Emergence of Self-Reproduction in Cooperative Chemical Evolution of Prebiological Molecules - Discover Life

In an earlier publication, steps of molecular self-organization leading to formation of a protocell with emphasis on the possible role of short peptides were discussed (Fishkis 2007). The proposed initial steps included polymerization and coevolution of short peptides and oligonucleotides on mineral surfaces with formation of a mutually autocatalytic molecular system. The next steps involved formation of vesicles by self-assembly of short peptides and other amphiphiles, entrapment of the peptide-oligonucleotide molecular systems in these vesicles, acquiring energy sources in the form of activated monomers or light absorbing polycyclic aromatic hydrocarbons (PAH’s), and further evolution of protocells.

In the present work, we concentrate on a critical step in this sequence: formation of a mutually autocatalytic self-replicating molecular system by the coevolution of short peptides and oligonucleotides.

One of the major problems in origin of life research is a very low probability of spontaneous formation of long functional biological molecules by chance (de Duve 1987; Shah 1994)Therefore, in order to make progress in our understanding of the origin of life, we need to address the following key questions:

• What made life's origin probable instead of random and highly improbable?

• How did molecular systems acquire functional aspects?

• What are the physical and chemical processes that produced complex systems, structural organization, and the emergence of new properties?

We define “system” here after Weiss’ definition (1970) as an aggregate of components, where the range of possible states of the components is significantly smaller than the corresponding range for the individual constituents, but larger than zero.

How did chemical systems arise, or in other words, what is a mechanism of selection that transforms chemical mixtures into systems?

Kauffman (1986, 1993, 1995) asserted that a collection of peptides at a certain critical level of molecular diversity may form a collectively autocatalytic molecular set. This set was assumed to be capable of reproducing itself as a whole, rather than reproducing separate molecules. Critics have argued that, without a replicating genetic molecule, such a system lacks heritable variations which are necessary for the evolutionary mechanism to function (Maynard Smith and Szathmary 1997, 1999).

Root-Bernstein and Dillon (1997) found molecular complementarily to be a powerful phenomenon that helps to answer the fundamental questions of the origin of life. These authors define complementarily as “non-random, reversible coupling of the components of a system. Molecular complementarily refers specifically to non-random, non-covalent interaction between molecules. Thus, van der Waals forces, hydrogen bonds, pi-pi stacking bonds, ionic bonds, charge-transfer complexes, and similar reversible chemical interactions can all be involved in molecular complementarity.”

Since complementarity involves non-random coupling between the components, it decreases variance in an aggregate and leads to its transformation into a System. Molecular complementarity provides a mechanism for the formation of stable and semi-stable functional subsystems having aggregate emergent properties not present in individual components (Root-Bernstein and Dillon 1997). In the next hierarchical level of organization, the subsystems interact to form a system with new emergent properties that stem from their interaction and that were not present in the component subsystems. Chemical co-evolution of prebiotic molecules on the early Earth can be viewed as a particular case of complementarity.

One prominent modern hypothesis for molecular evolution leading to the origin of life is the “RNA World” scenario. According to this hypothesis, the first molecular systems capable of replication and catalysis were based entirely on RNA molecules (Gilbert 1986; Maurel 1992; Weiner and Maizels 1991). One of the problems with this model is the difficulty of the prebiotic synthesis of nucleotides (Joyce 1989). The other major problem is the instability of RNA molecules (Orgel 1998). Due to the instability, in order to have a large enough number of molecules for continuing self-replication, replication of RNA molecules had to be efficient, which in turn requires longer and more efficient RNA enzymes. The availability of such molecules seems unlikely due to the imprecise character of replication with relatively short polynucleotides in the role of catalysts (Joyce 1989).

In all known life forms, the formation and replication of protein molecules requires RNA molecules and RNA synthesis and replication requires proteins. This fact, in combination with the continuity principle (Orgel 1968; Morowitz 1992; Weiner and Maizels 1991), makes the early collaboration of oligonucleotides and peptides a more attractive scenario. The various models that use this approach (Eigen 1971; Eigen and Schuster 1978; Brack and Barbier 1990; Lahav 1991, 1993) can be called “co-evolution hypotheses”.

The model developed in this work will show that early co-evolution between two types of polymers, one serving as a template (e.g. an oligonucleotide) and the other as a catalyst (e.g. a peptide), can lead to the formation of a stable functional collectively autocatalytic molecular System capable of growth and reproduction. In the model, this cooperative System outcompeted and replaced other molecules on mineral surfaces in the prebiotic environment.

Ideally, simulation of the molecular evolution leading to the origin of life requires direct modeling of the physics and chemistry of the involved molecular systems. However, such modeling far exceeds current computing capabilities (Hutton 2007).

Therefore, we are forced to use simulations that preserve the essential features of the modeled system, but require much less computing power. One such approach is artificial chemistry. Dittrich et al. (2001) defined artificial chemistry as “a man-made system that is similar to a real chemical system. More formally, it can be defined by a triple (S, R, A), where S is the set of all possible molecules, R is a set of collision rules representing the interaction among the molecules, and A is an algorithm describing the reaction vessel or domain and how the rules are applied to the molecules inside the vessel.” Artificial chemistry models exhibit various degrees of abstraction from real chemistry. The authors noted that “the knowledge accumulated in studying artificial chemistries will provide a fertile ground for new ideas about the origin of life and for prebiotic evolution.”

Molecular self-organization is one of many examples of emergent phenomena in complex systems (Troisi et al. 2005). Complex systems have components that are connected in a non-linear fashion. They often exhibit complex behavior on a global scale as a result of simpler interactions between their components. Agent-based computer modeling is widely used to study emergent phenomena in complex systems in ecology, epidemiology, and social sciences (see, for example, Berry et al. 2002; Bonabeau 2002; Lempert 2002). Recently, some progress has been made in the application of this technique to the study of multi-cellular biological phenomena (Bailey et al. 2007; Meyer-Herman and Maini 2006; Pierce et al. 2004; Segovia-Juarez et al. 2004).

The application of agent-based programming to study self-assembly of prebiotic molecules on mineral surfaces is described here.

The agents are autonomous entities that are allowed to interact with each other according to a defined set of rules. The great advantage of using this modeling technique is that a simple set of rules applied to each agent can lead to very significant and interesting emergent properties of the whole assembly; these are often unpredictable outside agent-based modeling. The rules often have ‘if-then’ conditions and frequently depend on the local environment. Agent-based modeling can be considered a generalization of Cellular Automata modeling (Vichniac 1984) since, in general, the model doesn’t have to be on a lattice. This approach provides a tool to gain insight into the detailed effects of various parts of the system on global behavior and vice versa.

For convenience, in the computer modeling of molecular self-assembly and interactions, we used the agent-based simulation framework Repast (REPAST.Sourceforge) that is widely applied in modeling social and economic systems (Samuelson and Macal 2006; North et al. 2006).

Our artificial chemistry agent-based model of early molecular co-evolution (EC model) is founded on the following assumptions discussed earlier in more detail (Fishkis 2007):

• Co-evolution of peptides and oligonucleotides

• Initial polymerization of amino acids and nucleotides on mineral surfaces

• Fluctuating environmental conditions

We assumed that the monomer molecules from which peptides and oligonucleotides could be constructed were available in the prebiotic milieu due to synthesis from atmospheric gases and minerals on the early Earth and the extraterrestrial delivery of organic material (Miller and Urey 1959; Cronin and Pizzarello 1983; Chyba et al. 1990; Chyba and Sagan 1992; Pierazzo and Chyba 1999).

We considered chemical reactions taking place on mineral surfaces since these surfaces can concentrate the reactants by adsorbing them and may catalyze polymerization reactions, as was experimentally shown by Ferris and co-workers (Ferris et al. 1989; Ferris 1993; Ertem and Ferris 1996; Ferris 2005, 2006). In fluctuating environmental conditions, during a cold/wet part of a cycle, the surfaces could be covered by a thin layer of liquid, which would accelerate surface diffusion of the molecules (Kohn 1999).

Polymerization of monomers to form peptides and oligonucleotides in solution is a thermodynamically unfavorable process. (Orgel 1998). High water concentration drives the process towards hydrolysis. Adsorption of the monomers on to mineral surfaces promotes polymerization by providing protection against hydrolysis (Wachtershauser 1988) and by partial ordering of molecules on the surface relative to their free movement in solution and thereby decreasing the entropy cost of polymerization (Penny 2005). Condensation reactions are also favored in anhydrous conditions when water produced by these reactions leaves the reaction zone. Hot and dry parts of the environmental cycles could provide such conditions (Deamer et al. 2006).

In addition, surface polymerization is promoted by the dependence of free energy of adsorption on the number of residues in the polymer (Orgel 1997). The longer polymer molecules have larger adsorption energy and as a result preferentially accumulate on mineral surfaces.

Catalysts including catalytic peptides decrease activation free energy of reaction and affect reaction rates of both the forward and reverse reactions. At equilibrium, they have no effect on reactant concentrations (Tinoco et al. 1985). Therefore the polymerization process requires input of energy in order to keep the System far from equilibrium (Lancet et al. 1994). Possible processes that could supply energy include dry heat (Deamer 1997; Zubay 2000; Deamer et al. 2006) , reaction of the amino acid with a pyrophosphate to produce a phosphate anhydride, reaction to form a thioester, or reaction with a condensing agent like cyanamide to yield a high energy intermediate (Hulshof and Ponnamperuma 1976; Orgel 1989)

In the EC model, the water pool in contact with the mineral surface contained two types of monomers: amino-acids and nucleotides. These molecules, for simplicity only of two varieties each, nucleotides A and B and amino acids C and D, were gradually deposited on a mineral surface at the edge of the shallow pool where conditions changed periodically from dry and warm to cold and wet. Some of them were adsorbed on the surface. Those that were close enough reacted with each other forming polymer chains containing gradually increasing numbers of unit molecules.

The model encompassed a mineral surface containing N_x × N_y lattice sites. Each lattice site was large enough to contain many molecules of various types at the same time. The lattice sites represent reaction areas. Only molecules that reside within each lattice site were considered to be close enough for chemical interactions; however, the sites could exchange molecules with each other by diffusion. They could also exchange molecules with the water pool. In modeling diffusion, we assumed toroidal topology to avoid edge effects.

Two distinct combinatorial families of chain molecules, oligonucleotides and peptides, were formed by slow random polymerization of monomers on the mineral surface. Peptides were not capable of self-replication, whereas oligonucleotides could be replicated and translated. Since the catalytic properties of peptides depend on their composition, peptides within a particular compositional range (as discussed below) were assumed to exhibit catalytic properties.

The catalytic properties of proteins are well known (Raven and Johnson 1995). Very short prebiotic peptides were not likely to form secondary or tertiary structures (Shen et al. 1990) However there exists experimental evidence that even very short peptides can have catalytic properties. For example, Shen and co-workers (1990) have found that histidyl–histidine (His–His) exhibits catalytic properties. They concluded that the dimer imidazole ring acts not only as a general base–acid catalyst but also functions as a proton relay that creates an activated intermediate for the formation and hydrolysis of phosphodiester bonds and peptide bonds.

Later, Bar-Nun and Kochavi (1994) showed that activated amino-acids can assemble themselves around substrate molecules through hydrogen bonds and van der Waals interactions and catalyze a variety of chemical reactions in the substrates. They also polymerized a group of amino acids in the presence of a substrate acting as template to form polypeptides. These short oligopeptides were found to be catalysts for the substrate molecules that directed their formation (Bar-Nun et al. 1994; Kochavi et al. 1997). According to the authors, reduction in an activation barrier for reaction by a peptide catalyst is based on low energy noncovalent interactions, such as hydrogen bonds, salt bridges, hydrophobic interactions, and van der Waals forces. Since each of these interactions is in the range of 0.1–1.0 Kcal/mol, the change in free energy per residue is too small to substantially change the activation barrier. However, accumulation of these changes in a cooperative way can modify the catalyst-substrate complex resulting in lowering the energy barrier.

In the EC model, the peptides having particular compositions as expressed by monomer sequences were considered to be catalysts for all replication and translation chemical reactions in the area (lattice site) where the catalytic molecules resided. However, due to a finite rate of diffusion, we expect a catalytic peptide molecule to be closer to the oligonucleotide template that produced it than to other molecules for a period of time after its formation. As a result, during this time it would catalyze reactions of the template self-reproduction and translation producing a reproductive advantage for the catalytic peptides and their templates.

It has long been established that polynucleotides adsorbed on mineral surfaces have inter-residue spacing that is very similar in size to the inter-residue spacing in polypeptides (Woese 1967; Soberby and Heckl 1998). Therefore polynucleotides on mineral surfaces may advantageously position and align amino acids for their polymerization.

Soberby and coworkers have found that nucleotides and nucleotide bases deposited on mineral surfaces can be selective in binding amino acids from a solution which they contact (Soberby et al. 2000, 2002). Preferential interactions of the specific amino-acid monomers with certain functional groups of the oligonucleotides were attributed to variation in the electronic configuration around the amino and carboxylate groups of amino acids. These interactions can, under certain conditions, lead to oligonucleotides functioning as templates for the synthesis of peptides from amino acids. Formation of peptide bonds between the aligned amino acids is facilitated by water evaporation during hot/dry period of the environmental cycle (Brack 1993; Lahav 1994; Rode 1999).

Some short oligonucleotides have been shown experimentally to serve also as templates for the synthesis of complementary oligonucleotides from a solution of nucleotide monomers. An example of this is the formation of oligo(G) up to the hexamer using oligo(C) of similar or higher length as templates (Grzeskowiak and Orgel 1986; Ertem and Ferris 1997; Ertem 2004).

In our model, oligonucleotides served as templates for self-reproduction and for translation into peptide sequences. They were reproduced by formation of the complementary sequences from free monomers. Complementary oligonucleotides were formed by Watson-Crick pairing, so that in a next step a replica of the original oligonucleotide was formed.

When an oligonucleotide served as a template for translation into a peptide sequence, we assumed a correspondence between a template nucleotide sequence and a peptide amino acid sequence. Translation rules assumed preferential bonding of the C amino acid to the A nucleotide and of the D amino acid to the B nucleotide. No recognition of specific template sequences by catalytic peptides was assumed.

We assumed periodic environmental conditions with hot/dry periods alternating with cold/wet periods (Orgel 1998). Molecules randomly deposited on the surface from the evaporating water. A time cycle in our model was defined as a period of time in which on average one molecule per cell was deposited on the surface.

For a polymer molecule the probability that the rate of chain cleavage or hydrolysis will exceed the rate of chain growth increases with the number of bonds between monomers [Orgel 1997, 1998]. This indicates that there is a limit to the size of spontaneously growing oligonucleotides and peptides on a mineral surface. In our model, we selected this limit to be five monomers as a compromise between catalytic and template properties of the oligomers (favoring greater length) and a rate of spontaneous formation on a mineral surface (favoring short molecules). The short length of templates allows a spontaneous dissociation of the formed copy from the template.

During each cycle, nucleotides A and B and amino acids C and D were randomly deposited from the pool and adsorbed by the mineral surface. The rate of deposition was the same for all four monomers, for the sake of simplicity. These spontaneously polymerized to form chain molecules of oligonucleotides and peptides up to five monomers long. After achieving this length oligonucleotides started functioning as templates for self-replication and translation.

During the hot/dry period, more molecules were deposited on the mineral surface from the evaporating water, followed by polymerization and separation of strands. During the cold/wet period, there was an increase in surface diffusion rates [Kohn 1999] and some hydrolysis and removal of molecules from the surface back into the water pool.

Compositions of peptides and oligonucleotides in our model were represented by sequences of A and B or C and D monomers. These sequences were coded by strings of binary digits with 0 representing A(or C) and 1 representing B(or D) and then each of the strings was replaced by equivalent decimal numbers from 0 to 31.

In the model it was assumed that peptides with molecular compositions in the range 16 to 19 exhibited weak catalytic properties. As mentioned above, due to a finite rate of diffusion, the catalytic peptides stayed in close proximity to the template molecules that coded for them for a period of time after their formation. As a result, during this time period, they had a stronger effect on the reactions involving their template molecules than on the reactions involving other molecules within the same vicinity.

Since concentration of the amino acids, nucleosides and other prebiotic molecules on the early Earth is not known, the model used relational values for the rates of polymer deposition, extension, hydrolysis/dissolution and surface diffusion. The length of a cycle was selected as a period of time necessary for deposition followed by adherence to the surface or by reaction with a surface bound molecule of one molecule per cell.

The rates of hydrolysis and transfer of molecules back to solution were adjusted to keep the surface molecular population constant in the state of dynamic equilibrium. Since polymer affinity to the mineral surface increased with the number of residues [Orgel 1997], the rate of return of the polymers back to the solution in the model was inversely proportional to the number of residues.

The other main parameters of the model were varied in order to study their effect on the accumulation of the catalytic molecules and their templates on the surface.

• Template replication error varied in the range 0–10%;

• Catalytic efficiency, defined as the reaction rate increase due to the presence of 1 catalytic molecule varied between 1.25 and 2; Due to the absence of experimental data, very conservative values were selected.

• Diffusion rates were introduced as a fraction of molecules that diffused to the neighbouring reaction zones during one cycle. In the model, they varied from 1/50 to 1/10.

The histograms in Figs. 1, 2, 3 and 4 show numbers of molecules of various compositions on the surface, at different stages of the co-evolutionary process.

Initial distribution of the oligonucleotide molecular compositions on the mineral surface

Full size image

Initial distribution of the peptide molecular compositions on the mineral surface

Full size image

Distribution of the oligonucleotide molecular compositions on the mineral surface after 400 reaction cycles

Full size image

Distribution of the peptide molecular compositions on the mineral surface after 400 reaction cycles

Full size image

Since the rates of deposition in the model were the same for all 4 monomer compositions, the initial distribution of molecular compositions for molecules on the mineral surface was roughly random (Figs. 1 and 2).

Not all molecules reproduced at the same rate. The oligonucleotides in close proximity to catalytic molecules had higher rates of replication and translation. Due to the finite rate of diffusion, these oligonucleotides were more likely to be the templates of catalyst molecules. A higher rate of reproduction of the catalyst templates led to a higher production of protein catalysts which, in turn, resulted in a higher production of oligonucleotide templates, and so on. This led to a predominant accumulation of these peptides and oligonucleotides on the mineral surfaces, while peptides and short nucleotides of other compositions were formed at lower rates and eventually became destroyed or dissolved in water. Therefore, the numbers of catalytic peptide molecules and the oligonucleotide molecules that coded for them gradually increased relative to the numbers of molecules of other compositions.

The histograms in Fig. 3 show that, after 400 cycles, the templates that code for the peptides that have catalytic properties (16–19 on the digital scale) and the templates of the complementary compositions (12–15 on the digital scale) accumulated on the mineral surfaces in preference to the peptides of other compositions.. The presence of a peak corresponding to the compositional sequences of the oligonucleotides complementary to the sequences of the catalyst templates is a consequence of the two step mechanism of the template replication. A similar pattern can be observed in Fig. 4 that presents the compositional distribution of peptides on the mineral surface. Peptides that exhibit catalytic properties and peptides of complementary compositions formed by translation of the complementary templates also were the prevalent peptides that stayed on the surface (Fig. 4).

Although the presented histograms give clear evidence of preferential accumulation of catalytic peptides, they show that smaller amounts of the peptides of other compositions were also present. Survival of molecules of various alternative compositions indicates that, when longer peptides form, there is also a pool of molecules with a variety of compositions available for further evolution.

The main parameters of the model are: accuracy of template reproduction and translation, catalytic efficiency of primitive enzymes and diffusion rates of molecules on the mineral surfaces. To investigate the effects of these parameters on the surface accumulation of primitive catalytic peptides and the templates that code for them, we plotted the concentration of the catalyst peptides and their templates, expressed as a mole fraction, as a function of the number of deposition cycles for various values of these parameters as shown in Fig. 5, 6,and 7.

Effect of template replication error on surface accumulation of the catalyst templates. e—template replication error. a e = 0; b e = 5%; c e = 10%

Full size image

Effect of diffusion rates on the concentration of catalyst templates on the surface. F—model parameter proportional to the rate of diffusion. a F = 1/10; b F = 1/25; c F = 1/50

Full size image

Effect of catalytic efficiency of peptides on the concentration of peptide templates. a Catalytic efficiency = 1.25; b Catalytic efficiency = 1.5; c Catalytic efficiency = 2

Full size image

The results indicate that template accuracy was the most critical parameter in our model. Figure 5 depicts the accumulation of catalyst templates on the surface at various template reproduction error rates. It shows that the System didn’t exhibit “error catastrophe” and that accumulation of mutually catalytic molecules took place even at template reproduction errors as high as 10%. We attribute this to the small size of the oligonucleotide and peptide molecules. Comparison of the plots in Fig. 5 indicates that the template reproduction error of 10% delayed preferential accumulation of catalyst templates on the surface and decreased final concentration of the catalyst templates. Final molar fraction without template error was above 0.9; with 10% error, the final fraction decreased to values near 0.6.

The effect of diffusion rates (Fig. 6) was twofold. On the one hand, diffusion rates influenced the rate of molecular exchange between lattice sites (interaction zones). On the other hand, diffusion rates affected the time of proximity of the newly formed catalytic peptide to the template that coded for it. Slower diffusion rates increase the reproductive advantage of the catalyst templates as compared to other molecules on the surface and therefore increase the rate of their accumulation (as well as the rate of catalytic peptide accumulation). This is evident from comparing the plots a, b and c in Fig. 6 that represent accumulation of catalyst templates for different values of the model parameter ‘f’ that is proportional to diffusion rate.

The influence of catalytic efficiency defined as reaction rate enhancement (Hazen et al. 2007) on the growth of the surface concentration of catalyst templates is shown in Fig. 7. In this model it was assumed that a non-specific catalytic efficiency applied to all reactions within the interaction area. However, due to the original proximity of the catalysts to their templates, there was an effect of the catalytic efficiency on the differential reproduction and translation rates of the templates. This effect manifested itself in higher accumulation rates of the catalysts and their templates for the catalysts having higher catalytic efficiency as evidenced by plots in Fig. 7.

The model illustrates how early cooperation between peptides and oligonucleotides polymerizing on a mineral surface could lead to the evolution of a molecular population resulting in the “emergence” of a non-random functional molecular System, capable of self-reproduction. The emerged molecular System had properties of multiplication, heredity and variability. It displayed a hereditary difference in reproduction rates and therefore could further evolve by natural selection.

Evolution requires not just heredity but what Szathmary and Maynard Smith (1995) call “unlimited heredity.” Unlimited heredity involves potential existence of a very large variety of replicating molecules. In our model, this can be achieved by gradual increase of the peptide and oligonucleotide molecular size as well as expansion of the variety of available monomers in the environment. With time, longer and more efficient peptide catalysts would emerge and, in turn, increase template accuracy. Some of the new template-catalyst autocatalytic systems with higher accuracy and rate of reproduction would come to predominate.

In all contemporary life forms, protein enzymes catalyze most molecular reactions, including the synthesis of nucleic acids, whereas nucleic acids carry the information necessary for the synthesis of proteins. The model described above outlines a possible origin of this relationship: it may have started with cooperative polymerization of short peptides and oligonucleotides on a mineral surface.

Follow up stages in this model of prebiotic evolution include the self assembly of simple amphiphiles and short peptides into vesicles (Segre et al. 2001; Santoso et al. 2002; Vauthey et al. 2002); the entrapment of the autocatalytic peptide–oligonucleotide systems within these vesicles (Deamer and Barchfeld 1982; Shew and Deamer 1985; Monnard et al. 1997), and the formation of protocells with semi-permeable membranes capable of primitive metabolism, growth and division (Chakrabarti et al. 1994; Yu et al. 2001; Treyer et al. 2002).

The model of early prebiological molecular coevolution presented in this paper suggests that cooperative interactions between very short oligonucleotides, serving as templates, and peptides having catalytic properties, could lead to the formation of collectively autocatalytic molecular sets capable of self-reproduction and molecular evolution.

This process could be the first step towards formation of a functioning protocell by encapsulation of the oligonucleotide-peptide molecular set in the semi-permeable membrane and incorporating light-capturing PAH derivatives or other energy sources (Deamer 1992, 1997).

In a broader sense, the model describes any molecular system that contains 2 types of polymers, with the one type serving as templates and the other as catalysts.

Early molecular co-evolution provides a mechanism for non-random evolution due to inherent self-selection. It may indicate a direction for resolving the problem of very low probability of chance spontaneous formation of functional biological molecules.

So far, there is no experimental data to support the co-evolution hypothesis. An experimental strategy could be to study the catalytic activities of small peptides and the template properties of oligonucleotides as a function of their size and composition.

Authors

1. Maya Fishkis

   You can also search for this author inPubMed Google Scholar