Introduction

Botrytis cinerea, the phytopathogenic fungus, is the main cause of gray mold in a wide range of fresh produce such as grapes, blueberries, eggplants, pepper, tomatoes, and strawberries (Elad et al., 2016). The infection can occur on several parts of the plant, including stem, flowers, leaves, seeds, and fruits, at different plant growth stages, starting from seedling to fruit ripening and even during storage and transportation (Dean et al., 2012). The broad range of host species and the huge losses in pre- and/or post-harvest stages make it the second major plant pathogen globally (Dean et al., 2012), and it is classified as a high-risk pathogen. Annual losses of $10 billion to $100 billion have been estimated for Botrytis diseases worldwide (Boddy, 2016). To control the B. cinerea infection, synthetic fungicides are applied on the field, which has been considered a fairly safe approach for plant protection. Nevertheless, pathogen resistance development, the potential risks of the residuals to consumer health, and environmental impacts are considered the main concerns of fungicide application on crops. Therefore, the food industry is in dire need of a green and sustainable alternative measure to control and reduce the incidence of B. cinerea infection in fresh produce during both production and postharvest storage and distribution.

Photodynamic treatment, also known as photosensitization, is one of the recently introduced decontamination technologies in the food industry. It consists of the illumination of a photosensitizer with light of an appropriate wavelength. This leads to the excitation of the photosensitizer, resulting in the generation of cytotoxic reactive oxygen species (ROS) such as hydroxyl, superoxide anion, and singlet oxygen. These, in turn, interact with the cellular structures of microorganisms (i.e., lipid membrane, proteins, and nucleic acids), causing oxidative cellular damage and ending up in cell death (Baltazar et al., 2015). These photodynamically generated short-lived radical species can be indirectly studied by electron paramagnetic resonance (EPR) spectroscopy coupled with the spin-trapping technique. This is facilitated by stable, paramagnetic EPR-detectable spin adducts that are generated by spin-trapping of short-lived radicals such as hydroxyl (OH•) and superoxide anion (O2•–) via addition to the double bond of nitrone or nitroso compounds (i.e., diamagnetic spin traps) (Villamena & Zweier, 2004). The paramagnetic characteristic of spin adducts allows them to be detectable by EPR and exhibit spectra with specific hyperfine splitting constants and g-value parameters of the trapped free radical (Rosen et al., 1999). The successful application of DMPO and PBN as trapping reagents for reactive oxygen species as well as producing persistent spin adducts has been demonstrated in the literature (Ferreira et al., 2017; Gonçalves et al., 2020). DMPO and PBN are nitrone spin traps in which the α-position carbon, relative to the nitrogen, scavenges the free radical species (Rosen et al., 1999). Generally, photodynamic treatment enjoys the advantages of low-energy requirement, no toxic chemical usage, flexibility of application for various purposes (e.g., microbial inactivation on the surface of food, packaging material, and food processing equipment), and no observed microbial resistance to the treatment (Penha et al., 2017).

Curcumin, the flavonoid isolated from the Curcuma longa roots, is a naturally occurring photosensitizer with maximum absorption at ca. 420 nm (i.e., blue light) with a penetration depth of hardly 1 mm (Clement et al., 2009) and high cell membrane permeability (Sun et al., 2008). Therefore, it is suitable for surface decontamination of food products with less adverse impact on the nutritional and chemical composition of the treated foods. Moreover, curcumin is recognized as safe to be consumed according to the Food and Drug Administration (FDA) of the USA because of its low toxicity (FDA, 2016). The phototoxicity of curcumin against a broad range of microorganisms such as Gram-positive bacteria, Gram-negative bacteria, viruses, yeasts, and molds has been studied, and a promising inactivation efficiency was reported (Al-Asmari et al., 2017; Chen et al., 2020; Nima et al., 2021). However, the interactive effects of treatment parameters on the phototoxicity of curcumin have not been studied, for which response surface methodology (RSM) was employed. RSM is widely used to study the relationship between factors (i.e., independent variables) and the investigating responses (i.e., dependent variables), as well as the inter-interaction of factors, which allows the simultaneous optimization of multiple factors (Myers et al., 2016). Therefore, this study looks at an I-optimal design of the curcumin formulations in different solvents and their in vitro phototoxicity against B. cinerea spores, the isolated microorganism from Australian-grown strawberries (Seidi Damyeh et al., 2021), at different illumination conditions. This probes the main factors affecting the photoinactivation efficiency of curcumin (viz., solvent, curcumin concentration, irradiance, and light dose), which was the primary goal of this study. Furthermore, the EPR-spin trapping technique was employed to characterize the amount and type of photogenerated ROS.

Materials and Methods

Design of Experiments

Randomized RSM was used to generate an I-optimal quadratic design (Design Expert, v.11.1.2.0, Stat-Ease, Minneapolis, MN, USA) to investigate the effect of curcumin concentration (13–130 μM), light dose (19.05–66.53 J cm−2), irradiance (31.75–44.35 mW cm−2), and solvent (ethanol, medium chain triglyceride (MCT) oil, ethanol-MCT) on the antimicrobial photodynamic activity of curcumin against B. cinerea spores by studying log reduction and viability of spores as dependent variables (Table 1). Since curcumin is sparingly soluble in water, the effect of solvent on the phototoxicity of curcumin was also considered. The levels of the independent variables were determined based on preliminary experiments to achieve acceptable antimicrobial activity (data not shown). The actual variables were coded to aid the multiple regression analysis.

Table 1 I-optimal quadratic design with the observed correspondent responses (i.e., log reduction and cell viability)
Full size table

Preparation of Curcumin Solutions

Curcumin (from Curcuma longa, ≥ 65%, Sigma-Aldrich, St Louis, MO, USA) stock solutions were prepared in three solvents, namely ethanol (99.8%, Fisher Chemical, Loughborough, UK), MCT oil (derived from coconut oil (C8, C10), 100%, Melróse, Mt Waverley, VIC, Australia), and ethanol-MCT oil mixture (1:1, v/v), and were kept at −20 °C until analysis. The purity of curcumin was considered in calculating the curcumin concentration in the final solutions. Ultrapure water (Milli-Q, Millipore, Burlington, MA, USA) was added to the respective stock solutions to have a 5% curcumin-solvent phase in the final working solutions (v/v). Tween 80 (Polyoxyethylene (20) sorbitan mono-oleate, BDH Chemicals, Kilsyth, VIC, Australia) was added to the solutions containing MCT oil as a surfactant to achieve a homogeneous solution. This was followed by vortexing the solutions for 5 min at 3000 rpm (Heathrow Scientific, Vernon Hills, IL, USA). The prepared solutions were kept at 4 °C for 2 days.

Absorbance and Fluorescence Spectra

UV–visible and fluorescence spectra of curcumin solutions (13, 52, 91, and 130 µM) were recorded with a spectrophotometer (Varioskan LUX 3020–80,884, Thermo Scientific, Singapore). The absorbance spectra were recorded within the wavelength range of 300–700- and 1-nm steps. The fluorescence emission spectra were acquired by exciting samples at 430 nm, which corresponds to the curcumin maximum absorption wavelength. The slit width was set at 1 nm and the spectra were recorded from 450 to 750 nm. All these measurements were carried out on freshly made solutions in triplicate and at room temperature.

Spore Suspension

The asexual spores of B. cinerea are conidia and responsible for easy transmission of the infection and were targeted in this work. The isolated B. cinerea from the commercial strawberry cultivar ‘Red Rhapsody’ grown in Queensland, Australia (Seidi Damyeh et al., 2021) was cultured at 25 °C on potato dextrose agar (CM0139, Oxoid, Basingstoke, Hampshire, UK). Spores were harvested from a 2-week-old culture by gently rubbing the surface flooded with sterile, ultrapure water, followed by filtering with a 2-layer sterile cheesecloth. The filtrate was centrifuged (4500 × g, 4 °C; 5804 R centrifuge, Eppendorf, Hamburg, Germany), washed with sterile ultrapure water, and centrifuged again. The resulting pellet was resuspended in sterile ultrapure water and counted with a Haemocytometer. The spore suspension was then kept at 4 °C and was used within a week.

Phototoxicity in Suspension Culture

To study the phototoxicity effect of curcumin solutions on the growth and viability of the spores, a 104 CFU ml−1 concentration of spore suspension was prepared in sterile ultrapure water and potato dextrose broth (PDB; Sigma-Aldrich, St Louis, MO, USA). Samples were prepared in sterile 96-well plates by adding 50 μl of curcumin solution and 50 μl of spore suspension. This was followed by illumination under LED light (420 nm, 200 W, 100–240 V; COB LED Light Fixture, HYG05-4, China) for the given time (Table 1). The photosynthetic photon flux density (PPFD) of LEDs was measured using a spectrometer (HR-450, HiPoint, Kaohsiung, Taiwan) at 10 different positions at a 10-cm distance from the light source. PPFD (µmol m−2 s−1) was converted to light irradiance (mW cm−2) by the following formula (Roh et al., 2016):

$$\mathrm{Light \;irradiance }\left(\mathrm{mW } \, {\mathrm{cm}}^{-2}\right)= \frac{h \times C\times A\times \mathrm{PPFD}\times {10}^{4}}{\lambda }$$

where h is Plank’s constant (6.63 × 10−34 J s−1), C is the light velocity (3 × 108 m s−1), A is the Avogadro constant (6.022 × 1023 photon mol−1), and λ is the wavelength (nm).

The dark control samples were incubated for 10 min in darkness at room temperature in the presence of different curcumin formulations. For a negative control representing 0 μM of curcumin, sterile ultrapure water, or PDB, was added.

Effect of Photosensitization on the B. cinerea Growth

To facilitate the B. cinerea colony counting, a selective medium was prepared according to the method given by Edwards and Seddon (2001). Aliquots of 30 μl were taken from the homogenized treated samples, spread-plated on Botrytis selective media, and incubated at 25 °C for 4 days. The number of viable colonies was then counted (colony counter, Stuart Scientific, Staffordshire, UK) and recorded as a Log10 (CFU ml−1) reduction.

Effect of Photosensitization on B. cinerea Viability

Tetrazolium dye (MTT) assay was used to evaluate the survival rate of the spores (Shamali et al., 2018). This assay measures the colored formazans produced through the reduction of colorless tetrazolium salts, which is an indication of the proliferation and viability of microbial cells (Stentelaire et al., 2001). After 24-h incubation at 25 °C, 10 μl of aqueous MTT solution (5 mg ml−1; EMD Millipore Group, Darmstadt, Germany) was added to each sample, mixed properly, and kept at 25 °C in an incubator for 2 h. The media were then removed, and 200 μl dimethyl sulfoxide (Sigma-Aldrich) was added, mixed, and incubated overnight at 25 °C to enhance the formazan dissolution. The fungal plug was then removed, and absorbance was read at 570 nm (Infinite M200, Tecan GmbH, Grödig, Austria). The viability was calculated as below:

$$\mathrm{Viability} \left(\%\right)=\frac{\mathrm{Absrobance \;of \;treated \;spore}-\mathrm{Absorbance \;of \;MTT \;in \;DMSO}}{\mathrm{Absorbance \;of \;the \;untreated \;spore}-\mathrm{Absorbance \;of \;MTT \;in \;DMSO}} \times 100$$

Electron Paramagnetic Resonance Spectroscopy

.A spin-trapping electron paramagnetic resonance (EPR) protocol was applied to detect, identify, and measure the ROS produced in the curcumin solutions during direct illumination in an X-band continuous wave (CW) EPR cavity (Fig. S.1) for 15 min at 440 nm using two different light irradiances (43.59 and 62.71 mW cm−2). EPR spectra were recorded with an ELEXSYS EPR spectrometer (Bruker, Rheinstetten, Germany) operated at 9.87 GHz, using a borosilicate glass capillary tube (Kimble Chase, Gerresheimer, Mexico). To detect the produced ROS, 5,5-dimethyl-pyrroline N-oxide (DMPO; 80 mM; > 97%, Chem-Supply, Gillman, SA, Australia) and alpha-phenyl N-tertiary-butyl nitrone (PBN; 80 mM; ≥ 98%, Sigma-Aldrich) dissolved in 50% ethanol were used as spin traps for hydroxyl (OH•) and superoxide (O2•–) radicals. Equal aliquots of the spin trap solution and the curcumin solution were mixed, which was used to determine the ROS photogeneration. The EPR signal intensity was recorded during 15 min illumination. Experimental parameters used for EPR operating in X-band were a modulation frequency of 100 kHz, microwave power of 19.82 mW, microwave frequency of ca. 9.87 GHz, a central field of 3518 G, sweep width of 110 G, sweep time of 41.94 s, a modulation frequency of 100 kHz, and a modulation amplitude of 3 G. All the EPR measurements were conducted at room temperature and in triplicate. The EPR spectra of the control samples (i.e., without curcumin) under illumination and the curcumin solutions under darkness were also recorded.

The concentration of photogenerated spin adducts was determined by double integration of the EPR spectrum using MATLAB software (R2020b, MathWorks Inc., Natick, MA, USA) based on known concentrations of ethanolic 4-hydroxy-2,2,6,6-tetramethyl-1-piperidimyloxy radical (TEMPO, 98%; Fujifilm Wako, Japan). Simulation and fitting of the experimental EPR spectra were performed using the garlic function of the EasySpin toolbox (MATLAB). This resulted in optimized spin-Hamiltonian parameters of the individual spin adducts, and subsequently, concentrations of the total adduct and identified radical adducts were determined from the EPR parameter by comparison to known literature values. The following parameters were used to simulate the PBN-hydroxyl radical adduct: aN = 15.5 G, aH = 2.75 G (Monteiro et al., 2016); DPMO-hydroxyl adduct: aN = 15 G, aH = 14.9 G (Lloyd et al., 1997); DMPO-superoxide anion adduct: aN = 14.2 G, aHβ = 11.34 G, aHγ = 1.25 G (Mottley et al., 1986); DMPO radical: aN = 14.6 G (Della Rocca et al., 2021); and DMPO-C5H5O2 adduct: aN = 15.8 G, aH = 22.3 G (Dodd & Jha, 2011).

Statistical Analysis

The analysis of the experimental data was performed by multiple regression analysis using the Design Expert v.11.1.2.0 software. The linear and quadratic effects of independent variables and their interactions involved in the model were analyzed by analysis of variance (ANOVA) at 95% level of confidence (p ≤ 0.05). To avoid multicollinearity in the models (due to the probable strong correlation of predictors), the stepwise regression procedure (2FI, Quadratic, Reduced Quadratic, and Linear) was applied. The adequacy of the model was determined by the adjusted coefficient of determination (R2) analysis and lack-of-fit test. The precision of each model tested was evaluated by comparing the obtained responses with the predicted ones from the fitted model. The surface responses were plotted based on the fitted polynomial regression equations to visualize the relationship between variables and responses. Maximizing log reduction and minimizing cell viability were the constraints to optimizing the photosensitization condition, which was validated by conducting four experiments.

Results and Discussion

Absorption Spectra and Fluorescence Emission Spectra of Curcumin

The visible absorption and fluorescence emission spectra for curcumin were measured in different aqueous environments and are presented in the Supplementary Material (Fig. S.2). Significant solvent-dependent shifts in maxima were observed in both spectra. The absorption maxima of curcumin were in the range of 410 to 430 nm, which is in line with previous studies (Santezi et al., 2021; Tsaplev et al., 2020). As shown in Fig. S.2a–c, increasing the curcumin concentration to 130 μM gave rise to a higher absorbance intensity in the three studied environments. Moreover, a slightly wider maximum absorbance was observed in aqueous ethanolic environments. For the fluorescence emission, the maxima were at around 560 nm in 13 μM curcumin in an aqueous ethanolic environment and increasing the concentration to 130 μM led to a slight “redshift” in the emission maxima to about 570 nm. However, the presence of MCT oil in the environment gave rise to a “blueshift” in the maximum emission at 500 nm (Fig. S.2d–f) with notably stronger fluorescence intensity. However, the lower fluorescence intensity in aqueous ethanolic environments can be related to the self-quenching phenomenon due to more aggregate formation of curcumin in low viscous environments (Iwaki et al., 1993). Therefore, we suggest that less intermolecular collision of curcumin molecules in viscous MCT-containing environments enhances fluorescence intensity, which can, in turn, improve ROS generation.

Effect of Treatment Factors on In Vitro Photoinactivation of B. cinerea Spores

The antifungal photodynamic activity of curcumin as a photosensitizer against B. cinerea spores is presented in Table 1, and the interactive effect of independent variables on the antifungal activity is reported in the Supplementary data (Tables S.2 and S.3; Figs. S.3 and S.4). The results showed that curcumin-mediated photodynamic treatment was able to inactivate B. cinerea spores at the in vitro level (Table 1). There was a considerable variation in the values of log reduction in B. cinerea spores’ survival (0.19–4.0 log10 (CFU ml−1)) as well as their viability (0.48–65.20%) across the various photodynamic conditions. The antimicrobial mechanism of photodynamic treatment is ascribed to the photochemical reactions that trigger upon activation of the photosensitizer by illumination and give rise to ROS generation. These ROS in turn interact non-specifically with various cellular components such as cell walls, cell membranes, enzymes, and DNA (D’Souza et al., 2015) and consequently cause fungal cell death. Wei et al. (2021) reported the occurrence of shrinkage and leakage in B. cinerea spores, deformations in cellular organelles, and complete destruction of the nucleus and mitochondria after curcumin-mediated photodynamic treatment (500 µM, 5 min). The authors maintained that the main photoinactivation mechanism is the ROS damage to the vital cellular components such as the nucleus and mitochondria, while the cell shape and size remained almost intact after the treatment (Wei et al., 2021). Another study on Penicillium expansum also showed severe damage to mycelia’s cell wall and membrane after curcumin-mediated photodynamic treatment. They observed the vacuolization occurrence on the cell walls using a lower curcumin concentration (30 µM), while increasing the concentration to 300 µM led to ruptures in the mycelia’s structures (Pang et al., 2022). Neither curcumin nor light alone showed a significant antifungal effect under the given experimental conditions, affirming that the combination of light and curcumin is necessary for the antifungal photodynamic treatment of B. cinerea spores.

A 2-factor interaction (2FI) model was selected to fit the experimental data (Tables S.2 and S.3) as it exhibited the highest R2 values with a non-significant lack-of-fit value. Results demonstrated the significant (p < 0.05) effect of three independent variables (solvent, curcumin concentration, and irradiance) on the antifungal activity of the treatment, whereas light dose showed to be insignificant (p > 0.05) (Tables S.2 and S.3). Comparing the F-values, solvent showed the greatest impact on phototoxicity of curcumin, followed by irradiance and curcumin concentration, and light dose exhibited the lowest impact in the studied range (Tables S.2 and S.3). In fact, the solvent can affect the photosensitizer’s affinity for the microbial cells via electrostatic and hydrophobic interactions, which are mainly ruled by the surface properties of spore cells such as hydrophobicity and surface charge (Dunlap et al., 2005). This in turn influences the adhesion of the photosensitizer to the microbial cell and, consequently, the antifungal photodynamic inactivation. Furthermore, irradiance and the concentration of photosensitizer can affect the quantum yield and the half-life of the triplet excited state (Spikes, 1989), which directly impacts the efficiency of treatment.

Figure 1 and Fig. S.5 and S.6 exhibit the three-dimensional surface plots of log reduction and cell viability as a function of independent variables at three different levels. Surface plots enabled the visualization of the interactive effects and inter-dependency of the four investigated variables. The higher photo-decontamination ability of curcumin was observed in aqueous ethanolic solution (Fig. 1), compared to aqueous MCT and ethanol-MCT environments (Figs. S.5 and S.6). The investigated variables and their interactions showed no significant (p > 0.05) effect on the antifungal photodynamic activity of curcumin in aqueous MCT and ethanol-MCT solutions (Fig. S.6a–f). Under all experimental conditions, when MCT-containing formulations were used, the viability of B. cinerea was reduced to 27.94–54.38%. The limited phototoxicity of curcumin in MCT oil-contained environments could be attributed to the encapsulation of most curcumin molecules in oil droplets as well as the low affinity of the spore hydrophobic cell wall to the lipophilic droplets containing curcumin. Therefore, curcumin may show a higher affinity for the core of emulsion droplets than the spore’s outer surface, and the probability of photoperoxidation of the emulsion droplets seems to be greater than the damage to spore cells. The free curcumin present in the MCT-containing environments could then contribute to the observed low antifungal effect of the treatment.

Fig. 1
figure 1

Surface graphs of different interaction effects in the aqueous ethanolic environment: curcumin concentration and light dose on a Botrytis cinerea log reduction (log (CFU ml−1)) and b cell viability at a constant irradiance of 38.05 mW cm−2; curcumin concentration and irradiance on c log reduction and d cell viability at a constant light dose of 42.79 J cm.−2; light dose and irradiance on e log reduction and f cell viability at a constant curcumin concentration of 71.50 μM (color scale: log reduction , cell viability ]

Full size image

On the other hand, curcumin demonstrated a higher antifungal photodynamic activity in the aqueous ethanolic environments, which enabled a complete reduction in the survival of B. cinerea spores (4 log10 (CFU ml−1)) (Table 1). The photochemical properties of curcumin and therefore its phototoxicity are substantially affected by the H-binding capacity of the environment (Chignell et al., 1994). It is possible that curcumin forms a complex with the outer surface of B. cinerea spores, which is induced by electrostatic interactions and/or H-bond formation in the hydrogen bond donating and accepting solvents such as ethanol. In fact, the phenolic OH group in the curcumin structure accepts hydrogen from the solvent or outer surface of microorganisms (Tønnesen, 2006). This encourages better adhesion of the photosensitizer to the spores and therefore results in better photoinactivation efficiency. Moreover, antifungal inactivation was shown to be affected by curcumin concentration and light irradiance. When the light irradiance was maintained constant at the central value, reducing curcumin concentration and light dose resulted in improved photoecontamination of B. cinerea spores in aqueous ethanolic environments (Fig. 1a). A similar trend was observed when the light dose was maintained constant and higher antifungal activity was achieved by decreasing the curcumin concentration, even though higher light irradiance was desirable for obtaining higher photoinduced death of fungal spores in higher curcumin concentrations, as shown in Fig. 1c–f, which was demonstrated by higher log reduction and lower spore viability. This could be attributed to the poor solubility of curcumin in aqueous environments, which consequently tends to aggregate in higher concentrations. This in turn has an adverse effect on not only the light absorption and excitation phenomenon but also its attachment to microbial cells (Konan et al., 2002), which contributes to the antimicrobial photodynamic activity of curcumin. Our results are in contrast with those reported in the literature, where the conventional step-by-step approach was used, and therefore the possible interactions between factors could not be studied. For example, Song et al. (2020) observed that increasing light dose led to an increase in curcumin phototoxicity against P. expansum spores, while they observed no significant differences between different curcumin concentrations (50–500 μM). In another study, complete inactivation of Listeria monocytogenes was obtained by adding 1 μM of curcumin to an aqueous ethanolic solution. The authors found that the photoinactivation ability of curcumin was enhanced by increasing the light dose (Huang et al., 2020).

This study optimized the curcumin-mediated photosensitization condition for four investigated variables, namely curcumin concentration, light dose, irradiance, and solvent to obtain higher phototoxicity of curcumin, as indicated by higher log reduction and lower viability. The optimized condition of curcumin-mediated photosensitization was 13 μM of aqueous ethanolic curcumin with illumination at 31.75 mW cm−2 and 19.05 J cm−2. A 3.25 Log (CFU ml−1) reduction and 7.65% spore viability were obtained under the optimized condition, which was in agreement with the predicted values of 3.25 log (CFU ml−1) and 9.91%, respectively. This in turn confirms the reliability of the optimized photosensitization condition, which is recommended for further application.

Generation and Evolution of Radicals (O2 • and OH•)

The photoexcitation of curcumin under the given experimental conditions led to the gradual generation of EPR spectra upon continuous illumination in the presence of a spin trap (i.e., DMPO or PBN). This dynamic process of trapping is presented in the Supplementary Material (Figs. S.7–S.12). It was shown that the signal intensity of DMPO and PBN adducts in the aqueous ethanolic environment is directly proportional to curcumin concentration, irradiance, and illumination time. A substantial rise in the characteristic quartet-line signal of DMPO and the six-line signal of PBN adducts was obtained by increasing the curcumin concentration and irradiance (Figs. S.7 and S.8, respectively). The MCT oil-containing formulations, on the other hand, were only studied at an irradiance of 62.71 mW cm−2 and in the presence of DMPO. Similarly, an increase in curcumin concentration enhanced the EPR signal intensity (Figs. S.11 and S.12). Figure 2 shows the evolution of experimental and simulated EPR spectra for DMPO and PBN adducts over the 15-min illumination. Compared to the PBN spectra, more intense and complex EPR spectra were obtained in the presence of DMPO. The simulation of the spectra according to the literature (Della Rocca et al., 2021; Dodd & Jha, 2011; Monteiro et al., 2016) revealed that four spin adducts, including DMPO-OH•, DMPO-O2•, DMPO-OC2H5•, and DMPO•, contributed to the obtained experimental EPR spectra with the partially overlapped spectra (Fig. 2a–c).

Fig. 2
figure 2

Experimental (black line) and simulated EPR spectra (red line) for aqueous ethanolic curcumin solutions (91 μM, 15 min illumination at 43.59 mW cm−2): DMPO-ROS adducts after 44.2 s illumination (a), after 909.3 s illumination (b), and the average of 15 min illumination (c) including DMPO-hydroxyl adduct (magenta), DMPO-superoxide anion adduct (blue), DMPO-hydroxyethyl adduct (green), and DMPO radical (cyan); PBN-ROS adducts after 44.2 s illumination (d), after 909.3 s illumination (e), and the average of 15 min illumination (f) including PBN-hydroxyl adduct (magenta)

Full size image

The variation in concentrations of total ROS and the identified radical adducts as a function of illumination time in different curcumin formulations at 43.59 and 62.71 mWcm−2 using DMPO and PBN as the spin trap is illustrated in Figs. 3, 4, and 5. The generation of ROS showed distinct patterns under different experimental conditions. Generally, three phases of ROS generation can be observed, as shown in Fig. 3. The first quantities of ROS generated are represented by a sharp line gradient (the first phase), which indicates the activation of readily accessible curcumin molecules in the environment (Figs. 3 and 5). As can be seen in Figs. 3–5, the starting time of ROS generation was rough once the illumination started. This demonstrates the instant light energy take-up by the curcumin molecules (< 30 s) to be excited and therefore rapid ROS generation in photodynamic treatment. This is followed by the second increasing phase with a slower rate, which indicates that the ROS generation is nearing completion. This phase was however observed in aqueous ethanolic environments only (Fig. 3), which probably represents the duration needed to level up the energy of curcumin molecules entrapped in the self-formed aggregates—due to curcumin insolubility in water—to their excitation point.

Fig. 3
figure 3

Evolution of total (experimental) adduct and simulated and fitted radical adduct concentration (µM) over 15 min illumination of different concentrations of curcumin in aqueous ethanolic solutions using DMPO as spin trap, a total radical adduct formation at 43.59 mW cm−2, b total radical adduct (µM) formation at 62.71 mW cm−2, c 13 µM curcumin at 43.59 mW cm−2, d 13 µM curcumin at 62.71 mW cm−2, e 52 µM curcumin at 43.59 mW cm−2, f 52 µM curcumin at 62.71 mW cm−2, g 91 µM curcumin at 43.59 mW cm−2, h 91 µM curcumin at 62.71 mW cm−2, i 130 µM curcumin at 43.59 mW cm−2, and j 130 µM curcumin at 62.71 mW cm−2. The results are the mean of at least three replicates. The error bar indicates the standard error of at least three replicates

Full size image
Fig. 4
figure 4

Evolution of total (experimental) adduct and simulated and fitted radical (•OH) adduct concentration (µM) over 15 min illumination of different concentrations of curcumin in aqueous ethanolic solutions using PBN as a spin trap. Total radical-adduct concentration at a 43.59 mW cm−2 and b 62.71 mW cm−2; fitted PBN–•OH adduct concentration at c 43.59 mW cm−2 and d 62.71 mW cm−2. The results are the mean of at least three replicates. The error bar indicates the standard error of at least three replicates

Full size image
Fig. 5
figure 5

Evolution of total (experimental) adduct and simulated and fitted radical adduct concentration (µM) over 15 min illumination of different concentrations of curcumin, dissolved in aqueous solutions of MCT oil and mixture of MCT and ethanol (MCT-Eth), at 62.71 mW cm−2 using DMPO as spin trap, a total radical adduct formation, MCT; b total radical adduct formation, MCT-Eth; c 13 µM curcumin, MCT; d 13 µM curcumin, MCT-Eth; e 52 µM curcumin, MCT; f 52 µM curcumin, MCT-Eth; g 91 µM curcumin, MCT; h 91 µM curcumin, MCT-Eth; i 130 µM curcumin, MCT; and j 130 µM curcumin, MCT-Eth. The results are the mean of at least three replicates. The error bar indicates the standard error of at least three replicates

Full size image

On the other hand, the first phase of ROS generation is indicated by a steeper upward line in MCT-contained environments (Fig. 5) that reached the maximal amount of generated ROS (6.46 μM (Fig. 5a) and 8.19 μM (Fig. 5b) in a considerably shorter time. This was followed by a sharp fall, indicating the completion of ROS generation. This can be due to the greater homogeneity of curcumin in these environments compared to aqueous ethanolic solutions, where curcumin shows a higher tendency to aggregate. The third stage in ROS generation is almost plateau, which demonstrates the least quantity of ROS is generated by the few amounts of curcumin molecules left in the environment and therefore yield is roughly constant with time (Fig. 3). This is where either there are no more curcumin molecules to be excited or the whole system has come to some kind of equilibrium that marks the end of the ROS generation process, which is then followed by a reducing slope. Nevertheless, only the first phase was observed for hydroxyl generation when PBN was used as the spin trap (Fig. 4). This can suggest that either there was enough PBN in the environment to react with the continuously generated hydroxyl radicals during the illumination or that the high stability of PBN-OH adducts led to their detection for a longer time. Furthermore, this shows that the applied illumination condition did not promote photolytic decay of the generated adducts.

The photogeneration of ROS was shown to be directly proportional to curcumin concentration as well as light intensity (Figs. 3–5). The latter, which was only studied in aqueous ethanolic environments, could contribute to the rapid activation of curcumin molecules and therefore reach the end of ROS generation faster (Fig. 3b). It was also shown that ROS generation in terms of DMPO and PBN adducts increased by 1.5 and 2 times, respectively, when the irradiance was increased from 43.59 to 62.71 mW cm−2. Moreover, a considerably low (< 1 μM) amount of spin adducts (i.e., ROS) was detected at the lowest curcumin concentration (13 μM) in aqueous ethanolic environments. However, this was increased by 12 times in an aqueous ethanolic environments and by 4 times in MCT-contained environments when the highest curcumin concentration (130 μM) was used (Figs. 3 and 5). Using PBN also showed a rise in photogeneration of spin adducts by increasing curcumin concentration from 13 to 91 μM; however, further increases in concentration did not improve the ROS generation at the two studied irradiances (Fig. 4a, b).

The simulation and identification of the radicals that contributed to the observed total ROS, demonstrated that O2•− (superoxide anion) rather than •OH (hydroxyl) was the dominant photodynamically generated ROS in the studied curcumin environments (Figs. 3–5). The efficacy and rate of O2•− trapping by DMPO are affected by a number of environmental factors, such as solvent (Villamena, 2016). The author also suggested that a high concentration of DMPO is required for O2•− trapping. Thus, 40 mM of DMPO used in this study could result in more efficient O2•− trapping, as indicated by stronger DMPO-O2• adduct signals (Fig. 2; reaction (3)). The observed rapid decay of EPR signals for the MCT-contained environments is possibly due to the high instability of DMPO-O2• adducts (τ1/2 = 60 s) (Tuccio et al., 1995), which was sufficient for the time frame of our measurements to detect the adduct. Therefore, we suggest that either there was not enough spin trap in the environment to react with the high amounts of generated ROS, or the observed spin-adduct concentration was the maximal amount that curcumin molecules in the environment could produce. The observed signal decay could also be caused by the radical scavenging ability of MCT oil containing lauric acid, caprylic acid, and caproic acid with reported antiradical (i.e., antioxidant) activity (Oh & Shahidi, 2018). In this case, we suggest that there might be a competitive environment between MCT oil and DMPO to react with the generated ROS, in which MCT oil seems to be more successful in scavenging the radicals. The signal decay could also have occurred due to the oxygen deficiency in the reaction environment, indicating the oxygen dependency of type I photosensitization. It should be mentioned that the control experiments showed no OH• and O2•− generation in aqueous MCT solutions without curcumin. This result is in accordance with the in vitro results against B. cinerea spores.

The observed low-intensity signals of DMPO-OH adduct (reaction (4)) could be attributed to the inhibitory effect of ethanol toward DMPO-OH formation by scavenging the •OH radicals and generating hydroxyethyl radicals (reaction (5)) (Ergaieg et al., 2008), which has also been reported in the Fenton system (Finkelstein et al., 1980). This reaction was confirmed by the slight increase in the concentration of DMPO-OC2H5 adducts during the 15-min illumination in all investigated environments (Figs. 3 and 5c–j). Therefore, higher concentrations of DMPO are required for comparable amounts of •OH radicals to react with both spin trap and ethanol and, in turn, to enable the detection of DMPO-OH adducts (Andersen & Skibsted, 1998). Furthermore, the possible formation of DMPO-OH from DMPO-OOH (resulting from DMPO-O2•) can be ruled out since the DMPO-OH signal did not rise with the decay of DMPO-O2•. This is in accordance with the findings of Finkelstein et al. (1982) and Buettner (1990) that less than 3% of DMPO-OOH converts into DMPO-OH. Using PBN as a spin trap, on the other hand, enabled us to detect the photogeneration of hydroxyl radicals in the studied curcumin environments (Fig. 4c, d). The six-line EPR spectrum observed arises from the coupling of the electron spin to the nuclear spins of the 14N (I = 1) and 1H (Hβ, I = ½) nuclei of the spin trap with characteristic hyperfine coupling values for a PBN-OH adduct (Buettner, 1987). These data thus reveal that •OH radicals are being generated. No other EPR signals of PBN adducts with other ROS were detected, which might be due to the high instability of other adducts or the higher efficacy of PBN in trapping •OH radicals.

It should be noted that prior to illumination, a small multicomponent paramagnetic signal attributed to DMPO• and DMPO-OC2H5 was detected in all samples containing DMPO and ethanol. The detection of DMPO-OC2H5 adducts in the absence of light or curcumin is attributed to the reaction between ethanol and DMPO that results in the formation of ethoxy radicals by hydrogen abstraction from ethanol (Brezová et al., 2004), which also leads to the generation of DMPO-OC2H5 (reaction (6)) adducts. The fourth identified DMPO adduct resulted from the degradation of DMPO molecules through their protonation (Della Rocca et al., 2021), and therefore DMPO• adducts were also observed within the multicomponent spectra (reactions (7) and (8); Fig. 2c). In the absence of light or curcumin, the two main components of photosensitization, no EPR signal relative to the OH• and O2•– adducts was observed (data not shown), indicating that the ROS formation is a photodynamic process.

Probable Mechanisms

From the EPR spin-trapping spectroscopy, a powerful tool to study short-lived radicals, it was concluded that the superoxide anion was the dominant ROS generated through photoexcitation of curcumin in different environments. Therefore, it is speculated that the continual generation of mainly superoxide anion in aqueous ethanolic solutions of curcumin leads to its observed phototoxicity against B. cinerea. The possible mechanism can be the activation of curcumin molecules (S0) by visible light, i.e., photoexcitation, resulting in an unstable singlet excited state (Sn), which is converted to a long-lived triplet state (T1) through intersystem crossing. The generated triplet state then undergoes two different pathways (type I or II) to produce cytotoxic ROS. Hydroxyl and superoxide anion are the results of the type I mechanism by transferring electrons to dissolved oxygen molecules, followed by their reduction. While a collision between the triplet state and the ground state oxygen molecule leads to an energy transfer to oxygen, producing singlet oxygen through the type II mechanism (DeRosa & Crutchley, 2002), notwithstanding this, electron transfer from an electron donor such as a photosensitizer molecule can result in singlet oxygen reduction and conversion into a superoxide anion, which has been reported in biological systems (Peters & Rodgers, 1980). To clarify the origin of the observed superoxide anion in the curcumin solutions, further studies are recommended using specific ROS inhibitors such as beta-carotene for 1O2 and superoxide dismutase for O2•− to study and measure one radical at a time. Overall, the generation of O2•− and OH• indicates the photodynamic activity of curcumin in the studied solvents. The mechanism of photosensitization of curcumin (CUR) and radical spin-trapping reactions can be described as follows:

$${}^{0}\mathrm{CUR }({S}_{0})\stackrel{hv}{\to }{}^{1}\mathrm{CUR }({S}_{n})\stackrel{\mathrm{Intersystem \;crossing}}{\to }{}^{3}\mathrm{CUR }({T}_{1})$$
(1)
$${}^{3}\mathrm{CUR }({T}_{1})\stackrel{\mathrm{Electron \;transfer}}{\to }\mathrm{ROS }\;({\mathrm{OH}}^{\bullet }, {{\mathrm{O}}_{2}}^{\bullet -})$$
(2)
$${{\mathrm{O}}_{2}}^{\bullet -}+\mathrm{DMPO}\to \mathrm{DMPO}-{{\mathrm{O}}_{2}}^{-}+{\mathrm{H}}^{+}\to \mathrm{DMPO}-\mathrm{OOH}$$
(3)
$${\mathrm{OH}}^{\bullet -}+\mathrm{DMPO}\to \mathrm{DMPO}-\mathrm{OH}$$
(4)
$$\begin{aligned}{\mathrm{OH}}^{\bullet }&+{\mathrm{CH}}_{3}{\mathrm{CH}}_{2}\mathrm{OH}\to {\mathrm{CH}}_{3}{\mathrm{CH}}^{\bullet }\mathrm{OH}\\&+\mathrm{DMPO}\to \mathrm{DMPO}-{\mathrm{CH}}_{3}{\mathrm{CH}}_{2}\mathrm{O}\end{aligned}$$
(5)
$${\mathrm{DMPO}}^{\bullet }+{\mathrm{CH}}_{3}{\mathrm{CH}}_{2}\mathrm{OH}\to \mathrm{DMPO}-{\mathrm{CH}}_{3}{\mathrm{CH}}_{2}\mathrm{O}+{\mathrm{H}}^{\bullet }$$
(6)
$$\mathrm{DMPO}+{e}^{-}\to {\mathrm{DMPO}}^{\bullet -}$$
(7)
$${\mathrm{DMPO}}^{\bullet -}+{\mathrm{H}}^{+}\to \mathrm{DMPO}-{\mathrm{H}}^{\bullet }$$
(8)

The spin-trapping reaction between radical (R•) and trap (T) is generally described by a second-order reaction that results in a spin-adduct ([SA]) formation. The amount of generated R• is reflected in the spin-adduct concentration (SA) if spin trapping is the only reaction to take place. Therefore, the rate of R• formation (F) is equal to the rate of spin-adduct formation. Nevertheless, the instability of spin adducts leads to their decay to EPR-silent products (P) at either an equal or higher rate (D, decay rate) than the spin-adduct formation rate. In general, the observed changes in spin-adduct concentration over illumination time can be due to spin-adduct formation and spin-adduct decay in the absence of other possible reactions (11).

$$F= \frac{d{R}^{\bullet }}{dt}=\frac{d[SA]}{dt}$$
(9)
$${R}^{\bullet }+T\to \mathrm{Spin \;adduct}\to P$$
(10)
$$\frac{d[SA]}{dt}=F-D$$
(11)

The spin-adduct decay is affected by the environment, such as the concentration of the spin trap (Roubaud et al., 1997). To find out the actual amount of trapped ROS upon illumination, we need to also study the decay process of spin-adducts in the absence of light, which was not performed in this study. Furthermore, spin trapping is not the only reaction that ROS undergoes. There are many other chemical reactions that the highly reactive free radicals easily participate in to form EPR silent products (XR). This in turn influences the spin-trapping efficiency (E), which can be described as the ratio of the spin-adduct formation rate to the rate of all other reactions (Vi) and free radical formation (13).

$${R}^{\bullet }+X\to \mathrm{XR}$$
(12)
$$E= \frac{\frac{d[SA]}{dt}}{{V}_{i}+ F}$$
(13)

Calculating trapping efficiency could become complicated if any reaction involved in Vi is other than a first-order one, leading to efficiency changes over time. Furthermore, the radical trapping process and achieving the actual spin-adduct concentration become more complicated if spin-adducts rapidly decompose by reacting with the surrounding free radicals, such as DMPO-OH adduct decay caused by a superoxide anion.

Conclusion

In this study, curcumin-based photodecontamination conditions were optimized using RSM, which enabled us to study the interactions between the factors influencing the photosensitization treatment. Among the studied curcumin formulations, aqueous ethanolic solutions showed a higher efficiency in B. cinerea photoinactivation. The concentration and irradiance dependency of curcumin phototoxicity were also investigated. Furthermore, the photodynamic generation of ROS in different curcumin aqueous environments was determined by EPR spectroscopy. The occurrence of type I photodynamic reaction was demonstrated by identifying OH• and O2•− radicals during the curcumin-based photosensitization. However, further studies are needed to elucidate the potential singlet oxygen generation, indicating a type II pathway of photodynamic reaction.