The parable of the caveman and the Ferrari: protein synthesis and the RNA world

Abstract

The basic steps of protein synthesis are carried out by the ribosome, a very large and complex ribonucleoprotein particle. In keeping with its proposed emergence from an RNA world, all three of its core mechanisms—aminoacyl-tRNA selection, catalysis of peptide bond formation and coupled translocation of mRNA and tRNA—are embodied in the properties of ribosomal RNA, while its proteins play a supportive role.

This article is part of the themed issue ‘Perspectives on the ribosome’.

1. Introduction

To mark the election of Sir Venki Ramakrishnan to President of the Royal Society, it is fitting that we honour him and the ribosome with this collection of papers. This paper aims to reflect on the remarkable inversion of the governing paradigm that has taken place during the decades that I have been an observer of the ribosome field.

In 1969, John Cairns, in his Forward to the Cold Spring Harbor Symposium on Quantitative Biology XXXIV (‘The Mechanism of Protein Synthesis’), wrote: ‘A science comes of age when the principles on which it was founded have been vindicated and are replaced, as an occupation, by the accumulation of detail’. [1]. Cairns's view of the mechanism of protein synthesis in 1969 was shared by most of the leading molecular biologists of the day. This understanding of the ribosome can be likened to that of cavemen trying to understand how the Ferrari works: the accelerator makes the car go faster, the brake pedal makes it slow down, the steering wheel makes it turn right or left, and so on. The cavemen could of course have been forgiven for mistaking macroscopic phenomenology for principles. Conversely, discovering that explosions of gasoline aerosols drive the movement of pistons to create torque on the crankshaft, which then drives the wheels, is not mere ‘accumulation of detail’, but the fundamental principles that made the automobile possible. After 47 more years of ribosomology, we do not yet understand protein synthesis with the same clarity with which we understand how cars work. But a great deal has been learned about what is under the hood, and many of the right questions are being asked. And this, together with the emergence of an impressive arsenal of powerful methods, fuels optimism that an understanding of the fundamental mechanisms of protein synthesis may become a reality in the near future.

A casual scan of the chapters from the 1969 symposium gives a sense of where the field stood at the time. Mechanistic studies were largely focused on isolation and characterization of translation factors, rather than on the ribosome itself. The ribosome was still being viewed in some quarters as a passive surface on which the ‘functional molecules’ carried out translation. Although some laboratories were already arguing that peptide bond formation is catalysed by the large ribosomal subunit, the question was: which protein was responsible? The discovery of mutations that confer dramatic changes in translational accuracy also pointed to the functional properties of the ribosome, but, again, to ribosomal proteins. Even after four more years of ribosome research, a chapter in the book from the 1973 Cold Spring Harbor ribosome meeting, contributed by the then leading ribosomal RNA laboratory, reads ‘The rRNAs appear to provide the framework to which the ribosomal proteins are attached during assembly. In the mature ribosome, they hold the proteins in a configuration which permits the particle to correctly carry out its functions in protein synthesis' [2]. This view was almost universally accepted at the time.

Six years later, early tremors signalled a seismic shift, reflected in the evangelical tone of the first chapter from the book from the 1979 Steenbock Symposium, which reads ‘It is now well established that rRNA also participates directly in the functioning of the ribosome, in addition to its undisputed contributions to structural organization. Among the functions for which there is evidence of rRNA participation are (1) mRNA selection; (2) tRNA binding; (3) antibiotic sensitivity/resistance; and (4) subunit association’ [3]. And now, 37 years down the road from Steenbock, the list of functional roles for rRNA in translation has expanded to eclipse virtually all of the basic mechanisms of protein synthesis.

2. Aminoacyl-tRNA selection

In spite of convincing evidence for functional roles for rRNA, the proteins did not go down easily. This was in large part due to the fact that bacteria have single copies of their ribosomal protein genes, but multiple copies of their rRNA genes. As a result, mutations in the r-proteins affecting ribosome function began to appear early on, while functional mutations in rRNA were not discovered until new genetic strategies were developed. Implication of ribosomal proteins in the mechanism of aminoacyl-tRNA selection was strongly supported by studies of antibiotic-stimulated misincorporation of amino acids. Most notable was the antibiotic streptomycin, which causes an increase in the translational error frequency by several orders of magnitude [4]. Mutations conferring streptomycin resistance were tracked down to ribosomal protein S12 [5]. Certain mutations in S12 even conferred streptomycin dependence, and ribosomes from these mutants were found to have a decreased error frequency relative to wild-type ribosomes in the absence of streptomycin [6]. Moreover, streptomycin dependence could be suppressed by mutations in ribosomal protein S4, which, when segregated from the S12 alleles, conferred an increased error frequency [7,8]. Interestingly, Gorini proposed, on the basis of these and other genetic studies, that the ribosome has a ‘screen’ that functions in discriminating between cognate and non-cognate (or near-cognate) aminoacyl-tRNAs [8]. In other words, tRNA selection was not simply codon–anticodon pairing, but in addition involved some process contributed by the ribosome. Inspired by his finding that streptomycin binds to purified 16S rRNA (and does not bind to protein S12 or S4), he even went on to propose that it causes miscoding by binding to a site in 16S rRNA that distorts the structure of the part of the ribosome that is responsible for the screen [9]. This proposal was not taken seriously by the ribosome community at that time.

Gorini's proposal, made in 1972, turns out to be uncannily correct. Streptomycin indeed binds to a specific site in the 30S ribosomal subunit made up almost exclusively of elements of 16S rRNA [10]. It appears to stimulate miscoding by distorting the structure of the decoding site of the 30S subunit (the structure responsible for the ‘screen’) [11]. This structure is composed mainly of universally conserved elements of 16S rRNA, including G530, A1492 and A1493 [12], which had been identified as elements of the binding site for aminoacyl-tRNA in the 30S subunit (the A site) [13,14]. The ribosome appears to screen against non-cognate tRNAs by a mechanism that centres on flipping these three bases into the minor groove of the codon–anticodon helix, creating a close van der Waals fit to authentic Watson–Crick base pairs, but not to mis-paired bases. The detailed mechanism by which this happens has been the subject of some controversy in recent years, prompted by the unexpected finding that near-cognate tRNAs containing wobble mismatches in the first or second codon positions are bound with the same Watson–Crick geometry as cognate tRNAs by the decoding site [15]. This finding suggests that the low error frequency of tRNA selection is based on the low frequency of keto–enol tautomerization by the RNA bases [16,17].

In any case, it is clear that ribosomal RNA plays a fundamental role in the mechanism of aminoacyl-tRNA selection. The mechanism involves formation of ‘A-minor’ interactions [12,18] between the rRNA bases and the minor groove of the codon–anticodon duplex [12]. It appears that cognate tRNAs may be discriminated primarily on the basis of the shape of the resulting codon–anticodon duplex, rather than its inherent thermodynamic stability [15,17,19]. A-minor interactions are widespread in the ribosome and in the structures of other functional RNAs [18], so are likely to have been commonplace in an RNA world. Such a simple mechanism could have arisen, prior to the evolution of the ribosome, for monitoring the accuracy of RNA replication, suggesting that the 30S ribosomal subunit may have evolved from an RNA replicase [20].

3. Catalysis of peptide bond formation

In vitro experiments, as early as 1967, using the antibiotic puromycin showed that peptide bond formation is not catalysed by some protein translation factor, but by the ribosome itself [21]. Moreover, it could be localized to the 50S ribosomal subunit [21]. Although ribosomal protein L16 was at first identified as a potential candidate for the catalytic molecule by partial reconstitution experiments [22], it was subsequently found that the rescue of activity by L16 was in fact due to its ability to stimulate the correct folding of 23S rRNA during in vitro assembly [23]. Attention was drawn to a possible catalytic role for 23S rRNA by numerous findings. The binding sites for antibiotics that inhibit peptidyl transferase mapped to 23S rRNA by chemical footprinting [24], and by mutations or methylations in 23S rRNA that confer resistance to these antibiotics (reviewed in [25]). Aminoacyl-tRNAs derivatized with reactive groups next to their amino groups reacted with high efficiency with conserved bases in 23S rRNA [26,27]. Finally, peptidyl transferase activity was shown to be unusually resistant to procedures that are normally used to extract, denature and digest proteins [28].

The lingering possibility that a ribosomal protein was providing the catalytic moieties for peptidyl transferase was at last put to rest by a high-resolution crystal structure of the 50S ribosomal subunit, which showed that there is no element of protein within 17 Å of the site of catalysis [29]. Mechanistic studies have shown that peptide bond formation is entropically driven, suggesting that the critical event is correct stereochemical placement and orientation of the reactants in the peptidyl transferase active site [30,31].

The ribosome is, therefore, a ribozyme, again no doubt reflecting its origins in an RNA world. But why would such a complex mechanism have evolved in the first place? The likelihood that protein synthesis—i.e. the coded (or even non-coded) synthesis of an enzyme, for example—could have emerged from an RNA world is virtually impossible [32]. But synthesis of small peptides, even random ones, could have played a beneficial role, for example in stabilizing complex RNA folds that are inaccessible to pure RNA [33]. The functional space of RNA is constrained by its structural limitations; RNA structure space (and thus its functional capabilities) can be expanded by interactions with peptides. It seems likely that peptidyl transferase emerged from the discovery of peptides in an RNA world.

4. Translocation

Perhaps, the most complex and dramatic function of the ribosome is its ability to move tRNA and mRNA, rapidly and accurately, following formation of each peptide bond. This process is catalysed by elongation factor EF-G, which is sometimes referred to as the ‘translocase’. But again, studies in Spirin's and Pestka's laboratories long ago showed that translocation can occur in the absence of EF-G, albeit very slowly [34,35]. The GTPase activity of EF-G has sometimes inspired the idea that GTP hydrolysis provides the energy for a ‘power stroke’ to drive translocation. But studies by Kaziro showed that translocation can occur even using a non-hydrolysable analogue of GTP [36]. Remarkably, the antibiotic sparsomycin, a peptidyl transferase inhibitor, was shown to trigger translocation in the absence of EF-G or GTP [37]. Once again, experiments began to point to the ribosome itself as the source of function.

A critical point is that translocation of mRNA must be rigorously coupled to translocation of tRNA, to avoid shifting of the translational reading frame. Unlike mis-sense errors, most of which are tolerated in living cells, frame-shift errors are potentially lethal events. In the two alternate reading frames, premature stop codons will be encountered, on average, every 20 codons, resulting in truncated polypeptide products, which can create dominant lethal effects. A central question, then, is how the mRNA and tRNAs move. A tRNA bound to the A site must dissociate from the A site before it moves to the P site, but during this movement correct codon–anticodon pairing must somehow be maintained, even though the codon–anticodon interaction is unstable at typical physiological temperatures in the absence of the ribosome. Another important point is that there is a strict directionality to translocation, from A to P to E site. For these reasons, it seems unlikely that translocation is based on simple diffusion. Although we do not yet have a complete understanding of the molecular mechanisms underlying translocation, many of the critical pieces have fallen into place. The evidence supports the view that translocation, like aminoacyl-tRNA selection and catalysis of peptide bond formation, is indeed a property of the ribosome itself. Moreover, it now appears that the ribosome is a complex molecular machine, and that its machine-like properties are based on its ribosomal RNA.

Translocation occurs in two main steps, both of which involve large-scale movement of structural domains of the ribosome. In the first step, the acceptor ends of the tRNAs move in the 50S subunit from the A and P sites to the P and E sites, respectively, while their anticodon ends remain in the A and P sites of the 30S subunit [38]. The tRNAs are said to be in the A/P and P/E hybrid states at this point. This step is coupled to intersubunit rotation of approximately 6° [39,40]. The first step can proceed spontaneously, and in the absence of EF-G or GTP, and can therefore be driven by thermal energy alone [38,41]. Although this rotational movement is often referred to as ‘ratcheting’, it has been observed to be reversible [41], and so is unlikely to be part of a true ratchet mechanism.

The second step is the movement of the anticodon ends of the tRNAs in the 30S subunit, coupled to movement of the mRNA, completing the translocation cycle. This step appears to be rate-limiting, and requires EF-G and GTP [38]. The second step is coupled to rotational movement of the head domain of the 30S subunit [42–47]. Besides overall movement of the 30S head, individual elements of the A and P sites of both ribosomal subunits rearrange during the second step of translocation. In the 30S subunit, the 966 and 1400 loops of the P site reach towards the A site to contact the A-site tRNA, guiding it into the P site, while in the 50S subunit, the A and P loops reach toward each other to simultaneously contact the CCA end of the A-site tRNA to escort it into the 50S P site [47]. A third dynamic element is the L1 stalk of the 50S subunit, which moves by nearly 60 Å to contact the elbow of the deacylated tRNA as it moves from the P/P state into the P/E hybrid state. The head of the stalk maintains contact with the tRNA elbow as it translocates from the P/E state to the pe/E chimeric hybrid state to the E/E classical state [48].

5. The roles of ribosomal RNA in protein synthesis

It has now become clear that all three core mechanisms of protein synthesis—aminoacyl-tRNA selection, catalysis of peptide bond formation and coupled translocation of mRNA and tRNA—are based on ribosomal RNA. In aminoacyl-tRNA selection, the conserved 16S rRNA bases G530, A1492 and A1493 in the 30S subunit create a three-dimensional template for sensing Watson–Crick pairing between the mRNA codon and tRNA anticodon. Catalysis of peptide bond formation is carried out by the peptidyl transferase active site of the 50S subunit, which is composed of universally conserved nucleotides of 23S rRNA. Finally, movement of dynamic elements of both 16S and 23S rRNA forms the basis of the mechanism of the coupled translocation of mRNA and tRNA. In retrospect, our understanding of the ribosome in 1969 was not only incomplete, but completely backwards. Thus, Crick's once heterodox conjecture that the first ribosomes were made of RNA [49] now seems plausible, if not inescapable.

Competing interests

I declare I have no competing interests.

Funding

This research is funded by NIH, ‘Ribosome Structure and Function’ (grant no. 1R35GM118156-01).