Experiments with the tRNA replicator

The replication scheme was realized experimentally and analyzed in detail. To this aim repli- cator sequences were designed from tRNA sequence and analyzed by fluorescence methods, gel electrophoresis and thermophoresis as elucidated in[46].

Sequence design

tRNA sequences are extracted from the Genomic tRNA database1.[12] In addition to se- quences, this database provides plots of the secondary structure of tRNAs. To test for the transition from tRNA to double hairpins, thermophilic species were chosen. This choice was made not so much to bring the experiments in the context of volcanic activities, but to yield replicator systems with melting temperatures higher than room temperature, which makes experiments easier to handle. The transition is not limited to hydrophilic species, as shown in figure 5.4. The tRNA was chosen to carry amino acids such as alanine or glycine that are considered to be old or arise from a chemistry that is thinkable on early earth.[62] Mutations

Figure 5.4.: Original tRNA sequences are plotted in first lines, modified double hairpins in second lines. Bases marked with ”>” pair with bases further downstream marked with ”<”. Bases marked with ”.” are single stranded. In third lines, ”m” labelles a mutation. Figure taken from supplementary material of [46] (see Appendix) are introduced in regions that are modified in comparative sequence studies.[24] The resulting sequence is tested by the RNA vienna package[37] online tool2 _{or other other software based}

on the Vienna algorithms such as the ”Oligo Analyzer”3_{. The transition is amazingly robust.}

In many cases, parts that are originally separated in the anticodon loop and the amino ac- ceptor stem hybridize with high complementarity, and only few mutations are necessary to destabilize the cloverleaf structure in favor of the double hairpin structure.

1_{URL:http://gtrnadb.ucsc.edu.} 2

http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi

As soon as the double hairpin conformation is stable, melting temperatures and hybridization energies are calculated. The experimental critera for a good model system are:

The melting transition of the codon region and its complement is well separated from the melting transition of the hairpin stem or the melting of duplexes and triplexes (see figure 5.6) Especially for longer successions this is necessary to gain clear signals in experiments. In principal, a partial opening of long successions is sufficient.

The loop size is long enough to store a sufficient amount of binding energy. To some extent this criteria is connected to the former. The number of monomers in the loop is the number of base pairs that a duplex gains compared to hairpin conformation. The energetic advantage increases with the loop length. A calculation of binding energies and entropic penalties is included in the supplement of [46]

The codon length is short enough to be separated, but long enough to allow for binding of a single codon at reachable temperatures and to provide selectivity.

Reaction partners are designed by means of complementarity

For the simplified replication reaction with four hairpins (see Figure 5.2), hairpin A was gained by cutting of tRNA of Methanobacterium Thermoautotrohpicum Ala TGC next to the anticodon (see figure 5.1) Reaction partners were generated by means of complementarity: With respect to hairpin A, hairpin a has a complementary codon, hairpin B has a comple- mentary hairpin loop and a codon that is not binding significantly to hairpin A, hairpin b has a codon complementary to the codon of hairpin B and a loop that does not allow hy- bridization to any already given sequence. As a consequence, hairpin a is then completely determined. The sequences designed for the replication of three codons as sketched in figure 5.3 was designed accordingly.

It is crucial to mention, that in principle a one-to-one complementarity of neighboring loops is not necessary, since RNA helices tolerate mismatches and bubbles to high extent. However, for a lab system, we choose a perfect match to yield interpretable signals and to gain reaction times smaller than the lifetime of a student, which is not necessary for evolutionary processes. Furthermore it is to mention that all reaction partners except hairpin Ado not directly come from an actual tRNA, but have a sequence with similar hybridization energies and secondary structures. It is a matter of former[76] and further investigations with bioinformatic methods to search for combinations of actual tRNAs that can be transformed to reaction partners for replication schemes. This does not influence the arguments about the similarities of tRNA and the replicator.

2-Aminopurine, a sensor for hybridization states

For fluorescence investigations of replication, labeling of hairpins b and r with 5-FAM (car- boxyfluorescin) and hairpin Q with Cy5 at the 5’-end were chosen. Both spectra can easily be distinguished and have no significant FRET-transfer or channel crosstalk (see 5.5). By means of electrophoresis, those labels enable us to analyze the distribution of species in the steady state after a reaction. To investigate the kinetic development, a fluorophore that reports the changes in hybridization state in real time is necessary. Considering the small size of the hairpin sequences in use, conventional FRET pairs or donor-quencher pairs may

Figure 5.5.: List of all sequences and modifications used.

disturb the system significantly. As an alternative, the base 2-Aminopurine (2-AP) is used. In terms of Watson-Crick base pairing, it behaves very much like its normal sister base adenine (6-Aminopurine).[27] The difference is a much longer fluorescence lifetime.[55] Interestingly, the fluorescence, which is in the UV region, is quenched by hybridization about a factor of 2₋3 [27] due to changed base stacking behavior.[41] Therefore, an adenine base in the loop of hairpinAand hairpinQas well as double hairpinU is replaced by 2-AP. A transition from a single stranded hairpin loop towards a double stranded duplex lowers fluorescence due to quenching. We define quenchingQfor all experiments as

Qt= 1−Ft/Ft=0 (5.1)

in whichFtis the time dependent fluorescence andFt=0is the fluorescence at the start of the

reaction.