TCI Chemicals provided the curcumin (C2302-5G, purity > 97%, Montgomeryville, PA, United States). The nonionic surfactants, namely Surfynol 465 (S465) and Tween 80 (T80), were purchased from Shenzhen Vtolo Industrial Co. Ltd (Shenzhen, Guangdong, China) and Sigma-Aldrich (P1754, St Louis, MI, United States), respectively. The 5 mM sodium citrate buffer was prepared using sodium citrate (#775538, Fisher-Scientific, Waltham, MA, United States) and citric acid monohydrate (C7129, Sigma-Aldrich). Pharmco (Brookfield, CT, United States) was the supplier of the absolute ethanol (#111000200).

Curcumin was dissolved in ethanol to prepare a 4 mM curcumin stock solution. The curcumin in ethanol stock solution was titrated at 2.5 mL/min into the surfactant solution (S465 or T80) and stirred with a magnetic stir bar at 125 rpm to obtain a 20 µM curcumin-surfactant stock solution. The surfactant solution was prepared by dissolving S465 or T80 in 5 mM sodium citrate buffer at pH 3.5 for 20 min with agitation. After titration, the stock curcumin micelle solution was stirred for additional 15 min to determine whether there was a high enough surfactant concentration to solubilize the curcumin (Kharat et al., 2017). If the concentration of surfactant were too low, shearing would induce nucleation and crystallization of the curcumin. Filter-sterilization of curcumin micelles was done using a 0.45 µM syringe filter (Cat# 02915-22, Cole-Palmer, Vernon Hills, IL, United States) and stored at 4°C. The concentration of surfactant and curcumin used to produce curcumin-surfactant stock solutions was selected so that 1 µM curcumin and surfactant concentration below, at, or above the critical micelle concentration (CMC) could be obtained after dilution. The curcumin S465 micelles and curcumin T80 micelles average size was about 6.3 and 15 nm respectively as reported by Ryu et al. (2021).

The stability of encapsulated curcumin was determined by diluting the 20 µM curcumin-surfactant stock curcumin to 1:20 in 5 mM sodium citrate buffer (pH 3.5) and measuring its absorbance at 10 min intervals over 1 h with a UV-visible spectrophotometer (Shimadzu Scientific, Kyoto, Japan). The solutions were placed into 1-cm light path quartz cuvettes (FireflySci Inc., Staten Island, NY, United States), and their absorbance was measured at 425 nm and 5-nm slit. The fluorescence intensity of the curcumin solution (1 µM) was used to assess the lack of precipitation and stability of the encapsulated curcumin to oxidation. The fluorescence emission spectra of the curcumin and the curcumin-surfactant solutions over time were collected at an excitation wavelength of 365 nm over an emission wavelength range of 375–600 nm using a spectrofluorometer (Fluoromax 4, Horiba Scientific Inc., Edison, NJ, United States). The excitation and emission slits were set at 3 and 4 nm, respectively.

E. coli O157:H7 (ATCC-43888, non-toxigenic strain) and L. innocua Seelinger (ATCC-51742) were obtained from the American Type Culture Collection (Manassas, VA, United States). A mixture of tryptic soy broth (TSB; Cat# DF0064-07-6, BD Diagnostic Systems, Berkshire, United Kingdom) and 25% v/v glycerol was used to prepare the E. coli O157:H7 suspension. The L. innocua Seelinger stock solution was prepared in a mixture of tryptic soy broth with 0.01% yeast extract (TSBYE) and 25% v/v glycerol. Both stocks were stored under freezing conditions, i.e., −80°C. A loopful of frozen stock was used to produce the working solutions. After the inoculum was transferred to TSB, it was incubated at 37°C overnight. Then, the working stocks were streak plated onto MacConkey Sorbitol Agar (MCS; Cat# 279100, BD Diagnostic Systems) or Modified Oxford Agar (Cat# 222530, BD Diagnostic Systems) and stored at 4°C for a week.

In order to conduct the photoinactivation assay, a colony of E. coli O157:H7 or L. innocua was inoculated into TSB or TSBYE, respectively, and incubated at 37°C for 18 h on a 125 rpm shaker. For each experiment, after 18 h, the cultures were diluted, and their optical density at 600 nm (OD600) was measured and adjusted to 0.15 cm^-1 to verify that the initial culture contained about 9 log CFU/mL. Bacteria were then washed twice with PBS and centrifuged at 2000 g for 3 min. The OD600 of the suspensions was determined again after washing to confirm that no significant loss of bacteria took place. The bacterial counts were confirmed by dilution and plating on Tryptic Soy Agar (TSA; Systems Cat# 236920, BD Diagnostic Systems) using the spread plate method. The culture was diluted to obtain an initial inoculum of approximately 6 log CFU/mL.

Fluorescence Lifetime Imaging Microscopy (FLIM) was employed to study the characteristics of the local environment around the lumiphore (i.e., curcumin), which can help elucidate the partitioning of this PS into the microbial cells (Colaruotolo et al., 2021). 10 μL of each bacterial culture was placed on a 35 mm poly-D-lysine coated dish (P35GC-1.5-10-C, MatTek Life Sciences, Ashland, MA, United States). Cells were dried for 2 h under air circulation in a biological safety cabinet. Then, they were treated with 5 µM curcumin and S465 or T80 (near CMC). To assess curcumin partitioning, cells were washed with buffer after pre-incubation with 5 µM curcumin and S465 or T80 for 1 h and then dried for 10 min. The fluorescence lifetime measurements were recorded using a Nikon TiE confocal microscope with an A1 Spectral Detector (Nikon Instruments Inc., Melville, NY, United States) equipped with a FLIM/FCS module. Firstly, the traditional scanning confocal capability of the microscope was used to image the fluorescence of curcumin within each sample. Once a suitable field of view was identified and finely focused, the 405 nm pulsed input laser was selected, and the photons emitted were directed to the Becker-Hickl SPC-152/HPM-100–40 dual detector system (Boston Electronics). Curcumin fluorescence lifetime was measured using a 50 MHz pulse frequency. The recorded fluorescence decays were analyzed using the Becker & Hickl SPC Image fitting software. The mean fluorescence lifetime of the 256 × 256 pixel lifetime images was calculated, and the lifetime decay was characterized using a two-exponential model.

For irradiated or non-irradiated unwashed E. coli O157:H7 with curcumin or surfactants (S465 or T80) a live/dead cell assay was performed to measure the proportion of cells that exhibited a permeable membrane. Equal volumes of Syto 9 (S34854, Thermofisher, Waltham, MA, United States) and propidium iodide (25535-16-14, Sigma-Aldrich) were mixed in a centrifuge tube. 3 μL of the dye mixture was added to 1 mL of the bacterial suspension (7 log CFU/mL). The mixture was then incubated in the dark at room temperature for 15 min 20 μL of the sample was placed on a 35 mm poly-D-lysine coated dish. The fluorescence intensity of the cell-permeable dyes was acquired using a confocal microscope with structured illumination (A1R-SIMe, Nikon Instruments Inc., Melville, NY, United States). Once the appropriate field of view was identified and focused, a 488 nm laser was selected to excite the dyes. Obtained micrographs were analyzed using the NIS-Elements software (Nikon Instruments Inc.) to determine the proportion of cells in the sample with a permeable membrane as indicated by the emission intensity of the propidium iodide.

The percentage of injured E. coli O157:H7 cells in unwashed samples non-irradiated and after irradiation or non-irradiation with curcumin or surfactants (S465 or T80) was determined by comparing the number of colonies on selective (MSC and TSA+2.5%NaCl) and nonselective (TSA) agar. The percentage (%) of injury was calculated using Eq. 2 (Busch and Donnelly, 1992; Espina et al., 2016). P e r c e n t a g e   o f   i n j u r y ( % ) = 1 − #   o f   c o l o n i e s   o n   s e l e c t i v e   a g a r #   o f   c o l o n i e s   o n   n o n s e l e c t i v e   a g a r * 100 (2)

All experiments were performed in triplicate. All statistical analyses were done using p ≤ 0.05 to represent statistical significance.

Ryu et al. (2021) identified the synergistic photoinactivation of E. coli O157:H7 by curcumin when S465 was present in the 5 mM sodium citrate buffer at pH 3.5. Some of the potential mechanisms responsible for this synergistic effect are i) the increased solubilization of curcumin in the aqueous solution by the presence of the surfactant, which can cause the solution to become more transparent, reducing light scattering and increasing photoexcitation of the curcumin throughout the sample; ii) the improved stability of curcumin in the solution by the addition of the surfactant, prolonging its activity as a PS; and iii) the enhanced partition of curcumin inside the cell by the surfactant physically bringing the curcumin closer to the cell and also causing the cellular membrane to become more permeable. The absorbance of the curcumin-S465 micellar solution was significantly higher than that of curcumin dissolved in ethanol and diluted with buffer, which suggests less turbidity and scatter (Supplementary Figure S1). The surfactant micelles also prevented the nucleation of curcumin, causing an increase in its concentration in the solution and, consequently, its absorbance and fluorescence (Supplementary Figures S2, S3). Curcumin stability after dilution was assessed for 1 h based on the absorbance and the fluorescence emission intensity of diluted curcumin micelle solutions in 5 mM sodium citrate buffer (pH 3.5). The 20 µM stock curcumin micelle solution was diluted to 1 µM since this concentration was the one selected for the microbial inactivation assays. The encapsulated curcumin was more stable than the unencapsulated curcumin when either surfactant (S465 or T80) was present (Supplementary Figures S2, S3). Also, more curcumin was present in the solution when a higher concentration of either surfactant was added. It should be noted that the long term stability of the curcumin-surfactant solutions during storage was previously studied and reported, exceeding 30 days at room temperature and under refrigerated conditions (Ryu et al., 2021.) Having established that the presence of surfactants increased the stability and presence of curcumin in the testing solutions, in the consecutive sections, we focused on whether the micelles enhanced the partitioning of curcumin inside the cell.

The percentage of injured cells was evaluated by comparing the growth of the treated bacteria in TSA alone and in TSA with sodium chloride at 2.5%. Since L. innocua numbers after being treated with curcumin-S465 solutions and irradiated were below the limit of detection of the method, only E. coli O157:H7 was used for this part of the study. Cell injury was assessed to determine how many cells were still viable after treatments since the growth curves from oCelloscope™ could not distinguish whether the overall growth resulted from the contribution of one or many cells. The addition of curcumin increased the permeability of the cell membrane after irradiation, suggesting that the generated ROS reacted with components in the cellular membrane (Table 4). The irradiated cells treated with the solutions, including either curcumin or S465, had a higher percentage of injured cells than T80. However, it should be noted that cells treated with the curcumin-S465 micellar solution had the highest number of inactivated cells, indicated by the highest log reduction. Therefore, the reported overall number of cells that could be injured was affected by the overall higher lethality in the curcumin-S465 micellar solution. In general, the results were consistent with the rate and length of time for recovery of E. coli O157:H7 obtained using the oCelloscope™. Hence, the oCelloscope™ results correlate well with the proportion of cells that could have been inactivated or injured.

The contribution of the partitioned curcumin to microbial photoinactivation was studied by preincubating cells with curcumin-surfactant solutions for 1 h in the dark and washing the suspension with buffer before irradiation. During the washing step, the unbound or loosely-bound curcumin (or curcumin-surfactant) to the cell membrane was eliminated. The recovery of the washed cells was observed using the oCelloscope™. For all samples without curcumin, their growth parameters did not differ extensively after washing from those of the unwashed samples. In the case of the samples containing non-encapsulated curcumin or curcumin-T80 micellar solutions, the comparison of the parameters of the model (Eq. 1) revealed that the rate of recovery was slower, and the time required for recovery was longer for washed cells compared to that of the unwashed ones (Figures 2, 3; Supplementary Figures S5, S6; Tables 2, 3). A possible explanation for increased effectiveness in PDI in the washed samples treated with curcumin is that light absorption could have been higher due to the removal of extracellular curcumin, reducing scattering and increasing ROS generation and reactions with the cell. It should be noted that the curcumin average lifetime in the non-encapsulated and T80-curcumin samples before and after washing is very similar, which supports that some curcumin might have been internalized and cannot be removed by washing, remaining effective during growth (Figure 4). Conversely, in the case of the samples treated with curcumin-S465 micellar solutions, the recovery after washing is faster. This may have been due to the removal of curcumin encapsulated in surfactant micelles that could have been adjacent to the cell and the consequent interruption of its inactivating effect on the cells. This also aligns with a reduction of the curcumin average lifetime in the washed samples due to the removal of curcumin latched to the membrane, as will be further discussed in the next section. Therefore, the unbound curcumin or loosely-bound curcumin adjacent to the cell could have a significant role in the photoinactivation of bacteria. Again, the presence of surfactant increases the curcumin concentration in the proximity of the cells, which could be photoexcited and produce ROS. Dahl et al. (1989) evaluated whether the penetration of cell membrane by unencapsulated curcumin was necessary for microbial inactivation by irradiating Salmonella Typhimurium, E. coli, Sarcina lutea, and Staphylococcus aureus preincubated with curcumin for 60 or 90 min and subsequently washed. Their results indicate that washing removed unbound or loosely-bound curcumin and reduced inactivation efficacy, which is consistent with the results of the curcumin-S465 micellar solutions herein (Dahl et al., 1989).

Removing the unbound or loosely-bound portion of curcumin could also have prevented depletion of oxygen near the cells as photoinactivation depends on the presence of oxygen, which might have also contributed to the slower recovery in the non-encapsulated curcumin and the T80-curcumin samples. In the samples treated with curcumin-S465 micellar solution, the potential differential location of the PS (i.e., predominately in the membrane surface rather than internalized) resulted in a faster recovery in washed samples despite less oxygen depletion. The effect of S465 on the partitioning of curcumin in cell membranes was further investigated using FLIM and Live/Dead cell assay in the consecutive sections.

FLIM was used to gain insights into the location and state of the PS, i.e., curcumin, in washed and unwashed cells. Curcumin in the extracellular aqueous phase will exhibit a shorter lifetime than curcumin bound to or enclosed in cellular structures due to the restriction in movement of the fluorophore imposed by the surrounding structures. Curcumin in the extracellular aqueous phase is expected to have a shorter fluorescence lifetime as the excited curcumin would preferentially return to the ground state through non-radiative decay. Although the lifetimes of both curcumin bound to cellular structures and inside the cell are expected to be longer than those of curcumin in the extracellular matrix, a higher degree of restriction imposed by cellular structures in the former will determine longer lifetimes for curcumin molecules bound to, for example, the cell membrane. Hence, the longer fluorescence lifetimes suggest a different environment around curcumin than free ones in extracellular fluid, i.e., bound or inside the cell. For unwashed E. coli O157:H7 and L. innocua treated with curcumin or curcumin-surfactant solutions (near CMC), the averaged fluorescence lifetimes decreased in the following order: S465 > T80 >and unencapsulated curcumin, respectively (Figure 5; Supplementary Figure S7). Fluorescence lifetimes can be correlated to photoinactivation efficiency since longer residences in the excited state could result in a more sustained production of ROS (Turro, Ramamurthy and Scaiano, 2017). Therefore, the order of the length of fluorescence lifetime was consistent with the growth parameters reported in Tables 2, 3, e.g., less and slower growth for samples treated with curcumin-S465 micelles. The reason curcumin-S465 solutions had the longest fluorescence lifetime may be due to the different local environment around curcumin when combined with the different types of surfactants. Chignell et al. (1994) reported that the ¹O₂ phosphorescence generated by curcumin encapsulated in Triton X-100 micelle could be detectable, while there was no ¹O₂ phosphorescence emission in samples where curcumin was encapsulated in sodium dodecyl sulfate (SDS) micelles (Chignell et al., 1994). The authors attributed this effect to curcumin being located at the hydrophobic region of Triton X-100 micelle, as curcumin in Triton X-100 exhibited similar photophysical properties as when dissolved in toluene (Chignell et al., 1994). Therefore, the free volume and packing around curcumin could have been different when encapsulated in S465 or T80 since these surfactants have different structures and micellization properties, leading to differences in the local environment around curcumin. Also, they point out that ¹O₂ production by curcumin depends on the hydrophobicity and hydrophilicity of the environment as it influences the predominant tautomer formed (enol vs. keto) which affects its ability to generate ¹O₂ (Chignell et al., 1994).

After removing extracellular and loosely-bound curcumin by washing, curcumin remained active to a lesser degree, indicating that the remaining curcumin may have partitioned inside both gram-positive and gram-negative cells. The fluorescence lifetimes of curcumin in washed E. coli O157:H7 treated with unencapsulated curcumin were longer than those treated with curcumin-surfactant solutions produced with both T80 and S465 (Figure 4). However, the differences between fluorescence lifetimes of curcumin in L. innocua were not significantly different between the treatments (Supplementary Figure S7). This may have been due to the different degrees of restriction on partitioned curcumin imposed by different membrane structures in gram-positive and gram-negative cells. Also, the order of length of fluorescence lifetime of curcumin changed when cells were washed compared to when they were not (Unencapsulated > T80 = S465 vs. S465 > T80>Unencapsulated). Fluorescence lifetime is independent of the fluorophore concentration. This helps us assume that the observed changes in the photophysical properties are not related to an increase/decrease in curcumin concentration during the washing step. Demidova and Habmlin (2005) reported that the type of PS affects its ability to bond with cell membranes and could have different efficacy toward gram-positive and gram-negative bacteria (Lauro et al., 2002; Demidova and Hamblin, 2005). They show that PS-poly-L-lysine chlorin conjugates are still effective even after preincubation for 20 min and washing on both gram-positive and gram-negative cells. Other PS, such as Rose Bengal and toluidine blue, were rendered ineffective or became less effective after washing, respectively (Demidova and Hamblin, 2005). Therefore, surfactants or washing may have promoted the loss of unbound or loosely-bound curcumin adjacent to the cell, leading to inefficient photoinactivation when washed. We could conclude that loosely bound and unbound encapsulated curcumin adjacent to the cell may play a significant role when used for photoinactivation in the case of curcumin.

A Live/Dead cell assay was used to determine whether the cellular membrane became permeable or not after each treatment. Since L. innocua was vulnerable to S465 even without irradiation, only unwashed E. coli O157:H7 was used to observe the effect of different treatments on its membrane. E. coli O157:H7 treated with S465 had the highest number of cells with increased membrane permeability, whether the samples were irradiated or not (Figure 5; Table 4). Conversely, E. coli O157:H7 treated with T80 had the lowest number of cells with a permeable membrane. Data on the recovery of cells after treatment from the oCelloscope™ indicated that photoinactivation of cells by curcumin was non-lytic. Also, it showed that increased membrane permeability did not necessarily result in cell inactivation. If this was the case, the samples treated with S465 solutions should have also shown higher inactivation even before irradiation, according to the results in Tables 2, 3. As expected, the irradiated cells treated with curcumin had more permeable membranes than the non-irradiated ones (Figure 5). This indicates that ROS produced by curcumin could also disrupt the cell membrane, causing it to be permeable. Interestingly, when a curcumin-S465 solution was used to treat the cells, the number of cells with permeable membranes was less before irradiation than when cells were treated with curcumin or S465 alone. However, once irradiated, the number of cells with permeable membranes increased to a similar level as cells treated with S465 alone. This could indicate that when cells were exposed to a curcumin-S465 micellar solution, most of the curcumin or S465 did not react with the membrane as they formed the micelle. However, once irradiated, the micelles adjacent to the cell may have disrupted the cell membrane by producing ROS within a distance short enough for ¹O₂, the most stable form of ROS, to be effective. ¹O₂ half-life is typically lower than 3.5 microseconds, allowing it to diffuse distances shorter than 100 nm before being inactivated (Macdonald and Dougherty, 2001; Schweitzer and Schmidt, 2003; Dysart and Patterson, 2005). In addition, it may explain why the fluorescence lifetimes were longer for cells treated with the curcumin surfactant solution produced with S465, as the curcumin could have localized differently in the cell membrane that became more permeable. Cells treated with curcumin-T80 solution had the fewest cells with permeable membranes compared to other treatments. However, the results of a previous study and the data in Figure 5 indicate that the curcumin-T80 solution had similar antimicrobial efficacy to curcumin alone (Ryu et al., 2021). This suggests that cell membrane permeation may contribute but may not be necessary to inactivate the cells. Past studies also suggest that T80 has no inhibitory effect but could be used as a carbon source by microorganisms (Inouye et al., 2001; Ryu et al., 2018). Therefore, the reason curcumin-S465 solutions were effective against microorganisms may have been due to S465 being able to disrupt the membrane of the cell and facilitate the differential partition of the PS.