U373-MG human glioblastoma cells were obtained from Cell Lines Service (CLS, Eppelheim, Germany, No. 300366). Cells were routinely grown in RPMI 1640 medium supplemented with 10% fetal calf serum and antibiotics at 37 °C and 5% CO₂. In addition, U373-MG cells stably transfected with an activated TP53 suppressor gene were kindly provided by Prof. Jan Mollenhauer, Dept. of Molecular Oncology, University of Southern Denmark, Odense. Cell incubation with the fluorescent membrane marker 6-dodecanoyl-2-dimethylamino naphthalene (laurdan) [13] was performed in culture medium at a concentration of 8 μM for 1 h. For single molecule experiments highly diluted test samples of 10⁻¹⁰ M rhodamine 6G either dissolved in distilled water or embedded in agarose (1.5% in H₂O; Fluka Chemie, Buchs, Switzerland) were used.

To determine cell viability, single U373-MG cells were seeded at low density (10 cells/mm²) in 8-well microslides (μ-Slide; Ibidi, Martinsried, Germany) and irradiated under the microscope with variable light doses and wavelengths on the 3^rd day after seeding. Either a continuous wave (cw) argon ion laser operated at 514 nm (Innova 90, Coherent, Palo Alto, USA) or a cw helium-neon laser operated at 633 nm (Mod. 1248, Spectra Physics, Mountain View, USA) or a laser diode emitting a quasi continuous series of picosecond pulses (LDH 375 or LDH 400 with driver PDL 800-B, Picoquant, Berlin, Germany; wavelengths: 375 nm or 391 nm; pulse energy: 12 pJ, pulse duration: 55 ps, repetition rate: 40 MHz) was used for irradiation. CW and “quasi continuous” light sources were not further distinguished with respect to cell viability. Epi-illumination of whole cells was performed in an inverted microscope (Axiovert 200, Carl Zeiss Jena, Germany), whereas for TIR illumination of thin layers a prism based setup [14] adapted to an upright microscope (Axioplan 1, Carl Zeiss Jena, Germany) was used. Spatial coordinates of irradiated cells on a positioning table (Scan IM 130×100, Märzhäuser, Wetzlar, Germany) were stored. After re-incubation of the cells with RPMI 1640 medium and further cultivation until the 7^th day after seeding, colony formation, as depicted in Figure 1, was evaluated. Plating efficiency, i.e., the percentage of formed colonies per seeded cells, was determined, and cell survival was defined by less than 10% reduction of the plating efficiency in comparison with non-irradiated controls. For each light dose and each condition of irradiation the percentage of colony forming cells was determined after irradiation of 20 individual cells. Medians and the median absolute deviations (MADs) were calculated from four individual experiments in each case.

Wide field fluorescence microscopy of cultivated cells was performed in an upright microscope (s. above) upon epi-illumination or TIR illumination by one of the laser diodes described above at a wavelength of 375 nm or 391 nm and an irradiance of 100 mW/cm² (corresponding to solar irradiance). In contrast, single molecule experiments of test samples were carried out upon TIR illumination by the argon ion laser operated at 514 nm at a thousand-fold higher irradiance of 100 W/cm². In this latter case the TIR condenser described in [14] was slightly modified using a focusing lens and a glass prism of extremely low intrinsic fluorescence (N-PK 51). Fluorescence images were recorded with a 40×/1.30 oil or a 63×/0.90 water immersion objective lens, an appropriate long pass filter and an electron multiplying (EM-)CCD camera with Peltier cooling; its sensitivity is below 10⁻¹⁶ W/pixel, corresponding to less than 25 photons/pixel within a time frame of 100 ms (DV887DC, ANDOR Technology, Belfast, U.K.) [5]. For some measurements of brightly fluorescent laurdan this camera was replaced by a Sony 3CCD Colour Camera (Model MC-3254, AVT Horn, Aalen, Germany) and used in combination with the software AxioVision (Carl Zeiss Jena, Germany). In addition, fluorescence spectra of single cells were recorded with a custom made polychromator and an image intensifying detection unit (IMD 4562, Hamamatsu Photonics, Ichino-Cho, Japan), whereas fluorescence decay measurements were performed with an image intensifying camera system (Picostar HR 12 image intensifier coupled to a cooled CCD camera; LaVision, Göttingen, Germany). The whole setup is depicted in Figure 2.

Plating efficiency of non-incubated U373-MG glioblastoma cells as a function of light dose for various wavelengths of illumination is depicted in Figure 3. While the wavelength of 375 nm appears appropriate for excitation of intrinsic fluorescence, the longer wavelengths (514 nm, 633 nm) may be used e.g., for Raman microscopy. Plating efficiency is around 80% for non-irradiated cells, but decreases upon illumination. If cell viability is defined as less than 10% reduction of the plating efficiency, cells remain viable up to a light dose of about 25 J/cm² at an irradiation wavelength of 375 nm, 100 J/cm² at 514 nm and about 200 J/cm² at 633 nm. These doses correspond to 250 s, 1,000 s or 2,000 s of solar irradiance, respectively. Cells incubated with fluorescent dyes are generally more sensitive to irradiation, as demonstrated for the membrane marker laurdan. Upon incubation of U373-MG cells with laurdan (8 μM, 1 h) cells remained viable up to a light dose around 10 J/cm² (at an irradiation wavelength of 391 nm). A higher limiting dose of about 30 J/cm² was determined when only the cell surface (plasma membrane and adjacent cellular sites) was illuminated upon total internal reflection of a laser beam, as depicted in Figure 4.

Autofluorescence images of U373-MG human glioblastoma cells (excited at 375 nm and measured above 415 nm) are depicted in Figure 5, and the intracellular fluorescence patterns of unaltered U373-MG cells and U373-MG cells with an activated TP53 tumour suppressor gene are compared. While the unaltered tumour cells show a partly granular and partly diffuse fluorescence pattern, the granular pattern is predominant in cells expressing the activated suppressor gene. Therefore, the evaluation of autofluorescence patterns (in addition to fluorescence spectra and lifetimes) may be a helpful tool for a classification of cancer cells and for the evaluation of tumour cell viability, e.g., in response to anticancer treatment. Light doses of about 0.2 J/cm² were used for image acquisition by the highly sensitive EM-CCD camera. Even with additional detection of fluorescence spectra and decay kinetics the light dose could be limited to 10–12 J/cm², and cell viability of U373-MG cells was maintained. Cell viability of U373-MG glioblastoma cells was also maintained upon incubation with the membrane marker laurdan (8 μM, 1 h), however, for recording fluorescence images, spectra and decay curves the whole “budget” of light (10 J/cm²) was needed upon epi-illumination. In TIR microscopy the illumination dose always remained far below the limiting dose of 30 J/cm², even if about 10 variable-angle images were recorded for nanotomography of cell-substrate contacts [15].

Single molecule detection is based on repetitive excitation of fluorophores located at low concentration in a layer of about 100 nm thickness. To access those thin layers by light, TIR microscopy proved to be an ideal tool; however, an irradiance of about 100 W/cm² (e.g., 6.5 mW of laser power applied to an area of 6,500 μm²) revealed to be necessary to detect a sufficient number of fluorescence photons originating from each molecule. A series of successive single molecule images (each recorded within an exposure time of 100 ms) of rhodamine 6G at a concentration of 10⁻¹⁰ M in agarose is depicted in Figure 6 together with a 3-dimensional plot. Fluorescence intensities of most molecules differed between individual images due to blinking. The illumination time of individual samples was limited to 3–10 s corresponding to a total light dose of 300–1,000 J/cm².