Singlet Oxygen Generation from Polyaminoglycerol by Spin-Flip-Based Electron Transfer.

Reactive oxygen species have drawn attention owing to their strong oxidation ability. In particular, the singlet oxygen (1O2) produced by energy transfer is the predominant species for controlling oxidation reactions efficiently. However, conventional 1O2 generators, which rely on enhanced energy transfer, frequently suffer from poor solubility, low stability, and low biocompatibility. Herein, we introduce a hyperbranched aliphatic polyaminoglycerol (hPAG) as a 1O2 generator, which relies on spin-flip-based electron transfer. The coexistence of a lone pair electron on the nitrogen atom and a hydrogen-bonding donor (the protonated form of nitrogen and hydroxyl group) affords proximity between hPAG and O2. Subsequent direct electron transfer after photo-irradiation induces hPAG•+-O2 •- formation, and the following spin-flip process generates 1O2. The spin-flip-based electron transfer pathway is analyzed by a series of photophysical, electrochemical, and computational studies. The 1O2 generator, hPAG, is successfully employed in photodynamic therapy and as an antimicrobial reagent.

Introduction

Photosensitizer (PS)-mediated reactive oxygen species (ROS) generation has drawn significant attention owing to its extensive applications in various avenues such as photodynamic therapy, wastewater purification, antimicrobials, and photocatalytic organic synthesis.[1−8] Generally, ROS can be generated via two pathways: electron transfer (type I) or energy transfer (type II). Most oxygen radical anions, such as the superoxide radical (O2•–) and hydroxyl radical (OH•), are generated through electron transfer from an electron donor, whereas singlet oxygen (1O2) is generated through energy transfer from the triplet state of a photosensitizer.[6,9−12]1O2 is a predominant species among the ROS and is a powerful oxidant.[8,12] To accomplish efficient 1O2 generation, effective intersystem crossing (ISC) from singlet state to triplet state must be realized.[13,14] The ISC can be enhanced by increasing the coupling intensity (T|ĤSO|S1) or decreasing the energy gap (ΔES) between the S1 and T states. Therefore, strategies based on the utilization of heavy atoms to increase the spin–orbit coupling constant and on intramolecular donor–acceptor charge separation to reduce ΔES have been explored.[14−17] However, photosensitizers based on these principles suffer from dark toxicity, poor solubility, and low stability. In addition, the energy transfer is strongly dependent on the distance between the donor and the acceptor, rendering the estimation of its efficiency highly challenging due to the mobility of O2 (Scheme a).[18,19]

Scheme 1

Design Strategy and Mechanism of Electron Transfer-Mediated 1O2 Generation by Photosensitizers

(a) 1O2 generation based on “energy transfer” in conventional photosensitizers (PSs). (b) 1O2 generation based on “spin-flip-based electron transfer” in rationally designed polyaminoglycerol (hPAG). Nonbonding lone pair of hPAG (blue) interacts with O2 (red) within ∼3 Å. Simultaneously, hydrogen-bonding donors of hPAG increase the affinity to O2.

Herein, we report a molecular design strategy for 1O2 generation via spin-flip-based electron transfer. This pathway harnesses the nitrogen lone pair–O2–hydrogen-bonding interaction to minimize the distance between the photosensitizer and O2 (Scheme b). Initially, we chose an amine group as the target functional moiety because of the interaction between the π* orbitals of O2 and the lone pair of N, which facilitates the chemical adsorption of O2.[20,21] This is also the first step in the activation of O2 in the oxygen reduction reaction (ORR).[20,21] Furthermore, we considered the hydrogen bonding with protonated amine and hydroxyl groups (−NH2+– and −OH) as the additional factor for the strong affinity with O2. Taken together, we developed a polyaminoglycerol (hPAG) as a 1O2 generator, which includes the N atom within the framework of the highly biocompatible hyperbranched polyglycerol (hPG) molecule. As a result, we found that O2 was encapsulated within ∼3 Å in the polymer network by the virtue of the (i) interaction between O2 and the lone pair of N and (ii) hydrogen bonding (Scheme b). The polymer network allows direct electron transfer from the polymer to O2 upon photo-irradiation, suggesting spin-flip-based electron transfer mechanism for 1O2 generation. This is completely different from the conventional strategies for developing triplet photosensitizers (Scheme ). The ROS generated was investigated by various methods and successfully utilized to exert oxidative stress, triggering photo-induced cancer cell death and antimicrobial effect. This approach provides new insights into the design strategy for improving the dark toxicity, water solubility, and stability of photosensitizers.

Results and Discussion

Characterization of Synthesized hPAG

We first prepared Boc-protected aminoethanol glycidyl ether monomer (BAG) to synthesize hyperbranched hPAG, based on hPG backbone.[22] Owing to the unique three-dimensional architecture of hyperbranched hPG, which comprises a polyether backbone with several hydroxyl functional groups and exhibits excellent biocompatibility, it has gained significant attention in biological and biomedical applications. The Boc-protected hPAG was prepared by anionic ring-opening multibranching polymerization using trimethylolpropane as an initiator in the presence of potassium methoxide (Figure S1). Then, deprotection was carried out with 1 M HCl to yield the desired hPAGs. To examine the ROS generation capability as a function of the number of nitrogen atoms in the polymeric backbone, the molecular weight of each polymer was controlled by adjusting the monomer-to-initiator ratio (hPAG, 10,500 g/mol; small hPAG (s-hPAG), 2,300 g/mol) (Figure a, bottom and Figure S2). Parallelly, the control hyperbranched hPG containing only O atoms in the backbone was prepared by following an identical protocol using glycidol as a monomer (Figure a, top and Figure S2). The hyperbranched architecture of the synthesized polymers was confirmed, with the degree of branching being 0.40–0.55 (Figure S3). The characterization data and details of all polymers including their molecular weight, degree of branching, and photostability are provided in the Supporting Information (Figures S2–S4).

Figure 1

Identification of 1O2 generation and its pH dependency. (a) Chemical structure of hyperbranched polyglycerol (hPG; top) and polyaminoglycerol (hPAG; bottom) interacting with O2. (b) Validation of 1O2 generation by utilizing 9,10-anthracenediyl-bis(methylene) dimalonic acid (ABDA). Absorbance of ABDA is attenuated after the reaction with 1O2. Conditions [hPG or hPAG] = 10 μM, [ABDA] = 100 μM, irradiation power = 12.5 mW/cm2. (c) Tracking the 1O2 generation by hPAG at different pHs. The pH was adjusted using 1 mM NaOH and HCl. Conditions: [hPAG] = 10 mM, [ABDA] = 100 mM, light source = solar simulator, irradiation power = 25 mW/cm2.

The photophysical properties of hPG, s-hPAG, and hPAG were analyzed by ultraviolet–visible (UV–vis) (Figure S5) and photoluminescence (PL) spectroscopy (Figure S6). Interestingly, s-hPAG and hPAG exhibited an absorption band in the range of 250–450 nm (approximately 2.8–4.9 eV), whereas hPG exhibited no absorption in the same range. hPAG, with a higher molecular weight, exhibited stronger absorption than the corresponding s-hPAG, which implied that this absorption originated from the polymer network of hPAG. Moreover, the forbidden phosphorescence (PL) emission of the heavy-atom-free hPAG polymer could be observed at a low temperature of 77 K, which suppresses the nonradiative process by establishing an environment in which molecular motion is frozen. The following phosphorescence of hPAG was observed in the range of 400–500 nm (Figure S6), while no significant emission was observed in the same range of hPG (Figure S7). In addition, the phosphorescence of hPAG was quenched under O2 bubbling compared to Ar bubbling (Figure S7a). Quenching by O2 implies that the low-temperature PL emission of hPAG is derived from the triplet state. In contrast, no change of emission intensity was observed in the PL spectra of hPG under O2 bubbling (Figure S7b).

This intrinsic optical property might originate from the nanocluster formed through hydrogen bond or electrostatic interactions in the branched polymer network.[23,24] In addition, the pH-titration curve of hPAG revealed the protonation behavior of hPAG in an aqueous solution (Figure S8). The titration curve of hPAG shows buffered region from polymer network showing variable pKa values for secondary amine-like branched PEI.[25−27] This behavior is attributed to the hydrogen bonding between nitrogen lone pair and hydroxyl group or protonated amine in the hyperbranched hPAG network. Based on the results, it can be concluded that the optical properties of hPAG, endowed by the interactions in the polymer network, facilitate the utilization of its excited state, which can initiate photo-induced electron transfer from the lone pair of N to an appropriate target receiving its electron near the polymer. Considering various examples, where the N atom in PEI interacts with the O atom of a DNA strand or where N atom insertion on the edge of graphene enables O2 adsorption, the optical properties of hPAG, engendered by the hyperbranched polymer architecture with an additional amine group, have prompted us to investigate the interaction of hPAG with O2 and the following electron transfer to produce ROS after photo-activation.

Investigation of ROS Generation

We investigated the ROS generation ability of the compounds using three representative methods: dihydrorhodamine 123 (DHR123) assay for type I ROS,[28,29] 9,10-anthracenediyl-bis(methylene) dimalonic acid (ABDA) assay for 1O2,[28,30] and electron paramagnetic resonance (EPR) spectroscopy with spin trapping.[32−34] The enhanced emission from DHR123 after photo-activation supports type I ROS generation by electron transfer, with yields decreasing in the order hPAG > s-hPAG > hPG (Figure S9). To confirm the generation of superoxide radical (O2•–), we obtained EPR signals using 5,5-dimethyl-1-pyrroline N-oxide (DMPO), which exhibits a characteristic spectral change depending on the specific chemical modification induced by the ROS (Figure S10).[31,32] Only the characteristic EPR signal of DMPO-OOH was observed, providing clear evidence of O2•– generation. The concentration of O2•– was also dependent on the photo-irradiation time, ensuring that hPAG generated O2•– via electron transfer (Figure S10). Moreover, OH• was not identified in hydroxyphenyl fluorescein (HPF) assay (Figure S11).

Additionally, the attenuated absorbance of ABDA was investigated to identify type II ROS (1O2) generated upon photo-irradiation of the polymer. While there was negligible 1O2 generation from hPG, the absorbance of ABDA with hPAG was remarkably attenuated compared to that of s-hPAG (Figures b and S12b). Control experiments with diethanolamine (a fragment of hPAG) and BAG (monomer of hPAG) did not show significant changes in ABDA absorbance (Figure S13) relative to that of hPAG, and this establishes the importance of polymer networks for photo-induced ROS generation. The addition of NaN3 as a 1O2 scavenger successfully quenched the 1O2 generated by hPAG, thus, changes in the absorbance of ABDA did not occur (Figure S12).[33] Moreover, we monitored the absorbance of ABDA using photo-activated hPAG at various pH because we assumed that the protonation state of a polymer could influence its interaction with O2. The 1O2 generation capability increased as the pH decreased from 11 to 7 to 3, and we correlated this to the enhanced interactions with O2 upon the creation of positive charge and formation of hydrogen bond donor–protonated amine (−NH2+−) species under acidic conditions (Figure c). We further analyzed the 1O2 generation from hPAG using 2,2,6,6-tetramethylpiperidine (TEMP) by an EPR spin-trap method (Figure S14).[34] The generation of 1O2 from hPAG is surprising because hPAG does not possess the appropriate energy bandgap (ca. 2.21 eV) to produce 1O2 via the conventional energy transfer mechanism (type II) (Figure S6).[29] This suggests the possibility of an alternative mode for 1O2 generation through O2•– via a sequential electron transfer cascade for O2 (type I).[35,36] The 1O2 generation pathway through electron transfer and further mechanistic analyses on hPAG and hPG are assessed in detail in later sections.

Electron Transfer from hPAG to Molecular Oxygen (O2)

Cyclic voltammetry (CV) was used to examine the electron transfer from hPAG to produce O2•– using ferrocene as the external standard (+0.63 V vs NHE) (Figures and S15).[37−39] To identify the ROS generated through electron transfer, current changes from hPAG and hPG under O2 or N2 bubbling were analyzed (Figure a). In contrast to the CV spectrum of hPAG under saturated N2 conditions, new peaks at +0.94 and −0.87 V (vs NHE) appeared under saturated O2 conditions under a negative sweep (reduction followed by oxidation). The potential for O2•– generation ranges from 0 to −1.0 V (vs NHE), depending on the electron donor.[37] Therefore, the peak at −0.87 V (marked with an asterisk) was consistent with the generation of O2•– from hPAG (Figure a, left). The other new peak at +0.94 V (double asterisks) indicates the conversion of O2•– to H2O2 (Figure a, right).[38,39] However, this peak did not appear under a positive sweep (oxidation followed by reduction) (Figure S16). This indicates that O2•– generation through reductive potential sweep must occur first for the conversion to H2O2. In addition, potential sweep with photo-irradiation resulted in much clearer peaks and higher current peaks at −0.87 and +0.94 V than that in the case without photo-irradiation, supporting the photo-activated generation of ROS (Figure b). No distinguishable current peak was observed under N2 bubbling. Furthermore, we monitored the current recovery of the peak at +0.94 V under cyclic bubbling of O2 and N2; the current was found to be significantly altered for hPAG (Figures c and S17a), whereas hPG did not exhibit any change (Figures c and S17b). Diethanolamine, a simple fragment of hPAG, exhibited a slight current change under O2 bubbling; however, its magnitude was considerably smaller than that of hPAG (Figure S17c). This result demonstrates that the electron transfer capability of hPAG is highly affected by the presence of O2 and polymer network.

Figure 2

Evidence of electron transfer from the polymer to O2. (a) Cyclic voltammetry (CV) curves of the aqueous polymer solution saturated with N2 or O2. Asterisks indicate the two different species of ROS generated by electron transfer. Current at +0.94 and −0.87 V (vs NHE) corresponds to O2•– → H2O2 (right) and O2 → O2•– (left), respectively. (b) CV analysis of the effect of photo-irradiation on hPAG under different atmospheres. (c) Recording of peak oxidative current at +0.94 V upon cyclic bubbling of O2 and N2. Conditions: bubbling time = 20 min, scan rate = 50 mV/s, counter electrode = Pt, reference = Ag/AgCl (std. KCl), working electrode = polymer-coated FTO, and electrolyte = 0.1 M t-butylammonium hexafluorophosphate (TBA+PF6–).

Density Functional Theory (DFT) Calculations for Establishing the Plausible Mechanism of 1O2 Generation

To further understand the mechanism of O2•– and 1O2 generation, we performed density functional theory (DFT) calculations for the natural transition orbitals (NTOs) of hPG-O2 and hPAG-O2 to clarify the electronic transitions.[40] The NTOs indicate that a hole orbital with p-orbital character localizes on an O atom for hPG and on a N atom for hPAG, while an electron orbital exhibits π* character in O2 (Figure a,b). This indicates that the transition occurs from the O atom of hPG or the N atom of hPAG to O2, further supporting ROS generation.

Figure 3

Density functional theory (DFT) calculations for analyzing 1O2 generation properties of hPG and hPAG via spin-flip-based electron transfer. (a, b) Natural transition orbitals (NTOs) of (a) hPG-O2 and (b) hPAG-O2 at T0 optimized geometry. Carbon, hydrogen, oxygen, and nitrogen in the molecular structure correspond to black, gray, red, and blue. (c) Relative energy diagram for singlet and triplet states in the minimum-energy structure of hPG-O2 and hPAG-O2 complexes at the triplet ground state (T0) by mixed-reference spin-flip time-dependent density functional theory (MR-SF-TDDFT). The black Gaussian curve describes the incident photon energy band (λ = 360 nm with a bandwidth of 100 nm). (d) Process of 1O2 generation by photo-activation. S1 and S2 states are degenerate, corresponding to the doubly degenerate singlet oxygen state (1Δg). Singlet states (S1/S2) for both complexes are 0.66 eV above T0, almost identical to the T0 and S1 energy differences for isolated oxygen. The energy gap between S3 and S1/S2 is 0.69 eV, which coincides with the energy gap between 1Δg and 1Σg+ of isolated oxygen, implying that the S3 state corresponds to the 1Σg+ state of isolated oxygen. (e) Schematic illustration of molecular orbital configurations and overall transition for the singlet and triplet energy states. Blue spin represents the n lone pair electron from the nitrogen of hPAG. Red spin represents the electron from the π* antibonding orbital of O2.

Mixed-reference spin-flip time-dependent density functional theory (MR-SF-TDDFT) calculations were performed for a detailed photodynamic analysis of hPG and hPAG with O2 (Figures c–e, S18, and Table S1).[41,42] The ground electronic states of hPG-O2 and hPAG-O2 have the T0 state, where each π* orbital of O2 is singly occupied (Figure c,e). The two lowest triplet excited states (T1/T2) were nearly degenerate, corresponding to the charge transfer excitation from hPG/hPAG to the respective two π* orbitals of O2 (hPG/hPAG•+-O2•–) (Figure c–e). Surprisingly, the T0 → T1/T2 transition energies for hPG and hPAG were substantially different (4.91 eV (252 nm) for hPG-O2 and 3.86 eV (321 nm) for hPAG-O2) and were consistent with the UV–vis absorption energy (Figure S5). The difference in excitation energies originates from the different orbital energies of the localized state of hPG and hPAG because electrons of the nitrogen atoms exhibit weaker electron–nuclear interactions than the electrons of the oxygen atoms, resulting in efficient electron transfer from the nitrogen atoms to O2. In addition, hPAG exhibited a more stable oxidized state than hPG (ΔG°hPG – ΔG°hPAG ≈ 20 kcal/mol), in agreement with the above-mentioned transition energy difference (4.91 vs 3.86 eV) (Figure S19).

Moreover, the singlet states of hPAG-O2 complexes generated after the T0 → T1/T2 excitation were further evaluated to obtain mechanistic insights into this process (Figure d–e and Table S1). The S4/S5 states of the hPAG-O2 complex are nearly degenerate, and their energies (3.89 eV) are very close to the T1/T2 energies (3.86 eV). Therefore, we can expect efficient intersystem crossing, which, in turn, enhances the spin-flip process to form the S3/S4 states after the triplet excitations (T1/T2, hPAG•+-O2•–). Subsequently, these S4/S5 states undergo transition to the S3 state (hPAG-1O2(1Σg)) via internal conversion, followed by the generation of S1/S2 states (hPAG-1O2(1Δg)) (Figure d). Finally,1O2 is released in hPAG-1O2(1Δg) state for a favorable energy state (Figure S20). The detailed electronic configurations of hPAG-O2 are shown in Figure e.

Finally, to explain the origin of the improved ROS generation yield of hPAG under acidic conditions, we considered the optimized geometries of hPAGH+-O2-hPAG and hPAG-O2-hPAG, where hPAGH+ indicates hPAG bearing protonated nitrogen atoms (Figure and Table S2). The H–O2–H configuration depicts interacting O2 intercalated between the protonated amines in hPAGH+ and hydrogen of −NH– in hPAG (Figure a, left), while the N–O2–H configuration depicts the interaction between the protonated amine in hPAGH+ and the N lone pair in −NH– in hPAG (Figure a, right). In the hPAGH+–O2–hPAG network, the structure of N–O2–H is more stable than that of H–O2–H by −0.045 eV (Table S2). The hPAG-O2-hPAG configurations are nearly degenerate, while the H–O2–N configuration is the lowest-energy structure, with an energy of −0.005 eV (Figure b and Table S2). The complex system containing one hPAGH+ has stronger binding with O2 than that containing only hPAGs (Table S2). In the optimized geometry, moreover, the distance between the polymer backbone and O2 of hPAGH+-O2-hPAG is 3.21 Å, which is shorter than that of hPAG-O2-hPAG (3.44 Å) (Figure ; magnified region). These results suggest that the origin of the improved ROS generation lies in the strong binding with O2 under acidic conditions.

Figure 4

Close O2 binding in the protonated hPAG network. (a) Molecular configurations of H–O2–H and N–O2–H in hPAGH+ and hPAG, respectively. (b) Molecular configurations of H–O2–H, N–O2–N, and H–O2–N in two hPAG fragments. The magnified structures in (a) and (b) show the lowest-energy structures of the hPAGH+-O2-hPAG and hPAG-O2-hPAG configurations and interaction distance between O2 and hPAG.

Suggested 1O2 Generation Mechanism via Spin-Flip-Based Electron Transfer

A plausible mechanism for 1O2 generation by the electron transfer cascade in the hPAG network with O2 is depicted in Figure . (1) The nitrogen (N) of the hPAG polymer first interacts with the ground state oxygen under ambient conditions. (2) The photo-irradiation then induces charge separation in the network leading to the hPAG•+-O2•– conformation, a critical factor in the ROS generation. (3) The subsequent spin-flip process and primary/secondary internal conversions trigger the O2•–-to-1O2 conversion. (4) 1O2 is finally generated from the hPAG–1O2 complex by relaxation.

Figure 5

Plausible mechanism of 1O2 generation via spin-flip-based electron transfer between hPAG and O2.

Photo-Induced Cancer Cell Death and Antimicrobial Activity by 1O2 Generation

Type I ROS assay with 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA) was performed to identify whether the ROS produced by the suggested mechanism can be applied in the intracellular environment.[43] Fluorescence enhancement after photo-irradiation was substantial for hPAG, whereas the change in fluorescence was negligible for hPG (Figure S21). Furthermore, an intracellular superoxide radical assay revealed stronger red fluorescence from hPAG than that from hPG upon photo-irradiation (Figure a). These results are in good agreement with those of the in vitro ROS assays.

Figure 6

Intracellular O2•– and 1O2 generation by spin-flip-based electron transfer and therapeutic applications. (a) Dihydroethidium (DHE) assay for O2•– and the fluorescence signal quantification for each condition (n = 9; three regions of interest (ROIs) from three different images). Scale bar = 40 μm. Data are presented as mean ± standard deviation, *p < 0.05, **p < 0.01. (b) Live/dead assay with Calcein-AM and propidium iodide (PI). Note that the green signals represent live cells, and the red signals represent dead cells. Scale bar = 100 μm. (c) Cell cytotoxicity of hPG and hPAG with or without photo-irradiation. Data are presented as mean ± standard deviation (n = 4). Conditions: 405 nm photoreactor, irradiation time = 600 s, light intensity = 28 mW/cm2, total energy = 16.8 J/cm2, and incubation time = 24 h. (d) Antimicrobial experiment of hPG and hPAG with or without photo-irradiation. (e) Colony visualization and quantification from antimicrobial experiment. Conditions: solar simulator irradiation time = 600 s, light intensity = 100 mW/cm2, total energy = 60 J/cm2, and incubation time for the colony formation unit (CFU) assay = 24 h. Data are presented as mean ± standard deviation (n = 3), *** p < 0.005, and **** p < 0.0005.

In addition, cell death by photo-induced ROS was visualized by the live/dead and MTT-assays (Figure b,c). In the live/dead assay, red signals from dead cells were significant in the hPAG-treated cells upon photo-irradiation; however, the signal was not observed in the hPG-treated cells and negative control (Figure b). Additionally, we quantified the cell viability upon photo-irradiation for 10 min (light dose = 16.8 J/cm2 with 405 nm LED array of 28 mW/cm2 intensity; 60 J/cm2 with solar simulator of 100 mW/cm2) via the MTT assay (Figures c, S22, and Table S3). The viability of HeLa cells treated with hPAG (≥30 μg/mL) significantly decreased under light irradiation, while hPG did not exhibit a considerable change in cell viability.

Furthermore, the antimicrobial effect was assessed to confirm the 1O2 generation by photo-activated hPAG (Figure d).[44] The bacterial growth, monitored by optical density measurement at 600 nm (OD600), was significantly decelerated with hPAG under photo-irradiation. It was also visualized as colonies on an agar plate. The number of colonies with hPAG was less compared to that with hPG, and its quantification indicated that the ROS generation by hPAG could impede bacterial growth (Figure e).

Collectively, these results indicate that 1O2 generation using this molecular design strategy based on O2 interaction in the polymer network could be potentially applied to photodynamic therapy and antimicrobial effect.

Conclusions

In summary, we report a new 1O2 generation process by the spin-flip-based electron transfer from an N-containing aliphatic polyglycerol-based polymer (hPAG) to O2. The main 1O2 generation process is related to the hyperbranched polymer network and a nitrogen atom, which strengthens O2 interaction and stabilizes hPAG•+-O2•– (T1/T2 state). hPAG showed significant biocompatibility under dark conditions and induced cell death upon photo-irradiation. These results suggest that the ROS generation capability of hPAG via spin-flip-based electron transfer could induce the photo-dependent death of cancer cells and antimicrobial effect. The developed molecular design strategy for biocompatible and O2-interactive photosensitizer is expected to contribute to the advancement of a new type of 1O2 generator, and it might break the existing limitations of previous hydrophobic photosensitizers with a narrow range of applications.