Thermal stability of the wildtype and the modified EBS1·IBS1 interactions

interactions

In order to confirm the low stability of the wildtype EBS1·IBS1 interaction as found by NMR, UV melting studies were performed. Melting temperatures Tm

give an estimate of the thermal stability of

oligonucleotides and of the homogeneity of the sample. At the melting temperature Tm 50%

of the oligonucleotide is melted, i.e. is single stranded. To investigate the stability of the wildtype as well as the altered EBS1-IBS1 interactions, three constructs were designed, which include only the EBS1·IBS1 interaction without the stem (Figure 32A-C). The melting temperatures are summarized in Table 4. The UV melting studies of the EBS1·IBS1 interaction (Figure 32A) reveal a melting point Tm of 15.5 °C (100 mM KCl). Higher

concentrations of KCl do not lead to a considerable increase of Tm (Table 4), thus indicating a

poor stability of the EBS1·IBS1 complex in the absence of the covalent exon-intron linkage and/or EBS2·IBS2. These results confirm the observations made by NMR (see Section 2.1). NMR studies are normally carried out between 20 °C and 30 °C, thus the wildtype EBS1·IBS1 interaction with its low stability is not suitable for NMR studies as at usual measurement temperatures IBS1 is not completely bound to EBS1.

The altered EBS1*·IBS1* and EBS1*·IBS1*-GC interactions result in significantly higher melting temperatures. Indeed, an increase in Tm by a factor of more than two to 34.3 °C (100

mM KCl, Table 4) was achieved for EBS1*·IBS1*. Even when reducing the KCl

Figure 32 Secondary structures of the wt-EBS1·IBS1 (A), EBS1*·IBS1*

(B) and EBS1*·IBS1*-GC (C) constructs as used in the UV melting studies. Altered nucleotides are highlighted.

Table 4 Melting temperatures Tm [°C] of EBS1·IBS1, EBS1*·IBS1*, EBS1*·IBS1*-GC and d3'-EBS1*·IBS1* (see also Figure 32 and 30B) at different KCl concentrations [mM].

[KCl]/mM Tm (EBS1·IBS1) Tm (EBS1*·IBS1*) Tm (EBS1*·IBS1*-GC) Tm (d3'-EBS1*·IBS1*)

10 nd 27.5 ± 0.3 nd 27.9 ± 0.5 / 51.0 ± 0.2

100 15.5 ± 1.0 34.3 ± 0.2 49.9 ± 0.1 40.6 ± 0.2 / 63.5 ± 0.2

150 17.0 ± 1.1 nd nd nd

200 17.5 ± 2.0 nd nd nd

concentration to 10 mM, Tm is with

27.5 °C still higher than the one of the wildtype interaction. Such a stabilizing effect by KCl is no surprise as the higher ionic strength reduces the charge repulsion of the two negatively charged RNA strands. A further small increase in stability of the EBS1*·IBS1* interaction is achieved by including the d3'-stem, i.e. having the corresponding full-length d3'- EBS1*·IBS1* interaction as found in the full D135GC ribozyme: Melting temperatures of 28 °C (10 mM KCl) and 40 °C (100 mM KCl) are reached. The melting curve of this RNA dimer shows a second transition at higher temperature, which originates from the dissociation of the hairpin stem (Figure 33, Table 4). Taken together, the modified d3'-EBS1*·IBS1* construct shows a very good thermal stability even in the absence of divalent metal ions and was therefore used in all further experiments.

The construct EBS1*·IBS1*-GC (Figure 32C) has with 49.9 °C (100 mM KCl) an even higher melting temperature than EBS1*·IBS1*. But as already mentioned in Section 2.2.2 the change from wobble to Watson-Crick base pairing results in very poor activity in vitro.

In order to determine the thermodynamic parameters for the EBS1*·IBS1* and EBS1*·IBS1*-GC interaction the melting curves were converted into plots showing the percentage of folded fraction (α) versus temperature as it has been previously described for other RNAs(294) and fit with a two-state transition model using linear sloping of the upper and lower baselines (Figure 34A).(295,296) Because of the low thermal stability of the wildtype EBS1·IBS1 interaction, the linear region of the RNA duplex could not be reached. This lower baseline is needed for the calculation of ∆H0-values. Hence, a van't Hoff analysis was only performed for the EBS1*·IBS1* interaction in the presence of 10 mM and 100 mM KCl and for EBS1*·IBS1*-GC (100 mM KCl). A van't Hoff plot (lnK vs 1/T) was drawn by using the equilibrium constant K as determined for each temperature using equation 1 for bimolecular transitions (Figure 34B):(294)

Figure 33 UV-melting curves for d3'-EBS1*·IBS1* in the presence

of 10 mM KCl (○) and 100 mM KCl (■). The curves were acquired

by measuring the absorbance at 260 nm. The first transition corresponds to the melting of the EBS1*·IBS1* interaction and the second to the one of the d3'-stem (see also Table 4).

Results and Discussion 49

(

)

where css is the concentration of each single

strand. This expression is used for non-self- complementary associations. K values are most precise in the α-range between 0.2 and 0.8,(296,297) and hence only the values within this region were used (Figure 34A). For a concentration-independent monomolecular transition as in the case of the d3'-stem K is determined with equation 2.(295,298)

1 K α α = − equation 2 ∆H0 is assumed to be independent of temperature for a two-state transition.(296) Therefore, the van't Hoff plot is linear with – ∆H0

/R as the slope and ∆S0/R as the axis intercept. All standard errors for ∆H0 and ∆S0 are in the range of 5-8%. As the errors of ∆H0 and ∆S0 compensate, the analysis of optical melting curves provides a precise way of

determining ∆G37 yielding standard errors of ±2-5%.(297) The thermodynamic data for the

EBS1*·IBS1* and EBS1*·IBS1*-GC interaction are summarized in Table 5. ∆H0 for the melting of EBS1*·IBS1* decreases from –313.8 ± 15.7 kJmol–1 to –334.3 ± 16.7 kJmol–1 when raising the KCl concentration from 10 mM to 100 mM. At the same time, ∆G37 only

slightly decreases (–20.3 ± 1.4 kJmol–1 vs –28.8 ± 2.0 kJmol–1) illustrating that the thermodynamic stability of EBS1*·IBS1* is only little stabilized by increasing the

Figure 34 (A) UV-melting curves for EBS1*·IBS1*

in the presence of 10 mM KCl (○) and 100 mM KCl (■). The curves were acquired by measuring the absorbance at 260 nm. (B) Van't Hoff plot (ln K vs

1/T) for EBS1*·IBS1* at 10 mM KCl (○) and 100

mM KCl (■). For the final calculation of the thermodynamic parameters, only the data points between 20 and 80% folded were used, i.e. the points shown in grey were omitted.

Table 5 Thermodynamic parameters for the folding of EBS1*·IBS1* at 10 and 100 mM KCl and of

EBS1*·IBS1*-GC at 100 mM KCl as determined by UV melting experiments. The errors given correspond to one standard deviation.

Construct ([KCl]/mM) ∆H0/kJ·mol–1 ∆G37/kJ·mol–1 ∆S0/J·mol–1 K–1

EBS1*·IBS1* (10) –313.8 ± 15.7 –20.3 ± 1.4 –946.2 ± 66.2

EBS1*·IBS1* (100) –334.3 ± 16.7 –28.8 ± 2.0 –985.2 ± 69.0

concentration of monovalent cations. However, illustrated by the construct EBS1*·IBS1*-GC with a ∆G37 of –43.3 ± 3.0 kJmol–1, the exchange of a G-U wobble pair by a G-C base pair

leads to a more stable interaction. The compiled thermodynamic parameters in Table 6 illustrate that the d3'-stem has no influence on the EBS1*·IBS1* interaction: The ∆H0 and ∆S0

values are the same within the error limits and also ∆G37 is not significantly different for

the melting of isolated EBS1*·IBS1* and the same sequence embedded within d3'- EBS1*·IBS1*.

Taken together, the thermodynamic parameters show that the stability of double helices can be increased rather by replacing A-U and G-U pairs, respectively, with G-C base pairs than by increasing the ionic strength.

Table 6 Thermodynamic parameters for the folding of d3'-EBS1*·IBS1* at 10 mM and 100 mM KCl as

determined by UV melting experiments. The errors given correspond to one standard deviation.

[KCl]/mM ∆H0/kJ·mol–1 ∆G37/kJ·mol–1 ∆S0/J·mol–1 K–1

d3'-stem EBS1*·IBS1* interaction 10 10 –346.3 ± 17.3 –317.1 ± 15.9 –19.8 ± 1.4 –17.7 ± 1.2 –1053.0 ± 73.7 –965.0 ± 67.6 d3'-stem EBS1*·IBS1* interaction 100 100 –447.2 ± 22.4 –336.3 ± 16.7 –41.3 ± 2.9 –35.3 ± 2.5 –1308.7 ± 91.6 –965.1 ± 67.6

Results and Discussion 51