Photochemical generation of the 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) radical from caged nitroxides by near-infrared two-photon irradiation and its cytocidal effect on lung cancer cells

Novel caged nitroxides (nitroxide donors) with near-infrared two-photon (TP) responsive character, 2,2,6,6-tetramethyl-1-(1-(2-(4-nitrophenyl)benzofuran-6-yl)ethoxy)piperidine (2a) and its regioisomer 2b, were designed and synthesized. The one-photon (OP) (365 ± 10 nm) and TP (710–760 nm) triggered release (i.e., uncaging) of the 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) radical under air atmosphere were discovered. The quantum yields for the release of the TEMPO radical were 2.5% (2a) and 0.8% (2b) in benzene at ≈1% conversion of 2, and 13.1% (2a) and 12.8% (2b) in DMSO at ≈1% conversion of 2. The TP uncaging efficiencies were determined to be 1.1 GM at 740 nm for 2a and 0.22 GM at 730 nm for 2b in benzene. The cytocidal effect of compound 2a on lung cancer cells under photolysis conditions was also assessed to test the efficacy as anticancer agents. In a medium containing 100 μg mL−1 of 2a exposed to light, the number of living cells decreased significantly compared to the unexposed counterparts (65.8% vs 85.5%).

Near-infrared (NIR) photons are excellent light sources in physiological studies as this wavelength of light is less harmful to living tissue than ultraviolet irradiation. Deeper penetration of NIR photons into biological samples is possible using NIR radiation with wavelengths of 650-1050 nm (= 27-44 kcal mol −1 ). However, in general, chromophores do not absorb at such long wavelengths and the photon energy is too low for bondcleavage reactions to generate (i.e., uncage) functional molecules. For example, the bond-dissociation energy of the weak PhCH 2 -OPh, linkage is reported to be 52.1 kcal mol −1 [43]. These issues can be solved by using the NIR-two-photon (TP) Scheme 1: Photochemical generation of TEMPO radical. excitation technique [44], in which a molecule is electronically excited to the same state generated by one-photon (OP) excitation in the UV-vis region [45]. In addition to the advantages of TP excitation, three-dimensional control of the electronic excitation is possible because the probability of TP excitation is proportional to the square of the light intensity [46]. The lightinduced generation of nitroxides using the TP excitation technique, i.e., the concentration jump of nitroxides, is one promising method of exploring the role of these species in life phenomena [47][48][49][50][51][52][53][54] and of promoting site-selective chemical reactions such as polymerization. Very recently, Guillaneuf and co-workers reported the two-photon-induced release of nitroxides in a materials science study [55].
In the last decade, we developed a TP-responsive photo-labile protecting group [56][57][58] with simple cyclic stilbene structures such as 2-(4-nitrophenyl)benzofuran (NPBF) that absorb in the NIR region of 710-760 nm for the uncaging of bioactive substances such as glutamate and Ca 2+ [59][60][61][62][63][64]. Herein, we report the synthesis of new caged nitroxides (nitroxide donors) 2a and 2b having the TP-responsive NPBF chromophore and the NIR TP-triggered generation of the 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) radical under atmospheric conditions using these species (Scheme 1). Because free radicals are cytotoxic due to their strong DNA-damaging activity [65], they play important roles as anticancer therapeutic agents [66]. Among the free radicals, nitroxides including the TEMPO radical have unique properties, where they can act not only as radical scavengers, but also as anticancer agents [67]. Due to the unique properties described above, nitroxides are not toxic to normal host cells and exhibit toxicity only to tumor cells. Thus, nitroxides are ideal candidates as anticancer therapeutic agents. Based on this knowledge, the cytocidal effect of the   c Fluorescence quantum yields. The standard sample 9,10-diphenylanthracene (Φ f = 0.91) was used for determining the quantum yields. d Fluorescence lifetime monitored at 560 nm. The concentrations were the same as those used for the fluorescence measurements. e Each contribution is 57% and 43%, respectively. f Each contribution is 70% and 30%, respectively. radical released from compound 2a on lung cancer cells was tested in vitro, in addition to the fundamental study.
The photophysical data for the new compounds 2a,b and 5a,b are summarized in Table 1. The absorption maxima of com-pounds 2 and 5 were observed at ≈370 nm with a molar extinction coefficient ε ≈20000 M −1 cm −1 in both benzene and DMSO. The emission profile showed a significant solvent effect. The fluorescence quantum yields in DMSO of 5a and 5b were determined to be 16.1 and 8.6%, respectively, although no emission was observed from these compounds in non-polar benzene, indicating that the excited state has zwitterionic character. The charge transfer transition was supported by time-dependent density functional theory (TD-DFT) calculations for 5a at the CAM-B3LYP/6-31G(d) level of theory (Supporting Information File 1, Figure S2). The fluorescence quantum yields of caged-TEMPO 2a and 2b were found to be 2.9 and 2.2% in DMSO, which are much smaller than those of 5a and 5b, respectively, suggesting the chemical reactivity of the singlet excited states of 2a and 2b.
OP photolysis of 2a (5 mM) was first conducted in benzene at ≈298 K using 365 nm light (6.02 × 10 15 photons s −1 ) under atmospheric conditions ( Figure 1). Clean release of the TEMPO radical was confirmed by measuring the electron paramagnetic resonance (EPR) signals of the typical nitroxide, A N = 15.5 G (g = 2.00232, Figure 1 and Figure 2c). The first-order rate constant for generation of TEMPO in the bulk photoreaction was found to be k = 1.6 × 10 -5 s −1 . The amount of photochemically released TEMPO radical was determined by comparing the EPR intensity with the calibration curve of the standard TEMPO sample (Supporting Information File 1, Figure S4). The chemi- cal yield of TEMPO was 80% after 10 min irradiation in benzene under air atmosphere (Figure 2g). Secondary photoreaction of TEMPO gradually decreased the chemical yield of TEMPO. The quantum yield (Φ) for photochemical release of the TEMPO radical was 2.5% at ≈1% conversion in the photolysis of 2a in benzene under atmospheric conditions. Similar photochemical generation of the TEMPO radical was conducted with 2b (5 mM, Supporting Information File 1, Figure S5 and Figure 2d,h). The clean generation of the TEMPO radical was also observed during photolysis under 365 nm irradiation in benzene at ≈298 K under atmospheric conditions, although the reaction was slower than that of 2a, k = 5.5 x 10 -6 s -1 ; Φ = 0.8% at ≈1% conversion of 2b. However, the chemical yield of TEMPO was also high (81% after 20 min irradiation under the same conditions), although slow photochemical decomposition of TEMPO was observed with prolonged irradiation (Figure 2h). In DMSO, the quantum yield for the formation of TEMPO increased significantly to 13.1% (from 2a) and 12.8% (from 2b) at ≈1% conversion of 2 under atmospheric conditions (Figure 1). The notable effect of the solvent on the TEMPO generation may be due to the increase in the lifetime of the excited states. Photochemical decomposition of TEMPO in DMSO was found to be faster than that in benzene, but the chemical yield of TEMPO (56% from 2a and 58% from 2b after 40 s irradiation) was found to be lower than that obtained in benzene (Figure 1).
To obtain insight into the mechanism of generation of the TEMPO radical, the photolysis of 2 was conducted under degassed conditions using the freeze-pump-thaw (FPT) method (Figure 2a,b). Interestingly, the generation of the TEMPO radical was highly suppressed under the photolysis conditions (Figure 2a,b). Under air conditions, however, the photochemical release of TEMPO was detected in benzene, as shown in Figure 2c,d. Faster formation of TEMPO was observed when O 2 atmosphere was used instead of an air atmosphere (Figure 2e,f). Therefore, the O 2 molecule may play an important role in clean generation of the TEMPO radical during photolysis. Indeed, the compounds oxidized at the benzylic carbon, 6 and 7, were isolated in 15% (15%) and 56% (42%) yield in the photolysis of 2a and 2b under atmospheric conditions, respectively (Scheme 3), indicating that under degassed conditions, the photochemically generated radical pair returns to the starting compound 2 with rapid radical recombination. Over 70% of the caged TEMPO 2a and ≈85% of 2b were recovered after 2 h of irradiation under degassed conditions. The retarded formation of TEMPO after 5 min of irradiation is due to the decrease in the relative absorbance of 2a to those of primary photoproducts (Figure 2c,e).
The TP photolysis of 2a (10 mM) and 2b (10 mM) was carried out in benzene under atmospheric conditions using 710, 720, 730, 740, 750, and 760 nm near infrared light from a Ti:sapphire laser (pulse width 100 fs, 80 MHz) emitting at an average of 700 mW (Figure 3 for 2a and Supporting Information File 1, Figure S6 for 2b). The typical EPR signals of TEMPO were also observed after TP excitation of 2a and 2b (Supporting Information File 1, Figure S7). The formation of TEMPO at 740 nm, k 740 = 4.9 × 10 −6 s −1 in the bulk photoreaction, was the fastest in the TP-uncaging reaction of 2a (Figure 3). For the uncaging reaction of 2b, the rate of consumption under 730 nm irradiation, k 730 = 1.6 × 10 -6 s −1 , was larger than those at other wavelengths (Supporting Information File 1, Figure S6). The TP action spectra of 2a and 2b in benzene are shown in Figure 4, where the spectra were obtained by extrapolation from the absolute TP cross-section of the parent NPBF (18 GM) at Scheme 4: Isodesmic reaction from BRa and 5b to 5a and BRb. 720 nm [58]. The TP cross-section of 2a was higher than that of 2b by ≈15 GM. This higher GM value may be due to the stronger donor-acceptor character of 2a relative to that of 2b, because the electron-donating alkyl group is located at the paraposition of the p-nitrophenyl group in 2a.
As observed in the OP uncaging reaction at 365 nm, the efficiency of the TP-induced TEMPO uncaging reaction of 2a was almost three times higher than that of 2b in benzene. This is attributed to the substituent effect of the meta-alkoxy group on the reactivity in the electronically excited states [70]. Moreover, the relative stability of radicals BRa and BRb generated by the photolysis of 2a and 2b had an important impact on the uncaging efficiency. The isodesmic reaction shown in Scheme 4 suggests the radical BRa derived from 2a was 2.04 kcal mol −1 more stable than BRb generated from 2b based on DFT calculations at the B3LYP/6-31G(d) level of theory. In DMSO, no significant difference between 2a and 2b was observed for the photochemical release of the TEMPO species, although the solvent effect is not clearly explained.
As mentioned above, the spatiotemporally controlled generation of the radical pair of TEMPO and BR was confirmed in the photolysis of compounds 2a and 2b. Because free radicals play important roles as anticancer therapeutic agents, the cytocidal effect of the radical released from compound 2a was also tested in vitro using lung cancer cells. One hundred thousand Lewis lung carcinoma (LLC) cells were seeded into 24-well plates (medium: DMEM) and incubated overnight at 37 °C under an atmosphere of 95% air and 5% CO 2 . The medium was replaced with fresh phenol-red free DMEM containing various concentrations of 2a (0, 10, 100 μg mL −1 ) and further incubated for 4 h under the same conditions. Without exposure to light, 2a itself exhibited slight cytotoxicity based on trypan blue exclusion, and ≈80-90% living cells remained in the medium containing of 100 μg mL −1 of 2a (Supporting Information File 1, Figure S8).
The cytocidal effect of the radicals released from compound 2a on LLC cells was also tested. Four hours after 1 min exposure to 360 nm light in various concentrations of 2a-containing medium, the number of living cells decreased in a 2a concentration-dependent manner (Supporting Information File 1, Figure  S9). After exposure of the cells in the medium containing 100 μg mL −1 of 2a, the number of living cells decreased significantly compared to that without exposure (66.5% vs 87.8%, Supporting Information File 1, Figure S10). An irradiationtime-dependent decline in the viability of the LLC cells was also observed ( Figure 5). To evaluate whether the cytocidal effect was due to photochemical radical generation, cells exposed to 360 nm light for 1 min and the unexposed congeners were stained by using a ROS-ID oxidative stress detection kit (Enzo Life Sciences, Farmingdale, NY, U.S.A.). Reactive oxygen species (ROS) were detected in the cells irradiated in the 2a-containing medium, but not in the non-irradiated cells in  2a-containing medium or the irradiated cells without 2a-containing medium ( Figure 6). Thus, the preliminary analyses indicated that the photochemical generation of radicals from 2a induced cancer cell death in vitro, although no in vivo study was performed because of the low water solubility of 2a. At this point, we cannot rule out generate of ROS by photosensitization of the chromophore in the presence of O 2 for the cytotoxicity.

Conclusion
In the present study, novel caged nitroxides 2a and 2b having a TP-responsive chromophore were synthesized, and OP-and TP-induced generation of the TEMPO radical with these species was examined. The quantum yields for generation of the TEMPO radical from 2a and 2b were determined to be 2.5% and 0.8% in benzene, respectively. The quantum yields in DMSO were found to be higher than those in benzene, 13.1% and 12.8%, respectively. The OP-uncaging efficiency (ε × Φ) was found to be 480 and 175 for 2a and 2b, respectively, at 360 ± 10 nm, in benzene, and 3026 and 2995 in DMSO, respectively. The TP efficiency of the TEMPO uncaging reaction was found to be 1.1 GM at 740 nm for 2a and 0.22 GM at 730 nm for 2b in benzene. The TP-induced clean release of the TEMPO radical is expected to be applicable to further physiological studies and site-selective polymerization reactions.

Experimental
All reagents were purchased from commercial sources and were used without additional purification, unless otherwise mentioned. Caged nitroxides 2a and 2b were prepared according to the methods described previously (Scheme 2) and were isolated by silica gel column chromatography and GPC. 1 H and 13 C NMR spectra were reported in parts per million (δ) by using CDCl 3 . IR spectra were recorded with a FTIR spectrometer. UV-vis spectra were taken by a SHIMADZU UV-3600 Plus spectrophotometer. Mass spectra were measured by a Mass Spectrometric Thermo Fisher Scientific LTQ Orbitrap XL, performed by the Natural Science Center for Basic Research and Development (N-BARD), Hiroshima University.

Photoirradiation of LLC cells with compound 2a
One hundred thousand Lewis lung carcinoma (LLC) cells were seeded into a 24-well plate (medium: DMEM) and incubated overnight at 37 °C in an atmosphere of 95% air and 5% CO 2 . The medium was replaced with fresh phenol-red free DMEM containing 100 µg/mL of compound 2a. Four hours after various irradiation time of 360 nm light (0, 10, 30, 60, 90, 120, and 150 s) using a fluorescence microscope (BIOREVO BZ-9000, Keyence, Osaka, Japan), cell viability was determined by trypan blue exclusion. Bars represent the mean ± standard deviation (n = 4).

Detection of intracellular ROS in irradiated LLC cells with 2a-containing medium
Fifty thousand Lewis lung carcinoma (LLC) cells were seeded into 24-well plate (medium: DMEM) and incubated overnight at 37 °C in an atmosphere of 95% air and 5% CO 2 . The medium was replaced with fresh phenol-red free DMEM containing 0 or 100 µg/mL of compound 2a. Thirty minutes after 1 min or no exposure of 360 nm light using a fluorescence microscope (BIOREVO BZ-9000, Keyence, Osaka, Japan), intracellular ROS were detected using the ROS-ID Oxidative Stress Detec-