General model of prebiotic evolution
We assume that early earth had widespread biophysical characteristics that will result from would permit prebiotic Darwinian dynamics among self-replicating polynucleotides. Central to our hypothesis is simply the day/nigh cycle which causes an external, diurnal energy fluctuations sufficient [21, 22] to provide alternating sources and sinks of Gibbs free energy which has both enthalpic and entropic components (ΔG = ΔH -TΔS). Thus, during daylight, we assume thermal energy (or enthalpy – energy at constant pressure) was sufficient to overcome energetic barrier (i.e. the hydrogen bonds between strands), whereas at night, the energy release from stable covalent bond formation (during synthesis of a daughter strand) dissipated as heat. We note that many primordial energy sources have been proposed [22, 23] and hydrothermal vents have been recently favoured. While the energy changes during day/night cycles would likely have been smaller than those associated with thermal vents [23], we favour this scenario because diurnal cycles impose regular variations that are necessary for self-replication as well as temporal constraints that impose Darwinian selection for optimizing the dynamics of self-replication. Furthermore, the UV radiation in sunlight could catalyse formation of nucleotides bases [4] which would then be available for strand synthesis at night. Finally, diurnal fluctuations of temperature provide a relatively constant frequency while the amplitude of the temperature variations may vary because of local weather and seasonal changes. The former provides a regular informational metric on time while the latter may impose potential threats or opportunities. Thus, self-replicators could, for example, measure local conditions and, by using their “knowledge” of the current time within the diurnal cycle, “predict” and adapt to conditions during the remainder of the cycle. The role of cyclical forcing functions in promoting dynamic self-assembly and network formation has been noted in multiple physical systems [24].
Modelling autocatalytic polymers in prebiotic conditions
Schrödinger first noted self–replicating polymers capable of information storage [25], requiring monomers that form two kinds of bonds among themselves. Stronger, thermodynamically near-irreversible bond between monomers are necessary for polymerizing monomers. Weaker, thermodynamically reversible bonds between monomers of the template and daughter strands permit autocatalytic self-replication. Furthermore, auto-catalysis requires that monomers forming new strands do so preferentially on a template compared to spontaneous synthesis. Self-replication on a template is favoured when hydrogen bonds of two contiguous monomers binding to the template catalyse the formation of covalent bonds between them.
The requirement for symmetry breaking
The technical details of our computer simulations have been published [26]. The evolutionary dynamics outlined above will include two sets of conflicting demands. First, optimal self-replicating polymers must balance stability which favours longer polymers [27] and short replication times which favours short strands. Second, long polymers with short replication times will need to simultaneously select for both monomer acquisition and monomer retention. However, monomer acquisition is favoured by lowering the hydrogen bond kinetic barrier, which maximizes the probability that a monomer in solution will attach to the template. However, monomer retention is favoured by increasing the kinetic barrier which decreases the probability of monomer separation from the template strand. This is similar to the dynamical models of cytoskeletal filament polymerization [28,29,30], although important differences exist between these two models. The filamental molecules of actin or microtubule are structurally constrained to grow or shrink only at the ends, whereas, DNA polymerization can happen anywhere on the template strand, but is similarly constrained only through our model.
As Anderson observed [31, 32], a physical system typically responds to imposed, incompatible forces by breaking symmetries. Similarly, our computational model (Fig. 1, adapted from [26]) shows how a left-right asymmetry in monomers and their resulting polymers satisfies the two conflicting selection forces. An asymmetric monomer, upon forming a hydrogen bond with its counterpart on the template strand, asymmetrically influences the hydrogen bond kinetic barrier for adjacent nucleotides. In forming a new strand, the kinetic barrier for bond formation/dissociation is lower on one side and higher on the other [33, 34]. Decreasing the barrier for the nucleotide bonding/dissociation to the right (say) of a pre-existing bond would increase the probability of a monomer in solution binding to the template (speed of association). Similarly, increasing the barrier for the nucleotide on its left would reduce the probability of separation of an already attached monomer (stability of retention). Finally, monomer symmetry breaking also imposes strong directionality on the self-replicating polymer so that the addition of free nucleotides to the complementary strand must occur from right to left (or vice-versa) rather than haphazard and simultaneous binding at all sites which, a priori, might appear to be a faster mechanism of self-replication.
Experimental observations of directional asymmetry in modern RNA
If the ribozyme properties of RNA evolution preceded the development of full living systems, we anticipate the “fittest” and most abundant extant RNA species would likely have been integrated into primordial life forms. This is supported by the following observations:
Nucleotide properties
The most obvious prediction of our hypothesis is that all nucleotides in modern RNA should show right-left directional asymmetry. It lowers the free nucleotide binding energy on one side and increases it on the other. Note that rotation of a nucleotide along its strand would reverse this effect and prevent the sequential addition of nucleotides on the growing strand. To prevent this rotational disruption, nucleotides must bind with their counterpart on the template strand with two or more hydrogen bonds (Fig. 2). In fact, all constituent nucleotides in modern RNA and DNA form 2 or 3 hydrogen bonds with their counterpart on the opposite strand.
Strand directionality
As shown in Fig. 2, the asymmetric effects of monomers on the hydrogen bonds [35] lowers the kinetic barrier at the adjacent empty site increasing the probability that a diffusing monomer will bind. This produces a directionality of strand self-replication that is maintained in all modern organisms. Both the “reading” and “duplication” of double stranded RNA and DNA proceed only along the 3′-5′ direction of the template and not in the kinetically improbable reverse direction.
Evolutionary dynamics in self-replicators
As noted above, self-replicating polynucleotides in a constrained environment such as a tide pool would have competed for monomers. In addition, cycling temperatures which promoted the sequence of strand separation and replication also imposed selection for replicative speed and accuracy. That is, complete autocatalysis required the new strand to fully replicate each nucleotide on the template prior to the onset of warmer daytime temperatures, which would produce strand separation. Furthermore, erroneous integration of a nucleotide without asymmetric properties would reduce the replicative speed of the daughter strand reducing the probability it will subsequently replicate removing it from the lineage. Note, however, there is no restriction on mixing nucleotides that possess the necessary asymmetric effects during replication. In total, these dynamics produce both environmental selection forces, a mechanism of inheritance that is imperfect thus introducing heritable variation into the replicating population.
In the Darwinian competition among autocatalytic replicators, mathematical models find that directional strand replication will be faster than will any other method of replication. For example, while simultaneous binding of nucleotides to all sites on the empty strand would intuitively seem to allow faster strand synthesis, the dwell time of each nucleotide on the template would be too low to allow covalent bonds to form consistently.
Curiously, double-stranded RNA today only exists in the form of viruses indicating it can store information in a manner that we propose for the first replicators. The absence of double stranded RNA in prokaryotic and eukaryotic life probably reflects subsequent evolutionary optimization in which DNA evolved to provide a more biochemically stable molecule for storing information while RNA became specialized to translate this information into specific sequences of amino acids in a polypeptide. As described below, the proposed evolutionary dynamics would include competition between autocatalytic RNA and DNA for 3 of the 5 nucleotides. This would allow scenarios of stable co-existence so that both “species” of polynucleotides would have been available to be incorporated in early living systems.
Experimental demonstration of nucleotide asymmetry, DNA/RNA unzipping energies
A simple prediction of the proposed right-left asymmetric influence on adjacent monomer bond energy is corresponding asymmetry in bond breaking. Experimental observations [35,36,37] demonstrated that the average force required to unzip double stranded RNA and DNA is significantly greater from one end than the other, consistent with the expected asymmetry.
Evolution of anti-parallel strand orientation and heteromolecular base-pairing
As strand length increases, the time for unidirectional replication will similarly increase thus imposing additional selection pressures. We propose, as with prior adaptive strategies, autocatalytic nucleotides “solved” the trade-off of polymer length and replication time through symmetry breaking. As shown in Fig. 3, anti-parallel strand orientation, together with heteromolecular base-pairing results in sequence-dependence of asymmetric effects on adjacent hydrogen bond kinetic barriers allowing replication to occur simultaneously at multiple locations (see [33] for full details). The first three base-pairs of Fig. 3 (d) are right-asymmetrically cooperative, whereas the next three base-pairs are left-asymmetrically cooperative. This allows these two portions of the strand to replicate simultaneously and independently of each other, thus increasing the rate of replication. The alternative option would be an RNA/DNA parallel strand duplex that has no local switching of the modes of asymmetric cooperativity. Such parallel-stranded DNA have been demonstrated to form under physiological conditions. Such a replicator would be evolutionarily inferior because the parallel-strand RNA would have to unzip in a single, continuous, and sequential order. Replication of such a polymer would take far longer and the acquisition of monomers would be correspondingly slower. Such a replicator, while possible, would be outcompeted by those with different modes of asymmetric cooperativity.
Quadruplet alphabet
The unzipping of strands of a duplex heteropolymer will be faster if strands are populated predominantly by one type of nucleotide as shown in Fig. 3. Any scrambling of this arrangement will adversely affect unzipping kinetics. Thus, fast unzipping selects for single strands composed of the same monomer. Clearly, the information content of DNA/RNA composed of a single nucleotide is insufficient for translation to proteins. It is possible that mixing of nucleotides occurred only after replicators evolved the next level of complexity involving chemical specialization between polypeptides and polynucleotides for catalysis and information storage, respectively. We hypothesize that evolution solved this issue by introducing another pair of nucleotides. This allowed RNA/DNA strands to simultaneously set the mode of asymmetric cooperativity and store information in a single strand.
Here we speculate that the primary selection force is thermal energy needed for melting of the double strand in a changing environment. The regular diurnal cycles importantly allowed a predictable temporal variation in temperatures that likely represented an initial information source for the replicators. However, while the frequency of the cycle remained relatively constant, the amplitude would change frequently due to weather conditions and seasonal effects. It seems reasonable to expect changes in the minimum and maximum temperatures during cycles would alter the separation and replication kinetics of DNA and RNA.
In considering the possible benefit of polymers formed of different nucleotides, it is notable that the energy needed to separate Adenine and Uracil (and later Thymine) with two hydrogen bonds is less than that for Cytosine and Guanine with 3 hydrogen bonds. This leads to the hypothesis that mixtures of these two types of nucleotides permitted heritable phenotypic plasticity that allowed the polynucleotides to undergo autocatalysis despite seasonal and regional variations in day and night temperatures. Furthermore, these variations allow replicators with a quadruplet alphabet to adapt to depth-dependent variations in temperature fluctuations in pools of water and perhaps variations in pH and concentrations of ions and minerals within the tidal pools. Note that this primitive information storage could lead to other phenotypic properties of single strands determined by the sequence of monomers such as 3-dimensional folding to perform enzymatic function in response to environmental changes.
While this hypothesis is speculative, it does make experimentally testable predictions about the evolutionary roles of multiple nucleotides in self-replicators.
Homochirality
Homochirality is one of the most intriguing properties of RNA/DNA molecules. We propose [38] chirality in self-replicators emerged from the requirement that monomers bind to the growing daughter strand in a specific orientation to maintain unidirectionality. Fast daughter strand synthesis requires directional monomers to arrive in a specific orientation on the template for hydrogen bonding. Thus, achiral monomers would be capable of bonding in multiple possible orientations with respect to the template strand, some of which will not allow for unidirectional strand elongation. Elimination of such symmetries would favour one invariant configuration of a chiral monomer. Hence, the strand could be D only or L only, but not a combination.
Co-evolution of DNA and RNA replicators
The proposed evolutionary dynamics allows formation of both DNA and RNA replicators. As noted above, each “species” will have different biophysical properties. Nevertheless, they will need to compete for 3 of the 5 constituent nucleotides while being subjected to identical environmental conditions and selection forces. We note that, under some scenarios, these pre-biotic evolutionary dynamics would permit equilibrium states in which both species co-existed. Furthermore, under some conditions of changing environmental conditions, they may have co-evolved mutually beneficial interactions that promoted cooperative dynamics that optimized survival for both species. These could have served as the precursors for the information dynamics of DNA and RNA in fully living systems. We anticipate this will be a subject of future investigations.