Automated screening of small molecule libraries for telomeric G-quadruplex

To identify new candidates or lead structures of putative telomeric G-quadruplex binders, a FRET-based screening approach, developed by Dr. Armin Benz, was performed (in detail ²⁴⁸). This screening was shown to be a reliable approach to identify molecules interacting with the telomeric G-quadruplex. Therefore, the screening was carried out using the same settings as described by Benz et al.

(2011) ²⁴⁸. Out of two small molecule libraries (ChemBioNet library (CBN) and the library of Prof. Dr. E. Daltrozzo (D), University of Konstanz) about 3800 structurally diverse compounds were automatically screened for their potential to stabilize G-quadruplex formation. For this screening approach an oligonucleotide composed of the sequence GGG(TTAGGG)3 (HT) and labelled

with fluorescein (FAM) at the 5’ end and tetramethylrhodamine (TAMRA) at the 3’ end was used (Fig. 3.4.1 A). It has been shown that as long as the oligonucleotide is unstructured, emission of FAM (measured at 535 nm) can be measured after excitation at 485 nm wavelength. However in case of G-quadruplex formation, either induced by the presence of a monovalent cation like potassium or induced by a small molecule, FAM and TAMRA come into close vicinity, thereby enabling a fluorescence resonance energy transfer (FRET) between FAM and TAMRA (Fig. 3.4.1 B) ^248,249. Therefore, G-quadruplex formation can be measured by the decrease of FAM emission and the increase of TAMRA emission (measured at 590 nm) after excitation of FAM at 485 nm wavelength.

Fig. 3.4.1: FRET-based screening approach for the identification of putative telomeric G-quadruplex binders. A: Sequence of the HT-DNA oligonucleotide with the chemical structures of the 5´ coupled fluorescein (FAM) and 3´ coupled tetramethylrhodamine (TAMRA). B: Scheme of the underlying mechanism of the screening approach. In the absence of an appropriate ligand, the FAM/TAMRA labeled HT-DNA (FHT-DNA) is unstructured and therefore the two fluorophores FAM and TAMRA are situated away from each other. In this case the excitation of FAM with light of 485 nm wavelength is detectable by its emission at 535 nm wavelength.

However in terms of a ligand-induced G-quadruplex formation, the two fluorophores come into close vicinity, therebyenabling fluorescence resonance energy transfer (FRET) between FAM and TAMRA. Therefore, the formation of a G-quadruplex can be measured by the decrease in

3. Results and Discussion

FAM emission (measured at 535 nm) and the increase in TAMRA emission (measured at 590 nm) after excitation at 485 nm wavelength. (Figure modified from A. Benz ²⁵⁰)

The screening was performed in a 384 well format using a robot system of the Tecan Freedom EVO^® series and fluorescence was measured with a Tecan Infinite F500 plate reader. 0.2 µM of the FAM/TAMRA labelled HT (FHT) was incubated with ~ 95 µM compound in 10 mM Tris-HCl buffer (pH 7.5) for 20 min at RT, followed by detection of FAM and TAMRA emission after excitation at 485 nm wavelength in the fluorescence reader. As controls, FHT was incubated in the presence (positive control) or absence (negative control) of 100 mM KCl.

The ratio of FAM/FRET for the negative control was set to 1 and the ratio of FAM/FRET for the positive control was set to 0. As a criterion for compound-induced G-quadruplex formation the FAM/FRET ratio was defined to be smaller than 0.4.

Via this automated screening 31 putative G-quadruplex binders were identified and further characterized by concentration dependency experiments. To that end, 0.2 µM of the FHT oligonucleotide was incubated together with a serial dilution of the identified compounds ranging from 100 µM to 1.6 µM. Like in the automated screen described above, the compound-induced G-quadruplex formation within FHT was analyzed via FRET measurement. The results of the 10 most promising compounds are depicted in Fig. 3.4.3; within these compounds seven hits (CBN1-7) were out of the ChemBioNet library (CBN) and three hits (D1-D3) out of the Daltrozzo library (D). Their corresponding chemical structures are shown in Fig. 3.4.2.

Fig. 3.4.2: Chemical structures of the putative telomeric G-quadruplex binders CBN1-7 and D1-D3 identified in an automated screening of 3800 compounds of the ChemBioNet (CBN) and Daltrozzo (D) libraries.

All 10 identified compounds seem to show a concentration-dependent ability to induce G-quadruplex formation within the FHT oligonucleotide, as indicated by the changes in the FAM/FRET ratio (Fig. 3.4.3). These data suggest that all of

3. Results and Discussion

the depicted molecules are able to maximally induce G-quadruplex formation at 100 µM and even show an effect at 1.6 µM. The obtained results imply that the molecules found within the ChemBioNet library (CBN1-7) are more effective compared to the molecules found within the Daltrozzo library (D1-3).

Fig. 3.4.3: Concentration-dependent ability of the 10 identified compounds to induce G-quadruplex formation within the FHT-DNA. Out of a screen using about 3800 compounds of the ChemBioNet (CBN) and Daltrozzo (D) libraries, the 10 most promising compounds (CBN1-7 and D1-3) were further characterized with the focus on their ability to induce G-quadruplex formation within the FHT-DNA in the absence of KCl. Different compound concentrations (from 1.6 µM to 100 µM) and a FHT-DNA concentration of 0.2 µM were tested in 10 mM Tris-HCl (pH 7.5). As controls, the FHT-DNA was measured in the absence (Ctrl – KCl) or in the presence (Ctrl + KCl) of 100 mM KCl.

Approximately similar results were obtained for the compounds CBN2, CBN3, CBN4 and CBN5. This outcome is very likely due to their shared chemical substructure (Fig. 3.4.2 and Fig. 3.4.3). CBN6 is the berberine derivative palmatine which already has been described as a G-quadruplex binder ²⁵¹. This finding supports the principal functionality of the used screening approach for the identification of telomeric G-quadruplex-interacting molecules. Interestingly, all detected hits contain a quinoline-like and/or a naphthyridine-like substructure which are characteristic features of already known G-quadruplex binders

239,252,253

. Although CBN1, CBN6 and CBN7 showed the best results (Fig.

3.4.3), only the compounds D1, D2 and D3 were further characterized due to their availability (provided by Prof. Dr. E. Daltrozzo, University of Konstanz).

3.4.2 Characterization of the putative G-quadruplex ligands D1-D3 via CD