The Problem of the Origin of Nucleic Acid Replication.

It would be nice to know how long sequences of nucleic acid could have come to self-replicate because we would then understand the origin of a system capable of transmitting effectively infinite amounts of information from parent to child, i.e. the origin of unlimited heredity. Information is used in the formal sense here, and is a measure of the number of distinct states that can be transmitted from sender (parent) to reciever (child). Systems capable of unlimited heredity can be counted easily on the fingers of one hand. The first is the genetic system, in which long sequences composed of 4 letters are replicatable. Next is the human system of language. Language is capable of transmitting an effectively infinite number of different 'meanings' faithfully, both horizontally (between living humans) and vertically (to humans that have not yet been born). The origin of language is also not understood. We can try to think of other systems of unlimited heredity in between these two.

In the section on chemical evolution, the origin of nucleotides and early template replication was discussed. It is not known under what conditions non-enzymatic template replication could have evolved since there has been no demonstration of artificial chemical evolution capable of producing replicating polymers. Leslie Orgel believes that "to understand the origin of life one only has to understand the origin of the RNA world" (Orgel, 2003), however, Orgel admits that protocells, and a non-nucleic acid system of template heredity are likely to have been a necessary prerequisite for the RNA world.

In the Chemoton model described previously, it is assumed that nucleic acid templates are able to replicate without enzymes, inside a replicating protocell with a Formose cycle metabolism. Of-course Ganti does not make it clear how non-enzymatic template replication is actually to occur. The chemoton model assumes that nucleic acids undergo template polycondensation. This is a kind of polymerization in which a side-product is formed and released in addition to the extended template product. We saw the only artificial physical model of template polymerization by Jarle Breivik (Breivik, 2001) in the section on artificial life approaches to self-replication. Another inorganic example of template polymerization occurs in montmorillonite clay replication (Weiss, 1980). The relationship between clay replication and nucleic acid replication will is considered elsewhere in this thesis (Ertem, 2004). Ganti's original model considers the simplest case of homologous pairing of monomers, (see Figure 2.17 and 2.19 from Chemoton theory Vol I reproduced below) , that form templates. He assumes that polymers breathe at their ends due to thermal aggitation by water.

At a critical threshold concentration of monomer, V*, the breathing ends become occupied by monomers which then block the re-zippering of the breathing ends. Further monomers attach rapidly (due to stacking interactions) and Ganti assumes that this process continues until the templates are completely replicated. This process is abstracted in the kinetic equations into a rate-limiting initiation step and a fast chain propogation set of steps that result in the increase of template concentration (see the 'Whole Cell Models' chapter). If this model were accurate, then non-enzymatic template replication would be straightforward, involving only a high enough monomer concentration, for any length of template. Unfortunately, the kinetics of real non-enzymatic template replication are not so simple. Conclusions about protocell behaviour that depend critically on this threshold initiation condition are probably artifactual (see Chapter 'Whole cell chemoton models' for further discussion). We investigate below, the dynamics that seem to be most important for RNA replication, so that we can make a more valid model which may be of help in discovering how long RNA template replication may have originated.

 

The Properties of Nucleic Acid Polymers.

 

Discuss the primary and secondary structures of RNA.

The dynamics that are crucial for RNA replication seem to be those determining how template-directed replication can occur, i.e. the formation and breakage of phosphodiester (or equivalent covalent) bonds, and the formation and breakage of bonds responsible for denaturation and renaturation of dsRNA (double stranded) and ssRNA (single-stranged) respectively.

Nucleic acid polymers can be single or double stranded. Single strands are held together by phosphodiester bonds. These are strong covalent bonds between nucleotides, but also by base stacking (complex forces between adjacent nucleotides). Double strands are held together by additional Watson-Crick base pairs, and by base stacking.The phosphate bonds on double strands electrostatically repel each other. At the melting temperature (Tm) of a given strand, half the strands are dissociated, and half are associated. The bredth of the denaturation depends on the extent of cooperativity within the strand.

Phosphodiester bond formation.

The polymerisation reaction in DNA and RNA occurs by the 3'-hydroxyl group on the end of an existing primer strand forming a phosphodiester with a 5' phosphate on a nucleotide, displacing a pyrophosphate leaving group. The thermodynamic driving force is the cleavage of the weaker phosphatephosphate bond coupled with formation of a stronger phosphodiester bond and the entropy of freeing the pyrophosphate (Kool, 2001). A DNA polymerase enzyme decreases the activation energy of this reaction, so speeding it up. However, the reaction can occur slowly and with low sequence specificity without an enzyme.

Several forces act within the nucleic acid strand to bring the 3' hydroxy end and the 5' phosphate end together to allow polymerization. The net effect of these forces is to favour the DNA helix enthalpicly, so countering the strands entropic instability. The forces are: Watson-Crick hydrogen bonding which acts to stabilise the base pairing of nucleotides opposite each other; base stacking forces that act to stabilise adjacent nucleotides along the same strand; and electostatic forces between phosphate bonds that repel each other, both along a strand, and between opposite strands on a double helix. We consider these forces in turn.

Watson-Crick Hydrogen Bond Formation.

Base-Stacking

What is base-stacking?

These chairs can be stacked one on top of the other. In nucleic acids, base stacking refers to the interaction between nearest-naighbour nucleotides along the same strand. This is a very complex interaction that depends on Van der Waals forces, electostatic dipole forces between bases, and solvation effects, i.e. whether the DNA base is better bound to water, rather than to the adjacent base. “Aromatic stacking” refers both to the geometry of face-to-face juxtaposition of two aromatic molecules so that the pi-systems are in direct contact, and to the forces that favor this geometry energetically (Kool, 2001). Bases in DNA are in near-maximal face-to-face contact. Electrostatic interactions account for the large differences in stacking efficiency of DNA bases depending on the neighboring base (ibid. p10), but on average the elecrostatic effect is small.

Oligonucleotide Properties.

Experiments that compare the behaviour of oligonucleotides can identify the forces between specific nucleotides depending on their structural context. In order to derive individual forces from melting temperatures of oligonucleotides, it is necessary to have a model of how oligonucleotides denature and renature, and also to conduct experiments with many different types of oligonucleotide. Some of these experiments are described below.

Q1. How does the melting temperature depend on strand length? Experiments with homologous series consisting of AU repeats, showed that melting temperature increased with length according to the relationship, 1/Tm = A + B/N where A and B are oligomer specific constants, Tm is the melting temperature and B is the strand length. The stabilizing effect of an extra unstacked base further increases melting temperature slightly as shown above in Bommarito's experiments, however, the effect is small compared to an extra base-pair. Increasing the template concentration was also found to increase melting temperature according to the relation, 1/Tm = A' - B' ln(C), where A' and B' are oligomer specific constants and C is the template concentration. The figure below shows these melting temperature results from Chapter 23 of ? for AU sequences of different lengths.

Q2. How is melting temperature affected by base sequence? The base composition, and the base sequence influence the melting temperature. For example, the presence of a single G-C pair in an A-U duplexs increases melting temperature. The most useful model explaining this dependency has been the nearest neighbour model. The NN model assumes that the stability conferred to the duplex consists of an initiation energy due to the first base pair formation, followed by propogation energies due to continued base pair formation. The initiation energy is dependent on whether the initiating base pair is G-C or A-U, and the propogation energy is dependent on the identity of the adjacent base-pair and the base pair that is about to anneal. See diagram below of the 10 possible nearest neighbour interactions in the "dinucleotide cores" . The values for the change in free energy resulting from the formation of the adjacednt base pair are shown on the right.

 

SantaLucia (1998) integrated the work of many laboratories to determine the thermodynamics of nearest-naighbour interactions in DNA. The stability conferred to the double strand by the presence of a pair of adjacent bases can be determined, and is a function of both the Watson-Crick base-pair strength and the stacking interactions. The NN model for nucleic acids assumes (quite succesfully) that the stability of a given base pair depends on the identity and orientation of only the neighboring base pairs. The model assumes that the single strands come together by the formation of an initiating Watson-Crick base pair, and that subsequent propogation of the helix is by Watson-Crick and stacking interactions.

Unified NN parameters have been calculated that allow the calculation of the free energy of any duplex sequence at standard temperature. These are shown in Santalucia (1998), reproduced below.

The diagram below from SantaLucia 1998 shows how to use the table to calculate the free energy of any given sequence of DNA.

One simply goes through the 5'---> 3' strand one nucleotide at a time and sums the free energies that have been calculated for that particular adjacent pair of bases. There are some extra terms that depend on whether the template is self-complementary, and whether the sequence starts with a G.C or a T.A base pair. Note that the unified energies for the CG/GC and GC/CG pairs are quite similar.

The NN interactions differ depending on the type of the nucleotide, e.g. whether it is DNA, RNA, and one of their analogues. For example Xia 1998. Biochemistry, 37, 14719. Download pdf. presented another set of parameters for RNA, shown below.

These differ considerably from the DNA results. GC/CG has a much lower free energy of stacking than GC, in Xia's RNA results, in comparison to SantaLucia's DNA results. The melting temperature can be calculated from the free energy obtained from the NN model, according to the equation,

and so the nearest neighbour model encompases and agrees with the length dependence equations described above for AU strands.

Sequence effects due to CG and GC dimers.

Since our later experiments will concentrate on the differences between the behaviour of CG dimers and GC dimers of DNA and RNA analogues, and their polymers, the experiments of Sinclair et al 1984. Download pdf. will be considered in detail. Sinclair used NMR to study the melting behaviour of 6 self-complementary oligoribonucleotide 6-mers. At that time, Borer's NN measurements on RNA were the most recent. (Borer 1974), giving values of GG:CC -4.8, GC:GC -4.3 and CG:CG -3.0. These values differ from Xia's results above, which have GC:GC > GG:CC > CG:CG. Sinclair carried out experiments with the following 6 sequences. GGCC, CCGG, GCCG, CGGC, GCGC and CGCG.

He examined the secondary structures that they formed with themselves, at different termperatures. The diagram above from his paper shows that several secondary structures are possible depending on the sequence. a) We would expect GGCC and CCGG to anneal to form perfect minihelices since this is the only self-complementary secondary structure that is possible. b) We would expect CGGC and GCCG to anneal to form staggered helices since these are the only secondary structures possible that avoids non-complementary base pairing, i.e. the structure that allows free energy to be minimised. We would expect these staggered duplexes to form a 'loose' duplex joined by splint junctions.c) Finally we would expect CGCG and GCGC to anneal to form either staggered duplexes or perfect minihelices. Both these configurations are possible without non-complementary base pairing. If they anneal to form staggered duplexes we would expect that these could associate to form a 'loose' duplex joined by splint junctions.

Using NMR Sinclair deduced a relative ordering of stabilities that agreed with Xia's not Boron's model. The melting temperatures of the 6 6-mers are shown below.

These and chemical shift measurements show that GGCC, CCGG and GCGC are present as perfect helices, with high melting temperatures, and that CGCG, GCCG and CGGC are present as staggered helices with lower melting temperatures. The fact that GGCC has a slightly higher Tm than CCGG implies that the GC:GC core is stronger than the CG:CG core, consistant with Borer's results. Strangely, GCCG and CGGC both prefered to stagger with 3' dangling ends. The stability imparted by these ends outweighed the stability that would have resulted from a GC:GC core being possessed by CGGC had it bound with a 5' dangling end. GCGC favoured the formation of perfect duplxes, whilst CGCG associated in splint junction staggered complexes.

The explanation for the above findings was as follows. The perfect helix of GCGC contains two strong GC cores and one weak CG core, whereas the perfect helix of CGCG contains only one strong core and two weak cores. If aggregation is the driving force for the formation of staggered duplexes, then although the OVERALL core composition of aggregates is the same, at a splint junction, GCGC has a only a weak CG:CG core, and at a splint junction, CGCG has a strong GC:GC core. Thus the aggregate formed by CGCG would be more cohesive than the aggregate formed by GCGC. This explains why GCGC forms perfect helices and CGCG forms staggered helices.

Cruz et al (1982) have suggested that there exists interstrand and intrastrand stacking, with RR:YY and YR:YR tending to form interstrand stacks, and RY:RY tending to form intrastrand stacks. (Y pyrimidine, R purine). See diagram below.

This model accounts for Sinclair's findings of a preference for 3' dangling ends. GCGC bases stack internally, whereas CGCG bases stack across a splint junction.

Walter et al 1994. PNAS, 91, 9218. Download pdf.

 

Q3. How is melting temperature affected by mispairing that forms hairpins, internal loops, and bulge loops? The size and base composition of the internal loop or hairpin influences melting temperature. Tm decreases with increasing extents of mispairing. If the nearest neighbour model is used to calculate the free energy of formation (with respect to a single strand) of a particular secondary structure, then it is necessary to take into account changes of the free energy due to loops and hairpins, as well as contributions to the free energy provided by the duplex base pairs. One proposed model foer doing this is summerized in the table below, from Chapter 23..?

 

Q4. How is melting temperature affected by single-strand stacking?

The importance of stacking interactions for duplex stability is appreciated when one realizes that the stacking interaction between bases on the same strand actually stabilizes the double strand, i.e makes it less likely to dissociate. This is shown by experiments on the melting temperature of sequences with dangling ends, compared to sequences without dangling ends. Sequences with unpaired dangling ends had higher melting temperatures. The increase in melting temperature depended on the particular stacking interaction, i.e. which base the unpaired base was attached to, and what end it was attached on (3' or 5') (Bommarito et al 2000). The free energy of stacking at dangling ends of DNA is shown in the table below from Bommarito et al 2000.

[This may be modelled as a stabilizing effect that the dangling base has on the adjacent Watson-Crick bond]?

Q5. How is melting temperature affected by chain fragmentation?

 

Electrostatic Phosphate Repulsion.

 

The First Hurdle: Formation of Long Nucleic Acid Polymers.

 

Long Nucleic Acid Oligomers Have Been Synthesised.

 

Before polycondensation reactions could be achieved, all sorts of side-reactions had to be abolished. For example, non-regiospecific products with 2'-5' linkages often predominated in early attempts to produce template polycondensation(Orgel and Lohrmann, 1974), (Inoeu and Orgel 1981). Special activated monomers had to be used to increase the specificity of the 3'-5' condensation reaction.

Inoue and Orgel (1982,1983) showed that template directed condensation of 0.05M 2-MeImpG (an activated G monomer) on 0.01M poly(C) (a polymer consisting just of C monomers) was possible, producing long 3'-5' linked olgio(G)s, of mean length 15, and occasional polymers > 30, after 7 and 21 days. Condensation on Poly(U) was not possible because triple helices were formed, whilst Poly(G) formed stable self-structures. Poly(A) did not interact with U sufficiently to overcome the poor U-U stacking. Co-condensation of two or more types of base was also possible on random copolymer templates as long as the template had an excess of C (Joyce et al, 1984). These and other experiments highlight the fact that there is a huge diversity of chemical behaviours depending on the strand sequence, since stacking and folding properties of the nucleic acid strand are determined by its sequence. This 'phenotypic' variety consisting of the chemical kinetic and folding properties of the polymer, is useful if natural selection is to act on template sequences to select those with high fitness. [The stochastic model of templates that we will present later, has incorperated increasingly realistic chemical kinetics, e.g. sequence dependent stacking interactions, and has allowed the analysis of natural selection acting on sequences.]

Recent work has used 5'-phosphoro-2-methyl imidazolides as activated monomers (Kozlov and Orgel, 2000) with 10-mer template concentrations of 0.5mM and monomer concentrations of 100mM. 3'-5' phosphodiester linkages are the majority.

Zielinski and Orgel produced 30-mers after 12h by EDC self-condensation of GC dimers, but only 14-mers by EDC self-condensation of CG dimers (Zielinski and Orgel, 1987). One reason for the failure of CG elongation seems to be that CG dimers form P-bonds with themselves, rather than with other CG dimers (cyclization). 33-60%% of CG dimers were lost to cyclization, whereas only 14-28% of GC dimers were lost to cyclization. The figure below shows the yield of CG cyclization products (the side reaction) in red, and the yield of CG polymers in blue at 50mM concentration of dimers. Only 30% of the CG dimers polymerized, 60% underwent cyclization, and there were some other unknown side-reaction products.

Figure 0. From Zelinski and Orgal, (thanks to K. Achilles). Results obtained at 0.05M dimer concentration, and temperature 0 degrees C.

Why are GC dimers not cyclized as much as CG dimers? The explanation given by Zielinski and Orgel is that cyclization is only possible when dimers are not bound in a double-helix, and that GC dimers form more stable perfectly stacked 'mini-helices' than CG dimers (Sinclair et al, 1984), and so are present more often in the double-stranded state.

What molecular properties are responsible for this difference between GC and CG dimers? Sinclair et al (1984) used proton NMR to study the secondary structures of oligoribonucleotide tetramers consisting of C and G. They found that GGCC and CCGG formed perfect duplexes with melting temperature around 50 degrees C. This is because the non-staggered duplex is perfectly complementary, and the staggered duplex is non-complementary. The sequences GCCG and CGGC formed staggered duplexes with Tm apx. 30 degrees C, because perfect duplexes are non-complementary, and staggered duplexes are complementary. But the interesting results were that GCGC formed perfect duplexes with Tm = 50 degrees C, whilst CGCG formed staggered complexes with Tm = 37 degrees C.They accounted for these results by the conclusion that GC:GC pairs were more stable than GG:CC pairs which were more stable than CG:CG pairs.

GC:GC > GG:CC > CG:CG

Zielinski and Orgel used this to explain why GC dimers cyclized less readily than CG. They thought that the GC:GC mini-helices that Sinclair had shown were formed more readily than CG:CG mini-helices, would stabilize the GC dimers against cyclization. Zielinski and Orgel argued that elongation took place best with the GC dimers because fewer were lost to cyclization. However, according to Sinclair we would expect that, had fewer of the CG dimers been lost to cyclization, that CGCG tetramers would have formed more stable splint junctions than GCGC tetramers due to the presence of two strong GC:GC pairs in the junction and one weak CG:CG pair, compared to splint junctions formed between GCGC tetramers, which would a) have been less likely to form due to the stability of the perfect GCGC tetramer, and b) more unstable once formed due to the weaker splint junction. (See the diagram below).

The capacity of GCGC to form autocatalytic perfect complexes was supported by the findings of Zielinski and Orgel with 3'-amino-deoxynucleotides. When a GCGC template was added to GC dimers, the GCGC acted as a template to catalyse the condensation of GC and GC, forming further template. GC were present at concentration 0.001M. In a control reaction with no GCGC, GCGC was formed at a slow rate. Adding GCGC at increasing initial concentrations (0.25mM to 1mM) resulted in an increased rate of GCGC production. Increase in 4-mer concentration obayed a square root law dC/dt = a + b(C)^0.5, typical of template replication reactions. This was used to support the argument that staggered association was rare with GCGC templates. No controls were reported with CGCG templates in this paper (Zielinski and Orgel, 1987). Also, from the results of this paper, the formation of small but significant proportion of staggered GCGC duplexes has not been excluded.

The negative results for CGCG templates was published in another paper by Zielinski and Orgel in 1989 (Zielinski and Orgel, 1989) in which the efficient template effect of GCGC for the condensation of GC dimers was not demonstrated by CGCG for CG dimers. Adding increasing concentrations of CGCG template did not increase the rate of CGCG template production compared to controls. They write, "an estimate of the thermodynamic stability of helical oligoribonucleotide complexes can be made by summing the energies contributed by the helix by stacking of consecutive base pairs (Bubienko et al 1983). Recent studies have shown that GC stack contributes more to the stability of the helix than the CG stack (Sinclair et al 1984)." The explanation for the poor templating effect of CGCG compared to GCGC was that the complex formed by two dimers of CG on CGCG was unstable compared to the CGCG minihelix because a strong GC stacking energy was lost, whereas the complex formed by GC dimers on GCGC was not so destabilized relative to a GCGC minihelix since only a weak CG staking energy was lost (see diagram below). Another explanation they gave (again inspired by Sinclair) was that CGCG templates formed splint junction aggregates with themselves, and did not leave much template for CG dimers to condense onto. This is possible even when the ends of CGCG are protected.

The actual mechanism underlying Zielinski and Orgel's experiemant with the phosphoamidates was left unclear because it is not certain why CG dimers did not elongate, and it is one of the aims of this DPhil thesis to uncover what mechainsms could indeed explain their findings. Later work has shown that in a slightly different chemical system, CG dimers are actually able to make much longer splint junction attached aggregates than GC dimers, as indeed predicted and observed by Sinclair's NMR studies. This was possible because for some reason CG was not used up in the cyclization reaction as rapidly as in Zielinski and Orgel experiments.

 

Recent Experiments Demonstrate that CG Dinucleotide Building Blocks Can Elongate but Not Replicate at Low Temperatures.

 

Recent work by Thoennessen published in his PhD thesis (2003), extended the work of Zelinski and Orgal (Z&O) by reducing the extent of the cyclization reactions by using a different kind of nucleic acid (a deoxynucleotide) to that used by Z&O (a ribonucleotide). Cyclization was much less than with Zelinski's experiment, only 5% after 22h at 0.86mM [dimer], and was abolished completely at 50mM. However, Cyclization dominates at 30 degrees C and low concentrations, since it is a unimolecular reaction compared to non-templated ligation which is a bimolecular reaction and templated ligation which is a trimolecular reaction.

The figures below show that when a reactor is initialized with only dimers at 30 degrees C, that CG dimers and 4-mers are used up most rapidly. The figure on the left shows the dimer concentration in mM, and the figure on the right shows the 4-mer concentration in mM. Blue triangles represent the CG dimers and 4-mers. It is remarkable that even at 30 degrees when duplex formation of dimers and 4-mers does not occur to an appreciable extent, that CG dimers and 4-mers are used up so rapidly, apparently contributing to even longer strands. Some cyclization occurs but cyclization is equal for all dimer types, and so cannot account for the rapid decrease in CG compared to all other dimers. The experiments are carried out at 20mM, a high concentration of dimer and so cyclization is not a signficiant factor.

One hypothesis to explain these curves is that CG dimers undergo non-templated ligations more rapidly than other types of dimer, so forming long templates more rapidly, which can then serve to capture dimers and 4-mers rapidly, and form long strands. Another hypothesis is that the rate of non-templated ligations is the same for all dimer types but that more staggered duplxes of CG are formed than for other dimer types because of the arguments given by Sinclair et al 1984 (see above), and that these subsequently form stronger splint junctions than other dimer types.

Figure 1. From O. Thoennessen, Thesis, Bochum 2003. Examples of concentration-time-profiles of the 4 dimer-systems.   a) Dimer consumption b) Tetramer formation, all reactions at 30 °C

Whatever the mechanism, at low temperatures (2 degrees C) and concentrations of 20mM initial dimer, Thoennessen observed that there was elongation of CG dimers to greater than 100-mers, compared to Orgels mere 14-mers. It is not possible to determine chemically exactly how long the polymers are. Figure 2 below shows a stack plot of polymer lengths after various times up to 3 days, after which polymers are longer than can be detected by this analysis method.

Figure 2. From O. Thoennessen, Thesis, Bochum 2003 : HPLC-Stackplot of a CG-Oligomerization after 241min and after 3 d. 20 mM n CG p , 2°C, 0.4 M EDC. This plot is produced by IE-chromatography.

The proposed mechanism to account for the high reaction rates, was the "formation of large multimolecular hybridized aggregates of oligomers with sticky, double-dangling CG-ends". Such complexes have been suggested above (Chen et al 1985), (Sinclair et al 1984) (Rokita et al 1989). The proposed mechanism of the staggered olgiomer formation was the "shifting or sliding process of oligomers in normal duplexes (in a bulb migration) to generate shifted duplexes with double dangling ends on both sides. The resulting double-helical structure with interrupted backbone can support the ligation of reactive neighbored groups and thus leads to an increased reaction rate." (Von Kiedrowski et al, in prep). The slidomer mechanism is shown in the figure below.

Figure 3. From O. Thoennessen, Thesis, Bochum 2003. "Example of a short slidomer, consisting of multiple slidomer-duplexes with overlapping double-dangling CG-ends." (Von Kiedrowski et al, in prep). Sliding by bulge migration is the proposed mechanism for the production of staggered duplexes.

Evidence for the Slidomer Mechanism.

1. The original argument for splint junction based aggegation of staggered ribonucleotide duplexes as a mechanism for elongation was put forward by Zielinski and Orgel (1989). They found that the rate of CGCG production from CG dimers did not increase as initial CGCG template concentration was increased, suggesting that autocatalysis was not occuring. Z&O adopted Sinclair's explanation based on experiments showing that CGCG tended to form aggregates whereas GCGC tended to form perfect duplexes, due to differences in the relative strengths of dinucleotide cores in the perfect and staggered states (see above).

2. Similar experiments were carried out in the deoxynucleotide system, and it was found that similar results were obtained as Z&O above.

3. Calculations using Xia's NN model confirmed that the staggered association mechanism was correct.

Relavence

The relavence of the above mechanism is that it suggests that early nucleic acid polymers that were rich in CG may have been able to elongate by the slidomer mechanism so allowing the potential for their replication and selection. However, there are some unfortunate complexities to this mechanism as a means of information transmission.

The Second Hurdle: Replication of Long Nucleic Acids.

 

Recognized Obsticles to Self-Replication.

In general, the template-directed ligation/cleavage step is rate limiting for template replicating systems, allowing only short sequences to replicate. These short sequenes are thought not to have much catalytic activity and so cannot feed back on the system sufficiently to allow evolution to act to select for sequences that benefit more effective replication.

 

Replication of a 6-mer Has Been Achieved.

 

Non-enzymatic, short, non-hereditary, and transient, nucleic acid replication was demonstrated by Guenter Von Kiedrowski in 1986 (Von Kiedrowski, 1986). He used a deoxyribonuclotide oligomer of sequence *CCGCGG* with protected 5' and 3' ends, shown with an asterix (so that it was incapable of undergoing elongation side-reactions). The building blocks were hemi-protected *CCG and CGG* trimers, so that the system was capable of ligation only between the unprotected ends of *CCG and CGG*, to produce the original template (since it is palindromic).

Sigmoidal template concentration increase, although expected of an autocatalytic reaction, was not observed. Only 12% conversion of trimer to 6-mer was observed after 4 days. It is likely this was due to hydrolysis of an activating agent (CDI) necessary for polycondensation to occur, and also due to the production of a side-product (CCGGCC). The common obsticle of product inhibition also contributed to poor yield of template, dissociation of the newly formed 6-mer from the template being rate-limiting and resulting in parabolic growth dynamics (Von Kiedrowski, 1993). Because of this problem, evidence for autocatalysis had to be obtained from the initial rates of production of 6-mer. Von Kiedrowski showed that adding template increased the initial rate of reaction, so supporting the autocatalytic hypothesis. (See diagram below).

Robertson et al (2000) have suggested that rather than considering the system autocatalytic, "a better interpretation would be that the sequence is replicated, but the system displays limited autocatalysis".

Later work by Von Kiedrowski et al investigated the sequence dependence of non-enzymatic template replication with different trimers, and used a faster ligating system consisting of phosphoamidate linkages catalysed by EDC (Von Kiedrowski et al 1989). This chemical system is the same as that described above in the elongation section. A sigmoidal concentration increase was demonstrated, but there was still product inhibition. Sigmoidal concentration increase was also demonstrated by Von Kiedrowski 1991 with 3'-5' phosphoamidates, and lead to the development of selection theories based on parabolic replicators.

One solution to the problem of product inhibition was developed by Luther, Brandsch and Von Kiedrowski (1998). Template was immobilised on a surface. Periodic reduction in 'feedstock' and removal of newly synthesised product from the templates could in theory result in continued synthesis of template.

Theory of Minimal Replicators and Parabolic Growth.

 

So far it has been possible to obtain replication of short strands (6-mers). What are the consequences for selection acting on parabolically replicating strands? How would we expect evolution to act on a population of non-enzymatic template replicators? One of the aims of this thesis is to answer this question within the context of the chemoton model, and using a realistic model of template dynamics. In particular what processes would select for long templates, and template diversity?

Szathmary and Gladkih (1998) showed that different oligonucleotide sequences should coexist under parabolic growth, since any varient can invade from rarity if templates grow subexponentially. The case of competition in a population of replicators with different subexponential orders of replication p, was not examined [Check this is true. It seems that such is the case when templates of different lengths are replicating together in a flow reactor. The short templates outcompete the longer ones. What is more, short templates are actually produced by the longer templates].

 

 

Other Approaches to Self-Replication of Nucleic Acids.

 

Temperature and Salt Cycling as a Means of Relieving Product Inhibition.

Several papers have speculated on the possibility of temperature cycling as a means of relieving product inhibition. An interesting approach by Braun and Libchaber (2004) is to search for a non-equilibrium mechanism that could provide a source of activated nucleotides. They show that long DNA strands can be concentrated in preferecne to short strands, by thermophoresis on a surface. They consider thermal convection in rock pores near hydrothermal venting systems as a mechanism for the accumulation and replication of nucleic acids. In an experiment using PCR, they show that thermal convection provides a means of rapid temperature cycling ( see figure below).

However, they have not demonstrated that temperature cycling works in non-enzymatic systems, which would require much lower temperatures, and longer periods.

Richard Lathe (2003) recently proposed that tidal cycling causing dilution and reduction in salt concentration could have acted to relieve product inhibition. When the tide comes in, salt concentration is decreased, and inter-phosphate repulsion on opposite strands promotes denaturing. When the tide goes out, salinity increases as does the concentration of monomers promiting templated ligations. There is no experimental evidence for such a mechanism. One of the aims of this thesis is to consider such cycling mechanisms and see if they can work.

 

Other Models of Polymer Self-Replication.

A large number of theoretical models of polymer self-replication exist. Here we examine what these models have achieved.

Monteiro and Piqueira (1998, 1999) have examined the distribution of polymer lengths at steady state in a model of RNA polymer self-replication, as a function of monomer concentration, and rate constants. They want to know what circumstances favour long polymers, and specifically whether monomer concentration oscillation makes a difference. Their model is described below.

0. Monomers are of one type only.

1. Oligomers are formed spontaneously by non-instructed association of monomers.

2. Polymers can be extended by ligation of monomers at their ends.

3. Polymers above a threshold length catalyse the formation of other polymers.

4. Polymers decompose into fragments.

The equations of the model are shown below.

m is the monomer concentration, and pi is the concentration of polymer of length i. Non-instructed formation is described by the term with coefficient Ai. It is assumed only dimers are formed by non-instructed monomer association, so A1 = A = constant, and alpha1 = 2. Templated polymer formation is descrived by the term with the B coefficient. Bii = constant above a threshold of length Nthreshold. The capture of monomers at chain ends is described by the term with the C coefficient. The monomer decomposition rate is D. Polymer decomposition rate is described by the term with the E coefficient. The flux function is the first term in the monomer equation. A effects of both a constant, and a rectangular flux function are studied. The polymer degredation function is the final term in the polymer equation and is set to zero. The average length of the polymer population is the dependent variable studied. A maximum and minimum polymer length is imposed by altering the lambda term, Nmax = 7, Nmin = 2, and Nthreshold = 5. It is assumed that (B = 100) > (C = A = 10) > (D = E = 1). Monomer flux is set to 10.

The results obtained showed that average polymer length, L, behaved as follows.

1. If A was increased, L decreased, because monomers were lost to dimer formation, rather than being available for strand elongation.

2. If C was increased, L increased, because monomers were used at a greater rate for strand elongation.

3. If D is increased, there are decreased concentrations of all species, and a slight reduction in L.

4. If E is increased, since polymer degredation is proportional to length, L decreases.

5. If B is increased, L increases. If B =0, polymer concentrations decrease with increasing polymer length.

6. As monomer influx increases, L increases.

The time-evolution of polymer concentrations are shown below. Note that longer polymers have much lower concentrations than shorter polymers.

Experiments with rectangular monomer flux resulted in lower average polymer length, whatever the flux period. It will be interesting to compare these results with ours and to examine how and why they differ.

The Evolution of Ribozymes.

Yet again, the motief of natural selection acting to select systems with appropriate feedback is observed. Non-enzymaticly replicating sequences once folded, posses catalytic ability which feeds back and alters the probability of the replication of sequences. A similar effect may have occured when templates produced by metabolic systems fead back on the autocatalytic growth of the metabolic systems responsible for their production. Possibly a similar effect was observed in the origin of nucleic acid templates from mineral template replicators, with mutual aid bring provided by both replicators for each other. Processes of scaffolding, takeover, symbiosise, and cooperation are related. One of the aims of this thesis is to provide a model to investigate whether there is an underlying mechanism for these transitions.

Several groups have been working to try to produce an RNA replicase ribozyme, since this would have been the most important function for a ribozyme in the RNA world, devoid of proteins. At present, a ribozyme has not been produced that can copy itself.

 

 

Bibliography

GÖZEN ERTEM. MONTMORILLONITE, OLIGONUCLEOTIDES, RNA AND ORIGIN OF LIFE. Download pdf.

C. B. Chen, T. Inoue, L. E. Orgel, J. Mol. Biol. 1985 , 181 , 271.

D. Brown and L. Gold. RNA Replication by Q Beta Replicase: A Working Model. (1996) PNAS, 93, 11558-11562 Download pdf.

Bommarito, S. Peyret, N. Santalucia, J. Thermodynamic parameters for DNA sequences with dangling ends. (2000) Nucleic Acids Research , 28, 1929-1934. Download pdf.

Inoue, T. and Orgel, L.E. (1981) J.Am.Chem.Socm 103, 7666-7667.

Inoue, T. and Orgel, L.E. (1982) J.Mol.Biol, 162, 201-217.

Inoue, T. and Orgel, L.E. A Non-Enzyamtic RNA Polymerase Model. (1983) Science, 219, 859-862.

Kool, E.T. Hydrogen Bonding, Base Stacking, and Steric Effects in DNA Replication. (2001). Annu. Rev. Biophys. Biomol. Struct. 30, 1-22. Download pdf.

Kozlov, I.A. and Orgel, L.E. Nonenzymatic Template-directed Synthesis of RNA from Monomers. (2000), Molecular Biology, 34(6), 781-

Lehman, N. A cased for the Extreme Antiquety of Recombination (2003) J. Mol Evol. 56, 770-777. Download pdf

Luther, A. Brandsch, R. Von Kiedrowski, G. Nature, 1998, 396, 245.

Monteiro L.H.A. & Piqueira J.R.C. A Model for the Early Evolution of Self-replicating Polymers. J. Theor. Biol. 191, 237-248 (1998). Download pdf.

Monteiro L.H.A. & Piqueira J.R.C. Modeling Homopolymer Self-Replication: Implications for Early Competition. Modeling Homopolymer Self-replication: Implications for Early Competition. (1999) J. Theor. Biol, 196, 51-60. Download pdf.

Orgel, L.E. Lohrmann, R. (1974) Accounts of Chemical Research, 7, 368-377.

Orgel, L.E. Some consequences of the RNA World Hypothesis. (2003) Origins of Life and Evolution of the Biosphere. 33, 211-218.

Robertson, A. Sinclair, A.J. Philp, D. (1999) Minimal Self-Replicating Systems. Chem. Soc. Rev, 29, 141-152.

Rokita, S.E. , L. Romero-Fredes, Biochemistry 1989 , 28 , 9674.

Santos, M. Zintzaras, E. and Szathmary E. Origins of Sex Revisted (2003) Orig Life Evol Biosph, 33, 405-. Download pdf.

Sinclair, A. and D. Alkema, R. A. Bell, J. M. Coddington, D. W. Hughes, T. Neilson, P. J. Romaniuk, Biochemistry 1984 , 23 , 2656. Download pdf.

Von Kiedrowski, G. Angew. Chem Int. Ed. Engl. (1986), 25, 932. [1st biomimetic self-replicating 6-mer]

Von Kiedrowski, G. Wlotzka, B. Helbing, J. Sequence Dependence of Template-Directed Synthesis of Hexadenoxynucleotide Derivates with 3'-5' Pyrophosphate Linkage. (1989) Angew. Chem. Int. Ed. Engl. 28, 9, 1235-1237.

Von Kiedrowski, G. Wlotzka, B. Helbind, J. Angew. Chem. Int. Ed. Engl (1991), 28, 1235. [Sequence dependent 6-mer replication].

Von Kiedrowski, G. Wlotzka, B. Helbing, J. Matzen, M. Jordan, S. Parabolic Growth of Self-Replicating Hexadeoxynucleotide Bearing a 3'-5' Phosphamidate Linkage.

Von Kiedrowski, G. Bioorg. Chem. Front. (1993), 3, 113- [Parabolic Growth Dynamics]

W. Zielinski, L. Orgel, Oligomerization of activated derivates of 3'-amino-3'-deoxyguanosine on Poly(C) and poly(dC) templates. (1985) Nucl. Acids Res. 13(7), 2469-

W. Zielinski, L. Orgel, Oligoaminonucleoside phosphoramidates. Oligomerization of dimers of 3'-amino-3;-deoxynucleotides (GC and CG) in aqueous solution. (1987) Nucl. Acids Res. 15(4), 1699-

W. S. Zielinski, L. E. Orgel, The Template Properties of Triphosphraamidates Having CG Residues. J. Mol. Evol. 1989 , 29 , 281.

W. S. Zielinski, L.E. Orgel, Nature (1987), 327 , 346.

E. Zintzaras, Mauro Santos and Eors Szathmary. Living unver the chalenge of information decay: The Stochastic Corrector Model vs. Hypercycles. (2002) Journal of Theorietical Biology, 217, 167-181. Download pdf.

 

[Nearest-Neighbour Methods].

A. E. Walter, D. H. Turner, J. Kim, M. H. Lyttle, P. Müller, D. H. Mathews, M. Zucker, Proc. Natl. Acad. Sci. USA 1994 , 91 , 9218;

T. Xia, J. SantaLucia Jr., M. E. Burkard, R. Kierzek, S. J. Schröder, X. Jiao, C. Cox, D. H. Turner, Biochemistry 1998 , 37 , 14719;

J. SantaLucia Jr., Proc. Natl. Acad. Sci. USA 1998 , 95 , 1460.

[Early genes mainly CG].

M. Eigen, W. Gardiner, P. Schuster, R. Winkler-Oswatitsch, Scientific Am. 1981 , 244 , 78.

K. Ikehara, Viva Origino 2001 , 29 , 66.