2.1. Shape of the Ancestral tRNA
The minimum prerequisite from aminoacylated tRNAs involved in the prebiotic translation, would be their ability to line up side by side, being base-paired to an oligonucleotide serving as mRNA, while positioning the reacting amino acids in a proximity that allows for the formation of a coded peptide bond. The stereochemistry of a complex of two charged tRNAs attached to an mRNA was studied, using the high resolution structures of the modern 70S bacterial ribosomes (Materials and Methods). The length of a nucleotide-triplet, as obtained from measurements carried out on the mRNAs, lay in the range 13.8–18.2 Å, depending on the curvature of each codon, while the range of the maximal width of the double stranded AC stem, as obtained from measurements carried out on the tRNAs, is 20.6–21.7 Å (Figure S2
). Parallel accommodation of two tRNAs side by side, would have thus led to a severe clash between their AC stems. To circumvent this steric constraint, the reading of the two adjacent codons in the modern ribosome is accomplished by a significant kink of about 105° in the mRNA backbone between the A- and P-site codons, which results in an angle of about 26° between the planes of the A- and P-tRNAs [21
] (Figure 2
a), forming a rhombus-like configuration that prevents the collision. In order to bring the two amino acids attached to the 3′ends of the A-, P-tRNAs into the required proximity, the acceptor-TΨC arms of the modern A-, P-tRNAs approach each other in an inclined manner, up to a distance of 15.8±1.2 Å between the O3′ atoms of nucleotide A72 in A- and P-tRNAs (Figure S2
). The proximity allows the NCCA tails attached to these atoms to position the reacting amino acids in the PTC with the stereochemistry required for peptide bond formation. In the case of a moderate kink of 138° in the mRNA backbone, the juxtaposition of two tRNAs, as demonstrated by the AC stems of the P-, E-site tRNAs (PDB code 4V6F), leads to the deflection of the stems (Figure 2
b). When the mRNA backbone between the P- and E-sites is nearly straight (PDB code 1VY4), the E-site tRNA is not base paired to the mRNA at all.
Replacing the A-, P-site tRNAs by hairpins, in order to test the hypothesis that two hairpins attached to adjacent codons on mRNA could have induced the prebiotic polypeptide synthesis [6
], alters the stereochemistry of the reactants completely. The accommodation of two hairpins of four base-pairs on an mRNA, for example, will double the distance between the corresponding positions to which the 3′ end tails would be attached. The distance between O3′ of nucleotide 42 in the A- and P-tRNAs accommodated on the ribosome is about 37.1±0.3 Å (Figure S2
), and it would be further augmented for longer hairpins, precluding any proximity between their 3′ ends.
These observations, which are independent of any element that is specific to the contemporary translation system, indicate that hairpins would have failed to take part in the primitive translation. Avoiding the collision of the stems by a kink in the translated RNA or by deflection of the stems (Figure 2
a,b), would have set apart the reactants that are attached to the hairpins, obstructing any chemistry involving the aa-3′ ends. Moreover, the proximity between the 3′ends of the acceptor stems, which is made possible by the incline of the acceptor-TΨC stems towards each other, would be an impossible configuration for two parallel hairpins paired to neighboring codons. The stereoscopic difficulty was noticed by Orgel, even though the structural data available at the time was incomplete; “A striking feature of protein synthesis is the use of contiguous base triplets in the message, to code for contiguous amino acid. Our model provides no obvious reason why this should be true considering the crowding that would be expected at the junction between a message and 2 pre-tRNAs”. Correspondingly, he proposed a rather complex structure for the prebiotic tRNA that allowed the acceptor stems to slant towards one another [22
These geometrical restrictions, combined with evolutionary continuity, entail that the proto-tRNAs involved in the prebiotic translation process would have had an L-shape conformation. However, the intricate elbow region of the modern tRNA, having an angle of about 90° adjusted to the current stereochemistry of the ribosome, could have possibly evolved later. An elemental shape made up of two helices connected by a simpler elbow region, possibly similar to the elbow region of the DPR monomer encapsulating the PTC (Figure 3
a), would suffice for accomplishing the simultaneous interactions of the proto-tRNA with the ancient mRNA and with the PTC, which constitutes an indispensable prerequisite for translating a code into a peptide.
2.2. A Model of a Coded Proto-Ribosome
The universal nature of the contemporary ribosome, which employs a common mode of action in all life domains, together with its considerable level of structural and phylogenetic conservation, implies that the essence of the current translation mechanism should have already been present in the Last Universal Common Ancestor (LUCA). It follows that the vestige of the stand-alone primordial proto-ribosome, which is assumed to have continuously evolved into the modern ribosome, may still be inferred from the contemporary structure. A model of a coded proto-ribosome, i.e., an entity possessing the capability to translate a code embodied in an RNA template into a polypeptide, should therefore include the two active sites; an RNA element that accommodates mRNA and contains the decoding center, which is derived from the ribosomal RNA (rRNA) of the SSU i.e., a proto-SSU, and an element that forms the PTC pocket, derived from the rRNA of the LSU. The spatial gap between these elements necessitates the existence of a bridging element that guarantees the structural continuity of the model. Adherence to the current translation mechanism entails that the model should also retain the stereochemistry of the interactions of the two L-shape tRNA molecules with the active sites of both subunits. The minimal model for an initial system involved in a processive translation of RNA chains into polypeptides, will thus include three types of building blocks derived from the rRNA of the modern ribosome (Figure 3
a–c), two L-shape tRNAs (Figure 3
d), RNA chains acting as mRNA and amino acids.
The prebiotic PTC pocket will be approximated by the DPR (Figure 1
) derived from the LSU structure, due to the DPR’s compatibility with the fundamental expectation from a self-materializing entity, i.e., the realistic probability to emerge autonomously while retaining the PTC layout [15
]. The DPR is composed of two symmetrically positioned monomers, the A-, P-DPR, each being an L-shape RNA element (Figure 3
a,e), which, within the modern ribosome, are comprised of 60- and 61-mer sequences, respectively. The PTC walls are built chiefly from nucleotides belonging to the central loop of domain V, while the four helices that symmetrically radiate from it, i.e., H74, H89, H90, and H93, stabilize the pocket [19
]. Three of these helices, H74, H89, and H90, lack capping within the ribosome (Figures 3e and S1a,b). Completing them with tetraloops to achieve sequence continuity, would have resulted in 64- and 69-mer stand-alone monomers.
The ancient tRNA is assumed, according to the results of Section 2.1
, to have an L-shape conformation as well. It is possible that its sequence was shorter than the modern canonical 76 nucleotides, and that its elbow region was simpler, but for the present purpose, the ancient tRNA will be approximated by the modern one.
While the dimensions of the DPR and of the tRNA components are postulated, the extent of the two novel proto-ribosome building blocks, i.e., the proto-SSU that includes h44, h45, and the bridging element that includes H69–H71, were determined via their mutual interactions with the other elements. All of the nucleotides from these regions, which are within an interaction distance of 3.5 Å to the atoms in the other RNA elements composing the translation system model, were incorporated (Table S1–S4
). This resulted in a proto-SSU building block composed of nucleotides 1400–1419 and 1481–1504 from h44 (E. coli
numbering throughout), and 1505–1531 from h45 (Figure 3
c,g), forming a 71-mer RNA element. The bridging element contains nucleotides from domain IV, i.e., H69, H71, and the loop connecting them, a loop whose level of phylogenetic conservation nearly reaches that of the highly conserved central loop of domain V that assembles the PTC. Interactions with A-DPR, h44–45 proto-SSU, and with the A-, P-tRNA confine the bridging element to nucleotides 1906–1968, forming a rather complex 63-mer L-shape RNA element (Figure 3
b,f). The combination of the DPR with the H69–71 bridging element is suggested to constitute the entire proto-LSU, which together with the proto-SSU holds 267 nucleotides, i.e., less than 6% of the contemporary rRNA (Figure 4
The three types of L-shape building blocks derived from the rRNA of the contemporary ribosome, are interconnected (Figure 3
e–g, Tables S1–S4
). P-DPR is connected solely to A-DPR, A-DPR to the H69–71 bridging element, and the bridging element to h44–45 proto-SSU. These constituents together form the model of a coded proto-ribosome (Figure 5
). The linking among the building blocks is achieved via A-minor interactions [23
] that hook each building block to the neighboring ones (Table 1
, Figure 4
), as well as by a network of hydrogen bonds and stacking interactions.
This coded proto-ribosome would be stabilized by the interactions of the tRNA AC-loop with the mRNA accommodated on the h44–45 proto-SSU, and by the positioning of the aa-3′ends in the PTC pocket (Figure 5
). Additionally, a wide contact surface formed between the P-tRNA and the H69–71 bridging element (Table S4
) would have provided further stabilization, giving rise to the assembly of a compact system that contains all the RNA components crucial for translating a code prebiotically.
2.3. Spontaneous Formation of the Coded Proto-Ribosome Components
The fold of a hypothetical stand-alone RNA element, which is assumed to have materialized spontaneously in the prebiotic world, can be predicted by applying free energy minimization to its sequence. Predictions of the secondary schemes of the A- and P-DPR monomers, using sequences derived from structures of contemporary prokaryotic ribosomes, demonstrated that if any of these sequences existed in the prebiotic world, it would have been compelled to fold into an L-shape element matching the one found within the ribosome. Moreover, RNA sequences of about 70-mer, having the tendency to fold into DPR-monomer-like structures that conserve the reactant’s accommodation position in the PTC and dimerize in an energetic down-hill process, were shown to have a realistic probability to occur randomly [15
]. The DPR pocket, found in the core of the modern ribosome, could have therefore been assembled spontaneously in the prebiotic world.
Fold predictions of the proto-SSU and of the bridging element, using their E. coli
and Thermus thermophilus
sequences, were performed by Mfold [26
] (materials and methods), a program that screens the fold-landscape of an input sequence and produces a list of thermodynamically favorable secondary structures, along with their free energy of formation. The sequence of H69–71 bridging element, which is common to both bacteria, acquired a fold analogous to that which is found within the current ribosome (Figure S3
a,b). The sequences of the proto-SSU element, which differ slightly in E. coli
and Thermus thermophilus
, fold into an identical structure (Figure S3
c,d). The fold predicted for the h45 sequence is the same as the one found within the ribosome, but the base-pairing pattern of the section of h44 contained in the proto-SSU is rearranged, possibly because this region had to undergo massive mutational variation in order to fulfil its intricate contemporary role. The folding free energy of the bridging element was ΔG = −19.1 kcal/mol, and that of the proto-SSU ΔG = −34.4 ± 2 kcal/mol, indicating the self-foldability and stability of the obtained structures.
2.4. Evolutionary Path of the Translation System
A mandatory prerequisite for the spontaneous advent of the building blocks assembling the model of the initial translation system (Figure 5
), would be the prebiotic existence of random RNA chains of about 70-mer. Long oligonucleotides, reaching over 100-mer, were demonstrated, both experimentally and theoretically, to polymerize in solution under a thermal gradient [27
]. Long oligonucleotides, obtained in this manner or otherwise, could have autonomously folded into stable RNA molecules, mostly hairpins and RNA elements with two helical arms [28
]. Various types of L-shape RNA molecules could have therefore been floating around in the prebiotic environment, being self-folded products of the prebiotic chemistry. Among these, L-shape RNA elements resembling the DPR monomers, the H69–71 bridging unit, the h44–45 proto-SSU and the tRNAs could have been present. The sequences of the first three building blocks can be extracted from the conserved modern ribosomes, permitting the prediction of their L-shape secondary structures. The proto-tRNA molecules, whose sequences cannot be fully retraced, were suggested to have acquired an L-shape conformation via the association of two matching hairpins [29
], or due to the stability granted by this type of fold [32
]. During the review process, it was brought to my attention that a competing model for tRNA evolution, via the ligation of three minihelices, has been proposed [33
In the first stage of the ribosome evolution (Figure 1
and Figure 6
a) two DPR-monomers, similar in fold and nucleotide conformation but differing in sequence, could have coincidentally bumped into each other and dimerized, forming the PTC pocket. This step, which occurs in an energetically down-hill process, would have yielded the DPR, an early non-coded proto-ribosome that could have catalyzed the association of arbitrary amino acids, producing short peptides of random sequences [15
In the second stage (Figure 6
b) the two other building blocks, i.e., H69–71 bridging element and h44–45 proto-SSU, would have joined the DPR, primarily by hooking via A-minor interactions. This mode of assembly is consistent with the hypothesis that A-minor interactions were used by evolution to add new RNA elements to the evolving proto-ribosome [14
]. The modern translation process is initiated by base pairing of the initiator tRNA to the start codon on mRNA, followed by the association of the small and the large subunits to form a working ribosome. It is conceivable that the ancient initiation step could have been analogous, i.e., that base-pairing of an aa-proto-tRNA to a codon of an RNA oligonucleotide accommodated on the proto-SSU element, followed by the association with the proto-LSU, would have given rise to the first coded proto-ribosome apparatus. This RNA complex, in cooperation with aminoacylated tRNA molecules, would have constituted a prebiotic system comprising the essential components for translating a code into a polypeptide.
The next evolutionary stage (Figure 6
c) is likely to be concerned with the addition of the H92 and H80, stemming from the far ends of H90 and H74, respectively (Figure S1
), to the DPR, together assembling the entire symmetrical region [19
]. This expansion would have added stabilization to the initial model, due to the large binding interface formed between H92 of the expanded PTC region, and H71 from the bridging element (Table S5
). Moreover, the expansion would have stabilized the accommodation of the reactants in the PTC, allowing base pairing between C74, C75 from the tRNA 3′end, and the H92, H80 stem loops, i.e., the A-, P-loops.
The first protein to have been intercalated into the all-RNA coded proto-ribosome, according to the present model, would be L2. This universal r-protein was previously assumed to be among the most ancient ones [35
]. Its suggested primacy, in the context of the present study, stems from its being the only contemporary protein that interacts with more than a single building block of the coded proto-ribosome model. While as many as seven proteins interact with a single constituent of the model, L2 interacts with A-DPR, P-DPR, and with U1971 (Table S6
), a nucleotide located at the extension of the H69–71 bridging element tail (Figure 6
d). In Thermus thermophilus
, the L2 residues involved in the interactions with the coded proto-ribosome components are 228–243, which closely overlap the initial phase suggested for L2, i.e., residues 226–242 [36
]. This 16-mer peptide could have thus provided structural stabilization at the very early stage of ribosome evolution, and could have possibly been involved in the dynamics of the PTC [19
]. A translation system in which this peptide was synthesized by chance from a random RNA chain encoding for its sequence, would have therefore been granted with an evolutionary advantage, enhancing its translation fidelity and promoting its proliferation [37
Assembling a working translation system out of merely four types of L-shape building blocks, random RNA chains and amino acids, excluding any other enzymatic agents, seems to require the existence of two additional processes. An energy-free translocation process may be needed, due to the absence of a GTP center in the building blocks composing the coded proto-ribosome model, and a non-enzymatic process for aminoacylating tRNAs is required, due to the absence of a synthetase. Two experimental results reported in the literature may perhaps be pertinent. The existence of an energy-free translocation process may be implied by the polyU-directed elongation obtained in the absence of EF-Tu, EF-G, and GTP [38
], while charging tRNA with its cognate amino acid in the absence of a synthetase or ATP was reported to occur spontaneously under high pressure [39
]. However, the relevance of such non-enzymatic processes to prebiotic translation remains to be substantiated, and the absence of such processes would require the addition of enzymatic agents to the proposed initial translation system.
Sustainability of the initial translation system would have required the replication of its RNA components. An appealing non-enzymatic process, assumed to have preceded the enzymatic copying, is offered by the chemical copying of polynucleotides, where activated helper oligonucleotides were shown to catalyze the replication of an RNA template [40
]. Interestingly, copying the DPR monomers would have been particularly efficient, due to the sequence complementarity between nucleotides that constitute the two halves of the PTC cavity, observed in bacterial ribosomes. This complementarity implies that the sequence of each monomer could have acted as a template for the synthesis of its counterpart, eliminating the need for two replication steps for each copying event [20
]. Variations introduced during the primitive replication phase would have launched Darwinian evolution at a molecular level, promoting selection for the fittest proto-ribosomes. More efficient proto-ribosomes could have, in turn, translated longer random RNA chains, maybe yielding this way minimal versions of the polymerase and of the synthetase. The random RNA chains coding for these primitive enzymes would have been present in situ, allowing their re-translation and enabling a co-evolution process involving the extant ribozymes and enzymes [41
], opening the way to life as we know it. The model presented here put forward a continuous evolutionary path, from random RNA chains to life as we know it, initiated via a coded proto-ribosome that seems sufficiently simple to materialize spontaneously in the prebiotic world and perform programmed translation, yielding the first coded polypeptides.