X-ray induced singlet oxygen generation by nanoparticle-photosensitizer conjugates for photodynamic therapy: determination of singlet oxygen quantum yield

Singlet oxygen is a primary cytotoxic agent in photodynamic therapy. We show that CeF3 nanoparticles, pure as well as conjugated through electrostatic interaction with the photosensitizer verteporfin, are able to generate singlet oxygen as a result of UV light and 8 keV X-ray irradiation. The X-ray stimulated singlet oxygen quantum yield was determined to be 0.79 ± 0.05 for the conjugate with 31 verteporfin molecules per CeF3 nanoparticle, the highest conjugation level used. From this result we estimate the singlet oxygen dose generated from CeF3-verteporfin conjugates for a therapeutic dose of 60 Gy of ionizing radiation at energies of 6 MeV and 30 keV to be (1.2 ± 0.7) × 108 and (2.0 ± 0.1) × 109 singlet oxygen molecules per cell, respectively. These are comparable with cytotoxic doses of 5 × 107–2 × 109 singlet oxygen molecules per cell reported in the literature for photodynamic therapy using light activation. We confirmed that the CeF3-VP conjugates enhanced cell killing with 6 MeV radiation. This work confirms the feasibility of using X- or γ- ray activated nanoparticle-photosensitizer conjugates, either to supplement the radiation treatment of cancer, or as an independent treatment modality.


Results and Discussion
CeF 3 nanoparticles were prepared using a co-precipitation method followed by conjugation to commercially available VP (see Materials and Methods). Figure 2(a) shows the TEM image of the nanoparticles and the size histogram. The average size of the synthesized nanoparticle is 9 ± 2 nm. As shown in Fig. 2(b), there is a high degree of spectral overlap between the CeF 3 nanoparticle emission spectrum and the optical absorption spectrum of VP 19 , which is an important criterion for efficient photodynamic activation. The scintillation spectrum of CeF 3 upon irradiation with 8 keV X-rays is predominantly in the UV-A range, peaking at around 340 nm, and with 30% overlap with the Soret absorption band of VP.
We note that the Soret band is much stronger than the red Q-band (~690 nm) that is used for conventional visible light-mediated PDT. Varying concentrations of VP in the range 0-1 μ M were conjugated with 300 μ M of CeF 3 nanoparticles, and unconjugated VP was removed by washing. These concentrations of VP ensured zero order kinetics for the concentration of SOSG used here (4 μ M). The absorption spectra of the conjugates are shown in Fig. 3(a), where the peaks corresponding to VP indicate successful conjugation without any additional molecular linkage. The VP spectra in the conjugates are somewhat distorted compared to free VP, with an altered Q-to-Soret band ratio. Additional confirmation of the VP-CeF 3 attachment was obtained by FTIR spectroscopy Scientific RepoRts | 6:19954 | DOI: 10.1038/srep19954 (See Fig. 3(b)). Although the conjugation mechanism was not conclusively determined, we note that there is electrostatic interaction between the positively-charged CeF 3 nanoparticles and negatively-charged VP 41 . The concentration of VP in each conjugate sample was calculated from the absorption spectra (Supporting Information section S1) and it is shown in the insert to Fig. 3(a). From the VP and nanoparticle concentrations, as well as the size, density and molar mass of CeF 3 , we estimated that, on average, 31 VP molecules were conjugated to each nanoparticle in the case of Sample C. This sample had the highest concentration of conjugated VP of 0.9 μ M and the highest conjugation level per single nanoparticle. The corresponding values in Samples B and C were 13 and 4 VP molecules per nanoparticle, respectively.
The X-ray 1 O 2 quantum yield was then determined in several steps. Firstly, 1 O 2 generation from UV (365 nm) irradiation of VP and CeF 3 -VP conjugates in water was confirmed. This wavelength coincides with the VP absorption peak and also corresponds closely to the 340 nm peak emission wavelength of CeF 3 . 1 O 2 generation was confirmed using the SOSG probe by monitoring the enhancement of the fluorescence intensity at 488 nm excitation, integrated over the range 500-600 nm. This was done using the same concentration (4 μ M) of SOSG under conditions of zero-order kinetics, while varying the concentration of the photosensitizer (see Supporting Information Section S2 and Supplementary Fig. S1). The fluorescence intensity of the SOSG emission as a function of UV irradiation time is plotted in Fig. 4(a) for VP and in Fig. 4(b) for the conjugates. In interpreting these data it is necessary to take into account the complication that SOSG itself acts as a photosensitizer under UV irradiation and that the SOSG fluorescence decreases with irradiation due to photobleaching 42,43 . Hence, a control  sample of SOSG only was also included. The SOSG fluorescence intensities were also corrected for the inner-filter effect and for pH variations (see Supporting Information Section S3, S4, and Supplementary Fig. S2). Figure 4(a) shows that, for a fixed concentration of VP, the SOSG intensity increases linearly with UV exposure, demonstrating that 1 O 2 has been generated, the amount of which also increases proportionally with VP concentration. Figure 4(b) presents the corresponding results for the conjugates, as well as for unconjugated CeF 3 nanoparticles and the SOSG probe itself. Comparing the plots for pure SOSG ( Fig. 4(a)) and pure CeF 3 ( Fig. 4(b)), we conclude that that pure CeF 3 nanoparticles also act as a photosensitizer, which is reported here for the first time. However, CeF 3 -VP conjugates produce more 1 O 2 than pure CeF 3 and higher conjugation levels lead to increased 1 O 2 generation, as anticipated.
In order to quantify the generation of 1 O 2 under X-ray exposure and to determine the value of η , new conjugate samples containing the same amount of CeF 3 and VP as previously were mixed with 4 μ M of SOSG and exposed to X-ray irradiation. Figure 5(a) demonstrates that 1 O 2 is indeed generated during X-ray exposure, with a significant increase in the SOSG fluorescence compared to the nanoparticle-only and PS-only controls. In order to determine the X-ray singlet oxygen quantum yield we designed a new procedure, since the reference method 44 cannot be used to determine this due to the absence of applicable standards. Firstly, a 1 μ M concentration of the well-established photosensitizer protoporphyrin IX (PpIX) was combined with 4 μ M SOSG and the 1 O 2 produced under UV irradiation was measured using SOSG (Supplementary Fig. S3 (a)). The total number of UV photons absorbed by the PpIX was determined by standard methods (Supporting Information Section S5 and Supplementary Fig. S4 (b)). The SOSG fluorescence intensity was then related to the number of detected singlet oxygen molecules, based on the known 1 O 2 quantum yield of PpIX in water (0.56) under UV irradiation (Supplementary Fig. S3(d)). Finally, the total number of X-ray photons absorbed by the conjugate was determined (See Supporting Information Section S6 and Supplementary Fig. S4).
By combining the experimental results for UV-irradiated PpIX and X-ray-irradiated conjugates (see Supporting Information section S5 and S6 for the calculation), the number of 1 O 2 molecules generated by X-rays has been plotted in Fig. 5(b) as a function of number of X-ray photons absorbed by the nanoparticles. From the  slope of the best fit, the number of 1 O 2 molecules generated by each absorbed 8 keV X-ray was calculated as listed in Table 1 for each conjugate sample, together with the respective values for the X-ray 1 O 2 quantum yield, η . From Table 1 we conclude that, in order to produce one singlet oxygen molecule in sample C with the highest VP conjugation level achieved here, required 1.27 ± 0.08 eV of absorbed X-ray energy. For comparison, 0.98 eV is required to excite ground-state 3 O 2 to the singlet state 1 O 2 45 , which means that, in our case around 30% of the X-ray photon energy is lost through other radiative and non-radiative processes.
The values of the X-ray induced singlet oxygen quantum yield, η , can be used to estimate the 1 O 2 dose achievable with CeF 3 -VP conjugates during standard cancer radiotherapy. Several effects must be taken into account in this calculation. Firstly, the nanoparticles contain heavier elements than in tissue and interact more strongly with ionizing radiation, so that they receive a higher radiation dose than the tissue for the same total incident X-ray dose. To quantify this effect we determined the partitioning of the radiation dose between the nanoparticles in the tissue and the tissue itself (see Supporting Information Section S7). The fraction of radiation energy absorbed by the nanoparticles, F CeF 3 is given by: where V p is the volume fraction of nanoparticles in the tissue (proportional to nanoparticle concentration), ρ CeF 3 and ρ t are the density of CeF 3 and the tissue, respectively, and α CeF 3 and α t are the mass absorption coefficients of CeF 3 and the tissue, respectively. The mass absorption coefficient of CeF 3 for different X and γ -rays energies was obtained from the NIST database 23 for elemental Ce and F and for an example (lung) tissue. Figure 6 shows the values of F CeF 3 as a function of energy for different volume fractions, V p ; these are in agreement with earlier reports 24 . We then assumed a nanoparticle loading of = V NP 5% cell volume, as in ref. 24, noting that the relevant literature values vary from 0.1 to 33.7%. The photon energies used were based on current radiotherapy treatments; these were 6 MeV for high energy external-beam and 30 keV as representative of brachytherapy. As seen in Fig. 6, at 6 MeV the CeF 3 nanoparticles absorb 28% of the total absorbed energy and 72% is absorbed by the tissue, whereas at 30 keV 87% of energy is absorbed by the nanoparticles versus only 13% absorbed by the tissue.
Once the dose partition is known, the energy delivered to the nanoparticles can be determined based on the radiation dose delivered to the tissue. Assuming a therapeutic tissue dose over the course of fractionated treatment of 60 Gy, the radiation energy delivered per cell was calculated as follows. The cell is assumed to be a sphere of (10 μ m) 3 with water as main constituent and with a mass of 10 −12 kg. 60 Gy delivered dose means that an energy = E 375 l MeV is absorbed per cell. The energy per cell absorbed by the nanoparticles can be then found by the relation: . Using equation (2), the dose partition values in Fig. 6 and a 5% volume ratio yields 146 and 2510 MeV at photon energies of 6 MeV and 30 keV, respectively. It is further assumed that the number of 1 O 2 molecules, n keV , generated per keV of incident photon energy as derived above for 8 keV is the same at 30 keV, 6 MeV and all intermediate energies. This assumption is supported by the observation that the scintillation quantum yield across a wide range of materials is roughly constant (within ±20-40%) across this energy range 46 . Thus, using the previously obtained value of the X-ray induced singlet oxygen quantum yield and the value of energy absorbed by the CeF 3 per cell during radiotherapy treatment, the number of 1 O 2 molecules generated per cell n SO is given by: The calculated value of n SO for the most efficient conjugate measured (Sample C) is then (1.2 ± 0.7) × 10 8 for 6 MeV and (2.0 ± 0.1) × 10 9 for 30 keV. It is of interest to compare these values to literature estimates of the singlet oxygen dose required for cell killing. The most direct measurement of photodynamic cell killing dose 47 , carried out in vitro in leukaemia cells PpIX as the photosensitizer (following incubation with the prodrug aminolevulinic acid) and using direct near-infrared luminescence dosimetry, showed that ~5 × 10 7 1 O 2 molecules per cell result in 1/e clonogenic surviving fraction. Thus, 1.2 × 10 8 to 2.0 × 10 9 1 O 2 molecules per cell would correspond to ~10% and negligible surviving fraction, respectively. Other studies have estimated the concentration of 1 O 2 sufficient to cause tissue necrosis (in rat liver) to be 0.9 mM (~5 × 10 8 molecules per cell) 48 , while the threshold dose of singlet oxygen estimated for tumour spheroids was 0.323 mM (~2 × 10 8 molecules per cell) assuming no photosensitizer photobleaching 49 . These values are comparable to those obtained here for X-ray irradiation of the most efficient conjugates described here.
We validated our approach by a radiation-induced PDT experiment at 6 MeV conducted in cell cultures, where cells were treated with CeF3-VP conjugates prior to radiation treatment. Here, we used pancreatic cancer (Panc1) and HEK293 (control) cell lines. The viability of both types of cells with a different radiation (dose up to 6 Gy) and with different dilutions of the most efficient conjugate C was determined (Supporting Information Section S8 and Supplementary Figs S5 and S6). On this basis, the optimum concentration of conjugate C (80 μ M), for which both cancer and control cells have shown 100% viability, has been selected for radiation-induced PDT demonstration. The Panc 1 cells were treated with the conjugate C at 80 μ M. The treated Panc1 cells and controls (Panc1 with VP only) were incubated overnight and then exposed to radiation. Figure 7 shows the viability of cells which were treated with the conjugate and their controls for different radiation dose. The viability of cells treated with the CeF 3 -VP conjugate clearly decreases at different radiation doses. For example, at 6 Gy radiation dose 32% cells were killed, which is an indication of efficient PDT with γ -radiation.

Conclusions
Singlet oxygen generation from VP and from CeF 3 -VP conjugates was quantified using a fluorescent probe, SOSG, which is 1 O 2 specific, so that there was unequivocal generation of singlet oxygen upon X-ray exposure. The X-ray induced 1 O 2 quantum yield for the most efficient conjugate with 31 VP molecules per nanoparticle was 0.79 ± 0.05. With that information we estimate the concentration of 1 O 2 generated in nanoparticle-loaded tissue upon exposure to high energy (6 MeV) or low energy (30 keV) ionizing radiation. A radiotherapeutic dose of 60 Gy delivered to tissue containing a 5% volume fraction of PS-conjugated nanoparticles produced 1.2 × 10 8 to 2.0 × 10 9 1 O 2 molecules per cell. These values are within the range of significant cytotoxicity reported both in in vitro and in vivo for light-activated photodynamic therapy. Hence, it is conceivable that these nanoparticle conjugates could enhance the therapeutic efficacy of high-energy external-beam radiotherapy or low-energy brachytherapy through the complementary mechanism(s) of cell death between ionizing radiation (DNA damage) and photodynamic (membrane damage) treatments. This could then be exploited either to increase the anti-tumour effect or to reduce the normal tissue toxicity, especially if the conjugates have intrinsic preferential localization in tumour or are biomarker targeted. The alternative perspective is to develop X-ray activated PDT for treatment of larger inaccessible tumours that are not amenable to conventional light-activated PDT. Radiation induced PDT and cell killing has been demonstrated in a cell culture at 6 MeV radiation energy.

Materials and Methods
VP, protoporphyrin IX and DMSO were purchased from Sigma Aldrich (Australia) and used without further purification. SOSG was purchased from Invitrogen (USA). Stock solutions of VP (3 mM) and protoporphyrin IX (3.5 mM) were prepared by dissolving 2 mg photosensitizer molecules in 1 ml dimethylsulfoxide (DMSO) and then were kept in the dark below 4 °C. The stock solution of SOSG (500 μ M) was prepared by dissolving 100 μ g (1 vial) in 330 μ l methanol and then kept in frozen in the dark. CeF 3 nanoparticles were prepared using a simple co-precipitation method 50 . Briefly, 6 mmol of NH 4 F was dissolved in 20 ml of methanol and the solution was heated to 70 °C. 2 ml of methanol containing 2 mmol of CeCl 3 .7H 2 O was added drop wise to the above and the mixture was stirred at 600 rpm. After 5 hrs, the CeF 3 nanoparticles were cooled down and washed several times. Their average size was ~10 nm. A stock 5 mM suspension of nanoparticles was prepared by adding 1 mg of nanoparticles to1 ml of water.
To conjugate the nanoparticles with the photosensitizer, 500 μ M of CeF 3 and 0.5 μ M of VP were mixed in a rotator at room temperature for 6 h at 200 rpm. After 18 h the mixture was centrifuged at 15,000 rpm for 20 min. The supernatant was removed and washed twice. The same procedure was repeated to conjugate the same amount of CeF 3 with different concentrations of VP (1 and 1.5 μ M). All measurements were done under oxygenated conditions.
For singlet oxygen generation measurements, 2 ml conjugate and control samples (VP only, water and CeF 3 only) were placed in a quartz cuvette and 4 μ M of SOSG was added. The samples were excited at 488 nm and the emission of SOSG in the 500-600 nm range was measured before after the UV irradiation/X-ray radiation and it was plotted as a function of time.T he increase in the emission intensity is an indication of singlet oxygen generation.
The cells, normal and cancer, as well as appropriate controls (approximately 3 × 10 5 cells/ml) were seeded in the wells of a 96 well plate (100 μ l in each) and incubated over night. These wells were exposed to different doses of radiation (1 Gy, 2 Gy, 4 Gy and 6 Gy) radiation and incubated again for 24 hrs. The MTS assay testing cell viability was carried out according to the manufacturer protocol and the absorbance at 492 nm was measured after 2 hours using plate reader. Cell viability was then calculated as a percentage of the absorbance of the untreated control, which was set to 100%. To check the cell viability in the presence of the conjugate, Conjuate C (CeF 3 -320 μ M) and its 2 ,4 and 8 times dilution were added to a reduced serum medium. The normal (HEK293) and cancer cells (Panc1) were seeded in the wells as indicated earlier and incubated overnight. Then the medium was removed and added the conjugate with different dilution in a reduced serum medium and incubated overnight to ensure cell uptake. After 24 hours, the medium was removed, fresh medium was added and the MTS test was carried out as indicated earlier.
To perform PDT in cells, the optimised conjugate C at 80 μ M (the maximum concentration f the conjugate which showed negligible toxicity to both normal and cancer cells) and a control amount of VP were prepared in a reduced serum medium. The Panc 1 cells were seeded in the wells of the 96 well plate (5 different plates for different radiation doses) and incubated for 24 hours. Then the conjugate with different controls were added and incubated overnight. After 24 hours, the medium was changed and the wells were exposed to different radiation doses( 1 Gy, 2 Gy, 4 Gy and 6 Gy) . Afterwards, the cells were again incubated overnight and an MTS assay was carried out to check the viability.
TEM image of nanoparticles were taken with PHILIPS CM10 system with an accelerating voltage of 100 kV. Fluorescence measurements were carried out using a Cary Eclipse fluorescence spectrophotometer with 5 nm spectral resolution for both excitation and collection. The absorption spectra were measured using a UV/VIS/NIR Cary dual-beam spectrophotometer with paired 1 cm path length quartz cuvettes cleaned with ethanol. UV irradiation was performed with a 365 nm high power LED with 2.4 mWcm −2 incident power density. X-irradiation was performed using an XPert Pro system (PANalytical, Netherlands) operating at 45 kV/40 mA. The system produced Cu-K α radiation with a Ni filter to produce 8 keV X-rays. For the PDT experiment in cells with γ -ray radiation at 6 MeV, a linear accelerator (LINAC, Elekta AB, Sweden) was used to irradiate the samples. Each well in 96 well plates were CT-scanned and a radiation dose distribution was planned on an Elekta XiO planning system (Elekta AB, Sweden) to deliver a different dosage (1 Gy, 2 Gy ,4 Gy and 6 Gy) to each plate. Irradiation of the samples was carried out using 6 MV photons from anterior and posterior directed radiation fields. The absorbance in the 96 well are measured using a Fluorostar Galaxy plate reader by setting the wavelength at 492 nm with proper gain adjustment.