The anatase TiO₂ nanoparticles (Sigma-Aldrich, St. Louis, MO, USA; particle size <25 nm) were calcined under ammonia atmosphere with various calcination parameters, such as temperature, gas flow rate, and calcination time, and then cooled down in nitrogen flow to the room temperature. Three N-TiO₂ samples prepared with different calcination parameters were used in this work. Together with the pure TiO₂, they are denoted as listed in Table 1. The crystalline phases of these samples were determined by Raman spectra (LABRAM-1B; HORIBA, Jobin Yvon, Kyoto, Japan). To evaluate their absorptions in the visible region, the ultraviolet-visible (UV/Vis) diffuse reflectance absorption spectra of these samples were measured with a Jasco V550 UV/Vis spectrophotometer (Jasco, Inc., Tokyo, Japan)

Pure- and N-TiO₂ nanoparticles were dispersed in Dulbecco's modified Eagle's medium with high glucose (DMEM-H), respectively, at various concentrations between 50 and 200 μg/mL. To avoid aggregation, these suspensions were ultrasonically processed for 15 min before using.

The human cervical carcinoma cells (HeLa), human hepatocellular carcinoma cells (QGY), or human nasopharyngeal carcinoma cells (KB) procured from the Cell Bank of Shanghai Science Academy (Shanghai, China) were grown in 96-well plates or Petri dishes in DMEM-H solution supplemented with 10% fetal calf serum in a fully humidified incubator at 37°C with 5% CO₂ for 24 h. Then, the culture medium was replaced by TiO₂-containing medium and the cells were incubated for 2 h in the dark. After the TiO₂ nanoparticles deposited and adhered to the cells, the medium was changed to the TiO₂-free DMEM-H solution supplemented with 10% fetal calf serum for further study.

To examine the photokilling effect, the cells were irradiated with the visible light from a 150-W Xe lamp (Shanghai Aojia Electronics Co. Ltd., Shanghai, China). Two pieces of quartz lens were used to obtain a concentrated parallel light beam. An IR cutoff filter was set in the light path to avoid the hyperthermia effect. A 400-nm longpass filter was used to cut off the UV light. The visible-light power density at the liquid surface in cell wells was 12 mW/cm² as measured by a power meter (PM10V1; Coherent, Santa Clara, CA, USA). After irradiation with this visible light for 4 h, cells were incubated in the dark for another 24 h until further analysis were conducted. The cytotoxicity examinations were carried out with the same procedure as the photokilling effect examinations but without the light irradiation, i.e., the TiO₂-treated cells were incubated in the dark for 28 h.

The cell viability assays were conducted by a modified MTT method using WST-8 [2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H tetrazolium, monosodium salt] (Beyotime, Jiangsu, China). Each well containing 100 μL culture medium was added with 10 μL of the WST-8 reagent solution, and the cells were then incubated at 37°C with 5% CO₂ for 2 h. Subsequently, the absorbance was measured at 450 nm using a microplate reader (Bio-Tek Synergy™ HT; Bio-Tek^® Instruments, Inc., Winooski, VT, USA). The untreated cells were used as the control groups. The surviving fraction represents the ratio of the viable TiO₂-treated cells relative to that of the control groups. It should be noted that the TiO₂-containing DMEM-H solution will affect the absorbance value at 450 nm. Therefore, when measuring the cell viability, the absorbance values were measured as a reference before the WST-8 dyes were added. Each experiment was performed in triplicate and repeated three times.

The cells grown in Petri dishes were incubated with 50 μg/mL TiO₂ in DMEM-H for 10 h before the LSCM observation (Olympus, FV-300, IX71; Olympus, Tokyo, Japan). Hoechst 33342 (Beyotime) and BODIPY FL C₅-ceramide complexed to BSA (Molecular Probes; Invitrogen Corporation, Eugene, OR, USA) were used as the indicators for nucleus and Golgi complex, respectively. Hoechst 33342 (0.5 μg/mL) or Golgi complex marker (5 μM) was added into the growth medium for 15 to 30 min to stain the nuclei or Golgi complexes, respectively.

The reflection images of the intracellular TiO₂ nanoparticles and the fluorescence images of nuclei (or Golgi complexes) were simultaneously obtained by the LSCM in two channels with no filter for the reflecting light and a 585 to 640-nm bandpass filter for the fluorescence. A 488-nm continuous-wave (CW) Ar⁺ laser (Melles Griot, Carlsbad, CA, USA) or a 405-nm CW semiconductor laser (Coherent) was used as the excitation source. A 60 × water objective was used to focus the laser beam to a spot of about 1 μm in diameter. The differential interference contrast (DIC) micrographs to exhibit the cell morphology were acquired in a transmission channel simultaneously. The three-dimensional (3D) distributions of TiO₂ nanoparticles and nuclei (or Golgi complexes) were obtained using the z-scan mode of the microscope.

As shown in Table 1 and Figure 1a, the N-TiO₂ samples N-550-1 and N-550-2 with the calcination temperature of 550°C, as well as the pure TiO₂, exhibited a similar feature with five Raman peaks around 143, 197, 395, 514, and 640 cm^-1, corresponding to the Raman fundamental modes of the anatase phase 1516. The Raman peaks for rutile phase 16 around 238, 420, and 614 cm^-1 appeared when the calcination temperature was 600°C as shown in the spectrum of the sample N-600-1. It can be concluded that the phase of the TiO₂ nanoparticles would transform from anatase to rutile when the calcination temperature increased to 600°C. Such a phase transformation will result in a decrease of the photocatalytic ability for TiO₂ powders 1718. Therefore, we only used samples N-550-1 and N-550-2 for further studies.

Figure 1b shows the absorption spectra of the samples N-550-1 and N-550-2 and pure TiO₂. Compared to the pure TiO₂, the absorbances of N-550-1 and N-550-2 are higher in the visible region. However, the sample N-550-2 has the higher absorbance than N-550-1 in the region of 400 to 500 nm. Since N-550-1 and N-550-2 were calcinated at the same temperature and with the same amount of ammonia (flow rate times time), it seems that higher ammonia flow rate (N-550-2) could cause more absorption in the visible, which was expected to have higher photokilling efficiency of cells.

To evaluate the cytotoxicity of pure- and N-TiO₂ nanoparticles, the TiO₂-treated cells were further incubated in the dark for 28 h and the cell viability assays were then conducted. As shown in Figure 2a, all the surviving fractions of the treated HeLa cells were on the average values greater than 85% (with the concentration from 50 to 200 μg/mL). As shown in Figure 3, all the surviving fractions of the treated QGY or KB cells with the pure- or N-TiO₂ concentration of 200 μg/mL in the dark were greater than 85%. These results indicated that the cytotoxicities of pure- and N-TiO₂ nanoparticles were quite low. The cytotoxicities of these nanoparticles were quite similar, and there was no significant influence of the concentration on the cytotoxicity. Pure TiO₂ is biocompatible with primary and cancer cells 4. Nitrogen is an essential element of many biological molecules, such as proteins and nucleic acids. So, nitrogen is not toxic if it does not exceed the normal levels. It could be understood that a small amount of nitrogen doping would not lead to more cytotoxicity than pure TiO₂.

The photokilling effects were measured as described in the experimental section. The surviving fractions of HeLa cells under visible-light irradiations for 4 h in dependence on the concentrations of pure- and N-TiO₂ nanoparticles were shown in Figure 2b. As demonstrated in Figure 2b, the visible light showed very little photokilling effect on HeLa cells in the absence of any TiO₂ (pure or N-doped) (at the 0 concentration). The surviving fractions (compared to the control cells without irradiation) were around 93%, which might be caused by the light irradiation, the fluctuant temperature during irradiation, and the experimental procedures. The spectrum of the light irradiated on cells (with filters) is also shown in the figure as an inset. It should be noted according to the spectrum in Figure 1b that the pure TiO₂ nanoparticles still has some absorption around 400 nm though the band gap of TiO₂ was reported to be 3.2 eV (corresponding to a wavelength of 387 nm). Therefore, pure TiO₂ exhibited some photokilling effect under visible-light irradiation as shown in Figure 2b. However, the cells treated with N-TiO₂ were killed more effectively than that with pure TiO₂. The photokilling effects of samples N-550-1 and N-550-2 were quite similar although their absorption spectra showed some difference. It is also demonstrated in Figure 2b that the survival fractions decreased with the increasing concentrations of the TiO₂ samples. It decreased to 40% for the cells treated with sample N-550-2 at a concentration of 200 μg/mL.

The photokilling effects of sample N-550-2 at a concentration of 200 μg/mL on QGY and KB cells were also measured as shown in Figure 3. Similar with the photokilling effect on HeLa cells, the QGY and KB cells treated with N-550-2 were also killed more effectively than that with pure TiO₂ under the visible-light irradiation. The results revealed that the N-TiO₂ might be applied to different cancers as a photosensitizer for PDT.

The mechanism of the photokilling effect for cancer cells caused by TiO₂ nanoparticles is very complex. It has been identified that UV-photoexcited TiO₂ in aqueous solution will result in formation of various ROS, such as hydroxyl radicals (·OH), hydrogen peroxide (H₂O₂), superoxide radicals (·O₂^-) and singlet oxygen (¹O₂) 1920. The ROS will attack the cancer cells and finally lead to the cell death. In order to study the function of ROS on the photokilling effect, the L-histidine, a quencher for both ¹O₂ and ·OH 212223, was added into the 96-well plates (20 mM) 30 min before the cells were irradiated by light. In the presence of 20 mM L-histidine, all the surviving fractions of the cells treated with pure- and N-TiO₂ at a concentration of 200 μg/mL increased evidently as shown in Figure 4. These results are similar to the previous report for UV-photoexcited TiO₂ 14. It can be concluded that the ROS plays an important role on the photokilling effect, although we cannot tell which one played the main role. Further research is needed to figure out all the ROS influences.

As is well-known, light-excited TiO₂ generates the electron-hole (e^-/h⁺) pairs. The photogenerated carriers migrate to the particle surface and participate in various redox reactions there. Hence, the direct damage induced by photokilling effect would only occur at the sites of TiO₂ particles. Therefore, it is of importance to know if the TiO₂ nanoparticles were internalized into cells and how their intracellular distributions were. To find out the subcellular distribution of TiO₂ nanoparticles, the TiO₂-treated HeLa cells were stained with fluorescence indicators for Golgi complex and nucleus, respectively. Surprisingly, some TiO₂ nanoparticles were found inside the nuclei as shown in Figure 5, where the HeLa cells were treated with (N-550-2, 50 μg/mL) and stained with nuclear indicator. When these N-TiO₂-treated cells were irradiated by light from the Xe lamp with a 400-nm longpass filter (12 mW/cm²) for 4 h, some micronuclei were observed as shown in Figure 6. Since the TiO₂ nanoparticles had entered into the nuclei of cells, the photoactivation effect could occur directly inside the nuclei, which might cause chromosomal damage or nucleus aberration. Micronuclei are usually formed from a chromosome or a fragment of a chromosome not incorporated into one of the daughter nuclei during cell division. This is an evidence of the direct damage to the nucleus resulted from the photoexcited N-TiO₂ nanoparticles.

Figure 7 is the confocal micrographs to show the distributions of Golgi complexes (fluorescence image) and TiO₂ nanoparticles (reflection image) in HeLa cells. As shown in the merged image in Figure 7d, the TiO₂ particles were not only found on the cell membrane but also in the cytoplasm. Some TiO₂ nanoparticles aggregated around or in Golgi complexes. The co-localizations of TiO₂ with Golgi complexes (yellow color) were observed. The cell viability might be influenced by the localization of TiO₂ in Golgi complexes or other cell organelles, although there is no direct evidence found in this work.