This article was reviewed by Eric Bapteste, David Krakauer, Sergei Maslov, and Itai Yanai.
Evolution of the cosmos: eternal inflation, "many worlds in one", and anthropic selection
The "many worlds in one" (hereinafter MWO) model makes the startling prediction that all macroscopic, "coarse-grain" histories of events that are not forbidden by conservation laws of physics have realized (or will realize) somewhere in the infinite universe, and not just once but an infinite number of times [1,2]. There is, e.g., an infinite number of (macroscopically) exact copies of the earth with everything that exists on it, although the probability that a given observable region of the universe (hereinafter O-region) carries one of such copies is vanishingly tiny. This picture seems counterintuitive in the extreme but it is a direct consequence of eternal inflation, the dominant model of the evolution of the universe in modern cosmology [3-5].
Inflation is the period of the exponentially fast initial expansion of a universe [6]. In the most plausible, self-consistent inflationary models, inflation is eternal, with an infinite number of island (pocket) universes (hereinafter, simply, universes) emerging through the decay of small regions of the primordial "sea" of false (high energy) vacuum and comprising the infinite multiverse. To observers within each universe, it appears self-contained and infinite, and containing an infinite number of O-regions. For such observers (like us), their universe is expanding from a singularity (Big Bang) which corresponds to the end of inflation in the given part of the multiverse. Inflation is in excellent agreement with several crucial results of observational cosmology – above all, the flatness of space in our O-region, the overall uniformity of the cosmic microwave background radiation, and its local non-homogeneities [7]. Furthermore, the "populated landscape" version of string theory independently yields a very similar model of the multiverse [8-11]. Thus, although the model of eternal inflation cannot be considered proved, this is the strongly preferred current scenario of the cosmic evolution.
Garriga and Vilenkin showed that, in a finite time, the content of each O-region can assume only a finite number of states and, accordingly, any O-region has a finite, even if unimaginably vast (on the order of 10^10150), number of unique macroscopic, coarse-grain histories [1]. Effectively, the finiteness of the number of coarse-grain histories appears to be a straightforward corollary of quantum uncertainty [2]. The same conclusion is independently reached through a completely different approach, namely, the so-called holographic bound on the amount of entropy that can be contained in any finite region of the universe [1,11,12]. Combined, eternal inflation, the finiteness of the number of unique coarse-grain histories, and the inevitable quantum randomness at the Big Bing (the beginning of time for each universe) lead to the straightforward and striking conclusion that each history permitted by conservation laws of physics is repeated an infinite number of times in the multiverse and, actually, in each of the infinite number of infinite (island) universes [2,11].
The MWO model is tightly linked to the anthropic principle (anthropic selection), a controversial but increasingly popular concept among cosmologists. According to the anthropic principle, the only "reason" our O-region has its specific parameters is that, otherwise, there would be no observers to peer into the universe [13-15]. Of course, it should be emphasized that I only discuss here what is often called "weak" anthropic principle and is the only acceptable scientific rendering of this concept. The so-called "strong" anthropic principle is the teleological notion that our (human) existence is, in some mysterious sense, the "goal" of the evolution of the universe; as such, this idea does not belong in the scientific domain. It appears that the anthropic principle can be realistically defined only in the context of a vast (or infinite) multiverse [10]. In particular, in the MWO model, anthropic selection has a straightforward interpretation: the parameters of our O-region are selected among the vast number of parameter sets existing in the multiverse (in an infinite number of copies each) by virtue of being conducive to the emergence and sustenance of complex life forms.
Compared to older cosmological concepts that considered a finite universe, the MWO model changes the very notions of "possible", "likely", and "random" with respect to any historical scenario (see Table 1). Simply put, the probability of the realization of any scenario permitted by the conservation laws in an infinite universe (and, of course, in the multiverse) is, exactly, one. Conversely, the probability that a given scenario is realized in the given O-region is equal to the frequency of that scenario in the universe. From a slightly different perspective, the usual adage about the second law of thermodynamics being true in the statistical sense takes a literal meaning in an infinite universe: any violation of this law that is permitted by other conservation laws will happen – and on an infinite number of occasions. Thus, spontaneous emergence of complex systems that would have to be considered virtually impossible in a finite universe becomes not only possible but inevitable under MWO, even though the prior probabilities of the vast majority of histories to occur in a given O-region are vanishingly small. This new power of chance, buttressed by anthropic selection, is bound to have profound consequences for our understanding of any phenomenon in the universe, and life on earth cannot be an exception.
The central problem: the emergence of biological evolution, the inherent paradoxes of the origin of replication and translation systems, and the limitations of the RNA world
The origin(s) of replication and translation (hereinafter OORT) is qualitatively different from other problems in evolutionary biology and might be viewed as the hardest problem in all of biology. As soon as sufficiently fast and accurate genome replication emerges, biological evolution takes off. I use this general term to include Darwinian natural selection[16] along with other major evolutionary mechanisms, such as fixation of neutral mutations that provide material for subsequent adaptation [17], exaptation of "spandrels" (features that originally emerge as evolutionary by-products but are subsequently utilized for new functions) [18], and duplication of genome regions followed by mutational and functional diversification [19]. All these processes that, together, comprise biological evolution become possible and, actually, inevitable once and only once efficient replication of the genetic material is established.
The crucial question, then, is how was the minimal complexity attained that is required to achieve the threshold replication fidelity. In even the simplest modern systems, such as RNA viruses with the replication fidelity of only ~10-3, replication is catalyzed by a complex protein replicase; even disregarding accessory subunits present in most replicases, the main catalytic subunit is a protein that consists of at least 300 amino acids [20]. The replicase, of course, is produced by translation of the respective mRNA which is mediated by a tremendously complex molecular machinery. Hence the first paradox of OORT: to attain the minimal complexity required for a biological system to start on the path of biological evolution, a system of a far greater complexity, i.e., a highly evolved one, appears to be required. How such a system could evolve, is a puzzle that defeats conventional evolutionary thinking.
The commonly considered solution is the RNA world scenario, i.e., the notion that replication evolved before translation such that the earliest stage of life's evolution was a versatile community of replicating RNA molecules [21-23]. A central element of the RNA world is a replicase consisting of RNA. The RNA world concept is supported by the experimental discovery of diverse catalytic activities of ribozymes (catalytic RNAs) [24-27]. However, all the advances of ribozymology notwithstanding, the prospects of a bona fide ribozyme replicase remain dim as the ribozymes designed for that purposes are capable, at best, of the addition of ~10 nucleotides to a oligonucleotide primer, at a very slow rate and with fidelity at least an order magnitude below that required for the replication of relatively long RNA molecules [28,29]. As recently noticed by one of the leading RNA world explorers, "Despite valiant efforts,...it appears unlikely that this particular polymerase enzyme will ever be evolved to the point that it can copy RNA molecules as long as itself (~200 nucleotides)" [30]. Of course, it remains possible – and this is, indeed, the belief in the RNA world community – that other ribozymes are eventually evolved to that level; however, the evidence is lacking.
The second paradox of OORT pertains to the origin of the translation system from within the RNA world via a Darwinian evolutionary process: until the translation system produces functional proteins, there is no obvious selective advantage to the evolution of any parts of this elaborate (even in its most primitive form) molecular machine. Conceptually, this paradox is closely related to the general problem of the evolution of complex systems that was first recognized by Darwin in his famous discussion of the evolution of the eye [16]. The solution sketched by Darwin centered around the evolutionary refinement of a primitive version of the function of the complex organ; subsequently, the importance of the exaptation route for the evolution of complex systems has been realized [18]. However, origin of translation resists both lines of reasoning. Primitive translation in a protein-free system is conceivable as an intermediate stage of evolution (see below) but this does not resolve the paradox because, even for that form of translation to function, the core components must have been in place already. Speculative scenarios have been developed on the basis of the idea that even short peptides could provide selective advantage to an evolving system in the RNA world by stabilizing RNA molecules, affecting their conformations or enhancing their catalytic activities [31-33] (see Ref. [34] for an attempt of a synthesis on this direction in the study of translation origins). These ideas are compatible with observed effects of peptides on ribozyme activity [35] but none of the scenarios is complete or supported by any specific evidence, and all include reactions without precedent in modern biological or model systems.
All this is not to suggest that OORT is a problem of "irreducible complexity" and that the systems of replication and translation could not emerge by means of biological evolution. It remains possible that a compelling evolutionary scenario is eventually developed and, perhaps, validated experimentally. However, it is clear that OORT is not just the hardest problem in all of evolutionary biology but one that is qualitatively distinct from the rest. For all other problems, the basis of biological evolution, genome replication, is in place but, in the case of OORT, the emergence of this mechanism itself is the explanandum. Thus, it is of interest to consider radically different scenarios for OORT.
The transition from anthropic selection to biological evolution in the history of life and the no-RNA-World scenario
The history of life includes a crucial transition from chance to biological evolution (Fig. 1). Biological evolution cannot take off before there are polymers (most likely, RNA molecules) and means for their sustainable replication. Thus, the synthesis of nucleotides and (at least) moderate-sized polynucleotides could not have evolved biologically and must have emerged abiogenically, i.e., effectively, by chance abetted by chemical selection, e.g., preferential survival of stable RNA species. At the other end of the spectrum, there can be no reasonable doubt that the first cells were brought about by biological evolution. Somewhere in between is the transition, the threshold of biological evolution. Most often, since the advent of the RNA World concept, this threshold is (implicitly) linked to the emergence of replicating RNA molecules. Translation is thought to have evolved later via an unspecified or, at best, invented ad hoc selective process. As discussed in the preceding section, both the ribozyme-catalyzed replication and, especially, evolution of translation in the RNA world face formidable difficulties. The MWO model dramatically expands the interval on the axis of organizational complexity where the threshold can belong by making emergence of complexity attainable by chance (Fig. 1). In this framework, the possibility that the breakthrough stage for the onset of biological evolution was a high-complexity state, i.e., that the core of the coupled system of translation-replication emerged by chance, cannot be dismissed, however unlikely (i.e., extremely rare in the multiverse).
The MWO model not only permits but guarantees that, somewhere in the infinite multiverse – moreover, in every single infinite universe, – such a system would emerge. The pertinent question is whether or not this is the most likely breakthrough stage the appearance of which on earth would be explained by chance and anthropic selection. I suggest that such a possibility should be taken seriously, given the paradoxes of OORT. A central corollary to this hypothesis is that the RNA World, as it is currently pictured, i.e., a vast community of replicating RNA molecules possessing a variety of catalytic activities but no translation system and no genetically encoded proteins, might have never existed. Of course, as discussed below, this does not at all rule out the special importance of ribozymes in early biology, in particular, in the primordial translation system.
The modern translation apparatus shows clear signs of evolution by duplication and diversification in the essential, ubiquitous components, allowing one to glean some features of the putative breakthrough system. Analysis of the duplications of key proteins involved in translation suggests that the breakthrough system was an RNA-based machine, to a much greater extent than the modern translation system. Specifically, the aminoacyl-tRNA synthetases (aaRS) comprise two unrelated classes each of which evolved via a series of duplications[36,37]. Moreover, both classes of paralogous aaRS are relatively late elaborations within large classes of nucleotidases [38-40], strongly suggesting that the breakthrough system activated amino acids via an RNA-only mechanism. The same pertains to the translation factors that are relatively late products of evolution within the GTPase class of the P-loop NTPases; thus, the breakthrough system would not employ protein translation factors [41]. The phenomenon of mimicking of tRNA structures by some of the translation factors [42-44] further supports the notion that the ancestral translation system was RNA-centered. The experimentally demonstrated activities of ribozymes include, among others, those that are involved in the main chemical steps of translation, such as amino acid activation, RNA aminoacylation, and peptidyl transfer [45-48]. Self-aminoacylation of ribozymes selected for this activity is rapid and highly accurate, remarkably, even more so than the same reaction catalyzed by the cognate aaRS [49]. Perhaps, most importantly, the large subunit rRNA itself is a ribozyme that catalyzes the peptidyl transferase reaction [50,51]. Thus, an RNA-only translation system, although so far not demonstrated experimentally [52], appears to be a realistic possibility.
A notable and enigmatic feature of the modern translation machinery is the common structure and the presence of conserved sequence elements in the tRNAs of all specificities which suggests that all the tRNAs are ancient paralogs[53]. Thus, the current set of tRNAs, obviously, is a product of biological evolution. The breakthrough system, conceivably, would utilize adaptors that were simpler than tRNAs, with the latter taking over already at the biological evolution stage. These primordial adaptors would have to possess the crucial capacity that, in the modern translation system, belongs to aaRS, i.e., combining amino acids with the cognate anticodons [34].
Under the present model, the core elements of the translation system, namely, a RNA-only ribosome and the specific adaptors for, at least, a subset of the 20 modern protein amino acids emerged by chance and were anthropically selected (Fig. 2). The breakthrough system was a primitive, RNA-based translation machine that was capable of translating exogenous RNAs such that functional proteins, including a replicase, could be generated. The presence of a diversity of randomly synthesized RNAs, including one that encoded a protein with a replicase activity (however low, initially), on the early earth would be another anthropically selected feature. For such an ensemble of RNA molecules to exist, a natural "reactor" is required in which polynucleotides are produced at an adequate rate and chemical selection occurs such that stable molecules survive longer. Networks of inorganic compartments existing at hydrothermal vents might be plausible candidates for this role [54,55]. Interestingly, a recent study that combined simulation and experiment has shown that even a low rate of production of mononucleotides would lead to their significant concentration in the peripheral compartments of such networks, and should polynucleotide be formed, they could reach very high concentrations [56]. Thus, the existence of "RNA-making reactors" under prebiotic conditions could be quite realistic [57].
Under these conditions, the emergence of RNA-based translation machinery would lead to the production of the replicase, and, with the ensuing RNA replication, the fundamental transition from anthropic to biological selection would occur (Fig. 2). In principle, the start of biological evolution is imaginable with the replicase initially being the only active protein. However, given the requirement for a RNA-producing "reactor", it seems an attractive possibility that, upon the advent of translation, other random RNA sequences gave rise to prototypes of the other major protein folds, yielding several protein activities (e.g., RNA-binding proteins or primitive enzymes facilitating nucleotide synthesis) and so conferring the minimal required robustness to the emerging biological system. The emergence of these folds would comprise the "Big Bang" of the protein universe [39,58].
The modern, universal genetic code is much more robust than expected by chance with respect to mutational and, probably, also translational errors: it has been estimated that the probability to obtain a code with the same or greater robustness than the actual one is -6 [59-61]. This robustness is manifest in the non-randomness of the code structure such that amino acids with similar properties are, typically, encoded by codons that differ in a single position (e.g., all codons with a U in the second position encode hydrophobic amino acids) [62]. This is, typically, considered to be a result of evolutionary optimization of the code [60]. However, the MWO model suggests an alternative view under which the basic structure of the code results from anthropic selection inasmuch as only codes with a certain minimal level of robustness would allow the appearance of a functional replicase in the breakthrough system. Of course, this scenario for the emergence of the code does not preclude subsequent adjustments via biological evolution.
The proposal outlined above eliminates the paradoxes of OORT by postulating that replication and translation, in their most basic forms, have not evolved biologically but rather were brought about by chance abetted with anthropic selection. The MWO model seems to render this a viable, however counterintuitive, possibility.
Objections, implications, and falsification
The present proposal, the appearance, via anthropic selection alone, of a RNA-protein system sufficiently complex to couple translation with replication such that biological evolution could take off, might seem quite outrageous. However, there are several mitigating considerations. First, the postulated chance origin of the replication-translation system does not require any mysterious processes. On the contrary, the only reactions involved are regular ones, such as polymerization of nucleotides and amino acids, nucleotide phosphorylation/dephosphorylation etc, and the only interactions required are those that are common in chemistry and biochemistry. Interestingly, the elementary reactions required for translation (amino acid activation, RNA aminoacylation, and transpeptidation) are relatively easily modeled with ribozymes (see above), in a marked contrast to RNA replication. Second, any conceivable scenario of life's evolution necessarily requires combinations of highly unlikely conditions and events prior to the onset of biological evolution, including the abiogenic synthesis of fairly complex and not particularly stable organic molecules, such as nucleotides, the concentration of these molecules within appropriate compartments, and their polymerization yielding polynucleotides of sufficient size and diversity. Thus, anthropic selection appears to be an inevitable aspect of life's evolution (Fig. 1).
Here I invoke the MWO model to argue that the range of complexity that is open to anthropic selection could be much greater than previously suspected such that a primitive coupled replication-translation system might have emerged without biological selection (Fig. 1). This scenario seems to eliminate the paradoxes of OORT. The origin of a complex system capable of performing a biological function by chance might appear nonsensical. I believe, however, that this is, primarily, a semantic trap. Prior to the onset of biological evolution, there could be no function, just complexity, and the emergence of any level of complexity is guaranteed by the MWO model.
A crucial aspect of the framework developed here is brought about by a disturbing (almost nightmarish) but inevitable question: in the infinitely redundant world of MWO, why is biological evolution, and in particular, Darwinian selection relevant at all? Is it not possible for any, even the highest degree of complexity to emerge by chance? The answer is "yes" but the question misses the point. Under the MWO model, emergence of an infinite number of complex biotas by chance is inevitable but these would be vastly less common than those that evolved by the scenario that includes the switch from chance/anthropic selection to biological evolution. The onset of biological evolution canalizes the historical process by reducing the number of available trajectories to the relatively few robust ones that are compatible with the Darwinian mode of evolution of complex systems (Fig. 3). This leads to a much greater rate of change than achievable by chance such that, as soon as there is an opportunity for biological evolution to take off, anthropic selection is relegated to a secondary role in the history of life. Of course, "secondary" does not mean unimportant; contingency and randomness are crucial, especially, at transitional stages of evolution (e.g., [63,64] but the basic framework is Darwinian. Thus, in any reconstruction of the origin of life, the threshold should be mapped to the lowest possible point, i.e., to the minimally complex system capable of biological evolution.
The strong form of the present hypothesis, i.e., the notion that the breakthrough stage in the history of life was a primitive coupled replication-translation system (Fig. 2), is falsifiable. Such a system should be construed as the upper bound of complexity for the breakthrough stage (Fig. 1). As soon as the possibility of biological evolution at a lower level of complexity, e.g., in the RNA world, is demonstrated and the route from the RNA world to the translation system is established, either experimentally or, at least, in a compelling model, the origin of a complex system with coupled replication and translation by chance/anthropic selection will be, effectively, ruled out. A demonstration that life independently emerged on several planets in our O-region will have the same effect. In the Appendix, I provide a rough, toy calculation of the upper bound of the probability of the emergence of a coupled replication-translation system in an O-region – this probability is, indeed, vanishingly small. The converse prediction is that any life forms that might be discovered on Mars or, perhaps, Europa during future planetary explorations will have a common origin with life on earth.
Any of the above falsifications will refute the model shown in Fig. 2 but will not make the MWO worldview irrelevant for our understanding of the origin of life. Indeed, any such discovery (tremendously important in itself) will simply lower the threshold of biological evolution on the scale in Figure 1. This point can be illustrated by deliberately naïve, toy calculations of the upper bound of the probability of the emergence of different versions of the breakthrough system by chance. In the Appendix, I present such calculations for two versions, the RNA World with a ribozyme replicase, and the coupled translation-replication system. Under the assumptions of this toy model (idealized to the extreme in that an unrealistically high rate of abiogenic RNA production is assumed), the emergence, by chance, of a ribozyme replicase in a finite universe consisting of a single O-region like ours, in principle, could be considered. However, any significant increase in complexity would call for a different cosmological model. In particular, the emergence of a coupled replication-translation system is unlikely to the extent of being, effectively, impossible. For such a complex system to be a viable candidate for the breakthrough stage, an infinite multiiverse, such as the one depicted by MWO or, in the very least, a universe with a vast number of O-regions, is, indeed, a must.
Of course, the most straightforward and powerful falsification would be to disprove the MWO itself. Here, however, an important disclaimer is due. It is not, actually, crucial for the validity of the conceptual framework presented here that MWO is correct in all its details. Only two rather generic assumptions are essential: i) a spatially infinite universe such as any (island) universe in MWO; the multiverse, while integral to eternal inflation, is not actually required for my argument, ii) the finiteness of the number of macroscopic histories in any finite region of spacetime. The strong form of the hypothesis presented here will not be falsified if some specific details of the MWO turn out to be wrong but only if one of these general assumptions fails.
Answers to all your Biology Questions
