Carbon dot-assisted luminescence of singlet oxygen: the generation dynamics but not the cumulative amount of singlet oxygen is responsible for the photodynamic therapy efficacy

Abstract

A novel carbon dot-based luminescence probe for singlet oxygen (¹O₂) with a conventional optical detector has been implemented through the specific formation of electronically excited carbonyls from the breakdown of unstable endoperoxide intermediates, and its application in the real-time in vivo monitoring of ¹O₂ in photodynamic therapy (PDT) is achieved. More attractively, the relationship between the dynamics details of photosensitizer-generated ¹O₂ and the PDT efficacy has been established through a modified multiple-target survival model, enabling a direct and easy estimate of the surviving fraction of tumor cells from the generation dynamics of ¹O₂. Both in vitro and in vivo therapy results revealed that the rapid generation dynamics of ¹O₂ rather than its cumulative amount is responsible for better treatment efficacy in PDT. Overall, the deeper insight into the important roles of the generation dynamics of ¹O₂ in the PDT efficacy is irreplaceably advantageous in substantially reduced risks from deleterious treatment-related side effects by screening advanced photosensitizers and determining the light exposure end point.

---

New concepts

The real-time monitoring of singlet oxygen (¹O₂) generation in photodynamic therapy (PDT) represents a promising direction in the administration of a prescribed quantity of photosensitizer and light. However, the design of a specific probe for the real-time in vivo monitoring of ¹O₂ generation still remains a critical and unsolved challenge as it is ensured that the probe is easily-prepared, stable, biocompatible, water-soluble, and cell-permeable. In this contribution, we report a novel luminescence light-up nanoprobe for the real-time in vivo monitoring of ¹O₂ generation using imidazole-functionalized carbon dots. This work has provided solid evidence that the generation dynamics of ¹O₂ rather than its cumulative amount is a promising evaluation standard for the PDT efficacy. We believe that our success will be irreplaceably advantageous in substantially reduced risks from deleterious treatment-related side effects by screening advanced photosensitizers. This work also motivates scientists to take the role of functionalized carbon dots into account in the evaluation of ¹O₂ generation-related life processes.

Introduction

Photodynamic therapy (PDT) has received great attention in both fundamental research and cancer treatment over the past decades thanks to its high spatiotemporal selectivity and minimal invasiveness.^1–4 Upon light irradiation, photosensitizers employ tissue oxygen to generate highly cytotoxic singlet oxygen (¹O₂), which subsequently kills diseased cells through apoptosis or necrosis.^5–7 A standard clinical PDT protocol does not only rely on sufficient production of ¹O₂, but also on the administration of a prescribed quantity of photosensitizer and light.⁸ However, highly active photosensitizers and high-power light in photodynamic action can damage healthy cells and compromise therapeutic efficacy.⁹ Therefore, how to monitor the photosensitizer working in the body and how to evaluate the exact treatment end point are important. The real-time in vivo monitoring of ¹O₂ generation represents a promising direction in screening advanced photosensitizers and determining the light exposure end point in order to reduce the risks from deleterious treatment-related side effects. To address this problem, it is generally recognized that chemiluminescence (CL) probes are an ideal candidate for the real-time in vivo monitoring of ¹O₂ generation.¹⁰ However, to the best of our knowledge, a highly efficient CL probe of ¹O₂ meeting these requirements still remains an unsolved challenge.

A direct approach is to measure ¹O₂ generation in vivo by detecting the near-IR luminescence of ¹O₂ at a wavelength of 1270 nm owing to its inherent disadvantages (e.g., insufficient sensitivity and unconventional detector).¹¹ To overcome these hurdles, some strategies have been developed for the real-time monitoring of ¹O₂ generation. Almost all of the CL probes for ¹O₂ measurement are based on the strong enophile nature of ¹O₂ toward [2+2] or [2+4] cycloaddition reactions with double-bond structures.^10,12–14 However, the quenching of luminescence emission in water limits their practical applications in vivo. In addition, the current established CL probe-based fluoresceinyl cypridina luciferin analog (FCLA)¹² strategies suffer from cross-interference caused by other reactive oxygen species in biological systems, and their solutions are thermally unstable and must be stored at −80 °C. Excitingly, Shabat and co-workers recently pioneered a highly selective and sensitive CL probe for the detection of ¹O₂ through the formation of a new dioxetane luminophore.¹⁰ Unfortunately, such a developed CL probe could not be used for the detection of intracellular ¹O₂ before it was functionalized with a cell-penetrating peptide by a three-step synthesis. Accordingly, the complicated and time-consuming synthetic processes and high instability made it impractical and inconvenient for the real-time in vivo monitoring of ¹O₂. Furthermore, we have previously developed a CL method to monitor ¹O₂ generation using the tetraphenylethene-sodium dodecyl sulfonate surfactant with aggregation-induced emission characteristics.¹⁴ Disappointingly, in vitro and in vivo use of this probe in the detection of ¹O₂ faced an insurmountable obstacle because of its localization onto the cell membrane through hydrophobic interactions. Taking into account these facts, it is clearly understood that the design of a specific probe for the real-time in vivo monitoring of ¹O₂ generation is technically challenging as it has to be ensured that the probe is easily-prepared, stable, biocompatible, water-soluble, and cell-permeable.

Carbon dots (CDs) have been utilized as versatile scaffolds for sensing subcellular biomolecules due to their water solubility, high biocompatibility, excellent cell-permeability and good photostability.^15–17 It is generally recognized that the imidazole ring can specifically react with ¹O₂ to form an excited intermediate, which further emits strong luminescence when it is incorporated into conjugated structures.^18,19 In this contribution, we designed and synthesized a kind of imidazole-functionalized CDs (Im-CDs) by a one-step hydrothermal carbonization method using histidine as the sole precursor. Interestingly, the CL of the Im-CDs could be selectively triggered by ¹O₂ through the formation of electronically excited carbonyls from the breakdown of unstable endoperoxide intermediates (Fig. 1). The relationship between the generation dynamics of ¹O₂ and the PDT efficacy in vitro and in vivo was explored. It was found that the surviving fraction datum as a function of the generation dynamics of ¹O₂ were fitted much better with a single hit multiple target cell survival model (reduced χ² = 3.22), which can be used for a direct estimate of the surviving fraction of tumor cells from the generation dynamics of ¹O₂. The results revealed that the generation dynamics of ¹O₂ but not its cumulative amount plays a vital role in the PDT efficacy.

Fig. 1 Possible mechanism of the Im-CD-based luminescence probe for real-time monitoring of ¹O₂.

Results and discussion

The as-prepared Im-CDs exhibited excellent optical characteristics and good water-solubility (Fig. S1–S3 and Table S1, ESI†). Strong emission of the Im-CDs was observed at 402 nm under near ultraviolet 331 nm excitation (Fig. 2A). In the FT-IR spectra of the Im-CDs (Fig. 2B), the peaks in the range of 1400–1520 cm⁻¹ were assigned to the vibration of the imidazole backbone, and the peaks at 2855 cm⁻¹ and 2960 cm⁻¹ were attributed to the vibration of the alkyl side chain of the imidazole ring.²⁰ The FT-IR spectra confirmed the existence of imidazole groups on the surface of the Im-CDs. Moreover, the main contribution of the pyrrolic nitrogen (400.1 eV) in the XPS N 1s spectrum also revealed that the imidazole groups were embedded in the Im-CDs (Fig. 2C).^21,22 The Raman spectrum and XRD pattern of the Im-CDs further identified the presence of graphitic carbon, indicating the conjugated structures in the Im-CDs (Fig. 2D and Fig. S4, ESI†).²³ In conclusions, Im-CDs with conjugated structures were successfully synthesized.

Fig. 2 (A) Excitation–emission matrices of the Im-CDs. (B) FT-IR spectra of histidine and the Im-CDs, respectively. (C) XPS N 1s spectrum of the Im-CDs. (D) Raman spectrum of the Im-CDs.

The ¹O₂-triggered CL emission of the Im-CDs was investigated in a static injection CL setup (Fig. S5, ESI†). The results showed that strong CL emission of the Im-CDs was observed in the presence of ¹O₂ from the H₂O₂–IO₄⁻ system²⁴ in 50 mM phosphate buffer solution (PBS, pH 7.4) (Fig. S6, ESI†). The strong and fast CL emission is shown in the inset of Fig. S6 (ESI†). Moreover, the addition of 50 mM PBS (pH 7.4) had no influence on the CL emission of the Im-CD-¹O₂ system (Fig. S7, ESI†). In comparison, very weak emission was detected in the reaction of ¹O₂ with histidine, indicating that the conjugated structures of the CDs contributed to a significant enhancement of the CL emission. Moreover, serine-CDs were synthesized by using serine as the raw material. The excitation-dependent photoluminescence behavior of the serine-CDs is shown in Fig. S8 (ESI†). The reason for the behavior was that the different kinds of functional groups (e.g., C–OH, C–O, C–O–C, and C–H) on the surface of the CDs could lead to various energy levels and a series of emissive traps.²⁵ It was found that weak emission was aroused by the reaction between ¹O₂ and serine-CDs in the absence of imidazole groups in the control measurements. The results implied that the CL differences were attributed to the presence of the imidazole groups. Therefore, the imidazole groups in the conjugated structures of the CDs played a key role in the strong CL emission. As shown in Fig. S9 (ESI†), the emission of the Im-CD-¹O₂ CL system remained unchanged when the concentration of Im-CDs was higher than 0.5 mg mL⁻¹. 0.5 mg mL⁻¹ Im-CDs were utilized in the following measurements, and the optimum synthesis temperature was 240 °C (Fig. S10, ESI†). The as-prepared Im-CDs at both lower and higher synthesis temperature could induce a decrease in the CL intensity. The reason for this phenomenon was that a higher synthesis temperature could lead to a higher carbonization degree and a lower amount of functional groups on the surface of the CDs; on the contrary, a lower synthesis temperature could result in a lower carbonization degree and fewer conjugate structures of the CDs.²⁶ Furthermore, the CL emission of the Im-CDs upon the reaction with ¹O₂ lay in the band 400 to 630 nm (Fig. 3A, inset), suggesting that the emitting species was the electronically excited carbonyls from the [2+4] cycloaddition reaction between ¹O₂ and imidazole groups in the CDs.^19,20 On the other hand, the formation of carbonyl groups in the Im-CDs was further confirmed by the XPS C 1s spectrum. The XPS C 1s spectra for the pristine CDs and ¹O₂-treated CDs are shown in Fig. S11 (ESI†). Remarkably, the increased content of C=O (Table S2, ESI†) indicated the formation of carbonyl groups on the surface of the CDs.²² Meanwhile, the decreased content of C=C suggested that the double-bond structures of the Im-CDs were broken when they were attacked by ¹O₂. As expected, such carbonyl groups in the CDs were generated in the presence of ¹O₂, and the CL emission of the Im-CDs was aroused (Fig. 1). Therefore, we concluded that the strong CL emission was from the electronically excited carbonyls on the surface of the Im-CDs.

Fig. 3 (A) CL intensity of the Im-CDs in the presence of different ROS. (B) CL responses of cells loaded with the Im-CDs towards various concentrations of ¹O₂ from the IO₄⁻-H₂O₂ system. Inset: Calibration curve for ¹O₂. Error bars were defined as the standard deviation, n = 5. (C) Colocalization images of HeLa cells stained with the Im-CDs (C1, blue channel) and Lyso-Tracker Red (C2, red channel), and the overlay of the Im-CDs and Lyso-Tracker Red (C3). (D) Intensity profiles within the ROI (yellow line) of the Im-CDs and Lyso-Tracker Red across HeLa cells. Scale bars = 20 μm. The concentrations of ¹O₂ and other ROS were 10.0 μM and 100.0 μM, respectively.

The selectivity of the Im-CDs towards reactive oxygen species (ROS), including ¹O₂, H₂O₂, t-butylhydroperoxide (TBHP), ONOO⁻, ClO⁻, ˙OH, O₂˙⁻ and ROO˙, was investigated in PBS at pH 7.4. As shown in Fig. 3A, strong CL emission was produced in the presence of 10 μM ¹O₂, whereas no noticeable CL emission was observed when the imidazole-functionalized CD solution was mixed with other ROS. These observations revealed that the Im-CDs exhibited a high specificity towards ¹O₂. Moreover, an obvious decrease of the CL emission of the proposed system appeared in the presence of NaN₃ (a scavenger for ¹O₂) and ascorbic acid (a scavenger for ROS).²⁷ Thiourea (a scavenger for ˙OH) and nitro blue tetrazolium chloride (a scavenger for O₂˙⁻) had little effects on the CL emission (Fig. S12, ESI†). Note that Fe²⁺-catalyzed decomposition of TBHP could also act as a potential source of ¹O₂ in biological systems.¹² It was found that the Im-CDs could also enhance the CL intensity from the Fe²⁺-catalyzed decomposition of the TBHP system (Fig. S13 and S14, ESI†). These results further indicated that we could design a specific CL probe for ¹O₂ by using the imidazole groups in the conjugated structures of the CDs as a specific recognition element towards ¹O₂.

In view of the poor water-solubility of the reported organic probes for ¹O₂,¹¹ the as-prepared Im-CDs in this work exhibited good water-solubility, facile synthesis and high luminescence efficiency. These advantages made the Im-CDs an ideal probe for ¹O₂ measurement in vivo (Fig. S15–S17, ESI†). Note that hemolysis is an important concern for nanomaterials.^28,29 Therefore, a hemolysis assay was performed to investigate the interaction between the Im-CDs and blood components. The negligible changes in hemolysis percentage in the test concentration range (Fig. S18, ESI†) indicated the extremely low hemolytic property of the Im-CDs. It was concluded that the Im-CDs were considered low toxic to live cells. Finally, the effects of some typical intracellular interferents in biological samples were investigated on the CL intensity of the Im-CDs in the presence of ¹O₂, including three proteases, and some important reductants, cationic ions and anionic ions. As shown in Fig. S19 and Table S3 (ESI†), these substances had no influence on the detection of 10.0 μM ¹O₂. In conclusion, the Im-CDs could be used for the detection of ¹O₂ in biological environments.

The IO₄⁻–H₂O₂ system was used to produce exogenous ¹O₂ in HeLa cells.²³ Before incubation with various concentrations of H₂O₂, the cells were treated with 0.5 mg mL⁻¹ Im-CDs for 4 h.³⁰ Then, the CL intensities were recorded when 100.0 μM IO₄⁻ solution was injected into the above-mentioned cells. The CL intensities were found to be linear with concentrations of ¹O₂ in the range of 1.0–70.0 μM (R² = 0.9985) in HeLa cells (Fig. 3B). The final concentration of ¹O₂ was determined by H₂O₂,¹⁴ and the detection limit was 87.0 nM (S/N = 3). The relative standard deviation for three independent measurements of 1.0 μM ¹O₂ was 2.7%. The CL intensities decreased with the addition of NaN₃ (1.0 mM) into the present system (Fig. S20, ESI†). Moreover, the proposed CL probe was successfully used to monitor the generation dynamics of ¹O₂ in HeLa cells throughout the phorbol 12-myristate 13-acetate (PMA) stimulation period (Fig. S21, ESI†).³¹ These results demonstrated that the Im-CDs could be used for the monitoring of ¹O₂ generation in real biological systems.

¹O₂ is the major cytotoxic agent in type-II PDT (Fig. S22, ESI†).³² The chemical reaction pathways for the detection of ¹O₂ generation are exhibited in Fig. 1. Inspired by the real-time detection property of the CL detector,^33,34 the relationship of the CL intensity (I_CL) and ¹O₂ concentration could be expressed as

where [CDs′], [CDs] and [¹O₂] are the concentrations of the reacted Im-CDs, Im-CDs and ¹O₂, respectively; n is the order of the reaction with respect to the CDs; k₁ is the reaction rate constant of this recation; and φ is related to the quantum yield of CL. φ is assumed to be a constant in this study. When the reaction of the Im-CDs with ¹O₂ was an “n-zero reaction”, the CL intensity is proportional to the concentration of ¹O₂. Since 0.5 mg mL⁻¹ is the optimal concentration of Im-CDs for the toxicity test and CL assay, we need to verify whether the reaction of Im-CDs with ¹O₂ was an “n-zero reaction” at 0.5 mg mL⁻¹ Im-CDs. Afterwards, we investigated the fluorescence intensities of the Im-CDs before and after the photosensitized reaction. Fig. S23 (ESI†) shows that the fluorescence intensities of the Im-CDs in HeLa cells changed little during PDT, which indicated that the reaction between Im-CDs and ¹O₂ was n-zero at 0.5 mg mL⁻¹ Im-CDs. Therefore, the ¹O₂ generation in this reaction could be quantified by using Im-CDs as a CL probe.

The interval for CL measurement was set at 1.0 s per spectrum, and then the generation dynamics of ¹O₂ could be obtained directly from the CL intensity curve. Based on the calibration curve for the ¹O₂ generation in HeLa cells (Fig. 3B), the generation dynamics of ¹O₂ (r[¹O₂]) at moment t during PDT could be expressed as

where I_CL(t) is the moment t CL intensity recorded by the CL analyzer. This equation could be used for ¹O₂ concentrations in the range of 1.0–70.0 μM in the HeLa cell model. The cumulative amount of ¹O₂ and the CL intensity could be calculated from the r[¹O₂]–t curve and I_CL–t curve by using integral data processing software (Origin), respectively. The cumulative amount of ¹O₂ (Dose [¹O₂]) over a time T could be expressed as

Afterwards, the average generation dynamics of ¹O₂ (R[¹O₂]) over a time T was

Firstly, Fig. 3C and D showed that the Im-CDs were localized in lysosomes. Thus, we investigated the influence of the generation dynamics of ¹O₂ on the PDT efficacy in HeLa cells with three lysosomal-targeted type-II photosensitizers: methylene blue (MB), hypocrellin B (HB) and hypericin (HY).^35–37 The I_CL–t curves of the three photosensitizers are shown in Fig. S24 (ESI†), and the r[¹O₂]–t curves (Fig. 4A) were then obtained using the data in the I_CL–t curves and eqn (2). In order to investigate the relationship between the generation dynamics of ¹O₂ and the PDT efficacy in HeLa cells, the cumulative amount of ¹O₂ was controlled for various fluences with different treatment times. The integral area under the generation dynamics curve (Fig. 4A) from 0.0 to 25.0 min was taken as the cumulative amount of ¹O₂ (Fig. S25, ESI†). For the same cumulative amount of ¹O₂, the treatment time for HB, MB and HY was 4.5, 6.9 and 25.0 min, respectively. Therefore, the generation dynamics of ¹O₂ in HeLa cells was increased in the following order: HY < MB < HB during the treatment time. In addition, the PDT efficacy was evaluated by CCK-8 assays. After irradiation, HeLa cells were incubated at 37 °C for 24 h. As shown in Fig. 4B, the cell viability decreased in the following order: HB > MB > HY. Furthermore, the HeLa cells treated by PDT mediated by the three photosensitizers exhibited different degrees of apoptosis and necrosis. The proportions of HeLa cell death from apoptosis were 24.39% for HY, 57.49% for MB and 53.99% for HB, and the proportions of HeLa cell death from necrosis were 1.78% for HY, 1.50% for MB and 16.97% for HB, respectively (Fig. 4C). These results implied that the HB photosensitizer with the rapid generation dynamics of ¹O₂ could lead to an increased PDT efficacy, which exhibited serious cytotoxic damage and made the cancer cells undergo a rapid death (such as necrosis). In comparison, for the same cumulative amount of ¹O₂, the MB and HY photosensitizers with the slow generation dynamics of ¹O₂ could trigger a programmed cell death type response (Fig. 4D). As shown in Fig. S26–S28 (ESI†), the findings were also supported by the data from two other type-II clinical photosensitizers: chlorin e6 (Ce6) and meta-tetra(hydroxyphenyl)chlorin (mTHPC). The peaks in the generation dynamics curves may be due to photobleaching of the photosensitizers. mTHPC was used as the model photosensitizer. Afterwards, the photobleaching analysis of mTHPC was performed under the same conditions. The trend of the bleaching rate (Fig. S29, ESI†) was consistent with that in the generation dynamics curve (Fig. S27A, ESI†), indicating that the use of the Im-CDs to monitor the generation dynamics of ¹O₂ is valid.³⁸ On the other hand, the presence of the HY, MB, HB, Ce6 and mTHPC photosensitizers had little influence on the detection of ¹O₂ (Fig. S30 and S31, ESI†). The emission of the proposed CL probe could not excite these photosensitizers to produce ¹O₂. These results implied the reliability of the Im-CDs for the real-time monitoring of ¹O₂ in PDT with these photosensitizers.

Fig. 4 (A) The generation dynamics of ¹O₂ from the photosensitized reaction with HB, MB, and HY in HeLa cells during PDT. (B) Relative viabilities of HeLa cells after incubation with PBS, HY, MB and HB with or without 600 nm irradiation (60 mW cm⁻²). (C) Proportions of HeLa cell death from apoptosis and necrosis after incubation with PBS, HY, MB and HB with 600 nm irradiation (60 mW cm⁻²). (D) Relative viabilities of HeLa cells as a function of the average generation dynamics of ¹O₂ from the photosensitized reaction with HY, MB and HB in HeLa cells under the same cumulative amount of ¹O₂ generation. Error bars were defined as the standard deviation, n = 9. After irradiation, HeLa cells were incubated at 37 °C for 24 h. Control meant the photosensitized reaction without photosensitizer.

The generation dynamics of ¹O₂ in a tumor model was investigated to further explore the relationship between the generation dynamics of ¹O₂ and the PDT efficacy in vivo. Fig. 5A shows the generation dynamics of ¹O₂ from the photosensitized reaction with HB, MB, and HY. Since the generation dynamics of ¹O₂ in mouse tumors was proportional to the CL intensities based on eqn (1) and (2), the generation dynamics of ¹O₂ in mouse tumors was increased according to the following order: MB < HB < HY. The integral area under the I_CL–t curve (Fig. 5A) was taken as the cumulative CL intensity. The cumulative CL intensity was proportional to the cumulative amount of ¹O₂. Fig. 5B shows the cumulative CL intensity as a function of treatment time. The treatment time (t₁) for MB with slow generation dynamics of ¹O₂ was 20.0 min. For the same cumulative CL intensity, the treatment time (t₂ and t₃) for HY and HB was 5.5 and 9.5 min, respectively.

Fig. 5 (A) The generation dynamics of ¹O₂ from the photosensitized reaction with HB, MB, and HY in tumors during PDT. The maximum CL intensity was normalized to 100. (B) Cumulative CL intensity of ¹O₂ from the photosensitized reaction with HB, MB, and HY in tumors during PDT processes. Control in A and B meant the photosensitized reaction without photosensitizer. (C) Tumor growth curves of mice with the various treatments. Tumor sizes were normalized to their initial sizes. Error bars were defined as the standard deviation, n = 5 mice with three repeats. (D) Corresponding images of tumors taken on day 16 of treatment. (E) Representative photographs of mice with the various treatments.

Tumor sizes after PDT are recorded in Fig. 5C. The tumor growth was significantly inhibited for HY, HB and MB mediated PDT. In comparison with the control group (i.e., the mice injected with PBS), the mice injected with HY, HB and MB, respectively, showed a much smaller tumor size after 16 days of treatment (Fig. 5D and E). On the other hand, the relative tumor volume was increased in the following order: HY < HB < MB. Additionally, hematoxylin and eosin (H&E) stains of tumor slices from different groups are shown in Fig. S32 (ESI†). The degree of tissue necrosis was increased in the following order: MB < HB < HY. Interestingly, these variation tendencies were in accordance with that obtained from the generation dynamics of ¹O₂ in tumors: HY with the more rapid generation dynamics of ¹O₂ in tumors could lead to a significantly smaller tumor volume and severe tissue necrosis, in comparison to HB and MB with the slower generation dynamics of ¹O₂. At the same time, no obvious changes in the body weight of the mice were observed in the study period for all groups (Fig. S33, ESI†), implying that the employed Im-CDs and photosensitizers (i.e., HY, HB and MB) displayed no acute toxicity. Subsequently, the relationship between the generation dynamics of ¹O₂ and the PDT efficacy in vivo was also tested by monitoring the photosensitized reaction with two clinical photosensitizers (i.e. Ce6 and mTHPC) using a low-pass interference filter at 630 nm (Fig. 6 and Fig. S34, ESI†). As shown in Fig. 6D and E, the relative tumor volume of the mice in the treatment with mTHPC, Ce6 and PBS, respectively, was increased in the following order: mTHPC < Ce6 < PBS. The results from Ce6 and mTHPC were consistent with those from the other three photosensitizers, further confirming that the generation dynamics of ¹O₂in vivo played a vital role in the increased PDT efficacy. On the other hand, H&E stains (Fig. S35, ESI†) showed no obvious damage of major organs (i.e., the heart, liver, spleen, lung, and kidneys), indicating a satisfactory safety profile of the Im-CDs in vivo. Meanwhile, the complete blood count and serum biochemistry indicated that the Im-CDs exhibited no appreciable toxicity to mice at our tested dose within 16 days, and were safe for the real-time in vivo monitoring of ¹O₂ (Fig. S36 and S37, ESI†).

Fig. 6 (A) The generation dynamics of ¹O₂ from the photosensitized reaction with mTHPC and Ce6 in tumors during PDT. The maximum CL intensity was normalized to 100. (B) Cumulative CL intensity of ¹O₂ from the photosensitized reaction with mTHPC and Ce6 in tumors during PDT. Control in A and B meant the photosensitized reaction without photosensitizer. (C) Tumor growth curves of mice with the various treatments. Tumor sizes were normalized to their initial sizes. Error bars were defined as the standard deviation, n = 5 mice with three repeats. (D) Corresponding images of tumors taken on day 15 of treatment. (E) Representative photographs of mice with the various treatments. (F) H&E images of tumors treated with PBS, mTHPC with irradiation, and Ce6 with irradiation collected on day 15. Scale bar, 50 μm.

In order to clearly quantify the influence of the generation dynamics of ¹O₂ on the PDT efficacy, we investigated the relationship between the generation dynamics of ¹O₂ and the surviving fraction in vitro by using mTHPC as the model photosensitizer. In the experiments, we varied the concentrations of mTHPC and the treatment time (or the total fluence), while maintaining the same irradiation fluence rate. The generation dynamics of ¹O₂ as a function of treatment time for all of the experiments are shown in Fig. S38 (ESI†). The generation dynamics of ¹O₂ decreased with an increase in treatment time due to a combination of ³O₂ consumption and mTHPC photobleaching.³⁹ The average generation dynamics of ¹O₂ (Fig. 7A) and the cumulative amount of ¹O₂ (Fig. S39, ESI†) could be calculated from the data in Fig. S38 (ESI†). The results suggested that higher concentrations of mTHPC could result in more rapid generation dynamics of ¹O₂ and a larger cumulative amount of ¹O₂ generation. The cytotoxicity for a given fluence increased with increasing mTHPC incubation concentration. Subsequently, the data from Fig. 7B and Fig. S39 (ESI†) were combined to generate Fig. 7C, which showed the surviving fraction versus the integral generation dynamics of ¹O₂. The surviving fraction (SF) of cells as a function of the dose metric (D) is given by

SF = 1 − (1 − exp(−D/D₀))ⁿ, (5)

where D₀ is the dose required to reduce the surviving fraction by 1/e, and n is a measure of the shoulder width. Surprisingly, the data did not collapse (within experimental error) onto a universal dose–response curve and were not well fitted to the multiple-target survival model in eqn (5) (reduced χ² = 19.11). The results implied that the deviations were particularly at the slower generation dynamics of ¹O₂, indicating the presence of adaptive responses for the slow dose rates. The reason for this phenomenon was the presence of a cytoprotective autophagy pathway.⁴⁰ Autophagy could remove the toxic species through lysosomal degradation pathways, and it could be stimulated when proteins undergo irreversible oxidative damage. Meanwhile, the cumulation of oxidized macromolecules beyond the repair capacity of the cells would lead to failure in autophagy pathways.⁴¹ Therefore, for the slow generation dynamics of ¹O₂, autophagy may have enough time to remove the toxic species and improve the cell survival rate under an oxidative stress environment.

Fig. 7 (A) The average generation dynamics of ¹O₂ from the photosensitized reaction with various concentrations of mTHPC in HeLa cells during PDT. (B) Surviving fraction versus treatment time under a range of treatment conditions. (C) The fitting result of the surviving fraction as a function of the integral generation dynamics of ¹O₂. (D) The fitting result of the surviving fraction as a function of the modified integral generation dynamics of ¹O₂.

In light of the evidence for adaptive responses, the approach is to converge the integral generation dynamics of ¹O₂ for various generation dynamics of ¹O₂ by an adaptive response function. The cytotoxic effect of PDT was more closely related to the initial generation dynamics of ¹O₂.⁴¹ Accordingly, we tested an empirical adaptive response factor (A) to the integral generation dynamics of ¹O₂

where R[¹O₂] is the average rate of ¹O₂ at 5 min. The maximum R[¹O₂] at 5 min was set as a reference.

The dose metric is therefore:

Subsequently, a better fit of the multiple-target survival model (eqn (5)) was achieved (reduced χ² = 3.22) (Fig. 7D). From the fitting results, D₀ was 1.53 ± 0.04 mM, which implied that 1.53 ± 0.04 mM ¹O₂ was required to reduce the survival by 1/e. The data were consistent with those in a previous study.³⁸ Therefore, the surviving fraction of tumor cells could be estimated from the generation dynamics of ¹O₂ through the established multiple-target survival model. This investigation of a new dose metric for PDT could provide a deep insight into the relationship between the generation dynamics of ¹O₂ and the PDT efficacy. Our analysis could support the idea that the generation dynamics of ¹O₂ is responsible for the PDT efficacy.

Conclusions

In summary, we have designed a novel CL light-up probe for the real-time in vivo monitoring of ¹O₂ generation using Im-CDs with the advantages of facile synthesis, water-solubility, stability, biocompatibility and cell-permeability. The observation of the relationship between the generation dynamics of ¹O₂ and the PDT efficacy in both HeLa cells and tumor models demonstrated that the rapid generation dynamics of ¹O₂ could lead to an increased PDT efficacy in comparison with the slow generation dynamics of ¹O₂ in the presence of the same cumulative amount of ¹O₂. The reason for this phenomenon is that the presence of a cytoprotective autophagy pathway may reduce the biological effects at the slow generation dynamics of ¹O₂. The surviving fraction datum as a function of the generation dynamics of ¹O₂ are fitted much better with a modified multiple-target survival model (reduced χ² = 3.22), which could be applied to estimate the surviving fraction of tumor cells from the generation dynamics of ¹O₂. This work has provided solid evidence that the generation dynamics of ¹O₂ rather than its cumulative amount is a promising evaluation standard for the PDT efficacy. We believe that this work will lay the foundation for the feasible evaluation of ¹O₂ generation-related life processes.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (21705035, 21974008, 61520106015 and 61635014).

Notes and references

1. D. Manoharan, W.-P. Li and C.-S. Yeh, Nanoscale Horiz., 2019, 4, 557–578.
2. M. Li, H. Zhang, Y. Hou, X. Wang, C. Xue, W. Li, K. Cai, Y. Zhao and Z. Luo, Nanoscale Horiz., 2020, 5, 202–217.
3. F. Hu, D. Mao, X. Cai, W. Wu, D. Kong and B. Liu, Angew. Chem., Int. Ed., 2018, 57, 10182–10186.
4. S. Shrestha, J. Wu, B. Sah, A. Vanasse, L. N. Cooper, L. Ma, G. Li, H. Zheng, W. Chen and M. P. Antosh, Proc. Natl. Acad. Sci. U. S. A., 2019, 116, 16823–16828.
5. Y. Wan, G. Lu, J. Zhang, Z. Wang, X. Li, R. Chen, X. Cui, Z. Huang, Y. Xiao, J. Chelora, W. Zhang, Y. Liu, M. Li, H.-Y. Xie and C.-S. Lee, Adv. Funct. Mater., 2019, 29, 1903436.
6. G. Lan, K. Ni, S. S. Veroneau, X. Feng, G. T. Nash, T. Luo, Z. Xu and W. Lin, J. Am. Chem. Soc., 2019, 141, 4204–4208.
7. C. Ji, Q. Gao, X. Dong, W. Yin, Z. Gu, Z. Gan, Y. Zhao and M. Yin, Angew. Chem., Int. Ed., 2018, 57, 11384–11388.
8. Y. Gao, X. Wang, X. He, Z. He, X. Yang, S. Tian, F. Meng, D. Ding, L. Luo and B. Z. Tang, Adv. Funct. Mater., 2019, 29, 1902673.
9. T. Zhang, Y. Li, Z. Zheng, R. Ye, Y. Zhang, R. T. K. Kwok, J. W. Y. Lam and B. Z. Tang, J. Am. Chem. Soc., 2019, 141, 5612–5616.
10. N. Hananya and D. Shabat, Angew. Chem., Int. Ed., 2017, 56, 16454–16463.
11. Y. Wei, J. Zhou and D. Xing, J. Biomed. Opt., 2007, 12, 014002.
12. X. Li, G. Zhang, H. Ma, D. Zhang, J. Li and D. Zhu, J. Am. Chem. Soc., 2004, 126, 11543–11548.
13. J. Wang, D. Xing, Y. He and X. Hu, Cancer Lett., 2002, 188, 59–65.
14. F. Zou, W. Zhou, W. Guan, C. Lu and B. Z. Tang, Anal. Chem., 2016, 88, 9707–9713.
15. C. Ding, A. Zhu and Y. Tian, Acc. Chem. Res., 2014, 47, 20–30.
16. G. Hong, S. Diao, A. L. Antaris and H. Dai, Chem. Rev., 2015, 115, 10816–10906.
17. F. R. Baptista, S. A. Belhout, S. Giordani and S. J. Quinn, Chem. Soc. Rev., 2015, 44, 4433–4453.
18. K. Nakashima, Biomed. Chromatogr., 2003, 17, 83–95.
19. G. Lu, J. Wada, T. Kimoto, H. Iga, H. Nishigawa, M. Kimura and Z. Hu, Eur. J. Org. Chem., 2014, 1212–1219.
20. H. Huang, C. Li, S. Zhu, H. Wang, C. Chen, Z. Wang, T. Bai, Z. Shi and S. Feng, Langmuir, 2014, 30, 13542–13548.
21. S. Hu, A. Trinchi, P. Atkin and I. Cole, Angew. Chem., Int. Ed., 2015, 54, 2970–2974.
22. Y. Wang and A. Hu, J. Mater. Chem. C, 2014, 2, 6921–6939.
23. Y. Hu, J. Yang, J. Tian, L. Jia and J.-S. Yu, Carbon, 2014, 77, 775–782.
24. J. M. Lin and M. Yamada, Anal. Chem., 1999, 71, 1760–1766.
25. L. Tang, R. Ji, X. Cao, J. Lin, H. Jiang, X. Li, K. S. Teng, C. M. Luk, S. Zeng, J. Hao and S. P. Lau, ACS Nano, 2012, 6, 5102–5110.
26. W. Xue, Z. Lin, H. Chen, C. Lu and J.-M. Lin, J. Phys. Chem. C, 2011, 115, 21707–21714.
27. Y. Gong, B. Yu, W. Yang and X. Zhang, Biosens. Bioelectron., 2016, 79, 822–828.
28. T. Mizuhara, K. Saha, D. F. Moyano, C. S. Kim, B. Yan, Y.-K. Kim and V. M. Rotello, Angew. Chem., Int. Ed., 2015, 54, 6567–6570.
29. K. Dong, E. Ju, J. Liu, X. Han, J. Ren and X. Qu, Nanoscale, 2014, 6, 12042–12049.
30. W. Zhou, Y. Cao, D. Sui and C. Lu, Angew. Chem., Int. Ed., 2016, 55, 4236–4241.
31. D. Song, S. Cho, Y. Han, Y. You and W. Nam, Org. Lett., 2013, 15, 3582–3585.
32. W. Piao, K. Hamaoka, T. Fujisawa, S. Takeuchi, T. Komatsu, T. Ueno, T. Terai, T. Tahara, T. Nagano and Y. Urano, J. Am. Chem. Soc., 2017, 139, 13713–13719.
33. W. An, L. S. Ryan, A. G. Reeves, K. J. Bruemmer, L. Mouhaffel, J. L. Gerberich, A. Winters, R. P. Mason and A. R. Lippert, Angew. Chem., Int. Ed., 2019, 58, 1361–1365.
34. S. Gnaim, A. Scomparin, S. Das, R. Blau, R. Satchi-Fainaro and D. Shabat, Angew. Chem., Int. Ed., 2018, 57, 9033–9037.
35. J. S. Dysart, G. Singh and M. S. Patterson, Photochem. Photobiol., 1995, 61, 632–638.
36. K. J. Mellish, R. D. Cox, D. I. Vernon, J. Griffiths and S. B. Brown, Photochem. Photobiol., 2002, 75, 392–397.
37. T. A. Theodossiou, J. S. Hothersall, P. A. D. Witte, A. Pantos and P. Agostinis, Mol. Pharmacol., 2009, 6, 1775–1789.
38. J. S. Dysart, G. Singh and M. S. Patterson, Photochem. Photobiol., 2005, 81, 196–205.
39. S. Coutier, S. Mitra, L. N. Bezdetnaya, R. M. Parache, I. Geprgakoudi, T. H. Foster and F. Guillemin, Photochem. Photobiol., 2001, 73, 297–303.
40. J. J. Reiners, P. Agostinis, K. Berg, N. L. Oleinick and D. Kessel, Autophagy, 2010, 6, 7–18.
41. S. Kawauchi, Y. Morimoto, S. Sato, T. Arai, K. Seguchi, H. Asanuma and M. Kikuchi, Lasers Med. Sci., 2004, 18, 179–183.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental details, additional figures and table. See DOI: 10.1039/d0nh00128g

---