PDB Accession Code

This structure will be deposited at the RCSB.

Figure 4.10: Comparison of the electrostatic potential mapped on the molecular surface calculated with APBS (Adaptive Poisson-Boltzmann Solver^[160]) (red codes for negative, anionic and blue codes for positive, cationic potential) from A: wild-type 3KTQ at I614 B: at I614 from KlenTaqM747K and C: at K614 fromKlenTaqM747K, I614K.

Figure 4.11: Overlay at the position 614 of theKlenTaqwild-type in green with theKlenTaqM747K in blue and with theKlenTaqM747K, I614K in gray.

5.1 Abstract

Abasic sites or apurinic sites (AP) are DNA damages caused by spontaneous hydrolysis of the glycosidic bonds between the sugar and the nucleobase.^[15,173]Such errors occur 10 000 times per day in human cells (see Chapter 1.1.2). Abasic sites have lost the information stored in the nucleobase residues and are considered being non-instructive.^[173] In vitro and in vivo studies have shown that adenine (A) rather than guanine (G) is most frequently incorporated opposite of the lesion. Similarly, many polymerases also preferentially add adenine to the 3’- end of the primer strand when they have reached the 5’- end of a template strand.^[174] This behavior is known as the A-rule.[13,175–177]Until now it was not understood, why adenine is incorporated more efficiently than guanine and why the pyrimidines are not incorporated opposite of such a damage. Explanations for the A-rule are the ability of superior base stacking of the purine to the adjacentπ-system of the primer base and the better solvation.^[174]The relative stacking ability as well as the preference for insertion of the four natural bases opposite abasic site is A>G>C, T.^[174] To understand this mechanisms, theKlenTaqM747K, I614K was crystallized in ternary complex with DNA containing an AP site with a trapped ddATP opposite of the DNA lesion.

Well diffracting crystals were grown and the structure could be solved to 2.3 ˚A. Interestingly, the structure revealed a tyrosine mimicking the absent base opposite to the ddATP. Furthermore, to see how ddGTP is incorporated, crystals grown with ddATP were soaked with ddGTP and the structure could be solved to 2.0 ˚A with ddGTP bound opposite of the abasic site. From the crystal structure of thisKlenTaq M747K, I614K bound to an abasic site damaged template, a mechanism of translesion synthesis could be found which was not reported previously.

5.2 Introduction

Pre-steady state kinetic showed thatKlenTaqM747K, I614K is most efficient in incorporating nucleotides opposite of the abasic site and preferentially uses dATP rather than dGTP (Tab. 5.1).

This coincides with the known preferences for nucleotide insertion opposite of an abasic site (preferences: A>G>C,T^[174]). KlenTaqM747K, I614K incorporates adenine 10 times more frequently than guanine. In Table 5.1 are the kinetic measurements of the wild-type, theKlenTaq M747K, theKlenTaqI614K, theKlenTaqM747K, I614K and theKlenTaqY671W. The relative efficiency (kpol/KD) of dNTP incorporation efficiencies of these mutants were compared to those of the wild-typeKlenTaq.

The influence of the mutation on lesion bypass activity is mainly caused by the I614K mu-tation, whereas the M747K mutation has only a minor effect on lesion bypass (discussed in the introduction of the previous Chapter 4). The crystal structure should give information about the mechanisms of the translesion bypass synthesis of abasic sites.

Enzyme dNTP kpol KD kpol/KD relative

[s⁻¹x10⁻²] [µM] [(µM·s)⁻¹ efficiency^∗ x10⁻⁴]

wild-type^∗∗ A 2.7±0.2 149±35 1.8 1

wild-type G 0.4±0.09 66±5 0.65 0.4

wild-type T 0.13±0.02 424±125 0.03 0.02

M747K A 3.0±0.1 108±15 2.8 1.6

Table 5.1: Nucleotide incorporation opposite of the abasic site byKlenTaqwild-type and mutants. ^∗Relative efficiency compared to the efficiency (kpol/KD) of incorporation of dATP as determined for the wild-type enzyme. ^∗∗ Values for incorporation of dCTP were not accessible,^#Values for incorporation of dGTP were not accessible (kinetic measurements were done by Gloeckner and for the Y671W from Kranaster. Adapted from Schnuret al.^[173]).

The method used to obtain crystals with damaged DNA was similar to the previously reported method for the KlenTaq M747K, I614K bound to undamaged DNA. In addition to the previ-ously used 11-mer primer and 16-mer template, a shorter version, a 10-mer primer and a 15-mer template was used. Moreover, two different DNA damages were tested, either the abasic site (Fig. 1.5[1]) or the 8oxoG (Fig. 1.5,[3]). The following primer/template combinations were used:

• 16-mer template containing abasic site (X)and trapped with ddATP:

Primer : 5’-GAC CAC GGC GC-3’

Template1 : 3’-CTG GTG CCG CGXTAA A-5’

Template2 : 3’-CTG GTG CCG CGTXAA A-5’

• 15-mer template containing abasic site (X)and trapped with ddATP:

Primer : 5’-ACC ACG GCG C-3’

Template1 : 3’-TGG TGC CGC GXT AAA -5’

Template2 : 3’-TGG TGC CGC GTXAAA -5’

• 16-mer template containing 8oxoG (Y)and trapped with ddCTP:

Primer : 5’-GAC CAC GGC GC-3’

Template1 : 3’-CTG GTG CCG CGYGAA A-5’

Template2 : 3’-CTG GTG CCG CGGYAA A-5’

• 15-mer template containing 8oxoG (Y)and trapped with ddCTP:

Primer : 5’-ACC ACG GCG C-3’

Template1 : 3’-TGG TGC CGC GYG AAA -5’

Template2 : 3’-TGG TGC CGC GGYAAA -5’

The ternary complex was formed in a final concentration of 20 mM MgCl₂by mixing 3.0 mM duplex DNA with 7 mg/ml protein and after a reaction time of 1h at room temperature, the ddNTP was added in a molecular ratio of 1:6.7:50 (protein:DNA:ddNTP).

5.4 Results and discussion

Crystallization experiments with the previously used crystallization conditions and the clas-sic Nextal screens did not yield any crystals with the abaclas-sic template. From the structure of KlenTaqM747K, I614K in complex with undamaged DNA, a crystal contact between the de-oxyguanosine of the primer end and the tyrosine 339 of the next symmetry related complex was known (Figure 5.1). Since no crystals appeared in the setups with the abasic template, it was assumed that the damaged DNA was not permissive for crystal lattice formation due to its size when it got compressed or bent by the polymerase. Therefore, shorter oligomers were used for crystallization trials. However, no crystals appeared with the shorter oligomers either. The same happened upon crystallization with 8-oxoG. When returning back to the initial conditions with the 16-mer template and the 11-mer primer and using a new commercially available Nextal Nucleix screen (sparse matrix screen), a new condition was found which led to protein crystals.

The following primer and template were successful:

Primer : 5’-GAC CAC GGC GC-3’

Template : 3’-CTG GTG CCG CGT XAA A-5’

X= abasic site + ddATP

Figure 5.1: Example of one crystal contact of the 3’ guanosine (dG) of the primer with tyrosine 339 of a symmetry mate. The experimental density (2Fo-Fc) is shown as a blue mesh. This picture was made with Coot using the structure of theKlenTaqM747K, I614K and the undamaged DNA as described in Chapter 4.

This condition (Appendix 9.4, C3: 0.05 M Sodium cacodylate pH 6.5, 0.2 M Ammonium acetate, 0.01 M Magnesium acetate, 30 % PEG 8000) was optimized and the best crystals were obtained using the vapor diffusion hanging drop method against a reservoir containing 0.05 M Sodium cacodylate pH 6.5, 0.2 M Ammonium acetate, 0.01 M Magnesium acetate and 25 %

Figure 5.2: Crystal ofKlenTaqM747K, I614K with DNA containing abasic site and ddATP.

of the ternary complex with the abasic site, were collected at the beamline X06SA of the SLS (Swiss Light Source at the Paul Scherrer Institute, Villigen, Switzerland) at a wavelength ofλ=

1.000 ˚A. Data reduction was done with the XDS package^[142]. The structures were solved by molecular replacement using PHASER^[169] at a resolution of 2.2 ˚A in space group P3₁21, with cell dimensions a,b=107.49 ˚A, c= 89.41 ˚A. As search models 3KTQ^[124]and a modified version of 4KTQ^[124]were used. Refinement was performed with the program PHENIX^[170]and model rebuilding was performed using the program COOT^[154]. Figures were made using PyMOL^[129]

and electrostatic surfaces were made with APBS^[160]. The Ramachandran statistic and outliers were detected using MolProbity^[171,172]. Data statistic is shown in Table 5.2.