Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Electron Beam Susceptibility of Enteric Viruses and Surrogate Organisms on Fruit, Seed and Spice Matrices

Electron Beam Susceptibility of Enteric Viruses and Surrogate Organisms on Fruit, Seed and Spice... The objective of this study was to use high-energy electron beam (HEEB) treatments to find surrogate microorganisms for enteric viruses and to use the selected surrogates as proof of concept to investigate low-energy electron beam (LEEB) treatments for enteric virus inactivation at industrial scale on frozen blueberries. Six food matrices inoculated with HAV (hepatitis A virus), MNV S99 (murine norovirus), bacteriophages MS2 and Qβ, and Geobacillus stearothermophilus spores were treated with HEEB at 10 MeV using 4, 8 and 16 kGy doses. G. stearothermophilus spores showed the highest inactiva- tion on all matrices except on raisins, with a dose-dependent effect. HAV reached the maximum measurable log reduction (> 3.2 log ) when treated at 16 kGy on raisins. MNV showed the highest resistance of all tested microorganisms, inde- pendent of the dose, except on frozen blueberries. On frozen blueberries, freeze-dried raspberries, sesame seeds and black peppercorns, HAV showed a mean inactivation level in between those of MS2 and G. stearothermophilus. Based on this, we selected both surrogate organisms as first approximation to estimate HAV inactivation on frozen blueberries during LEEB treatment at 250 keV using 16 kGy. Reductions of 3.1 and 1.3 log were measured for G. stearothermophilus spores and MS2, respectively, suggesting that a minimum reduction of 1.4 log can be expected for HAV under the same conditions. Keywords Electron beam · HEEB · LEEB · Enteric viruses · Surrogates · Food Introduction pressure (HHP), ultra violet light (UV-C), cold plasma and irradiation, have the potential to achieve this goal (Knorr Hepatitis A virus (HAV) and human norovirus (hNoV) are and Watzke 2019). However, virus inactivation by UV-C responsible for food-borne outbreaks linked to fresh pro- light is very efficient in clear liquids such as water (Tree duce, ready-to-eat foods and shellfish worldwide (Alegbel- et al. 2005; Wiedenmann et al. 1993), but of limited use on eye et al. 2018; Bosch et al. 2018; Miranda and Schaffner complex food surfaces such as berries (Butot et al. 2018). 2019). Outbreaks are generally associated with food that Virus inactivation by HHP is only applicable to high mois- undergoes minimal processing, such as oysters, fresh-cut ture foods, such as oysters and fruit purees (DiCaprio et al. lettuce, frozen berries and very recently dried dates (Ethel- 2019; Kingsley and Chen 2009). For fragile or dry foods, berg et al. 2010; Tavoschi et al. 2015; Thebault et al. 2013; such as berries, herbs and spices, other technologies need Anonymous 2018). Consumer demands for minimally to be explored. processed foods with fresh-like quality are on the rise and Irradiation appears to be a good candidate technology, food processes must adapt to the preferences, acceptance but data describing the effectiveness of irradiation on food and needs of the consumer (Knorr and Watzke 2019). In against viruses, especially HAV, are scarce. Besides X-ray, this context, innovative approaches are needed to minimize two major irradiation technologies have been explored, the risk of viral food-borne outbreaks while retaining good gamma irradiation and high-energy electron beam (HEEB) food quality. Non-thermal processes, e.g. high hydrostatic (Farkas 1998; Moosekian et al. 2012; Pillai and Shayanfar 2017; Ravindran and Jaiswal 2019). Food matrix is likely to provide increased survival for viruses during irradiation, * Sophie Zuber but a major hurdle when performing inactivation studies sophie.zuber@rdls.nestle.com with viruses on food is the low recovery efficiency which is Nestlé Research, Institute of Food Safety and Analytical often encountered, leading to a low maximum measurable Science, 1000, 26 Lausanne, Switzerland Vol:.(1234567890) 1 3 Food and Environmental Virology (2021) 13:218–228 219 log reduction (Bosch et al. 2018; Butot et al. 2014). For food surfaces such as raisins (Etter et al. 2018). Inactivation example, at a HEEB dose of 12 kilo Gray (kGy), inactiva- of enveloped viruses such as influenza A in liquids has been tion of MNV (murine norovirus used as hNoV surrogate) in demonstrated (Etter et al. 2018), but no data are available on phosphate-buffered saline (PBS) reached 6.4 log , while on the effect LEEB has on non-enveloped enteric viruses such strawberries a maximum reduction of 2.2 log was observed as HAV and hNoV. The only industrial scale system cur- (Sanglay et al. 2011). Similarly, after a 11.9 kGy treatment rently available to the food industry is the Bühler AG Laatu in simple medium (DMEM) a 4.25 log reduction of Tulane system. As this system is an open system, it does not allow virus (TV) was measured, compared to a 2.1 log reduc- safe handling of pathogens and non-pathogenic surrogates tion after a treatment of 12.2 kGy on whole strawberries are needed to estimate virus inactivation by LEEB. (DiCaprio et al. 2016). The D values (dose required to The objective of this study was first to use HEEB treat- reduce virus titer by 1 log ) of rotavirus and poliovirus on ments to find surrogate microorganisms for enteric viruses lettuce were 1.0 and 2.3 kGy, respectively, suggesting virus- (HAV and hNoV) which are easy to grow to high titers and dependent susceptibilities to HEEB (Espinosa et al. 2012). use at industrial scale such as bacterial spores or bacterio- Using gamma irradiation, D values determined for HAV phages and second to use the surrogate organisms as proof in lettuce and strawberries were 2.7 and 3.0 kGy, respec- of concept to investigate LEEB treatments for enteric virus tively (Bidawid et al. 2000), while the HEEB dose required inactivation at industrial scale using frozen blueberries as a to reduce HAV by 1 log in whole oysters was 4.8 kGy model matrix. (Praveen et al. 2013). A new approach to irradiation processing of food, called low-energy electron beam (LEEB), uses electrons with ener- Materials and methods gies below 300 kilo electron Volt (keV), associated with a very limited penetration ability of around 500 μm (Gryczka HEEB Irradiation Treatment and Dose Evaluation et al. 2018; Hayashi et al. 1998; Pillai and Shayanfar 2017). In contrast, HEEB uses electrons with energies between 5 HEEB treatments (10 MeV) were performed at the LEONI and 10 Mega electron Volt (MeV) which penetrate foods Studer AG irradiation centre (Daeniken, Switzerland). Inoc- with high water content up to 3.9 cm (Farkas 2006) and ulated food samples were treated at three radiation doses, 4, has to be carried out at a dedicated irradiation facility with 8 and 16 kGy, to define critical dosages and screen for ade- appropriate safety measures. In contrast, the LEEB technol- quate surrogates. Pretests were performed for the different ogy is scalable to continuous processes and can be easily food matrices used in this study to verify dose distribution implemented in existing processing lines and offers ben- within the load. Based on the results, samples were pack- efits in cost reduction, environmental performance, and aged in 50 mL Falcon tubes with a maximum diameter of production flexibility (Zhang et al. 2018; Tetra Pak 2019). 3 cm which ensured equal irradiation doses throughout the As such, LEEB is an emerging irradiation technology that samples. The absorbed doses were measured using alanine performs surface and subsurface decontamination with a pellets (Aerial CRT, France) which were treated together minimal influence on food quality (Hertwig et al. 2018 ). with the samples (nine independent treatments performed on Since microorganisms reside mostly on the surface and three separate days) and the doses were calculated by com- subsurface of food if internalized through pores or dam- parison with a standard calibration curve established by Aer- aged tissue, the irradiation of the external layer should be ial CRT onsite at LEONI Studer AG irradiation centre and sufficient to eliminate food-borne microorganisms (De Lara traceable to NPL standard. The following mean values and et al. 2002; Hayashi et al. 1998). For example, Kikuchi et al. standard deviations were obtained: 3.91 ± 0.03, 7.89 ± 0.14 recommended LEEB treatment over gamma irradiation for and 15.85 ± 0.33 kGy. The results showed that the values for soybean decontamination, because it induces minimum or HEEB are close to the nominal doses, and thus, the nominal no quality deterioration, since the electrons do not reach the values are used in the graphs. internal matrix (Kikuchi et al. 2003). Similarly, LEEB was proposed as a method for microbiological decontamination LEEB Irradiation Treatment and Dose Evaluation of seeds, as it was shown not to negatively influence germi- nation (Fan et al. 2017). The response of Bacillus pumilus LEEB treatments were performed with Laatu designed by spores was found to be very similar when treated with HEEB Bühler AG (Uzwil, Switzerland). Inoculated frozen blueber- and LEEB (Tallentire et al. 2010). Another study investigat- ries were treated twice with 250 keV to reach a nominal dose ing Geobacillus and Bacillus spores also showed that the of 16 kGy. Samples were conveyed via an inlet channel and spore inactivation efficiency by LEEB was comparable to feeder table which control the throughput and distribution of that of other ionizing radiations (Zhang et al. 2018). LEEB the product. The berries were treated during free-fall through was also shown to reduce mould spores even on challenging the treatment zone which consists of two LEEB lamps facing 1 3 220 Food and Environmental Virology (2021) 13:218–228 each other (https://digit al.buhle r group.com/laatu / ). The dose for MS2. The MS2 and Qβ stock concentrations were 11.3 was measured with Riso B3 radiochromic film sensor pro- and 10.1 log plaque forming units (PFU)/mL, respectively. vided by Riso High Dose Reference Laboratory (HDLR) (Roskilde, Denmark). Due to the practical limitations of Inoculation of Food Matrices with Bacteria placing a dosimeter on the frozen blueberries, a B3 film strip and Viruses of 230 mm × 20 mm was attached along a polymer plate cov- ering the entire width of the treatment zone. The dosimeters Suspension (100 μL) of bacterial spores, viruses or bacterio- were exposed to the same beam conditions (voltage, current) phages were spotted (droplets of 5.2 ± 0.4 μL) directly on the as the inoculated frozen blueberries. The dosimetry system food surface of the seven selected food matrices (pumpkin, (including radiochromic film, sensor measurement software and sesame seeds, peppercorns, fresh blueberries, freeze- and calibration) is traceable to international standards in line dried raspberries and raisins) using a technique described with ISO/ASTM 51261 (ISO 2013). Dose measurements previously (Delbeke et al. 2015). For the HEEB treatment, using Riso B3 films determined a mean surface dose value 10 g of each food matrix were inoculated with bacterial of 15.51 ± 1.31 kGy which is close to the nominal dose, and spores, viruses or bacteriophages, respectively, except for thus, the nominal value is used in the graph. the freeze-dried raspberries, for which 5 g was used. The contaminated samples were allowed to dry in a biosafety cabinet at room temperature for 1 h, before being packaged Food Matrices and Pretreatments in a 50 mL tube and subjected to the HEEB treatment. The inoculated fresh blueberry samples were frozen at − 20 °C Six different food matrices were used in this study. Pumpkin, after the packaging. Following the same approach, blueberry and sesame seeds were kindly provided by the German Insti- samples were inoculated with G. stearothermophilus and tute of Food Technologies (DIL) (Quakenbrück, Germany), MS2 bacteriophages and frozen prior to the LEEB treatment. black peppercorns, fresh blueberries, freeze-dried raspberry flakes and raisins were purchased at Sabater (Spain), a local Recovery of Contaminants from Food Matrices distributor (Lausanne, Switzerland), Chaucer (UK) and Mariani Packing (USA), respectively. X-ray pretreatments To recover inoculated spores of G. stearothermophilus (25–50 kGy) of the matrices to kill background microflora from matrices, 5 or 10 g sample was aseptically transferred before inoculation with target organisms were performed to a stomacher filter bag, diluted 1:10 in buffered peptone at Synergy Health Daeniken AG (Daeniken, Switzerland). water and stomached for 1 min. From this initial suspension, 10 mL was submitted to a heat treatment at 80 °C for 10 min Geobacillus Spore Suspension to allow the spores to germinate. After the heat treatment, tubes were cooled down in cold water. Serial tenfold dilu- Ready to use suspension of Geobacillus stearothermophi- tions in tryptone salt solution were performed and plated on lus (ATCC 7953) spores, containing 7.7–8.7 log colony Plate Count Agar supplemented with 1 g/L of soluble starch forming units (CFU)/mL of viable spores, was purchased and incubated at 55 °C ± 1 °C for 48 h ± 4 h. The limit of from Merck Millipore (Burlington, USA) and was used to quantification (LOQ) was 2.00 log CFU/10 g for all food inoculate the different food matrices. matrices except for the freeze-dried raspberries, for which the LOQ was 1.98 log CFU/5 g. Viruses and Preparation of Suspensions HAV and MNV were recovered from 5 or 10 g samples using the ISO 15,216 virus extraction method (soft fruits The cytopathogenic HAV strain HM-175 (ATCC VR-1402), protocol) with slight modifications: no added process con- the murine norovirus strain MNV S99 used as proxy for trol, as untreated samples were analyzed to determine recov- hNoV and provided courtesy of the Friedrich-Loeffler-Insti- ery rate and the chloroform/butanol clarification step was not tut in Germany (Mueller 2007) were propagated, assayed and performed (ISO 2017). Before enumeration, concentrated titrated on FRhK-4 cells (ATTC CRL1688) and RAW 264.7 samples were decontaminated by sequential filtering through cells (ATCC TIB-71), respectively, as described previously 0.45 μm and then 0.22 μm spin centrifuge tube filters (Corn- (Butot et al. 2018). Viral stock titers were 6.6 and 7.6 log ing, New York) pretreated with 300 μL of phosphate-buff- 50% tissue culture infective dose (TCID )/mL for HAV and ered saline (pH 7.2 ± 0.2) containing 10% fetal calf serum MNV, respectively. MS2 bacteriophage (ATCC 15597-B1) followed by T CID titration as previously described (Butot was propagated, assayed and enumerated in Escherichia coli et al. 2018). The LOQ was 1.05 log TCID /10 g for all 10 50 K12 (ATCC 23631), as described previously (Butot et al. food matrices except for the freeze-dried raspberries, for 2018). The Qβ bacteriophage (ATTC 23631-B1) was propa- which the LOQ was 1.35 log TCID /5  g as the pellet 10 50 gated, assayed and enumerated using the same protocols as obtained after the polyethylene glycol precipitation was 1 3 Food and Environmental Virology (2021) 13:218–228 221 too viscous and therefore resuspended in 1 mL instead of on the values above maximum measurable log reduction 0.5 mL. to compute a pooled variance, common to all the measure- MS2 and Qβ were recovered from 5 or 10  g samples ments of a given microorganism in a given matrix. Notice using the protocol described previously (Butot et al. 2018). that since the width of the prediction interval depends only Briefly, the sample was transferred in a filter bag containing on the variance of the observations, the prediction intervals 50 mL of buffer (100 mM Tris, 50 mM glycine, 3% (m/v) have the same width for all the doses of a given microor- beef extract and 50 mM MgCl and adjusted to pH 9.5 with ganism and matrix. Finally, the mean log reductions com- NaOH solution) and stomached for 1 min. Serial dilutions puted in the presence of values at the maximum measurable (tenfold) were prepared and virus titer was quantified using log reduction are just a lower bound, since the real log ISO method 10705–1:1995(E), Annex C (ISO 1995). The reductions might be higher. This means that the presence LOQ was 2.78 log PFU/10 g for all food matrices except of values at the maximum measurable log reduction can 10 10 for the freeze-dried raspberries, for which the LOQ was 2.74 hide the effectiveness of the treatment, since the real mean log PFU/5 g. log reduction might actually be bigger than the estimated For all matrix/strain combinations, untreated inoculated one. However, this does not affect the size of the prediction triplicate samples were analyzed to determine the log intervals, which are computed using only values above the reductions, the maximum measurable log reduction cal- maximum measurable log reduction. 10 10 culated as log (LOQ/N ) and the recovery rates and un- We notice that the overall variability of the data is deter- 10 0 contaminated triplicate samples were analyzed as negative mined by different sources, for example, a variability in the controls following the procedures described above. recovery rates and in the irradiation values, and a biological variability in the response of the microorganisms. There- Statistical Analysis fore, the variance computed as described before takes into account all these sources of variability, that are also likely Reductions in spore and infectious viral particle counts to be found when applying the irradiation technology to real (inactivation) were calculated as log (N /N ), where N is production samples. 10 x 0 x the spore count or the viral titer recovered from treated food All statistical analyses were performed with R v3.6.1 matrices and N is the initial count or titer recovered from (R-Core Team 2019). untreated food matrices (mean of three replicates). Values of N below the LOQ were entered as being at that limit and this leads the log reduction to reach the maximum measur- Results able log reduction. A t test with a significance level at 0.05 and without cor - Recovery Efficiencies rection for multiplicity of test was used to test whether the mean log reduction of the replicates in one condition was To evaluate the efficiency of the methodologies employed, statistically different from the mean in another condition. the recovery of each microorganism was determined for Furthermore, it is also important to understand if the single all matrices tested as the ratio, in percentage, of N which measurements are consistently higher or lower. Thus, for corresponds to the initial count recovered from untreated each condition we also computed a prediction interval, cor- matrices and the calculated theoretical count using the responding to the interval where 95% of log reductions stock concentration and the volume inoculated (Table 1). measured in future experiments are expected to fall. If the G. stearothermophilus was recovered efficiently with mean prediction intervals of two conditions are non-overlapping, values ranging between 46.61 and 100.79% for all matrices, we can be confident that not only the mean log reductions are different, but also that the single observations in future Table 1 Recovery efficiencies of each microorganism–matrix combi- experiments are going to be well separated. nation expressed in percentage (%) In order to compute the t tests and the prediction inter- vals, the variance inside each condition is needed. However, Geobacillus HAV MNV MS2 Qβ as the maximum measurable log reduction was reached in Frozen blueberries 100.79 0.65 0.01 1.59 0.04 some conditions, the measured values are not representative Freeze-dried raspberries 25.72 0.40 – 0.03 0.05 of the real ones and must be discarded in the computation Raisins 92.86 4.22 0.06 0.09 2.97 of the variance. This leads to having conditions with very Peppercorns 82.54 0.89 16.24 0.80 0.03 few points (possibly zero) from which to estimate the vari- Sesame seeds 46.61 0.46 1.35 9.92 17.76 ance. To solve this issue, we assumed that the variance in Pumpkin seeds 63.92 0.12 0.60 0.06 7.26 the log reductions of a microorganism in a given matrix is independent from the dose. This allows us to use ANOVA – the missing value could not be determined due to technical issues 1 3 HAV MNV 222 Food and Environmental Virology (2021) 13:218–228 except for the freeze-dried raspberries where a rate of TCID /5 g, the maximum measurable log reduction was 50 10 25.72% was obtained. On the opposite, the bacteriophages 1.69 log . and the viruses showed lower recovery rates. The bacterio- We notice that a low recovery rate limits the maximum phages were recovered with efficiencies ranging from 0.03 measurable log reduction, therefore decreasing the power to 9.92% for MS2 and 0.03–17.76% for Qβ. Matrix effects of detecting strong reductions. However, even for the treat- were observed but no correlation was found between the ments where the low recovery leads to values below the two bacteriophages. Recovery rates above 1% were obtained LOQ, meaningful conclusions can still be drawn, as will be with sesame seed and frozen blueberry matrices for MS2 discussed in the following sections. Also, the low variabil- and with sesame seed, pumpkin seed and raisin samples for ity rates can increase the variance of the measured effects. Qβ. For viruses, the recovery rates oscillated from 0.12 and However, as explained in “Materials and Methods” section, 4.22% for HAV and from 0.01 to 16.24% for the hNoV proxy the computation of the t tests and of the prediction intervals MNV. Recoveries above 1% for HAV were reached only on take into account also any variation in the recoveries. raisins and for MNV only on sesame seeds and peppercorns. The freeze-dried raspberries showed low recoveries for all Reductions of HAV and MNV By HEEB Treatment the microorganisms (Table 1). In addition, this food product on Different Food Matrices induced a cytotoxic effect on the RAW 264.7 cells, making it impossible to report inactivation results for MNV. How- Inactivation values of HAV and MNV inoculated onto the ever, this effect was not observed on the FrhK4 cells used to different model food matrices (fruit, spice and seed) are measure the infectivity of HAV. shown in Fig. 1. The results are shown in log reduction The recovery efficiency is an important parameter as it with 95% prediction intervals (not confidence intervals) to has an impact on the maximum measurable log reduc- better represent the “minimal” effect of a certain treatment tion. Indeed, the lower the recovery, the lower the N and condition. A measure obtained from a future experiment will the maximum measurable log reduction will be. As an fall on this interval with a probability of 95%. The highest example, 5.45 log TCID of HAV were inoculated on reduction of HAV was observed when inoculated on rai- 10 50 5 g freeze-dried raspberries, but with a recovery efficiency sins (Fig.  1). On this matrix at 8 kGy, a mean reduction of 0.40%, only 3.04 log TCID /5 g of HAV were enu- of 1.9 log was measured, ensuring a minimal reduction 10 50 10 merated on the samples not treated (N ). Therefore, taking of only 0.7 log , according to the 95% prediction interval. 0 10 into consideration this N value and the LOQ of 1.35 log At 16 kGy the maximum measurable log reduction was 0 10 10 reached (> 3.2 log ), corresponding to a minimal reduction Freeze − driedraspberries Frozen blueberries Raisins Peppercorn Sesameseeds Pumpkinseeds −2 −4 −6 −2 −4 −6 48 16 48 16 48 16 48 16 48 16 48 16 Dose [kGy] Fig. 1 Mean inactivation of HAV and MNV on freeze-dried rasp- presence of values below the maximum measurable log reduction, berries, frozen blueberries, raisins, pepper corns, sesame seeds and we only have a lower bound on the mean reduction. For the predic- pumpkin seeds treated with HEEB at 4, 8 and 16  kGy. Different tion intervals, only the circles have been considered, and a common shadings show significant differences in the mean log reductions variance has been computed for a given microorganism-matrix com- between dosages within each microorganism–matrix combination. bination. Therefore, all the prediction intervals for a given combina- Bars represent 95% prediction intervals. Open circles represent single tion have the same widths, and they are not affected by having val- values, crosses represent single values below the maximum measur- ues below the maximum measurable log reduction. Supplementary able log reduction. To compute the mean log reduction, both cir- Table 1 contains the raw data used to generate Fig. 1 cles and crosses have been taken into account. For this reason, in the 1 3 log(N /N ) + 95% Prediction Interval x 0 Food and Environmental Virology (2021) 13:218–228 223 which is at least 2 log as given by the prediction inter- Reductions of Selected Surrogate Organisms By val. Notice that we talk about minimal reduction, because HEEB Treatment the real mean reduction might be bigger than the estimated one, as the maximum measurable log reduction has been Inactivation values of the three surrogate organisms (spores reached for all the three replicates (this does not affect the G. stearothermophilus and bacteriophages MS2 and Qβ) size of the prediction intervals, which are computed only inoculated onto the different model food matrices (fruit, from values above the maximum measurable log reduc- spice and seed) are shown in Fig.  2. MS2 and Qβ both tion). A dose-dependent inactivation of HAV was also showed a high resistance to HEEB treatments compared observed on frozen blueberries with mean reduction values to G. stearothermophilus, with mean reductions generally of 0.5, 1.4 and 2.4 l og at 4, 8 and 16 kGy, respectively. lower than 1.5 log . The only exceptions were raisins, where For comparison with raisins, the minimal log reduction at both bacteriophages showed reductions close to those for G. 16 kGy on frozen blueberries was estimated at 1.75 l og . stearothermophilus. We noticed that for some matrices, the The reductions of HAV inoculated on freeze-dried raspber- prediction intervals at two different irradiation doses (error ries subjected to the three HEEB dosages of 4, 8 and 16 kGy bars) were overlapping, meaning that the doses do not have a appeared not significantly different from each other (bars of statistically different effect. That is, we cannot conclude that the same colour for all three dosages), but this may be an future measurements at the higher dose will have a higher artefact which could be overcome if we had a better extrac- log reduction compared to the lower dose. The overlapping tion efficiency allowing to see a greater maximum measur - of prediction intervals can happen even when the means able log reduction than 1.7 log at 16 kGy. A significant are statistically different (the bars have a different shade of 10 10 dose-dependent inactivation of HAV subjected on HEEB grey), as for example for Qβ in sesame seeds treated at 4 treatment of 4, 8 and 16 kGy was seen on peppercorns and and 8 kGy. This shows that, although the mean reductions sesame seeds, with mean HAV reductions of 2.1 and 2.2 can be different at two different irradiation doses, for single log when treated at 8 kGy, respectively. At 16 kGy the future measurements we cannot conclude that the reduction maximum measurable log reduction was reached on both is lower at 4 kGy compared to 8 kGy. matrices, corresponding to 2.6 log for peppercorns and 3.1 On sesame seeds, however, MS2 showed a clear lower log for sesame seeds. The reductions of HAV inoculated reduction compared to Qβ, both considering the mean and on pumpkin seeds were lower than the reductions observed the prediction interval (Supplementary Table 2). Given the on sesame seeds (a mean inactivation of only 1.1 log at similarities between the two bacteriophages, and the fact 16 kGy) which may indicate complex matrix effects and that compared to Qβ, MS2 showed a higher recovery and a show that predictions of inactivation are risky, even for simi- dose-dependent reduction on frozen blueberries, we chose lar food matrices. to focus only on MS2 as candidate surrogate for HAV and Compared to HAV, MNV showed a higher resistance to hNoV for the LEEB study. This choice is especially relevant HEEB on all matrices, except for frozen blueberries where for HAV, as on all food matrices and doses, HAV showed a maximum measurable log reductions were reached in mean reduction which was either higher or non-statistically most cases, preventing a meaningful comparison (Fig. 1). different than the mean reduction for MS2 (Supplementary This highlights the high resistance of MNV to irradiation Table 3). by HEEB on food matrices (complementary Table 1). The Geobacillus stearothermophilus showed a dose-depend- inactivation was lowest on pumpkin seeds and peppercorns ent inactivation on all tested matrices with a generally higher with mean inactivation values showing no significant dif- inactivation than the tested viruses, for both mean reduc- ferences among the three dosages applied and ranging from tions and prediction intervals, except on raisins (Fig. 2). On 0.03 to 0.32 log . MNV inoculated on raisins and sesame frozen blueberries, for example, a 1.3, 3.3 and > 5.6 log 10 10 seeds did not show significant dose-dependent differences reduction was measured for HEEB treatments at 4, 8 and either, with reductions ranging from 1.35 to 1.44 log on 16 kGy, respectively. In comparison, at 16 kGy on the same raisins and 0.96 and 1.34 log on sesame seeds. A signifi- matrix, 2.4 log and 1.8 log reductions were measured 10 10 cantly higher inactivation at a higher HEEB dose could only for HAV and MS2, respectively (Figs. 1, 2). Similarly, on be measured for MNV on frozen blueberries with mean log sesame seeds, pumpkin seeds and peppercorns at 4, 8 and reductions of 0.35, > 0.76 and > 1.18 log for treatments at 16 kGy, HAV showed a lower mean inactivation level com- 4, 8 and 16 kGy, respectively. pared to G. stearothermophilus (Figs. 1, 2). This intermedi- ate ranking of HAV between G. stearothermophilus spores and MS2 determined by HEEB was ideal to select both sur- rogate organisms as first approximation to estimate HAV inactivation during LEEB treatments of frozen blueberries. 1 3 G. stearothermophilus MS2 phage Qβ Frozen blueberries 224 Food and Environmental Virology (2021) 13:218–228 Freeze − driedraspberries Pepper corn Pumpkinseeds Frozen blueberries Raisins Sesame seeds −2 −4 −6 −2 −4 −6 −2 −4 −6 48 16 48 16 48 16 48 16 48 16 48 16 Dose [kGy] Fig. 2 Mean inactivation of G. stearothermophilus spores, MS2 and intervals. Open circles represent single values, crosses represent sin- Qβ bacteriophages on freeze-dried raspberries, frozen blueberries, gle values below the maximum measurable log reduction. Notice raisins, pepper corns, sesame seeds and pumpkin seeds treated with that since a common variance has been computed for a given micro- HEEB at 4, 8 and 16  kGy. Different shadings show significant dif- organism-matrix combination, all the prediction intervals for that ferences in the mean log reductions between dosages within each combination have the same width. Supplementary Table  2 contains microorganism-matrix combination. Bars represent 95% prediction the raw data used to generate Fig. 2 bacteriophage on frozen blueberries were treated in the Inactivation of G. stearothermophilus Spores industrial Bühler AG Laatu system, as a proof of concept and MS2 During LEEB on Frozen Blueberries for validation of HAV and hNoV inactivation by LEEB. In Fig. 3, the LEEB results obtained from this experiment are Geobacillus stearothermophilus spores and MS2 Fig. 3 Mean inactivation of G. stearothermophilus MS2phage G. stearothermophilus spores, HAV and MS2 bacteriophage on frozen blueberries treated with LEEB at 16 kGy in −1 comparison with corresponding reduction generated by HEEB at 8 and 16 kGy. Open circles represent single values, crosses represent single values below the maximum measurable log −3 reduction −5 HEEB HEEB LEEB HEEB HEEB LEEB 8 kGy 16 kGy 16 kGy 8 kGy 16 kGy 16 kGy Treatment 1 3 log(N /N ) + 95% Prediction Interval x 0 log(N /N ) x 0 Food and Environmental Virology (2021) 13:218–228 225 compared with the data generated by HEEB. As the LEEB As reported by others, spores of G. stearothermophi- experiment was a proof of concept with a low amount of lus are highly resistant against irradiation and showed a data, we did not carry out statistical analysis on these data higher resistance during LEEB than the spores of B. pumi- points. Mean reductions of 3.1 l og and 1.3 l og were lus which is often suggested to be the biological indicator 10 10 measured after the LEEB surface treatment for G. stearo- for irradiation sterilization (Van Gerwen et al. 1999; Zhang thermophilus spores and MS2, respectively (Fig. 3). These et al. 2018). Therefore, G. stearothermophilus spores were values were similar to the log reductions determined for included in this study to compare their resistance to elec- both organisms in the HEEB trials at 8 kGy (3.3 log for tron beam with the one of HAV. We observed a higher sus- G. stearothermophilus spores and 1.1 log for MS2). The ceptibility of G. stearothermophilus to HEEB compared to similarity of the log reductions of the surrogates observed HAV (Figs. 1, 2). The data generated in this study for G. on frozen blueberries after a LEEB treatment at 16 kGy and stearothermophilus is new and valuable for future industrial a HEEB treatment at 8 kGy, suggests that if HAV underwent validation studies. It cannot be directly compared with the a LEEB treatment at 16 kGy, we might expect a reduction data present in the literature, as most studies have deter- similar to the one determined for the HEEB treatment at mined D values for electron beam treatments of B. pumilus 8 kGy (Fig.  1). Thus, we expect to see a log reduction and B. subtilis and to a lesser extent G. stearothermophilus of 1.4 on frozen blueberries for HAV treated by LEEB at spores not inoculated onto any food matrix (Tallentire et al. 16 kGy. 2010; Zhang et al. 2020, 2018). Nevertheless, the D-value of 3.1 kGy at 200 keV measured by Zhang and coauthors cor- responds to a 2.6 log reduction at 8 kGy which is similar Discussion to the reduction of 2.87 log we measured on peppercorn and sesame seeds. Additionally, the food matrix can have The results of the present study represent a proof of concept opposite effects on different microorganisms. For example, for validation of HAV and hNoV inactivation by LEEB, on raisins G. stearothermophilus showed a lower inactiva- using the industrial Bühler AG Laatu system, the only indus- tion than on the other matrices. Viruses, on the other hand, trial scale system currently available to the food industry. generally showed higher inactivation on raisins compared As this system is an open system not allowing safe handling to the other matrices, showing that extrapolation of effects of pathogens, the HEEB technology was used with inocu- from one matrix to another should not be attempted. lated and packaged food matrices to compare reduction of Low performance of the methods used to recover the HAV and hNoV with potential surrogates. The HEEB trials viruses and the bacteriophages impacted only a small part generated new data on the effect electron beam irradiation of the results, as the maximum measurable log reductions has on different viruses-food matrix combinations. Over - were in most cases sufficient to draw meaningful conclu- all, this study demonstrates the high resistance of HAV to sions. G. stearothermophilus spores showed good recoveries electron beam irradiation and an even higher resistance of on every matrix, whereas enteric viruses and bacteriophages MNV (Fig. 1). As different amino acids vary in their sus- displayed poor recovery rates. For enteric viruses, such as ceptibility to irradiation, crosslinking of proteins during HAV and MNV, this was expected as the methods avail- irradiation may damage portions of the capsid involved in able for the detection of viruses are well known for their receptor binding to varying degrees and may explain differ - complexities and low recovery efficiencies (Li et al. 2018; ences in susceptibility observed between different viruses Mäde et al. 2013; Perrin et al. 2015). In ISO 15,216, the (Hume et al. 2016; Stewart 2001). A comparison with hNoV minimal recovery efficiency required is 1% but this require- is needed in the future to determine which hNoV surrogate ment is often not reached (Li et al. 2018). For example, Per- can reliably be used as surrogate to evaluate the effect of rin and coauthors reported a recovery of 0.5% of MNV on electron beam irradiation. Using the porcine gastric mucin raspberries and Hida and coauthors reported MNV recov- magnetic bead (PGM-MB) binding assay, the level of hNoV ery efficiencies of 0.06% and 0.41% on lettuce and grapes, and TV RNA decreased by > 1 log and approximately 2 respectively (Hida et al. 2018; Perrin et al. 2015). The low log , respectively, following a 12.2 kGy HEEB treatment recovery efficiencies obtained with the bacteriophages were on whole strawberries (DiCaprio et al. 2016). In compari- more surprising as the method used is in principle simple son, at 8 kGy for MNV our results reveal lo g reductions and easy to apply. Shim and coauthors showed a correlation of 0.76 and 1.35 log on frozen blueberries and raisins, between the stronger adhesion and the lower recoveries of respectively. As shown by Ettayebi and coauthors using the MS2 on PVC, therefore one hypothesis of this low perfor- stem cell derived human enteroids assay, exposure to gamma mance could be a strong adhesion of the bacteriophages on irradiation of hNoV GII.3 and GII.4 stool suspensions inac- the matrices, impacting the recovery rates (Shim et al. 2017). tivated the viruses, but at this point in time no data are avail- One explanation of this stronger adhesion could be the time able with this system on food matrices (Ettayebi et al. 2016). between the inoculation and the analysis, as shown by Shim 1 3 226 Food and Environmental Virology (2021) 13:218–228 the Bühler AG team for the LEEB trials and dosimetry measurements and coauthors who extended this time from 1 min to 24 h to and Mireille Moser for initial support with data analysis. allow stronger MS2 attachment for their study (Shim et al. 2017). Author Contributions SZ was activity leader within EIT Food project The similarity of the log reductions observed between ID19087 and supervised the study. SZ and SB designed the study, HEEB at 8 kGy and LEEB at 16 kGy for both G. stearo- prepared and tested the samples and wrote the manuscript. LG per- formed the statistical analyses of the study and wrote the manuscript. thermophilus and MS2 on frozen blueberries may be due TP prepared, tested the samples and helped to write the manuscript. to several factors, one of them being the difference in the penetration depth of electrons between HEEB and LEEB. Funding This research was conducted under EIT Food project The HEEB measured dose is applied to the full depth of the ID19087with funding from the European Institute of Innovation and food matrix. The LEEB dose is a surface and subsurface Technology (EIT), a body of the European Union, under Horizon 2020, the EU Framework Program for Research and Innovation of the EIT measurement, where the average absorbed dose along the Food project ID19087. food matrix is depending on the depth dose distribution at a given energy, the matrix density and its chemical com- Compliance with Ethical Standards position (Shim et al. 2017). In the current study the beam penetration depth for frozen blueberries was not determined, Conflict of Interest The authors declare that they have no conflict of but in a matrix with water density at 250 keV, the absorbed interest. dose is maximum at the surface while it exponentially drops to zero at approximately 580 µm. If a certain number of Open Access This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- microorganisms are beyond the penetration ability of the tion, distribution and reproduction in any medium or format, as long electrons (at 250 keV) they will not be inactivated. Addi- as you give appropriate credit to the original author(s) and the source, tionally, it has been hypothesized that lower D value for provide a link to the Creative Commons licence, and indicate if changes HEEB versus LEEB treatments may be due to the additional were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated inactivation effect of reactive oxygen species if atmospheres otherwise in a credit line to the material. If material is not included in during the treatments contain different amounts of O . Thus, the article’s Creative Commons licence and your intended use is not when comparing HEEB and LEEB systems, the focus should permitted by statutory regulation or exceeds the permitted use, you will not be on comparing the dosimetry measurements. Instead, need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativ ecommons .or g/licenses/b y/4.0/. the impact of achieving the intended microbial reduction and the impact on food quality, such as nutrients and sensory attributes, should be evaluated. The quality of the HEEB and LEEB treated foods in this study were evaluated using sen- References sory attributes. There were no visible differences between HEEB and LEEB treated foods in this study and detailed Alegbeleye, O. O., Singleton, I., & Sant’Ana, A. S. (2018). Sources and sensory profiles and chemical properties are described else- contamination routes of microbial pathogens to fresh produce dur- where (Aisala et al. in preparation). ing field cultivation: A review. Food Microbiology, 73, 177–208. Based on our results for G. stearothermophilus and MS2, Anonymous. (2018). Swift investigation of hepatitis A outbreak facili- tated withdrawal of dates. Copenhagen: Statens Serum Institute. a log reduction of 1.4 can be expected for HAV on fro- Anonymous. (2020). Evaluation of the EU legal framework on food zen blueberries treated by LEEB at 16 kGy, indicating that irradiation. Brussels: EU. LEEB processing cannot completely eliminate the risk of Bidawid, S., Farber, J. M., & Sattar, S. A. (2000). Inactivation of hepa- viral illness on berries and emphasizes the critical impor- titis A virus (HAV) in fruits and vegetables by gamma irradiation. International Journal of Food Microbiology, 57, 91–97. tance of adhering to stringent application of hygiene control Bosch, A., Gkogka, E., Le Guyader, F. S., Loisy-Hamon, F., Lee, A., systems that limit enteric virus contamination at primary van Lieshout, L., et al. (2018). Foodborne viruses: Detection, risk production. In addition, the list of foods allowed for irradia- assessment, and control options in food processing. International tion treatment in the EU currently comprises only dried aro- Journal of Food Microbiology, 285, 110–128. Butot, S., Zuber, S., & Baert, L. (2014). Sample preparation prior to matic herbs, spices and vegetable seasonings and not berries, molecular amplification: Complexities and opportunities. Current but this may change in the future, as a legislative evaluation Opinion in Virology, 4, 66–70. related to the irradiation of food and food ingredients is on- Butot, S., Cantergiani, F., Moser, M., Jean, J., Lima, A., Michot, L., going within the EU (Anonymous 2020). et al. (2018). UV-C inactivation of foodborne bacterial and viral pathogens and surrogates on fresh and frozen berries. Interna- tional Journal of Food Microbiology, 275, 8–16. Supplementary Information The online version contains supplemen- De Lara, J., Fernández, P. S., Periago, P. M., & Palop, A. (2002). Irra- tary material available at https:// doi.org/10.1007/s12560-02 1-09463-3 . diation of spores of Bacillus cereus and Bacillus subtilis with electron beams. Innovative Food Science and Emerging Technolo- Acknowledgements We gratefully acknowledge Conrad Guenthard gies, 3, 379–384. (LEONI Studer AG irradiation centre in Daeniken, Switzerland) for support with the HEEB trial setup and the dosimetry measurements, 1 3 Food and Environmental Virology (2021) 13:218–228 227 Delbeke, S., Ceuppens, S., Jacxsens, L., & Uyttendaele, M. (2015). Kikuchi, O. K., Todoriki, S., Saito, M., & Hayashi, T. (2003). Efficacy Survival of salmonella and escherichia coli O157:H7 on strawber- of soft-electron (low-energy electron beam) for soybean decon- ries, basil, and other leafy greens during storage. Journal of Food tamination in comparison with gamma-rays. Journal of Food Sci- Protection, 78, 652–660. ence, 68, 649–652. DiCaprio, E., Phantkankum, N., Culbertson, D., Ma, Y., Hughes, J. Kingsley, D. H., & Chen, H. (2009). Influence of pH, salt, and tempera- H., Kingsley, D., et al. (2016). Inactivation of human norovirus ture on pressure inactivation of hepatitis A virus. International and Tulane virus in simple media and fresh whole strawberries Journal of Food Microbiology, 130, 61–64. by ionizing radiation. International Journal of Food Microbiol- Knorr, D., & Watzke, H. (2019). Food processing at a crossroad. Fron- ogy, 232, 43–51. tiers in Nutrition, 6, 85. DiCaprio, E., Ye, M., Chen, H., & Li, J. (2019). Inactivation of human Li, D., Butot, S., Zuber, S., & Uyttendaele, M. (2018). Monitoring norovirus and tulane virus by high pressure processing in simple of foodborne viruses in berries and considerations on the use of mediums and strawberry puree. Frontiers in Sustainable Food RT-PCR methods in surveillance. Food Control, 89, 235–240. Systems. https ://doi.org/10.3389/fsufs .2019.00026 . Mäde, D., Trübner, K., Neubert, E., Höhne, M., & Johne, R. (2013). Espinosa, A. C., Jesudhasan, P., Arredondo, R., Cepeda, M., Mazari- Detection and typing of norovirus from frozen strawberries Hiriart, M., Mena, K. D., & Pillai, S. D. (2012). Quantifying the involved in a large-scale gastroenteritis outbreak in Germany. reduction in potential health risks by determining the sensitivity Food and Environmental Virology, 5, 162–168. of poliovirus type 1 chat strain and rotavirus SA-11 to electron Miranda, R. C., & Schaffner, D. W. (2019). Virus risk in the food sup- beam irradiation of iceberg lettuce and spinach. Applied and Envi- ply chain. Current Opinion in Food Science, 30, 43–48. ronmental Microbiology, 78, 988–993. Moosekian, S. R., Jeong, S., Marks, B. P., & Ryser, E. T. (2012). X-ray Ethelberg, S., Lisby, M., Bottiger, B., Schultz, A. C., Villif, A., Jensen, irradiation as a microbial intervention strategy for food. Annual T., et al. (2010). Outbreaks of gastroenteritis linked to lettuce, Review of Food Science and Technology, 3, 493–510. Denmark, January 2010. Euro Surveillance, 15, 19484. Perrin, A., Loutreul, J., Boudaud, N., Bertrand, I., & Gantzer, C. Ettayebi, K., Crawford, S. E., Murakami, K., Broughman, J. R., (2015). Rapid, simple and efficient method for detection of viral Karandikar, U., Tenge, V. R., et al. (2016). Replication of human genomes on raspberries. Journal of Virological Methods, 224, noroviruses in stem cell-derived human enteroids. Science, 353, 95–101. 1387–1393. Pillai, S. D., & Shayanfar, S. (2017). Electron beam technology and Etter, D., Rupp, A., Prange, A., & Drissner, D. (2018). Inactivation other irradiation technology applications in the food industry. Top- of mould spores in a model system and on raisins by low-energy ics in Current Chemistry, 375, 6. electron beam. Food Control, 92, 357–361. Praveen, C., Dancho, B. A., Kingsley, D. H., Calci, K. R., Meade, G. Fan, X., Sokorai, K., Weidauer, A., Gotzmann, G., Rögner, F. H., & K., Mena, K. D., & Pillai, S. D. (2013). Susceptibility of murine Koch, E. (2017). Comparison of gamma and electron beam irra- norovirus and hepatitis a virus to electron beam irradiation in diation in reducing populations of E. coli artificially inoculated oysters and quantifying the reduction in potential infection risks. on mung bean, clover and fenugreek seeds, and affecting germi- Applied and Environmental Microbiology, 79, 3796–3801. nation and growth of seeds. Radiation Physics and Chemistry, R-Core Team. (2019). R: A language and environment for statistical 130, 306–315. computing. Vienna: R-Core Team. Farkas, J. (1998). Irradiation as a method for decontaminating food: A Ravindran, R., & Jaiswal, A. K. (2019). Wholesomeness and safety review. International Journal of Food Microbiology, 44, 189–204. aspects of irradiated foods. Food Chemistry, 285, 363–368. Farkas, J. (2006). Irradiation for better foods. Trends in Food Science Sanglay, G. C., Li, J., Uribe, R. M., & Lee, K. (2011). Electron-beam and Technology, 17, 148–152. inactivation of a norovirus surrogate in fresh produce and model Gryczka, U., Migdał, W., & Bułka, S. (2018). The effectiveness of the systems. Journal of Food Protection, 74, 1155–1160. microbiological radiation decontamination process of agricultural Shim, J., Stewart, D. S., Nikolov, A. D., Wasan, D. T., Wang, R., Yan, products with the use of low energy electron beam. Radiation R., & Shieh, Y. C. (2017). Differential MS2 interaction with food Physics and Chemistry, 143, 59–62. contact surfaces determined by atomic force microscopy and Hayashi, T., Takahashi, Y., & Todoriki, S. (1998). Sterilization of foods virus recovery. Applied and Environmental Microbiology, 83, with low-energy electrons (’Soft-electrons’). Radiation Physics e01881–e01917. and Chemistry, 52, 73–76. Stewart, E. M. (2001). Food irradiation chemistry. In R. Molins (Ed.), Hertwig, C., Meneses, N., & Mathys, A. (2018). Cold atmospheric Food irradiation: Principles and applications (pp. 37–76). New pressure plasma and low energy electron beam as alternative non- York: Wiley. thermal decontamination technologies for dry food surfaces: A Tallentire, A., Miller, A., & Helt-Hansen, J. (2010). A comparison of review. Trends in Food Science and Technology, 77, 131–142. the microbicidal effectiveness of gamma rays and high and low Hida, K., Papafragkou, E., & Kulka, M. (2018). Testing for human energy electron radiations. Radiation Physics and Chemistry, 79, norovirus and recovery of process control in outbreak-associated 701–704. produce items. Journal of Food Protection, 81, 105–114. Tavoschi, L., Severi, E., Niskanen, T., Boelaert, F., Rizzi, V., Liebana, Hume, A. J., Ames, J., Rennick, L. J., Duprex, W. P., Marzi, A., Tonk- E., et al. (2015). Food-borne diseases associated with frozen ber- iss, J., & Mühlberger, E. (2016). Inactivation of RNA viruses by ries consumption: A historical perspective, European Union, 1983 gamma irradiation: A study on mitigating factors. Viruses, 8, 204. to 2013. Eurosurveillance Weekly, 20, 21193. ISO. (1995). ISO 10705-1:1995: Water quality? Detection and enu- Tetra Pak (2019). Tetra Pak E3 with eBeam. Retrieved from https :// meration of bacteriophages ? Part 1: Enumeration of F-specific www.tetra pak.com/tetra -pak-e3-ebeam . Accessed 29 Oct 2020. RNA bacteriophages. Geneva: ISO. Thebault, A., Teunis, P. F., Le Pendu, J., Le Guyader, F. S., & Denis, J. ISO. (2013). ISO/ASTM 51261:2013: Practice for calibration of rou- B. (2013). Infectivity of GI and GII noroviruses established from tine dosimetry systems for radiation processing. Geneva: ISO. oyster related outbreaks. Epidemics, 5, 98–110. ISO. (2017). ISO 15216–1:2017: Microbiology of the food chain: Hori- Tree, J. A., Adams, M. R., & Lees, D. N. (2005). Disinfection of feline zontal method for determination of hepatitis A virus and norovirus calicivirus (a surrogate for Norovirus) in wastewaters. Journal of in food using real-time RT-PCR: Part 1: Method for quantifica- Applied Microbiology, 98, 155–162. tion. Geneva: ISO. Van Gerwen, S. J. C., Rombouts, F. M., Riet, K. V., & Zwietering, M. H. (1999). A data analysis of the irradiation parameter for bacteria 1 3 228 Food and Environmental Virology (2021) 13:218–228 and spores under various conditions. Journal of Food Protection, Zhang, Y., Moeller, R., Tran, S., Dubovcova, B., Akepsimaidis, G., 62, 1024–1032. Meneses, N., et al. (2018). Geobacillus and bacillus spore inacti- Wiedenmann, A., Fischer, B., Straub, U., Wang, C.-H., Flehmig, B., & vation by low energy electron beam technology: Resistance and Schoenen, D. (1993). Disinfection of hepatitis a virus and MS-2 influencing factors. Frontiers in Microbiology, 9, 2790. coliphage in water by ultraviolet irradiation: Comparison of UV- susceptibility. Water Science and Technology, 27, 335–338. Publisher’s Note Springer Nature remains neutral with regard to Zhang, Y., Huber, N., Moeller, R., Stülke, J., Dubovcova, B., Akepsi- jurisdictional claims in published maps and institutional affiliations. maidis, G., et al. (2020). Role of DNA repair in Bacillus subtilis spore resistance to high energy and low energy electron beam treatments. Food Microbiology, 87, 103353. 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Food and Environmental Virology Springer Journals

Electron Beam Susceptibility of Enteric Viruses and Surrogate Organisms on Fruit, Seed and Spice Matrices

Loading next page...
 
/lp/springer-journals/electron-beam-susceptibility-of-enteric-viruses-and-surrogate-EBQskTMsgj
Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2021
ISSN
1867-0334
eISSN
1867-0342
DOI
10.1007/s12560-021-09463-3
Publisher site
See Article on Publisher Site

Abstract

The objective of this study was to use high-energy electron beam (HEEB) treatments to find surrogate microorganisms for enteric viruses and to use the selected surrogates as proof of concept to investigate low-energy electron beam (LEEB) treatments for enteric virus inactivation at industrial scale on frozen blueberries. Six food matrices inoculated with HAV (hepatitis A virus), MNV S99 (murine norovirus), bacteriophages MS2 and Qβ, and Geobacillus stearothermophilus spores were treated with HEEB at 10 MeV using 4, 8 and 16 kGy doses. G. stearothermophilus spores showed the highest inactiva- tion on all matrices except on raisins, with a dose-dependent effect. HAV reached the maximum measurable log reduction (> 3.2 log ) when treated at 16 kGy on raisins. MNV showed the highest resistance of all tested microorganisms, inde- pendent of the dose, except on frozen blueberries. On frozen blueberries, freeze-dried raspberries, sesame seeds and black peppercorns, HAV showed a mean inactivation level in between those of MS2 and G. stearothermophilus. Based on this, we selected both surrogate organisms as first approximation to estimate HAV inactivation on frozen blueberries during LEEB treatment at 250 keV using 16 kGy. Reductions of 3.1 and 1.3 log were measured for G. stearothermophilus spores and MS2, respectively, suggesting that a minimum reduction of 1.4 log can be expected for HAV under the same conditions. Keywords Electron beam · HEEB · LEEB · Enteric viruses · Surrogates · Food Introduction pressure (HHP), ultra violet light (UV-C), cold plasma and irradiation, have the potential to achieve this goal (Knorr Hepatitis A virus (HAV) and human norovirus (hNoV) are and Watzke 2019). However, virus inactivation by UV-C responsible for food-borne outbreaks linked to fresh pro- light is very efficient in clear liquids such as water (Tree duce, ready-to-eat foods and shellfish worldwide (Alegbel- et al. 2005; Wiedenmann et al. 1993), but of limited use on eye et al. 2018; Bosch et al. 2018; Miranda and Schaffner complex food surfaces such as berries (Butot et al. 2018). 2019). Outbreaks are generally associated with food that Virus inactivation by HHP is only applicable to high mois- undergoes minimal processing, such as oysters, fresh-cut ture foods, such as oysters and fruit purees (DiCaprio et al. lettuce, frozen berries and very recently dried dates (Ethel- 2019; Kingsley and Chen 2009). For fragile or dry foods, berg et al. 2010; Tavoschi et al. 2015; Thebault et al. 2013; such as berries, herbs and spices, other technologies need Anonymous 2018). Consumer demands for minimally to be explored. processed foods with fresh-like quality are on the rise and Irradiation appears to be a good candidate technology, food processes must adapt to the preferences, acceptance but data describing the effectiveness of irradiation on food and needs of the consumer (Knorr and Watzke 2019). In against viruses, especially HAV, are scarce. Besides X-ray, this context, innovative approaches are needed to minimize two major irradiation technologies have been explored, the risk of viral food-borne outbreaks while retaining good gamma irradiation and high-energy electron beam (HEEB) food quality. Non-thermal processes, e.g. high hydrostatic (Farkas 1998; Moosekian et al. 2012; Pillai and Shayanfar 2017; Ravindran and Jaiswal 2019). Food matrix is likely to provide increased survival for viruses during irradiation, * Sophie Zuber but a major hurdle when performing inactivation studies sophie.zuber@rdls.nestle.com with viruses on food is the low recovery efficiency which is Nestlé Research, Institute of Food Safety and Analytical often encountered, leading to a low maximum measurable Science, 1000, 26 Lausanne, Switzerland Vol:.(1234567890) 1 3 Food and Environmental Virology (2021) 13:218–228 219 log reduction (Bosch et al. 2018; Butot et al. 2014). For food surfaces such as raisins (Etter et al. 2018). Inactivation example, at a HEEB dose of 12 kilo Gray (kGy), inactiva- of enveloped viruses such as influenza A in liquids has been tion of MNV (murine norovirus used as hNoV surrogate) in demonstrated (Etter et al. 2018), but no data are available on phosphate-buffered saline (PBS) reached 6.4 log , while on the effect LEEB has on non-enveloped enteric viruses such strawberries a maximum reduction of 2.2 log was observed as HAV and hNoV. The only industrial scale system cur- (Sanglay et al. 2011). Similarly, after a 11.9 kGy treatment rently available to the food industry is the Bühler AG Laatu in simple medium (DMEM) a 4.25 log reduction of Tulane system. As this system is an open system, it does not allow virus (TV) was measured, compared to a 2.1 log reduc- safe handling of pathogens and non-pathogenic surrogates tion after a treatment of 12.2 kGy on whole strawberries are needed to estimate virus inactivation by LEEB. (DiCaprio et al. 2016). The D values (dose required to The objective of this study was first to use HEEB treat- reduce virus titer by 1 log ) of rotavirus and poliovirus on ments to find surrogate microorganisms for enteric viruses lettuce were 1.0 and 2.3 kGy, respectively, suggesting virus- (HAV and hNoV) which are easy to grow to high titers and dependent susceptibilities to HEEB (Espinosa et al. 2012). use at industrial scale such as bacterial spores or bacterio- Using gamma irradiation, D values determined for HAV phages and second to use the surrogate organisms as proof in lettuce and strawberries were 2.7 and 3.0 kGy, respec- of concept to investigate LEEB treatments for enteric virus tively (Bidawid et al. 2000), while the HEEB dose required inactivation at industrial scale using frozen blueberries as a to reduce HAV by 1 log in whole oysters was 4.8 kGy model matrix. (Praveen et al. 2013). A new approach to irradiation processing of food, called low-energy electron beam (LEEB), uses electrons with ener- Materials and methods gies below 300 kilo electron Volt (keV), associated with a very limited penetration ability of around 500 μm (Gryczka HEEB Irradiation Treatment and Dose Evaluation et al. 2018; Hayashi et al. 1998; Pillai and Shayanfar 2017). In contrast, HEEB uses electrons with energies between 5 HEEB treatments (10 MeV) were performed at the LEONI and 10 Mega electron Volt (MeV) which penetrate foods Studer AG irradiation centre (Daeniken, Switzerland). Inoc- with high water content up to 3.9 cm (Farkas 2006) and ulated food samples were treated at three radiation doses, 4, has to be carried out at a dedicated irradiation facility with 8 and 16 kGy, to define critical dosages and screen for ade- appropriate safety measures. In contrast, the LEEB technol- quate surrogates. Pretests were performed for the different ogy is scalable to continuous processes and can be easily food matrices used in this study to verify dose distribution implemented in existing processing lines and offers ben- within the load. Based on the results, samples were pack- efits in cost reduction, environmental performance, and aged in 50 mL Falcon tubes with a maximum diameter of production flexibility (Zhang et al. 2018; Tetra Pak 2019). 3 cm which ensured equal irradiation doses throughout the As such, LEEB is an emerging irradiation technology that samples. The absorbed doses were measured using alanine performs surface and subsurface decontamination with a pellets (Aerial CRT, France) which were treated together minimal influence on food quality (Hertwig et al. 2018 ). with the samples (nine independent treatments performed on Since microorganisms reside mostly on the surface and three separate days) and the doses were calculated by com- subsurface of food if internalized through pores or dam- parison with a standard calibration curve established by Aer- aged tissue, the irradiation of the external layer should be ial CRT onsite at LEONI Studer AG irradiation centre and sufficient to eliminate food-borne microorganisms (De Lara traceable to NPL standard. The following mean values and et al. 2002; Hayashi et al. 1998). For example, Kikuchi et al. standard deviations were obtained: 3.91 ± 0.03, 7.89 ± 0.14 recommended LEEB treatment over gamma irradiation for and 15.85 ± 0.33 kGy. The results showed that the values for soybean decontamination, because it induces minimum or HEEB are close to the nominal doses, and thus, the nominal no quality deterioration, since the electrons do not reach the values are used in the graphs. internal matrix (Kikuchi et al. 2003). Similarly, LEEB was proposed as a method for microbiological decontamination LEEB Irradiation Treatment and Dose Evaluation of seeds, as it was shown not to negatively influence germi- nation (Fan et al. 2017). The response of Bacillus pumilus LEEB treatments were performed with Laatu designed by spores was found to be very similar when treated with HEEB Bühler AG (Uzwil, Switzerland). Inoculated frozen blueber- and LEEB (Tallentire et al. 2010). Another study investigat- ries were treated twice with 250 keV to reach a nominal dose ing Geobacillus and Bacillus spores also showed that the of 16 kGy. Samples were conveyed via an inlet channel and spore inactivation efficiency by LEEB was comparable to feeder table which control the throughput and distribution of that of other ionizing radiations (Zhang et al. 2018). LEEB the product. The berries were treated during free-fall through was also shown to reduce mould spores even on challenging the treatment zone which consists of two LEEB lamps facing 1 3 220 Food and Environmental Virology (2021) 13:218–228 each other (https://digit al.buhle r group.com/laatu / ). The dose for MS2. The MS2 and Qβ stock concentrations were 11.3 was measured with Riso B3 radiochromic film sensor pro- and 10.1 log plaque forming units (PFU)/mL, respectively. vided by Riso High Dose Reference Laboratory (HDLR) (Roskilde, Denmark). Due to the practical limitations of Inoculation of Food Matrices with Bacteria placing a dosimeter on the frozen blueberries, a B3 film strip and Viruses of 230 mm × 20 mm was attached along a polymer plate cov- ering the entire width of the treatment zone. The dosimeters Suspension (100 μL) of bacterial spores, viruses or bacterio- were exposed to the same beam conditions (voltage, current) phages were spotted (droplets of 5.2 ± 0.4 μL) directly on the as the inoculated frozen blueberries. The dosimetry system food surface of the seven selected food matrices (pumpkin, (including radiochromic film, sensor measurement software and sesame seeds, peppercorns, fresh blueberries, freeze- and calibration) is traceable to international standards in line dried raspberries and raisins) using a technique described with ISO/ASTM 51261 (ISO 2013). Dose measurements previously (Delbeke et al. 2015). For the HEEB treatment, using Riso B3 films determined a mean surface dose value 10 g of each food matrix were inoculated with bacterial of 15.51 ± 1.31 kGy which is close to the nominal dose, and spores, viruses or bacteriophages, respectively, except for thus, the nominal value is used in the graph. the freeze-dried raspberries, for which 5 g was used. The contaminated samples were allowed to dry in a biosafety cabinet at room temperature for 1 h, before being packaged Food Matrices and Pretreatments in a 50 mL tube and subjected to the HEEB treatment. The inoculated fresh blueberry samples were frozen at − 20 °C Six different food matrices were used in this study. Pumpkin, after the packaging. Following the same approach, blueberry and sesame seeds were kindly provided by the German Insti- samples were inoculated with G. stearothermophilus and tute of Food Technologies (DIL) (Quakenbrück, Germany), MS2 bacteriophages and frozen prior to the LEEB treatment. black peppercorns, fresh blueberries, freeze-dried raspberry flakes and raisins were purchased at Sabater (Spain), a local Recovery of Contaminants from Food Matrices distributor (Lausanne, Switzerland), Chaucer (UK) and Mariani Packing (USA), respectively. X-ray pretreatments To recover inoculated spores of G. stearothermophilus (25–50 kGy) of the matrices to kill background microflora from matrices, 5 or 10 g sample was aseptically transferred before inoculation with target organisms were performed to a stomacher filter bag, diluted 1:10 in buffered peptone at Synergy Health Daeniken AG (Daeniken, Switzerland). water and stomached for 1 min. From this initial suspension, 10 mL was submitted to a heat treatment at 80 °C for 10 min Geobacillus Spore Suspension to allow the spores to germinate. After the heat treatment, tubes were cooled down in cold water. Serial tenfold dilu- Ready to use suspension of Geobacillus stearothermophi- tions in tryptone salt solution were performed and plated on lus (ATCC 7953) spores, containing 7.7–8.7 log colony Plate Count Agar supplemented with 1 g/L of soluble starch forming units (CFU)/mL of viable spores, was purchased and incubated at 55 °C ± 1 °C for 48 h ± 4 h. The limit of from Merck Millipore (Burlington, USA) and was used to quantification (LOQ) was 2.00 log CFU/10 g for all food inoculate the different food matrices. matrices except for the freeze-dried raspberries, for which the LOQ was 1.98 log CFU/5 g. Viruses and Preparation of Suspensions HAV and MNV were recovered from 5 or 10 g samples using the ISO 15,216 virus extraction method (soft fruits The cytopathogenic HAV strain HM-175 (ATCC VR-1402), protocol) with slight modifications: no added process con- the murine norovirus strain MNV S99 used as proxy for trol, as untreated samples were analyzed to determine recov- hNoV and provided courtesy of the Friedrich-Loeffler-Insti- ery rate and the chloroform/butanol clarification step was not tut in Germany (Mueller 2007) were propagated, assayed and performed (ISO 2017). Before enumeration, concentrated titrated on FRhK-4 cells (ATTC CRL1688) and RAW 264.7 samples were decontaminated by sequential filtering through cells (ATCC TIB-71), respectively, as described previously 0.45 μm and then 0.22 μm spin centrifuge tube filters (Corn- (Butot et al. 2018). Viral stock titers were 6.6 and 7.6 log ing, New York) pretreated with 300 μL of phosphate-buff- 50% tissue culture infective dose (TCID )/mL for HAV and ered saline (pH 7.2 ± 0.2) containing 10% fetal calf serum MNV, respectively. MS2 bacteriophage (ATCC 15597-B1) followed by T CID titration as previously described (Butot was propagated, assayed and enumerated in Escherichia coli et al. 2018). The LOQ was 1.05 log TCID /10 g for all 10 50 K12 (ATCC 23631), as described previously (Butot et al. food matrices except for the freeze-dried raspberries, for 2018). The Qβ bacteriophage (ATTC 23631-B1) was propa- which the LOQ was 1.35 log TCID /5  g as the pellet 10 50 gated, assayed and enumerated using the same protocols as obtained after the polyethylene glycol precipitation was 1 3 Food and Environmental Virology (2021) 13:218–228 221 too viscous and therefore resuspended in 1 mL instead of on the values above maximum measurable log reduction 0.5 mL. to compute a pooled variance, common to all the measure- MS2 and Qβ were recovered from 5 or 10  g samples ments of a given microorganism in a given matrix. Notice using the protocol described previously (Butot et al. 2018). that since the width of the prediction interval depends only Briefly, the sample was transferred in a filter bag containing on the variance of the observations, the prediction intervals 50 mL of buffer (100 mM Tris, 50 mM glycine, 3% (m/v) have the same width for all the doses of a given microor- beef extract and 50 mM MgCl and adjusted to pH 9.5 with ganism and matrix. Finally, the mean log reductions com- NaOH solution) and stomached for 1 min. Serial dilutions puted in the presence of values at the maximum measurable (tenfold) were prepared and virus titer was quantified using log reduction are just a lower bound, since the real log ISO method 10705–1:1995(E), Annex C (ISO 1995). The reductions might be higher. This means that the presence LOQ was 2.78 log PFU/10 g for all food matrices except of values at the maximum measurable log reduction can 10 10 for the freeze-dried raspberries, for which the LOQ was 2.74 hide the effectiveness of the treatment, since the real mean log PFU/5 g. log reduction might actually be bigger than the estimated For all matrix/strain combinations, untreated inoculated one. However, this does not affect the size of the prediction triplicate samples were analyzed to determine the log intervals, which are computed using only values above the reductions, the maximum measurable log reduction cal- maximum measurable log reduction. 10 10 culated as log (LOQ/N ) and the recovery rates and un- We notice that the overall variability of the data is deter- 10 0 contaminated triplicate samples were analyzed as negative mined by different sources, for example, a variability in the controls following the procedures described above. recovery rates and in the irradiation values, and a biological variability in the response of the microorganisms. There- Statistical Analysis fore, the variance computed as described before takes into account all these sources of variability, that are also likely Reductions in spore and infectious viral particle counts to be found when applying the irradiation technology to real (inactivation) were calculated as log (N /N ), where N is production samples. 10 x 0 x the spore count or the viral titer recovered from treated food All statistical analyses were performed with R v3.6.1 matrices and N is the initial count or titer recovered from (R-Core Team 2019). untreated food matrices (mean of three replicates). Values of N below the LOQ were entered as being at that limit and this leads the log reduction to reach the maximum measur- Results able log reduction. A t test with a significance level at 0.05 and without cor - Recovery Efficiencies rection for multiplicity of test was used to test whether the mean log reduction of the replicates in one condition was To evaluate the efficiency of the methodologies employed, statistically different from the mean in another condition. the recovery of each microorganism was determined for Furthermore, it is also important to understand if the single all matrices tested as the ratio, in percentage, of N which measurements are consistently higher or lower. Thus, for corresponds to the initial count recovered from untreated each condition we also computed a prediction interval, cor- matrices and the calculated theoretical count using the responding to the interval where 95% of log reductions stock concentration and the volume inoculated (Table 1). measured in future experiments are expected to fall. If the G. stearothermophilus was recovered efficiently with mean prediction intervals of two conditions are non-overlapping, values ranging between 46.61 and 100.79% for all matrices, we can be confident that not only the mean log reductions are different, but also that the single observations in future Table 1 Recovery efficiencies of each microorganism–matrix combi- experiments are going to be well separated. nation expressed in percentage (%) In order to compute the t tests and the prediction inter- vals, the variance inside each condition is needed. However, Geobacillus HAV MNV MS2 Qβ as the maximum measurable log reduction was reached in Frozen blueberries 100.79 0.65 0.01 1.59 0.04 some conditions, the measured values are not representative Freeze-dried raspberries 25.72 0.40 – 0.03 0.05 of the real ones and must be discarded in the computation Raisins 92.86 4.22 0.06 0.09 2.97 of the variance. This leads to having conditions with very Peppercorns 82.54 0.89 16.24 0.80 0.03 few points (possibly zero) from which to estimate the vari- Sesame seeds 46.61 0.46 1.35 9.92 17.76 ance. To solve this issue, we assumed that the variance in Pumpkin seeds 63.92 0.12 0.60 0.06 7.26 the log reductions of a microorganism in a given matrix is independent from the dose. This allows us to use ANOVA – the missing value could not be determined due to technical issues 1 3 HAV MNV 222 Food and Environmental Virology (2021) 13:218–228 except for the freeze-dried raspberries where a rate of TCID /5 g, the maximum measurable log reduction was 50 10 25.72% was obtained. On the opposite, the bacteriophages 1.69 log . and the viruses showed lower recovery rates. The bacterio- We notice that a low recovery rate limits the maximum phages were recovered with efficiencies ranging from 0.03 measurable log reduction, therefore decreasing the power to 9.92% for MS2 and 0.03–17.76% for Qβ. Matrix effects of detecting strong reductions. However, even for the treat- were observed but no correlation was found between the ments where the low recovery leads to values below the two bacteriophages. Recovery rates above 1% were obtained LOQ, meaningful conclusions can still be drawn, as will be with sesame seed and frozen blueberry matrices for MS2 discussed in the following sections. Also, the low variabil- and with sesame seed, pumpkin seed and raisin samples for ity rates can increase the variance of the measured effects. Qβ. For viruses, the recovery rates oscillated from 0.12 and However, as explained in “Materials and Methods” section, 4.22% for HAV and from 0.01 to 16.24% for the hNoV proxy the computation of the t tests and of the prediction intervals MNV. Recoveries above 1% for HAV were reached only on take into account also any variation in the recoveries. raisins and for MNV only on sesame seeds and peppercorns. The freeze-dried raspberries showed low recoveries for all Reductions of HAV and MNV By HEEB Treatment the microorganisms (Table 1). In addition, this food product on Different Food Matrices induced a cytotoxic effect on the RAW 264.7 cells, making it impossible to report inactivation results for MNV. How- Inactivation values of HAV and MNV inoculated onto the ever, this effect was not observed on the FrhK4 cells used to different model food matrices (fruit, spice and seed) are measure the infectivity of HAV. shown in Fig. 1. The results are shown in log reduction The recovery efficiency is an important parameter as it with 95% prediction intervals (not confidence intervals) to has an impact on the maximum measurable log reduc- better represent the “minimal” effect of a certain treatment tion. Indeed, the lower the recovery, the lower the N and condition. A measure obtained from a future experiment will the maximum measurable log reduction will be. As an fall on this interval with a probability of 95%. The highest example, 5.45 log TCID of HAV were inoculated on reduction of HAV was observed when inoculated on rai- 10 50 5 g freeze-dried raspberries, but with a recovery efficiency sins (Fig.  1). On this matrix at 8 kGy, a mean reduction of 0.40%, only 3.04 log TCID /5 g of HAV were enu- of 1.9 log was measured, ensuring a minimal reduction 10 50 10 merated on the samples not treated (N ). Therefore, taking of only 0.7 log , according to the 95% prediction interval. 0 10 into consideration this N value and the LOQ of 1.35 log At 16 kGy the maximum measurable log reduction was 0 10 10 reached (> 3.2 log ), corresponding to a minimal reduction Freeze − driedraspberries Frozen blueberries Raisins Peppercorn Sesameseeds Pumpkinseeds −2 −4 −6 −2 −4 −6 48 16 48 16 48 16 48 16 48 16 48 16 Dose [kGy] Fig. 1 Mean inactivation of HAV and MNV on freeze-dried rasp- presence of values below the maximum measurable log reduction, berries, frozen blueberries, raisins, pepper corns, sesame seeds and we only have a lower bound on the mean reduction. For the predic- pumpkin seeds treated with HEEB at 4, 8 and 16  kGy. Different tion intervals, only the circles have been considered, and a common shadings show significant differences in the mean log reductions variance has been computed for a given microorganism-matrix com- between dosages within each microorganism–matrix combination. bination. Therefore, all the prediction intervals for a given combina- Bars represent 95% prediction intervals. Open circles represent single tion have the same widths, and they are not affected by having val- values, crosses represent single values below the maximum measur- ues below the maximum measurable log reduction. Supplementary able log reduction. To compute the mean log reduction, both cir- Table 1 contains the raw data used to generate Fig. 1 cles and crosses have been taken into account. For this reason, in the 1 3 log(N /N ) + 95% Prediction Interval x 0 Food and Environmental Virology (2021) 13:218–228 223 which is at least 2 log as given by the prediction inter- Reductions of Selected Surrogate Organisms By val. Notice that we talk about minimal reduction, because HEEB Treatment the real mean reduction might be bigger than the estimated one, as the maximum measurable log reduction has been Inactivation values of the three surrogate organisms (spores reached for all the three replicates (this does not affect the G. stearothermophilus and bacteriophages MS2 and Qβ) size of the prediction intervals, which are computed only inoculated onto the different model food matrices (fruit, from values above the maximum measurable log reduc- spice and seed) are shown in Fig.  2. MS2 and Qβ both tion). A dose-dependent inactivation of HAV was also showed a high resistance to HEEB treatments compared observed on frozen blueberries with mean reduction values to G. stearothermophilus, with mean reductions generally of 0.5, 1.4 and 2.4 l og at 4, 8 and 16 kGy, respectively. lower than 1.5 log . The only exceptions were raisins, where For comparison with raisins, the minimal log reduction at both bacteriophages showed reductions close to those for G. 16 kGy on frozen blueberries was estimated at 1.75 l og . stearothermophilus. We noticed that for some matrices, the The reductions of HAV inoculated on freeze-dried raspber- prediction intervals at two different irradiation doses (error ries subjected to the three HEEB dosages of 4, 8 and 16 kGy bars) were overlapping, meaning that the doses do not have a appeared not significantly different from each other (bars of statistically different effect. That is, we cannot conclude that the same colour for all three dosages), but this may be an future measurements at the higher dose will have a higher artefact which could be overcome if we had a better extrac- log reduction compared to the lower dose. The overlapping tion efficiency allowing to see a greater maximum measur - of prediction intervals can happen even when the means able log reduction than 1.7 log at 16 kGy. A significant are statistically different (the bars have a different shade of 10 10 dose-dependent inactivation of HAV subjected on HEEB grey), as for example for Qβ in sesame seeds treated at 4 treatment of 4, 8 and 16 kGy was seen on peppercorns and and 8 kGy. This shows that, although the mean reductions sesame seeds, with mean HAV reductions of 2.1 and 2.2 can be different at two different irradiation doses, for single log when treated at 8 kGy, respectively. At 16 kGy the future measurements we cannot conclude that the reduction maximum measurable log reduction was reached on both is lower at 4 kGy compared to 8 kGy. matrices, corresponding to 2.6 log for peppercorns and 3.1 On sesame seeds, however, MS2 showed a clear lower log for sesame seeds. The reductions of HAV inoculated reduction compared to Qβ, both considering the mean and on pumpkin seeds were lower than the reductions observed the prediction interval (Supplementary Table 2). Given the on sesame seeds (a mean inactivation of only 1.1 log at similarities between the two bacteriophages, and the fact 16 kGy) which may indicate complex matrix effects and that compared to Qβ, MS2 showed a higher recovery and a show that predictions of inactivation are risky, even for simi- dose-dependent reduction on frozen blueberries, we chose lar food matrices. to focus only on MS2 as candidate surrogate for HAV and Compared to HAV, MNV showed a higher resistance to hNoV for the LEEB study. This choice is especially relevant HEEB on all matrices, except for frozen blueberries where for HAV, as on all food matrices and doses, HAV showed a maximum measurable log reductions were reached in mean reduction which was either higher or non-statistically most cases, preventing a meaningful comparison (Fig. 1). different than the mean reduction for MS2 (Supplementary This highlights the high resistance of MNV to irradiation Table 3). by HEEB on food matrices (complementary Table 1). The Geobacillus stearothermophilus showed a dose-depend- inactivation was lowest on pumpkin seeds and peppercorns ent inactivation on all tested matrices with a generally higher with mean inactivation values showing no significant dif- inactivation than the tested viruses, for both mean reduc- ferences among the three dosages applied and ranging from tions and prediction intervals, except on raisins (Fig. 2). On 0.03 to 0.32 log . MNV inoculated on raisins and sesame frozen blueberries, for example, a 1.3, 3.3 and > 5.6 log 10 10 seeds did not show significant dose-dependent differences reduction was measured for HEEB treatments at 4, 8 and either, with reductions ranging from 1.35 to 1.44 log on 16 kGy, respectively. In comparison, at 16 kGy on the same raisins and 0.96 and 1.34 log on sesame seeds. A signifi- matrix, 2.4 log and 1.8 log reductions were measured 10 10 cantly higher inactivation at a higher HEEB dose could only for HAV and MS2, respectively (Figs. 1, 2). Similarly, on be measured for MNV on frozen blueberries with mean log sesame seeds, pumpkin seeds and peppercorns at 4, 8 and reductions of 0.35, > 0.76 and > 1.18 log for treatments at 16 kGy, HAV showed a lower mean inactivation level com- 4, 8 and 16 kGy, respectively. pared to G. stearothermophilus (Figs. 1, 2). This intermedi- ate ranking of HAV between G. stearothermophilus spores and MS2 determined by HEEB was ideal to select both sur- rogate organisms as first approximation to estimate HAV inactivation during LEEB treatments of frozen blueberries. 1 3 G. stearothermophilus MS2 phage Qβ Frozen blueberries 224 Food and Environmental Virology (2021) 13:218–228 Freeze − driedraspberries Pepper corn Pumpkinseeds Frozen blueberries Raisins Sesame seeds −2 −4 −6 −2 −4 −6 −2 −4 −6 48 16 48 16 48 16 48 16 48 16 48 16 Dose [kGy] Fig. 2 Mean inactivation of G. stearothermophilus spores, MS2 and intervals. Open circles represent single values, crosses represent sin- Qβ bacteriophages on freeze-dried raspberries, frozen blueberries, gle values below the maximum measurable log reduction. Notice raisins, pepper corns, sesame seeds and pumpkin seeds treated with that since a common variance has been computed for a given micro- HEEB at 4, 8 and 16  kGy. Different shadings show significant dif- organism-matrix combination, all the prediction intervals for that ferences in the mean log reductions between dosages within each combination have the same width. Supplementary Table  2 contains microorganism-matrix combination. Bars represent 95% prediction the raw data used to generate Fig. 2 bacteriophage on frozen blueberries were treated in the Inactivation of G. stearothermophilus Spores industrial Bühler AG Laatu system, as a proof of concept and MS2 During LEEB on Frozen Blueberries for validation of HAV and hNoV inactivation by LEEB. In Fig. 3, the LEEB results obtained from this experiment are Geobacillus stearothermophilus spores and MS2 Fig. 3 Mean inactivation of G. stearothermophilus MS2phage G. stearothermophilus spores, HAV and MS2 bacteriophage on frozen blueberries treated with LEEB at 16 kGy in −1 comparison with corresponding reduction generated by HEEB at 8 and 16 kGy. Open circles represent single values, crosses represent single values below the maximum measurable log −3 reduction −5 HEEB HEEB LEEB HEEB HEEB LEEB 8 kGy 16 kGy 16 kGy 8 kGy 16 kGy 16 kGy Treatment 1 3 log(N /N ) + 95% Prediction Interval x 0 log(N /N ) x 0 Food and Environmental Virology (2021) 13:218–228 225 compared with the data generated by HEEB. As the LEEB As reported by others, spores of G. stearothermophi- experiment was a proof of concept with a low amount of lus are highly resistant against irradiation and showed a data, we did not carry out statistical analysis on these data higher resistance during LEEB than the spores of B. pumi- points. Mean reductions of 3.1 l og and 1.3 l og were lus which is often suggested to be the biological indicator 10 10 measured after the LEEB surface treatment for G. stearo- for irradiation sterilization (Van Gerwen et al. 1999; Zhang thermophilus spores and MS2, respectively (Fig. 3). These et al. 2018). Therefore, G. stearothermophilus spores were values were similar to the log reductions determined for included in this study to compare their resistance to elec- both organisms in the HEEB trials at 8 kGy (3.3 log for tron beam with the one of HAV. We observed a higher sus- G. stearothermophilus spores and 1.1 log for MS2). The ceptibility of G. stearothermophilus to HEEB compared to similarity of the log reductions of the surrogates observed HAV (Figs. 1, 2). The data generated in this study for G. on frozen blueberries after a LEEB treatment at 16 kGy and stearothermophilus is new and valuable for future industrial a HEEB treatment at 8 kGy, suggests that if HAV underwent validation studies. It cannot be directly compared with the a LEEB treatment at 16 kGy, we might expect a reduction data present in the literature, as most studies have deter- similar to the one determined for the HEEB treatment at mined D values for electron beam treatments of B. pumilus 8 kGy (Fig.  1). Thus, we expect to see a log reduction and B. subtilis and to a lesser extent G. stearothermophilus of 1.4 on frozen blueberries for HAV treated by LEEB at spores not inoculated onto any food matrix (Tallentire et al. 16 kGy. 2010; Zhang et al. 2020, 2018). Nevertheless, the D-value of 3.1 kGy at 200 keV measured by Zhang and coauthors cor- responds to a 2.6 log reduction at 8 kGy which is similar Discussion to the reduction of 2.87 log we measured on peppercorn and sesame seeds. Additionally, the food matrix can have The results of the present study represent a proof of concept opposite effects on different microorganisms. For example, for validation of HAV and hNoV inactivation by LEEB, on raisins G. stearothermophilus showed a lower inactiva- using the industrial Bühler AG Laatu system, the only indus- tion than on the other matrices. Viruses, on the other hand, trial scale system currently available to the food industry. generally showed higher inactivation on raisins compared As this system is an open system not allowing safe handling to the other matrices, showing that extrapolation of effects of pathogens, the HEEB technology was used with inocu- from one matrix to another should not be attempted. lated and packaged food matrices to compare reduction of Low performance of the methods used to recover the HAV and hNoV with potential surrogates. The HEEB trials viruses and the bacteriophages impacted only a small part generated new data on the effect electron beam irradiation of the results, as the maximum measurable log reductions has on different viruses-food matrix combinations. Over - were in most cases sufficient to draw meaningful conclu- all, this study demonstrates the high resistance of HAV to sions. G. stearothermophilus spores showed good recoveries electron beam irradiation and an even higher resistance of on every matrix, whereas enteric viruses and bacteriophages MNV (Fig. 1). As different amino acids vary in their sus- displayed poor recovery rates. For enteric viruses, such as ceptibility to irradiation, crosslinking of proteins during HAV and MNV, this was expected as the methods avail- irradiation may damage portions of the capsid involved in able for the detection of viruses are well known for their receptor binding to varying degrees and may explain differ - complexities and low recovery efficiencies (Li et al. 2018; ences in susceptibility observed between different viruses Mäde et al. 2013; Perrin et al. 2015). In ISO 15,216, the (Hume et al. 2016; Stewart 2001). A comparison with hNoV minimal recovery efficiency required is 1% but this require- is needed in the future to determine which hNoV surrogate ment is often not reached (Li et al. 2018). For example, Per- can reliably be used as surrogate to evaluate the effect of rin and coauthors reported a recovery of 0.5% of MNV on electron beam irradiation. Using the porcine gastric mucin raspberries and Hida and coauthors reported MNV recov- magnetic bead (PGM-MB) binding assay, the level of hNoV ery efficiencies of 0.06% and 0.41% on lettuce and grapes, and TV RNA decreased by > 1 log and approximately 2 respectively (Hida et al. 2018; Perrin et al. 2015). The low log , respectively, following a 12.2 kGy HEEB treatment recovery efficiencies obtained with the bacteriophages were on whole strawberries (DiCaprio et al. 2016). In compari- more surprising as the method used is in principle simple son, at 8 kGy for MNV our results reveal lo g reductions and easy to apply. Shim and coauthors showed a correlation of 0.76 and 1.35 log on frozen blueberries and raisins, between the stronger adhesion and the lower recoveries of respectively. As shown by Ettayebi and coauthors using the MS2 on PVC, therefore one hypothesis of this low perfor- stem cell derived human enteroids assay, exposure to gamma mance could be a strong adhesion of the bacteriophages on irradiation of hNoV GII.3 and GII.4 stool suspensions inac- the matrices, impacting the recovery rates (Shim et al. 2017). tivated the viruses, but at this point in time no data are avail- One explanation of this stronger adhesion could be the time able with this system on food matrices (Ettayebi et al. 2016). between the inoculation and the analysis, as shown by Shim 1 3 226 Food and Environmental Virology (2021) 13:218–228 the Bühler AG team for the LEEB trials and dosimetry measurements and coauthors who extended this time from 1 min to 24 h to and Mireille Moser for initial support with data analysis. allow stronger MS2 attachment for their study (Shim et al. 2017). Author Contributions SZ was activity leader within EIT Food project The similarity of the log reductions observed between ID19087 and supervised the study. SZ and SB designed the study, HEEB at 8 kGy and LEEB at 16 kGy for both G. stearo- prepared and tested the samples and wrote the manuscript. LG per- formed the statistical analyses of the study and wrote the manuscript. thermophilus and MS2 on frozen blueberries may be due TP prepared, tested the samples and helped to write the manuscript. to several factors, one of them being the difference in the penetration depth of electrons between HEEB and LEEB. Funding This research was conducted under EIT Food project The HEEB measured dose is applied to the full depth of the ID19087with funding from the European Institute of Innovation and food matrix. The LEEB dose is a surface and subsurface Technology (EIT), a body of the European Union, under Horizon 2020, the EU Framework Program for Research and Innovation of the EIT measurement, where the average absorbed dose along the Food project ID19087. food matrix is depending on the depth dose distribution at a given energy, the matrix density and its chemical com- Compliance with Ethical Standards position (Shim et al. 2017). In the current study the beam penetration depth for frozen blueberries was not determined, Conflict of Interest The authors declare that they have no conflict of but in a matrix with water density at 250 keV, the absorbed interest. dose is maximum at the surface while it exponentially drops to zero at approximately 580 µm. If a certain number of Open Access This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- microorganisms are beyond the penetration ability of the tion, distribution and reproduction in any medium or format, as long electrons (at 250 keV) they will not be inactivated. Addi- as you give appropriate credit to the original author(s) and the source, tionally, it has been hypothesized that lower D value for provide a link to the Creative Commons licence, and indicate if changes HEEB versus LEEB treatments may be due to the additional were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated inactivation effect of reactive oxygen species if atmospheres otherwise in a credit line to the material. If material is not included in during the treatments contain different amounts of O . Thus, the article’s Creative Commons licence and your intended use is not when comparing HEEB and LEEB systems, the focus should permitted by statutory regulation or exceeds the permitted use, you will not be on comparing the dosimetry measurements. Instead, need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativ ecommons .or g/licenses/b y/4.0/. the impact of achieving the intended microbial reduction and the impact on food quality, such as nutrients and sensory attributes, should be evaluated. The quality of the HEEB and LEEB treated foods in this study were evaluated using sen- References sory attributes. There were no visible differences between HEEB and LEEB treated foods in this study and detailed Alegbeleye, O. O., Singleton, I., & Sant’Ana, A. S. (2018). Sources and sensory profiles and chemical properties are described else- contamination routes of microbial pathogens to fresh produce dur- where (Aisala et al. in preparation). ing field cultivation: A review. Food Microbiology, 73, 177–208. Based on our results for G. stearothermophilus and MS2, Anonymous. (2018). Swift investigation of hepatitis A outbreak facili- tated withdrawal of dates. Copenhagen: Statens Serum Institute. a log reduction of 1.4 can be expected for HAV on fro- Anonymous. (2020). Evaluation of the EU legal framework on food zen blueberries treated by LEEB at 16 kGy, indicating that irradiation. Brussels: EU. LEEB processing cannot completely eliminate the risk of Bidawid, S., Farber, J. M., & Sattar, S. A. (2000). Inactivation of hepa- viral illness on berries and emphasizes the critical impor- titis A virus (HAV) in fruits and vegetables by gamma irradiation. International Journal of Food Microbiology, 57, 91–97. tance of adhering to stringent application of hygiene control Bosch, A., Gkogka, E., Le Guyader, F. S., Loisy-Hamon, F., Lee, A., systems that limit enteric virus contamination at primary van Lieshout, L., et al. (2018). Foodborne viruses: Detection, risk production. In addition, the list of foods allowed for irradia- assessment, and control options in food processing. International tion treatment in the EU currently comprises only dried aro- Journal of Food Microbiology, 285, 110–128. Butot, S., Zuber, S., & Baert, L. (2014). Sample preparation prior to matic herbs, spices and vegetable seasonings and not berries, molecular amplification: Complexities and opportunities. Current but this may change in the future, as a legislative evaluation Opinion in Virology, 4, 66–70. related to the irradiation of food and food ingredients is on- Butot, S., Cantergiani, F., Moser, M., Jean, J., Lima, A., Michot, L., going within the EU (Anonymous 2020). et al. (2018). UV-C inactivation of foodborne bacterial and viral pathogens and surrogates on fresh and frozen berries. Interna- tional Journal of Food Microbiology, 275, 8–16. Supplementary Information The online version contains supplemen- De Lara, J., Fernández, P. S., Periago, P. M., & Palop, A. (2002). Irra- tary material available at https:// doi.org/10.1007/s12560-02 1-09463-3 . diation of spores of Bacillus cereus and Bacillus subtilis with electron beams. Innovative Food Science and Emerging Technolo- Acknowledgements We gratefully acknowledge Conrad Guenthard gies, 3, 379–384. (LEONI Studer AG irradiation centre in Daeniken, Switzerland) for support with the HEEB trial setup and the dosimetry measurements, 1 3 Food and Environmental Virology (2021) 13:218–228 227 Delbeke, S., Ceuppens, S., Jacxsens, L., & Uyttendaele, M. (2015). Kikuchi, O. K., Todoriki, S., Saito, M., & Hayashi, T. (2003). Efficacy Survival of salmonella and escherichia coli O157:H7 on strawber- of soft-electron (low-energy electron beam) for soybean decon- ries, basil, and other leafy greens during storage. Journal of Food tamination in comparison with gamma-rays. Journal of Food Sci- Protection, 78, 652–660. ence, 68, 649–652. DiCaprio, E., Phantkankum, N., Culbertson, D., Ma, Y., Hughes, J. Kingsley, D. H., & Chen, H. (2009). Influence of pH, salt, and tempera- H., Kingsley, D., et al. (2016). Inactivation of human norovirus ture on pressure inactivation of hepatitis A virus. International and Tulane virus in simple media and fresh whole strawberries Journal of Food Microbiology, 130, 61–64. by ionizing radiation. International Journal of Food Microbiol- Knorr, D., & Watzke, H. (2019). Food processing at a crossroad. Fron- ogy, 232, 43–51. tiers in Nutrition, 6, 85. DiCaprio, E., Ye, M., Chen, H., & Li, J. (2019). Inactivation of human Li, D., Butot, S., Zuber, S., & Uyttendaele, M. (2018). Monitoring norovirus and tulane virus by high pressure processing in simple of foodborne viruses in berries and considerations on the use of mediums and strawberry puree. Frontiers in Sustainable Food RT-PCR methods in surveillance. Food Control, 89, 235–240. Systems. https ://doi.org/10.3389/fsufs .2019.00026 . Mäde, D., Trübner, K., Neubert, E., Höhne, M., & Johne, R. (2013). Espinosa, A. C., Jesudhasan, P., Arredondo, R., Cepeda, M., Mazari- Detection and typing of norovirus from frozen strawberries Hiriart, M., Mena, K. D., & Pillai, S. D. (2012). Quantifying the involved in a large-scale gastroenteritis outbreak in Germany. reduction in potential health risks by determining the sensitivity Food and Environmental Virology, 5, 162–168. of poliovirus type 1 chat strain and rotavirus SA-11 to electron Miranda, R. C., & Schaffner, D. W. (2019). Virus risk in the food sup- beam irradiation of iceberg lettuce and spinach. Applied and Envi- ply chain. Current Opinion in Food Science, 30, 43–48. ronmental Microbiology, 78, 988–993. Moosekian, S. R., Jeong, S., Marks, B. P., & Ryser, E. T. (2012). X-ray Ethelberg, S., Lisby, M., Bottiger, B., Schultz, A. C., Villif, A., Jensen, irradiation as a microbial intervention strategy for food. Annual T., et al. (2010). Outbreaks of gastroenteritis linked to lettuce, Review of Food Science and Technology, 3, 493–510. Denmark, January 2010. Euro Surveillance, 15, 19484. Perrin, A., Loutreul, J., Boudaud, N., Bertrand, I., & Gantzer, C. Ettayebi, K., Crawford, S. E., Murakami, K., Broughman, J. R., (2015). Rapid, simple and efficient method for detection of viral Karandikar, U., Tenge, V. R., et al. (2016). Replication of human genomes on raspberries. Journal of Virological Methods, 224, noroviruses in stem cell-derived human enteroids. Science, 353, 95–101. 1387–1393. Pillai, S. D., & Shayanfar, S. (2017). Electron beam technology and Etter, D., Rupp, A., Prange, A., & Drissner, D. (2018). Inactivation other irradiation technology applications in the food industry. Top- of mould spores in a model system and on raisins by low-energy ics in Current Chemistry, 375, 6. electron beam. Food Control, 92, 357–361. Praveen, C., Dancho, B. A., Kingsley, D. H., Calci, K. R., Meade, G. Fan, X., Sokorai, K., Weidauer, A., Gotzmann, G., Rögner, F. H., & K., Mena, K. D., & Pillai, S. D. (2013). Susceptibility of murine Koch, E. (2017). Comparison of gamma and electron beam irra- norovirus and hepatitis a virus to electron beam irradiation in diation in reducing populations of E. coli artificially inoculated oysters and quantifying the reduction in potential infection risks. on mung bean, clover and fenugreek seeds, and affecting germi- Applied and Environmental Microbiology, 79, 3796–3801. nation and growth of seeds. Radiation Physics and Chemistry, R-Core Team. (2019). R: A language and environment for statistical 130, 306–315. computing. Vienna: R-Core Team. Farkas, J. (1998). Irradiation as a method for decontaminating food: A Ravindran, R., & Jaiswal, A. K. (2019). Wholesomeness and safety review. International Journal of Food Microbiology, 44, 189–204. aspects of irradiated foods. Food Chemistry, 285, 363–368. Farkas, J. (2006). Irradiation for better foods. Trends in Food Science Sanglay, G. C., Li, J., Uribe, R. M., & Lee, K. (2011). Electron-beam and Technology, 17, 148–152. inactivation of a norovirus surrogate in fresh produce and model Gryczka, U., Migdał, W., & Bułka, S. (2018). The effectiveness of the systems. Journal of Food Protection, 74, 1155–1160. microbiological radiation decontamination process of agricultural Shim, J., Stewart, D. S., Nikolov, A. D., Wasan, D. T., Wang, R., Yan, products with the use of low energy electron beam. Radiation R., & Shieh, Y. C. (2017). Differential MS2 interaction with food Physics and Chemistry, 143, 59–62. contact surfaces determined by atomic force microscopy and Hayashi, T., Takahashi, Y., & Todoriki, S. (1998). Sterilization of foods virus recovery. Applied and Environmental Microbiology, 83, with low-energy electrons (’Soft-electrons’). Radiation Physics e01881–e01917. and Chemistry, 52, 73–76. Stewart, E. M. (2001). Food irradiation chemistry. In R. Molins (Ed.), Hertwig, C., Meneses, N., & Mathys, A. (2018). Cold atmospheric Food irradiation: Principles and applications (pp. 37–76). New pressure plasma and low energy electron beam as alternative non- York: Wiley. thermal decontamination technologies for dry food surfaces: A Tallentire, A., Miller, A., & Helt-Hansen, J. (2010). A comparison of review. Trends in Food Science and Technology, 77, 131–142. the microbicidal effectiveness of gamma rays and high and low Hida, K., Papafragkou, E., & Kulka, M. (2018). Testing for human energy electron radiations. Radiation Physics and Chemistry, 79, norovirus and recovery of process control in outbreak-associated 701–704. produce items. Journal of Food Protection, 81, 105–114. Tavoschi, L., Severi, E., Niskanen, T., Boelaert, F., Rizzi, V., Liebana, Hume, A. J., Ames, J., Rennick, L. J., Duprex, W. P., Marzi, A., Tonk- E., et al. (2015). Food-borne diseases associated with frozen ber- iss, J., & Mühlberger, E. (2016). Inactivation of RNA viruses by ries consumption: A historical perspective, European Union, 1983 gamma irradiation: A study on mitigating factors. Viruses, 8, 204. to 2013. Eurosurveillance Weekly, 20, 21193. ISO. (1995). ISO 10705-1:1995: Water quality? Detection and enu- Tetra Pak (2019). Tetra Pak E3 with eBeam. Retrieved from https :// meration of bacteriophages ? Part 1: Enumeration of F-specific www.tetra pak.com/tetra -pak-e3-ebeam . Accessed 29 Oct 2020. RNA bacteriophages. Geneva: ISO. Thebault, A., Teunis, P. F., Le Pendu, J., Le Guyader, F. S., & Denis, J. ISO. (2013). ISO/ASTM 51261:2013: Practice for calibration of rou- B. (2013). Infectivity of GI and GII noroviruses established from tine dosimetry systems for radiation processing. Geneva: ISO. oyster related outbreaks. Epidemics, 5, 98–110. ISO. (2017). ISO 15216–1:2017: Microbiology of the food chain: Hori- Tree, J. A., Adams, M. R., & Lees, D. N. (2005). Disinfection of feline zontal method for determination of hepatitis A virus and norovirus calicivirus (a surrogate for Norovirus) in wastewaters. Journal of in food using real-time RT-PCR: Part 1: Method for quantifica- Applied Microbiology, 98, 155–162. tion. Geneva: ISO. Van Gerwen, S. J. C., Rombouts, F. M., Riet, K. V., & Zwietering, M. H. (1999). A data analysis of the irradiation parameter for bacteria 1 3 228 Food and Environmental Virology (2021) 13:218–228 and spores under various conditions. Journal of Food Protection, Zhang, Y., Moeller, R., Tran, S., Dubovcova, B., Akepsimaidis, G., 62, 1024–1032. Meneses, N., et al. (2018). Geobacillus and bacillus spore inacti- Wiedenmann, A., Fischer, B., Straub, U., Wang, C.-H., Flehmig, B., & vation by low energy electron beam technology: Resistance and Schoenen, D. (1993). Disinfection of hepatitis a virus and MS-2 influencing factors. Frontiers in Microbiology, 9, 2790. coliphage in water by ultraviolet irradiation: Comparison of UV- susceptibility. Water Science and Technology, 27, 335–338. Publisher’s Note Springer Nature remains neutral with regard to Zhang, Y., Huber, N., Moeller, R., Stülke, J., Dubovcova, B., Akepsi- jurisdictional claims in published maps and institutional affiliations. maidis, G., et al. (2020). Role of DNA repair in Bacillus subtilis spore resistance to high energy and low energy electron beam treatments. Food Microbiology, 87, 103353. 1 3

Journal

Food and Environmental VirologySpringer Journals

Published: Feb 10, 2021

There are no references for this article.