Dissection of DNA damage and repair pathways in live cells by femtosecond laser microirradiation and free-electron modeling

Abstract

Understanding and predicting the outcome of the interaction of light with DNA has a significant impact on the study of DNA repair and radiotherapy. We report on a combination of femtosecond pulsed laser microirradiation at different wavelengths, quantitative imaging, and numerical modeling that yields a comprehensive picture of photon-mediated and free-electron-mediated DNA damage pathways in live cells. Laser irradiation was performed under highly standardized conditions at four wavelengths between 515 nm and 1,030 nm, enabling to study two-photon photochemical and free-electron-mediated DNA damage in situ. We quantitatively assessed cyclobutane pyrimidine dimer (CPD) and γH2AX-specific immunofluorescence signals to calibrate the damage threshold dose at these wavelengths and performed a comparative analysis of the recruitment of DNA repair factors xeroderma pigmentosum complementation group C (XPC) and Nijmegen breakage syndrome 1 (Nbs1). Our results show that two-photon-induced photochemical CPD generation dominates at 515 nm, while electron-mediated damage dominates at wavelengths ≥620 nm. The recruitment analysis revealed a cross talk between nucleotide excision and homologous recombination DNA repair pathways at 515 nm. Numerical simulations predicted electron densities and electron energy spectra, which govern the yield functions of a variety of direct electron-mediated DNA damage pathways and of indirect damage by ^•OH radicals resulting from laser and electron interactions with water. Combining these data with information on free electron-DNA interactions gained in artificial systems, we provide a conceptual framework for the interpretation of the wavelength dependence of laser-induced DNA damage that may guide the selection of irradiation parameters in studies and applications that require the selective induction of DNA lesions.

Keywords: DNA strand breaks; nonlinear photodamage; oxidative DNA damage; reductive DNA damage; wavelength selectivity.

Fig. 1.

DNA microirradiation of live cells at different wavelengths. (A) Scheme of the multiwavelength fs laser system delivering pulses of 80 fs duration at 515 (green), 620 (yellow), 775 (red), and 1,030 nm wavelength (brown) into a dual-scanner confocal microscope. Pulse trains at 1,550 nm are depicted by black arrows. HNF: highly nonlinear fiber, SFG: sum-frequency generation, SHG: second-harmonic generation, DL: optical delay line, Si and SF10: silicon and SF10 glass prism compressors. Fluorescence imaging is performed via continuous-wave (CW) excitation. (B and C) Exemplary images of CPD- and γH2AX-specific immunofluorescence of HeLa cell nuclei irradiated with fs pulses at 515 nm and 1,030 nm.

Fig. 2.

Dose-response curves, I_IF(I₀), for the immunofluorescence signal intensity reflecting fs laser-induced DNA damage (CPDs and γH2AX) at different wavelengths, (A) in linear representation and (B), as log-log plots. Data points are mean ± SD of ≥48 cells from at least three independent experiments. The shaded regions indicate irradiance ranges producing pan-nuclear γH2AX staining. The I_ref irradiance values were taken as a reference value for the recruitment studies and the numerical simulations of plasma formation shown below. The dashed lines in (B) represent power-law fits to the I_IF(I₀) graphs where the linear slopes (red and blue numbers) in this double-logarithmic representation indicate the exponents, respectively.

Fig. 3.

Live recruitment of Nbs1 unveils different mechanisms of DNA strand break generation at 515 nm and 1,030 nm. Hela cells expressing the indicated DNA repair proteins fused to GFP were microirradiated at the indicated wavelengths and the respective irradiance values I_ref (marked in Fig. 2). The graphs display the time evolution of the relative increase of fluorescence signal at the irradiated sites. (A) Recruitment of XPC-GFP nm indicates the presence of photoproducts at 515 nm only. (B) Recruitment of Nbs1-GFP signals the presence of DNA strand breaks. The different accumulation kinetics at 1,030 nm and 515 nm point to different pathways of strand break generation at the two wavelengths. (C), (D) Recruitment of Nbs1-GFP is affected by the antioxidant Trolox only at 1,030 nm, indicating direct DNA damage with an involvement of oxidative species at this wavelength. The blue shaded areas mark image frames taken before and during microirradiation. The dip in fluorescence signal is due to the local initial bleaching of the fluorophores by the fs laser pulses and does not carry information about the recruitment dynamics. Data are mean ± SD of ≥48 cells from at least three independent experiments.

Fig. 4.

Inhibition of NER suppresses Nbs1 recruitment to fs laser–induced damage at 515 nm in nonreplicating cells. Hela cells were microirradiated at 515 nm and I_ref in the presence of solvent or cycloheximide (10 μg/mL), spironolactone (8 μM), or a combination of the two. The red curve shows the results for G phases only. S-phase cells were excluded based on PCNA-specific immunofluorescence. Data are mean ± SD of ≥48 cells from at least three independent experiments. Blue shaded area: see legend to Fig. 3.

Fig. 5.

Wavelength dependence of electron density, energy, and CBE spectra in the irradiance range between I_th and I_pan. (A) Irradiance dependence of total electron density, n_e,total, and average electron energy, ε_avg. The shaded area indicates the irradiance range for pan-nuclear staining. The dashed line indicates the electron density corresponding to one free electron per focal volume. From 515 nm to 620 nm, n_e,total jumps up by two orders of magnitude, but for λ ≥ 620 nm, it decreases with increasing wavelength, while ε_avg strongly increases. (B) Spectra n_e (ε) of the kinetic energy of CBEs at the end of the fs pulse calculated for I_th, I_ref, and I_pan. To enable proper visualization of all spectra in one graph, the n_e,total values at I_th were multiplied by a factor of 100 or 10 as indicated. The energy spectra exhibit pronounced changes with increasing wavelength but are similar within the irradiance range investigated at each individual wavelength.

Fig. 6.

Schematics of fs laser generation of DNA-damaging reagents in water, the wavelength dependence of average free-electron kinetic energy, and the corresponding DNA damage pathways. (A) Water is a transparent dielectric with a conduction band (CB) located ≈9.5 eV above the valence band, and an intermediate energy level at 6.5 eV (28). The lower edge of CB is diffuse because adiabatic molecular rearrangement can help crossing the band gap, and 9.5 eV is, therefore, an approximate value for the band gap. Excitation into the CB is a multiphoton process, whereby direct excitation from the valance band across the entire band gap (dashed line) is unlikely due to the high order of the multiphoton process. A two-step pathway involving an intermediate energy level is energetically more favorable. In the first step, an electron is abstracted from a multiphoton-excited water molecule, H₂O*, which then prehydrates, e_pre⁻, and settles into a preexisting trap to form a long-lived solvated electron, e_solv⁻, at 6.5 eV. This process and a competing water dissociation process form also hydrated hydroxyl radicals, ^•OH_aq, as well as hydronium ions, H₃O⁺, and molecular hydrogen, H_2aq (60) (SI Appendix, SI3). From the solvated state, the electron can be easily upconverted by low-order multiphoton excitation into the CB located ≈3 eV above the intermediate level. Plasma formation can be initiated not only through multiphoton excitation of water molecules but also through multiphoton ionization of DNA bases, especially of guanine, which has an ionization energy of only 5.8 eV (32). The first step is the formation of a (base⁺:e⁻) pair with a lifetime of a few ps, followed by electron abstraction, hydration, and solvation (61, 62). Subsequent processes are the same as for e_solv⁻ originating from water. Once electrons have reached CB, their energy increases via inverse bremsstrahlung absorption (IBA). When the threshold for impact ionization at 14.25 eV is exceeded, avalanche ionization sets in. The kinetic energies of laser-produced CB electrons thus lie in the same range as those of the low-energy secondary electrons in radiation chemistry (LEEs). CB electrons can interact with DNA molecules either directly or indirectly via ^•OH radicals arising from collisions of energetic CB electrons (≥6 eV) with water molecules (63). (SI Appendix, SI4). Hot electrons loose energy by elastic and inelastic collisions and hydrate within about 300 fs after the end of the laser pulse (57). All DNA-damaging reagents produced during the laser-induced excitation/dissipation chain are marked in red. (B) Average kinetic energy of CB electrons, ε_avg, at irradiance I_ref. Because avalanche ionization is favored by long wavelengths and high irradiance, ε_avg strongly increases with wavelength. (C) Overview of DNA damage pathways at energy levels within the conduction band of water and below.

Fig. 7.

Dependence of average kinetic energy, ε_avg, of laser-produced CBEs at different wavelengths plotted as a function of free-electron density, n_e,total. The circles correspond to the (n_e,total, ε_avg) data pairs at the I_ref values in (Figs. 1 and 5) and (Tables 1 and 2). The opposite trends of decreasing n_e,total and rising ε_avg with increasing wavelengths ≥620 nm are well visible in the (n_e,total,ε_avg) map. The minimum energy required for DSB generation by CBEs from ref. is indicated as a dashed horizontal line, and the electron density corresponding to the bubble threshold (optical breakdown threshold) is demarcated as vertical dashed line. The electron energies show large variations with laser wavelength at moderate electron densities, which progressively vanish with increasing n_e,total. This corresponds to a loss of wavelength specificity of laser effects.