaged by heat stress.

Figure 11.26

Molecular chaperones of Escherichia coli. (A) Hsp70 chaperones bind to hydrophobic regions in unfolded polypeptides, including those that are still being translated, and hold the protein in an open conformation until it is ready to be folded. (B) The (more...)

The chaperonins work in a quite different way. GroEL and GroES form a multi-subunit structure that looks like a hollowed-out bullet with a central cavity (Figure 11.26B; Xu et al., 1997). A single unfolded protein enters the cavity and emerges folded. The mechanism for this is not known but it is postulated that GroEL/GroES acts as a cage that prevents the unfolded protein from aggregating with other proteins, and that the inside surface of the cavity changes from hydrophobic to hydrophilic in such a way as to promote the burial of hydrophobic amino acids within the protein. This is not the only hypothesis: other researchers hold that the cavity unfolds proteins that have folded incorrectly, passing these unfolded proteins back to the cytoplasm so they can have a second attempt at adopting their correct tertiary structure (Shtilerman et al., 1999).

Eukaryotic proteins equivalent to both the Hsp70 family of chaperones and the GroEL/GroES chaperonins have been found, but it seems that in eukaryotes protein folding depends mainly on the action of the Hsp70 proteins. This is probably true also of bacteria (Ellis, 2000) even though the GroEL/GroES chaperonins play a major role in the folding of metabolic enzymes, and proteins involved in transcription and translation (Houry et al., 1999).

Cleavage of the ends of polypeptides

Processing by cleavage is common with secreted polypeptides whose biochemical activities might be deleterious to the cell producing the protein. An example is provided by melittin, the most abundant protein in bee venom and the one responsible for causing cell lysis after injection of the bee sting into the person or animal being stung. Melittin lyses cells in bees as well as animals and so must initially be synthesized as an inactive precursor. This precursor, promelittin, has 22 additional amino acids at its N terminus. The pre-sequence is removed by an extracellular protease that cuts it at 11 positions, releasing the active venom protein. The protease does not cleave within the active sequence because its mode of action is to release dipeptides with the sequence X-Y, where X is alanine, aspartic acid or glutamic acid, and Y is alanine or proline; these motifs do not occur in the active sequence (Figure 11.28A).

Figure 11.28

Post-translational processing by proteolytic cleavage. (A) Processing of promelittin, the bee-sting venom. Arrows indicate the cut sites. For the one-letter abbreviations of the amino acids see Table 3.1. (B) Processing of preproinsulin. See the text (more...)

A similar type of processing occurs with insulin, the protein made in the islets of Langerhans in the vertebrate pancreas and responsible for controlling blood sugar levels. Insulin is synthesized as preproinsulin, which is 105 amino acids in length (Figure 11.28B). The processing pathway involves the removal of the first 24 amino acids to give proinsulin, followed by two additional cuts which excise a central segment, leaving two active parts of the protein, the A and B chains, which link together by formation of two disulfide bonds to form mature insulin. The first segment to be removed, the 24 amino acids from the N terminus, is a signal peptide, a highly hydrophobic stretch of amino acids that attaches the precursor protein to a membrane prior to transport across that membrane and out of the cell. Signal peptides are commonly found on proteins that bind to and/or cross membranes, in both eukaryotes and prokaryotes.

Proteolytic processing of polyproteins

In the examples shown in Figure 11.28, proteolytic processing results in a single mature protein. This is not always the case. Some proteins are initially synthesized as polyproteins, long polypeptides that contain a series of mature proteins linked together in head-to-tail fashion. Cleavage of the polyprotein releases the individual proteins, which may have very different functions from one another.

Polyproteins are not uncommon in eukaryotes. Several types of virus that infect eukaryotic cells use them as a way of reducing the sizes of their genomes, a single polyprotein gene with one promoter and one terminator taking up less space than a series of individual genes. Polyproteins are also involved in the synthesis of peptide hormones in vertebrates. For example, the polyprotein called pro-opiomelanocortin, made in the pituitary gland, contains at least ten different peptide hormones. These are released by proteolytic cleavage of the polyprotein (Figure 11.29), but not all can be produced at once because of overlaps between individual peptide sequences. Instead, the exact cleavage pattern is different in different cells.

Figure 11.29

Processing of the pro-opiomelanocortin polyprotein. Abbreviations: ACTH, adrenocorticotropic hormone; CLIP, corticotropin-like intermediate lobe protein; ENDO, endorphin; LPH, lipotropin; ME, met-encephalin; MSH, melanotropin.