Introduction
With the increasing awareness of fresh produce safety and demand for convenient fresh products with high-quality nutritional and sensory attributes, non-thermal technologies have gained growing attention in food research in both academia and the commercial sector (Fortuny et al., 2014; Gómez-López et al., 2021). Non-thermal atmospheric plasma (NTP) is one such recent example. NTP is defined as partially ionized gas, which consists of a plethora of highly reactive species, such as free radicals, electrons, positive and negative ions, excited and non-excited atoms, gas atoms, molecules, photons, and quanta of electromagnetic radiation (UV photon and visible light) at approximately room temperature (Niemira, 2012). Accordingly, NTP has gained prominence in food research in light of its suitability to treating heat-sensitive materials (Jadhav et al., 2021); as examples, dielectric-barrier discharge (DBD) and plasma jet have been commonly used for food study because of their simple configuration and ease of control (Misra et al., 2016). During NTP treatment, reactive oxygen species (ROS) (e.g., •OH, O2−•, HO2•, 1O2, H2O2, O3) and reactive nitrogen species (RNS) (e.g., •NO, OONO−, NO2−, NO3−, NH4+) are considered to have a crucial role in antimicrobial processes (Zhou et al., 2020). Applications of NTP have been examined and show promising results against microorganisms on different types of matrices (Deng et al., 2007; Handorf et al., 2021; Liao et al., 2020; Min et al., 2016; Niemira and Sites, 2008; Pasquali et al., 2016; Rothwell et al., 2021; Smet et al., 2017; Zhou et al., 2015).
In food processing, NTP generation systems are classified by how the food is exposed to the generated plasma (Sarangapani et al., 2018): direct treatment and indirect treatment, with the most popular indirect treatment being plasma-activated water (PAW). In general, the antimicrobial efficacy of NTP depends on several parameters, including plasma source design, operating conditions, feeding gas, exposure time, type of microorganisms, and the matrices to which the microorganisms are attached (Guo et al., 2015). For PAW, in particular, the properties of the liquid, and how the plasma is introduced to the water or liquid samples (submerged beneath the water or exposed to the liquid surface) also influence the antimicrobial activity of the NTP (Bradu et al., 2020). These factors will further affect the reactions of reactive species; hence, the chemical composition of NTP is quite complex.
The high diffusivity of direct plasma treatment (e.g., gliding arc) might alter the food surface and introduce some side effects, including color deterioration and nutrient loss, resulting from plasma-induced surface etching and degradation of bioactive compounds (Baier et al., 2015). In regards to PAW research, previous reports have used plasma treatments to obtain PAW by action from above the water surface, generally relying on long-lived active species (Kamgang-Youbi et al., 2007). PAW application for microbial decontamination has demonstrated promising results, in both in vitro studies (Traylor et al., 2011) and in decontamination of fresh produce (Ma et al., 2015). However, ensuring uniform exposure of plasma species to food products, particularly fresh produce, with complex surface morphology (e.g., pores and rough surfaces) is a challenge that remains to be solved (Chen et al., 2020).
In previous studies, a plasma-bubbling system was introduced to enhance the distribution of reactive species in the liquid system, thereby obtaining superior uniformity of the antimicrobial effect inside the plasma-treated sample (Aparajhitha and Mahendran, 2019; Lokeswari and Mahendran, 2022). This technology has the potential to improve the effectiveness of microbial control. However, studies reporting plasma treatment with a bubble-assisted approach are limited. In particular, the comparison between the direct treatment of plasma-bubbling and PAW from a single plasma reactor has not been clear until now. In addition, while previous studies have acknowledged the bioactivity of plasma-bubbling for water purification (El Shaer et al., 2020) and pathogen inactivation (Suenaga et al., 2022), the contribution of the specific generated reactive species has not been thoroughly investigated. Therefore, this study aimed to evaluate the in vitro bactericidal effect of the plasma-bubbling system in comparison with PAW. In addition, the involved reactive species of the bactericidal activity were investigated using electron spin resonance (ESR). The results obtained in this study are expected to provide the basic findings needed to investigate more effective treatment methods for food sanitization in future studies.
Materials and Methods
Bacterial strain and culture conditions E. coli O157:H7 (CR-3, isolated from bovine feces) was used as a model organism. One loop of E. coli O157:H7 was inoculated into 30 mL tryptic soy broth (TSB) (Nissui Pharmaceutical Co., Ltd., Tokyo, Japan) and cultivated at 37 °C for 24 h. The cells were harvested using a refrigerated centrifuge at 4 400 rpm for 10 min at 4 °C and washed with sterile sodium chloride solution (0.85 % NaCl). This process was repeated twice. Finally, the pelleted cells were resuspended in 30 mL sterile 0.85 % NaCl.
The concentration of the bacterial suspension was around 8.0 log colony-forming units (CFU)/mL, which was enumerated by spread plating 0.1 mL of the tenfold serial dilutions on tryptic soy agar (TSA) plates (Nissui Pharmaceutical Co., Ltd.), which were incubated at 37 °C for 24 h.
Plasma-bubbling system setup A non-thermal multi-gas plasma jet at atmospheric pressure was used as the plasma source, operated at 9 kV and 16 kHz, and provided by an AC power supply (Plasma Concept Tokyo, Inc., Tokyo, Japan). Pure oxygen was used as the feeding gas at a flow rate of 8 L/min. The plasma jet was connected to a cylindrical gas injection tube (diameter × length = φ8 mm × 180 mm, filter size = φ15 mm × 20 mm; AS ONE Corporation, Osaka, Japan) to create the bubbling system. Bubble sizes were measured by ImageJ software (ver 1.54d) after capturing images of the bubbles using a digital camera (IXY 200; Canon Inc., Tokyo, Japan), according to the method of Aparajhitha and Mahendran (2019) with some modifications. Twenty randomly selected bubbles were analyzed and the average bubble size was 0.7 ± 0.2 cm. A schematic diagram of the plasma jet device is shown in Fig. 1.
Plasma treatment methods for bacterial suspensions The plasma treatment was performed using two processes. For both treatments, the exposure time was measured starting from the plasma ignition. The first approach was defined as the direct treatment of plasma-bubbling. A 50-mL aliquot of E. coli suspension added to a 100-mL glass beaker was treated directly by the plasma-bubbling system for 30, 60, 120, and 180 sec. The distance between the filter tip and the base of the beaker was kept at approximately 5 mm. An untreated E. coli suspension was used as the control. A 1-mL aliquot of the plasma-treated bacterial suspension and the control were centrifuged at 13 000 rpm for 5 min at room temperature. The supernatant was subsequently used for hydrogen peroxide measurement and ESR analysis.
The other treatment involved the production of plasma-activated water (PAW). This method was performed by placing the injection tube under distilled water and exposing the water to the plasma-bubbling system. The distance from the filter tip to the bottom of the beaker was kept constant at approximately 5 mm. In this study, 50 mL of distilled water was placed in a 100-mL glass beaker. Plasma activation was performed for 3, 5, 7, and 10 min. After the prescribed time, the plasma treatment was stopped, and the obtained aqueous solution was defined as PAW. A 9-mL aliquot of PAW was immediately pipetted into a sterile 15-mL centrifuge tube, to which 1 mL of the E. coli suspension was added, resulting in an initial concentration of 7.0 log CFU/mL. This mixture was left for 5 min on a high-speed shaker (AS ONE ASCM-1, Osaka, Japan) at room temperature. For this treatment, 1 mL of E. coli suspension mixed with 9 mL of distilled water was used as the control.
Physicochemical properties of PAW The oxidation-reduction potential (ORP), pH value, temperature, and dissolved ozone concentration were measured immediately after the plasma activation process. The pH and temperature were measured using a hand-held pH meter (Testo 206 pH2; Testo SE & Co., KGaA, Schwarzwald, Germany). ORP levels were measured using an ORP meter (Lutron Electronic, Taipei, Taiwan).
Colorimetric test kits were chosen to measure the concentrations of dissolved ozone and hydrogen peroxide in the PAW. Dissolved ozone concentration was analyzed using a dissolved ozone meter (O3-3F; KRK, Saitama, Japan), and hydrogen peroxide was measured using a hydrogen peroxide assay kit (Cosmo Bio Inc., Tokyo, Japan). Both tests were conducted according to the protocols provided by the kit manufacturers.
Determination of E. coli viable counts To assess bacterial inactivation following plasma treatments, untreated and treated samples were serially diluted in phosphate-buffered saline (3M, Seoul, Korea) following each plasma treatment and plated on TSA plates. The number of surviving viable cells was determined by plate counts following incubation at 37 °C for 24 h.
ESR analysis of plasma-treated bacterial suspension and PAW To detect short-lived species contained in PAW and the plasma-treated bacterial suspension, two different spin-trapping reagents were used: 2-(5,5-dimethyl-2-oxo-2λ5-[1,3,2] dioxaphosphinan-2-yl)-2-methyl-3,4-dihydro-2H-pyrrole 1-oxide (CYPMPO) to detect •OH radicals, and 2,2,6,6-tetramethyl-4-piperidone (TMPD) to detect 1O2.
Four milligrams of CYPMPO and 3 mg of TMPD were added to 200 μL of each PAW sample and the supernatant from the plasma-treated bacterial suspension. Those mixtures were analyzed using an X-band EPR spectrometer (EMX-Plus, Bruker BioSpin, Karlsruhe, Germany) with measurement conditions according to Kameya et al. (2019) as follows: room temperature (23−25 °C); magnetic field range, 3 521 ± 100 G; data points, 1 000; microwave power, 1 mW; time constant, 20.48 μs; conversion time, 20 ms; sweep time, 20 s; scan, 10 times; modulation intensity, 1.00 G.
Scavenger assay A scavenger assay was conducted to verify the contribution of ROS to the bactericidal effect of plasma treatments. Filter-sterilized chemical scavengers were added to the bacterial suspension and distilled water before plasma treatment: d-mannitol (scavenger for •OH) and l-histidine (scavenger for 1O2) (Fujifilm Wako Pure Chemical Corporation, Osaka, Japan). The concentration of both scavengers used for PAW and direct treatment was 15 mM and 30 mM, respectively.
The type and final concentration of the scavengers were adapted from previous studies (Aboubakr et al., 2016). A scavenger control test was also conducted to ensure there was no bactericidal effect of the scavenger itself on the bacteria, in which the bacteria were suspended in the scavenger solution without exposure to plasma treatment. The number of surviving viable cells after each treatment was calculated to assess the suppressive effect of each scavenger.
Direct treatment: E. coli suspended in 30 mM d-mannitol, 30 mM l-histidine, and 0.85 % NaCl (without scavengers) were treated with direct plasma-bubbling for 3 min.
PAW treatment: 15 mM d-mannitol, 15 mM l-histidine, and distilled water were activated by plasma for 5 min. Then, 1 mL of the E. coli suspension was homogenized with 9 mL plasma-treated scavengers for 5 min.
Statistical analysis All experiments were carried out in triplicate. Results are shown as the mean and standard deviation of these determinations. The results were subjected to one-way analysis of variance (ANOVA), followed by the Tukey HSD test using RStudio version 4.0.2 (The R Foundation for Statistical Computing, 2020). Differences among mean values were considered significant at p < 0.05.
Results and Discussion
Physicochemical properties of PAW
a) pH The pH of PAW was slightly modified after exposure to plasma-bubbling for 10 min, from 5.17 ± 0.32 to 4.65 ± 0.20, showing that the plasma treatment did not acidify the water significantly (Fig. 2A). Our study used pure oxygen as the primary feeding gas. In the case of oxygen plasma, the pH level does not change significantly due to the lack of nitrogen species (Bolouki et al., 2021). Burlica et al. (2006) reported similar findings, in which the use of oxygen as the feeding gas resulted in the slowest rate of pH decrease compared to air or nitrogen carrier.
However, our finding was inconsistent with some previous studies, which suggested that plasma treatment decreases the pH of water. For example, some studies using atmospheric pressure plasma jet (APJJ) and Ar/O2 as the working gas reported that the pH value of PAW decreased with a more extended activation time (Ma et al., 2015; Shen et al., 2016). Those studies suggested that this was caused by nitric acid, peroxynitrous acid, and hydrogen peroxide contained in PAW (Liu et al., 2021; Oehmigen et al., 2010). These discrepancies could be attributed to differences in plasma generators and variations of the feeding gas (Bruggeman and Leys, 2009; Tian et al, 2015).
b) ORP An ORP value indirectly indicates the overall level of ROS in a solution (Tian et al., 2015; Wu et al., 2017). As shown in Fig. 2B, compared to the control (untreated distilled water), whose value was approximately 176 ± 21 mV, the ORP of PAW reached 381 ± 21 mV after plasma activation for 3 min. Plasma activation significantly affected the ORP value until 5 min (p < 0.05). The ORP value of PAW increased to 557 ± 24, 595 ± 20, and 556 ± 23 mV after 5, 7, and 10 min, respectively. This increase in ORP value was attributed to water ionization and short- and long-lived ROS. Other similar findings were also reported in previous studies (Guo et al., 2017; Lin et al, 2020).
c) Dissolved ozone concentration Ozone concentration was measured using the diethyl-p-phenylenediamine (DPD) method (Okubo et al., 2019). DPD reacted with the produced ozone from the plasma treatment, resulting in a color change in the PAW. Compared to the control, plasma-bubbling treatment increased the dissolved ozone concentration (p < 0.05) (Fig. 2C). The maximum ozone concentration was 1.09 ± 0.23 mg/L after 7 min of treatment.
As described by Whitehead (2016), competitive reactions between the formation of ozone molecules and the subsequent dissociation reaction into oxygen gas are expected to occur, yielding the depleted dissolved ozone concentration at 10 min of treatment. Previous research by Pavlovich et al. (2013) and Patange et al. (2019) used air plasma generated from indirect dielectric barrier discharges. Both showed that the ozone concentration of the PAW remained constant after 120 sec of plasma treatment, reaching 3 mg/L and 6 mg/L, respectively. Notwithstanding the similar trend in results, the different configurations of plasma devices and their varied design and operation can also lead to different results with different mechanisms.
d) Temperature Figure 2D shows the change in temperature of PAW after plasma-bubbling exposure. After 10 min of treatment, the temperature of PAW was approximately 19.1 ± 0.2 °C. The temperature was similar to that of the untreated water, which was 20.7 ± 0.4 °C. While no temperature increase was observed, it could be confirmed that the reduction in bacterial numbers in this study was not caused by a heating effect from the experimental setup.
Detection of short-lived reactive oxygen species Radicals with unpaired electrons are paramagnetic and have short lifespans (in micro or millisec). To undertake this lifespan issue, electron paramagnetic resonance (EPR) or electron spin resonance (ESR) is conjugated with the spin-trap technique. The radicals react with a specific compound, called a spin-trap reagent, forming a long-lived adduct to be analyzed using ESR spectroscopy. The signal intensity of the spin adducts is directly proportional to the concentration of the formed free radicals R• (Janzen, 1971).
The spectra obtained from ESR are further used to identify the type of ROS by the shape and numerical parameters, such as g-value and hyperfine coupling constant (hfcc) (Kohno, 2010). The g-value is calculated from the resonance magnetic field and the resonance frequency observed as ESR signals. hfcc values are characterized as the splitting of an ESR spectrum and determined by analyzing the splitting of the spectrum of the spin adduct.
CYPMPO was added to each sample of PAW and the supernatant of plasma-treated bacteria. The mixture was analyzed further using ESR. From the ESR spectra, the spin adduct CYPMPO-OH (g = 2.008) was able to be distinguished from the consecutive peaks in the spectrum with the following hfcc: isomer 1: Ah = 1.36 mT, An = 1.37 mT, Ap = 4.89 mT; isomer 2: Ah = 1.23 mT, An = 1.35 mT, Ap = 4.70 mT. These hfcc values are similar to the •OH radical adduct spectrum in previous studies (Kamibayashi et al., 2006; Oowada et al., 2012). In addition, using similar protocols, 1O2 was also identified using TMPD as the spin trap on both samples of the two plasma treatment approaches (g = 2.004). According to the obtained ESR spectra, the hfcc for 1O2 were as follows: Ah = An = Ap = 1.45 mT, which are comparable to the reported values (Gromov et al., 2021). Therefore, these findings suggested that •OH and 1O2 were confirmed to exist in PAW and the supernatant of plasma-treated bacterial suspension in our study. However, their signal intensities and behavior over time differed. According to the ESR spectra (data not shown), the peaks of both ROS adducts were identified more clearly in PAW compared to the direct treatment samples. This observation implied that the amount of •OH and 1O2 was more abundant in PAW than in the direct treatment.
Figure 3 shows an increasing trend of signal intensity of CYPMPO-OH and TMPD-1O2 adducts after plasma activation in PAW of up to 2.20 ± 0.16 and 2.00 ± 0.07, respectively, at 7 min. On the other hand, the intensity of both ROS adducts decreased at 10 min of plasma activation time. Therefore, it could be assumed that at this activation time, •OH and 1O2 still existed in PAW. However, some of the species might have transformed or contributed to the formation of other reactive long-lived species, such as ozone or hydrogen peroxide.
In contrast, for the direct treatment (Fig. 4), the signal intensities of CYPMPO-OH and TMPD-1O2 increased significantly with the plasma treatment time. •OH began to appear on the spectrum after 30 sec of plasma treatment, while 1O2 was subsequently identified after 60 sec. The intensity of both ROS adducts continued to increase up to 1.09 ± 0.11 for CYPMPO-OH and 0.45 ± 0.09 for TMPD-1O2 at 180 sec. Due to their high reactivity, one might speculate that these reactive species attacked the bacteria in the suspension immediately following plasma discharge. For this reason, these ROS could not be identified by ESR before 60 sec and 120 sec of plasma treatment for •OH and 1O2, respectively.
Hydrogen peroxide concentration H2O2 is considered to be associated with antimicrobial properties in several non-thermal plasma designs and reactors (Traylor et al., 2011; Vlad et al., 2019; Zhang et al., 2013). The concentration of H2O2 in the liquid sample was quantified colorimetrically based on complex formation between xylenol orange and ferric iron. This complex was produced by peroxide-dependent oxidation of ferrous iron (Gay et al., 1999).
H2O2 was not detected in the PAW samples (no significant differences with the untreated samples) regardless of the activation time. Vaka et al. (2019) reported that the initial volume of water, generation mode (i.e., the introduction of the bubbling system), and the relatively long activation time in this study might affect the presence of H2O2 in PAW. For instance, hydrogen peroxide was identified (120 μM) from PAW generated by a similar reactor geometry operating at 9 kV and 16 kHz for 3 min using a gas mixture of He and 3 % O2 (Kawano et al., 2018). However, a significantly lower treatment volume (100 μL) was used compared to the values used in this work. In addition, Gorbanev and co-workers (2016) demonstrated that water vapor in the feed gas might also contribute to the amount of H2O2 in PAW generated by a plasma jet. Their observation showed that increased humidity in the feeding gas also increases the concentration of H2O2. Their finding suggests that H2O2 is not formed by dissociation of H2O followed by recombination of the aqueous hydroxyl radicals. Instead, H2O2 is formed in the gas phase and diffuses into the liquid sample (Qi et al., 2020). Since no water vapor was introduced in the feeding gas in this study, the latter mechanism is also likely to occur, and a minor amount might be present or below the detection limit.
In contrast to the PAW results, H2O2 was detected in the bacterial suspension supernatant after direct treatment, and the concentration increased with more prolonged plasma exposure. However, as illustrated in Fig. 5, the detected H2O2 was at a low concentration (< 0.05 mg/L) and was insufficient to kill the bacteria. The presence of low H2O2 concentrations might be due to damaged bacterial cells following plasma-bubbling treatment; the compound might be released from the dead cells as a byproduct (Seaver and Imlay, 2001). While it is commonly known that H2O2 acts as an antimicrobial agent, concentrations of mM to hundreds of mM are typically required (Watts et al., 2003). Therefore, these findings might suggest that the low H2O2 dose had no bactericidal effect in our experiment.
Bactericidal effect of different plasma-bubbling treatments The presence of microorganisms, including bacteria, is a crucial factor in the contamination and spoilage of fresh produce (Tournas, 2005). Therefore, we evaluated bacterial survival following exposure to PAW and direct treatment with the plasma-bubbling system. Figure 6 presents the number of E. coli O157:H7 surviving plasma treatment (log CFU/mL) in relation to plasma activation time (min). PAW, the result of treating distilled water with the plasma-bubbling system for 3–10 min, significantly reduced E. coli viable cell counts from 7.24 log at 0 min to 3.91 log after 5 min contact time. As a powerful oxidant, the presence of ozone in PAW induced the highly positive ORP (Fig. 2B and C). This synergistic property was responsible for the decrease in E. coli viable cell counts. The high ORP indicated that a significant amount of plasma species was generated in PAW. These strong oxidizing species pull electrons away from the bacterial cell membrane (McPherson, 1993; Suslow, 2004), compromising cell integrity, causing leakage of cellular components, and accelerating cell death (Gómez-López et al., 2021).
As for the direct treatment approach, the time necessary to eliminate bacteria was shorter than that for PAW. As shown in Fig. 7, treatment of E. coli with direct plasma-bubbling for 120 sec led to a 5-log reduction of the viable cell counts from 8.21 log at 0 sec. In addition, the viable cell counts were under the detection limit after 180 sec of treatment. This finding agrees with the results from the previous section, where increasing treatment time generates more plasma-generated species, further decreasing the bacterial population.
The significant differences in the results for bacterial inactivation by the two plasma-bubbling approaches were related to the reactions with the available reactive species. When the plasma discharge was switched off after producing the PAW, some short-lived species might decay or be involved in secondary processes and yield ozone as the secondary species. These relatively long-lived species are suggested to be more critical in microbial inactivation.
On the other hand, direct treatment with plasma-bubbling generates more active species, including free electrons, monoatomic compounds, and charged particles. In addition, other physical mechanisms might also be involved in bacterial inactivation. This includes UV/VUV photons and ions through charge exchange reactions and photolysis (Laroussi, 2005) and changes in osmotic pressure (Lukes et al., 2012). More importantly, there was no delay time between exposing the bacterial suspension to the plasma generated species compared to PAW treatment.
There have been several studies demonstrating the antibacterial activity of plasma against E. coli as the model microorganism, in which different configurations of plasma reactors were used with variations in process parameters (Ikawa et al., 2016; Oehmigen et al., 2010; Oehmigen et al., 2011; Suwal et al., 2019; Zhou et al., 2018). Results from previous research using a similar plasma jet reactor also agreed with this study, where the plasma treatment resulted in a 6 log reduction in E. coli numbers within 1 min (Takamatsu et al., 2015). Their higher antibacterial activity compared to that of PAW in this study could be attributed to differences in sample size, working gas, and experimental setup.
In the tests with bacterial suspensions, the direct treatment with plasma-bubbling showed higher bactericidal performance compared to PAW. Therefore, when attempting to sanitize food products using NTP treatment, plasma-bubbling would have the potential to significantly reduce bacterial populations. On the other hand, food products contain a large amount of organic matter, which may interfere with the bactericidal effect of plasma-generating species, and further experiments using a variety of food products should be conducted.
Scavenger assay In order to demonstrate the role of •OH and 1O2 in bacterial inactivation by the plasma-bubbling system, d-mannitol and l-histidine were used to scavenge •OH and 1O2, respectively, and none of the scavengers possessed bactericidal effects. If the addition of d-mannitol and l-histidine led to increased bacterial survival, it was concluded that the particular species contributes to the bactericidal effect of the plasma treatment. A higher concentration of scavengers was introduced for direct treatment with plasma-bubbling in light of the more significant bactericidal effect shown in the previous result.
As depicted in Fig. 8A, d-mannitol and l-histidine suppressed the bacteria inactivation of PAW. However, with direct treatment, only l-histidine was capable of hindering bacterial activity (Fig. 8B). This result indicates that both •OH and 1O2 play a significant role in the bactericidal effect of PAW, whereas 1O2 was found to be the main functional species in direct treatment with plasma-bubbling.
•OH and 1O2 are highly reactive with biological substances, and have been reported to be important in microbial inactivation by non-thermal plasma (Aboubakr et al., 2016; Kondeti et al., 2018; Takamatsu et al., 2012; Xu et al., 2020). As the most reactive among all reactive oxygen and nitrogen species, •OH can react with proteins and lipids on the cell membrane, damaging the components of intracellular materials, such as DNA. 1O2 is also reactive and can initiate lipid oxidation (Pryor, 1986). While •OH is reported to react non-selectively with many chemical bonds, 1O2 might be more crucial in eliminating bacteria in this plasma-bubbling system. Wu et al. (2012) suggested that 1O2 has a relatively long lifespan compared to •OH and more easily diffuses into the bacterial cell membrane. In addition, it has been reported that 1O2 is often found in higher concentrations compared to •OH in several plasma generation systems, particularly plasma jet and DBD (Guo et al., 2018).
