Reversible photoswitching of the DNA-binding properties of styrylquinolizinium derivatives through photochromic [2 + 2] cycloaddition and cycloreversion

It was demonstrated that styrylquinolizinium derivatives may be applied as photoswitchable DNA ligands. At lower ligand:DNA ratios (≤1.5), these compounds bind to duplex DNA by intercalation, with binding constants ranging from Kb = 4.1 × 104 M to 2.6 × 105 M (four examples), as shown by photometric and fluorimetric titrations as well as by CD and LD spectroscopic analyses. Upon irradiation at 450 nm, the methoxy-substituted styrylquinolizinium derivatives form the corresponding syn head-to-tail cyclobutanes in a selective [2 + 2] photocycloaddition, as revealed by X-ray diffraction analysis of the reaction products. These photodimers bind to DNA only weakly by outside-edge association, but they release the intercalating monomers upon irradiation at 315 nm in the presence of DNA. As a result, it is possible to switch between these two ligands and likewise between two different binding modes by irradiation with different excitation wavelengths.


General procedure 2 (GP2) for the photodimerization of 2styrylquinolizinium derivatives 3b and 3c
A suspension of the styrylquinolizinium derivative (0.50 mmol) in H2O (150 mL) was irradiated at ca. 450 nm for 6 h with thorough stirring. The product was extracted with MeNO2 (3 × 70 mL), and the organic layers were combined and dried with Na2SO4. After filtration, the filtrate was concentrated in vacuum. The product was precipitated from MeNO2 using Et2O and recrystallized from MeOH or H2O.

Absorption and Emission Spectroscopy Preparation of solutions for spectrometric analyses
The ct DNA was dissolved in BPE buffer and stored at 4 °C for at least 24 h and filtered. The concentration (in base pairs, bp) was determined photometrically (λmax = 260 nm, ε = 12824 cm −1 ⋅M −1 ). BPE buffer was prepared from biochemistry-grade chemicals (Fluka BioChemika Ultra) and epure water: c = 6.0 mM Na2HPO4, c = 2.0 mM NaH2PO4, and c = 1.0 mM Na2EDTA; c = 16 mM total Na + . Prior to use, the buffer solutions were filtered through a PVDF membrane filter (pore size 0.45 μm).
Stock solutions were prepared by dissolving 3a-d in acetonitrile or water (4b) to give a concentration of 1.0 mM. The solutions were stored at 4 °C. All measurements were performed in thermostated quartz cuvettes with a path length of d = 1 cm at 20 °C, if not stated otherwise.
The absorption and emission spectra were recorded with a scan rate of 120 nm/min. The detection wavelength range of the absorption spectra was 200-600 nm with slit widths of 2 nm, and the recorded range of the emission spectra was 400-800 nm. Every experiment was performed at least two times. The fluorescence spectra were recorded with a detection voltage of U = 700 V and an excitation and emission slit of 5 nm, if not stated otherwise. The excitation wavelength for 3a was at the isosbestic point at λ = 370 nm, the excitation wavelength of 3b was λ = 410 nm and 3c λ = 404 nm. The binding constants were determined according to published procedures [3] from the spectrophotometric titrations ( Figure S1) and fitting of the experimental binding to a theoretical S5 model considering noncompetitive binding [3], respectively (Figure S1A-C). Standard deviations (SD) of Kb values were calculated from Equation 1.

SD (Kb) = ((SD of A/A)/A)/clig
(1) SD = standard deviation A = value of the absorption SD of A is used as provided by the Origin 7.5 software.

Circular Dichroism (CD) and Flow Linear Dichroism (LD) Spectroscopy
The analyte solutions were prepared by pipetting different aliquots of the stock solution of the ligands into Eppendorf vials and then the solvent was removed. Buffer solution and a solution of ct DNA in BPE buffer were added to adjust the concentration and the ligand-DNA values. The S6 spectra were recorded at λ = 230−600 nm with a band width of 1 nm and a scan rate of 1 nm⋅s −1 with a time per point of 0.5 s. The samples were recorded three times and averaged. The spectra were smoothed with the Savitzky-Golay method and implemented in the Chirascan software, with a polynomial order of 5.

Photometric Monitoring of Photoreactions
Aliquots of the stock solutions of compounds 3a-d were pipetted in Eppendorf vials, the solvent was evaporated, and the residue was dissolved in MeCN or H2O to obtain a final concentration of c = 20-25 µM. The samples were irradiated at ca. 530 nm (3a) and ca. 450 nm (3b, 3c) with an LED lamp or a high-pressure Hg lamp with cut-off filter (>395 nm, 3d). The reactions were monitored photometrically with a scan rate of 120 nm/min in a range of 200-600 nm ( Figure S2).

Photodimer 4b
Data were measured using  and  scans of 0.5° per frame for 90 s with MoK radiation (microfocus sealed X-ray tube, 50 kV, 0.99 mA). The total number of runs and images was based on the strategy calculation from the program APEX3. The maximum resolution that was achieved was  = 22.469° (0.93 Å).
A multiscan absorption correction was performed using SADABS-2016/2 (Bruker, 2016/2) and used for absorption correction. wR2(int) was 0.0879 before and 0.0599 after correction. The ratio of minimum to maximum transmission was 0.7849. The /2 correction factor is not present. The absorption coefficient  of this material was 0.114 mm −1 at this wavelength ( = 0.71073 Å), and the minimum and maximum transmissions were 0.773 and 0.985.
The structure was solved in space group P21/n with the XT [5] structure solution program using the intrinsic phasing solution method and by using Olex2 [5] as the graphical interface.
The material crystallized with significant disorder. The orientation of one quinolizinium unit was found in two different orientations. Each fragment was refined using similarity restraints on bond lengths and angles. In addition, one BF4 anion was modelled in three orientations, with restraints on all B-F bond lengths. Finally, the material crystallized with disordered solvent in the lattice. This solvent could not be modeled, therefore, the PLATON/SQUEEZE program [6] was employed to generate a 'solvent-free' data set. All nonhydrogen atoms were refined anisotropically. Hydrogen atom positions were calculated geometrically and refined using the riding model.

Photodimer 4c
Single colorless irregular shaped crystals of 4c were recrystallized from water by slow evaporation. The structure was solved with the XT structure solution program using the intrinsic phasing solution method and by using Olex2 [5] as the graphical interface. The model was refined with version 2018/3 of XL using least-squares minimization.
Data were measured using  and  scans of 0.5° per frame for 30 s using MoK radiation (microfocus sealed X-ray tube, 50 kV, 0.99 mA). The total number of runs and images was based on the strategy calculation from the program APEX3. The maximum resolution that was achieved was  = 26.468° (0.80 Å).
The diffraction pattern was indexed and the unit cell was refined using SAINT (Bruker, V8.38A, after 2013) on 7034 reflections, 51% of the observed reflections. Data reduction, scaling and absorption corrections were performed using SAINT (Bruker, V8.38A, after 2013). The final completeness was 100% out to 26.468° in .
A multiscan absorption correction was performed using SADABS-2016/2 (Bruker, 2016/2) and used for absorption correction. wR2(int) was 0.0913 before and 0.0629 after correction. The ratio of minimum to maximum transmission was 0.8424. The /2 correction factor is not present. The absorption coefficient  of this material was 0.119 mm− 1 at this wavelength ( = 0.711 Å), and the minimum and maximum transmissions were 0.628 and 0.745.

S9
The structure was solved and the space group P-1 (# 2) determined by the XT [5] structure solution program using intrinsic phasing and refined by least squares using version 2018/3 of XL [5]. The material crystallized with a small fraction (ca. 7%) of near whole molecule disorder, with the two fragments related by a 180° rotation and oriented such that the methoxyphenyl and quinolizinium moieties were nearly overlapping. Modeling of the minor fraction required the use of restraints and constraints to maintain reasonable geometries. All nonhydrogen atoms were refined anisotropically. Hydrogen atom positions were calculated geometrically and refined using the riding model. The value of Z' was 0.5. This means that only half of the formula unit was present in the asymmetric unit, with the other half consisting of symmetry-equivalent atoms.