Modelling of the Behaviour of Salmonella enterica serovar Reading on Commercial Fresh-Cut Iceberg Lettuce Stored at Different Temperatures

The aim of this study was to model the growth and survival behaviour of Salmonella Reading and endogenous lactic acid bacteria on fresh pre-cut iceberg lettuce stored under modified atmosphere packaging for 10 days at different temperatures (4, 8 and 15 °C). The Baranyi and Weibull models were satisfactorily fitted to describe microbial growth and survival behaviour, respectively. Results indicated that lactic acid bacteria (LAB) could grow at all storage temperatures, while S. Reading grew only at 15 °C. Specific growth rate values (μmax) for LAB ranged between 0.080 and 0.168 h−1 corresponding to the temperatures 4 and 15 °C while for S. Reading at 15 °C, μmax = 0.056 h−1. This result was compared with published predictive microbiology models for other Salmonella serovars in leafy greens, revealing that predictions from specific models could be valid for such a temperature, provided they were developed specifically in lettuce regardless of the type of serovars inoculated. The parameter delta obtained from the Weibull model for the pathogen was found to be 16.03 and 18.81 for 4 and 8 °C, respectively, indicating that the pathogen underwent larger reduction levels at lower temperatures (2.8 log10 decrease at 4 °C). These data suggest that this Salmonella serovar is especially sensitive to low temperatures, under the assayed conditions, while showcasing that a correct refrigeration could be an effective measure to control microbial risk in commercial packaged lettuce. Finally, the microbiological data and models from this study will be useful to consider more specifically the behaviour of S. Reading during transport and storage of fresh-cut lettuce, elucidating the contribution of this serovar to the risk by Salmonella in leafy green products.


Introduction
In recent years, the technological progress in food industry and consumer demand for a healthy lifestyle have led to a rise in the production and consumption of fresh produce, specially minimally processed pre-cut fruits and vegetables. In this regard, the fresh-cut industry has experienced a faster growth rate than other sectors of the fruit and vegetable market [1]. Particularly, packaged lettuce accounts for about 50% the fresh-cut market volume [2]. At the same time, there has been an increase in the number of foodborne outbreaks associated with such food commodities. The main challenge is that fresh-cut products are generally ready-to-eat (RTE) and consumed raw without any heat treatment or other microbial inactivation method applied prior to consumption that eliminates fully foodborne pathogens.
Contamination of vegetables in the field might be due to, among other factors, faecally contaminated irrigation water being a potential source of enteric pathogens [3]. The use of chlorine in post-harvest washing and modified atmosphere packaging (MAP), together with refrigerated storage, are the most typical preservation technologies used to ensure the food safety and quality of fresh-cut fruits and vegetables [4]. Despite that, washing water, cutting, packaging and food handlers can be sources of microbial contamination [5,6]. Indeed, the process of post-harvest washing can potentially lead to cross-contamination events if appropriate concentrations of sanitizer are not maintained [7].
According to the latest European Food Safety Authority (EFSA) and European Centre for Disease Prevention and Control (ECDC) EFSA/ECDC report on zoonoses and zoonotic agents, Salmonella enterica is the second most common cause of foodborne outbreaks after Campylobacter in the European Union (EU) [8]. In addition, Salmonella has been reported to be the second largest contributor to produce-associated outbreaks both in the United Stated (US) and the EU [9]. Particularly, leafy greens eaten raw as salads have been identified as the most frequent food commodity of non-animal origin implicated in Salmonella human infections in the EU [10]. S. Enteritidis is the predominant serovar in human cases acquired in the EU [8]. The most reported Salmonella serovars causing leafy vegetable-associated outbreaks in the US during 1973-2012 were Enteritidis and Typhimurium, followed by Javiana and Newport [11]. In the past, the serovar Reading has been uncommonly associated with human illness and rarely involved in foodborne outbreak. However, during the last years there has been a change in this trend with a rise of reported cases of S. Reading associated with different contaminated food items [12]. Lienemann et al. [13] investigated a nationwide outbreak caused by S. Newport and S. Reading in Finland and suggested pre-cut iceberg lettuce as the source of the outbreak. Robertson et al. [14] isolated S. Reading from processing water used in the production of mung bean sprouts. More recently, multidrug-resistant S. Reading has caused a multistate outbreak linked to raw turkey products in the US during 2017-2019, involving 358 cases in 42 states [15].
Predictive microbiology models are used to predict bacterial kinetics in specific food environments. Over the last decades, mathematical models have become a valuable information source to conduct risk assessment studies and support risk managers concerning microbial food safety and quality issues. A number of predictive models are available for Salmonella on leafy greens [16][17][18][19][20][21]. These models have been mostly developed for the serovars Enteritidis and Typhimurium, and at present, the individual survival and growth behaviour of S. Reading has not been reported in any food.
The objective of this study was to assess and model the growth and survival of S. Reading on fresh-cut RTE iceberg lettuce stored at different temperatures (4, 8 and 15 • C) for 10 days.

Sample Preparation and Microbiological Analysis
Fresh-cut RTE iceberg lettuce (Lactuca sativa L.) in bags (200 g) under MAP was obtained on the day of production from a local lettuce processing plant located in the Oslo area (Norway). The bags were collected at the processing plant and immediately transported to the laboratory for inoculation and initial testing.

Preparation of Inoculum
Salmonella Reading VI 51763, previously isolated from spent irrigation water used for sprout production [14], was plated on two blood agar plates (OXOID, Basingstoke, UK) from glycerol stocks kept at −80 • C and incubated at 37 • C for 24 h. A single colony was transferred to nine ml of buffered peptone water (BPW) (OXOID, Basingstoke, UK) and incubated for 24 h at 37 ± 1 • C. After incubation, the culture was serially diluted in peptone saline (PS) (0.1% peptone and 0.85% NaCl). The initial inoculum was quantified by plating 100 µL of the appropriate dilution on blood agar plates in triplicate, which were incubated at 37 ± 1 • C for 18-24 h before enumeration. The concentration of the inoculum varied between log 10 4.83 and log 10 5.08 cfu/mL.

Inoculation of Lettuce Bags
Lettuce bags were inoculated with 100 µL from the appropriate dilution following a similar protocol to the work by Carrasco et al. [22]. Briefly, 100 µL was inoculated through penetrating an adhesive silicon patch on each lettuce bag, using a 1 mL syringe and 0.6 × 25 mm needles. The area around the patch was marked in order to facilitate subsequent sampling.
Inoculated and control (non-inoculated) lettuce bags were stored at 4, 8 and 15 • C for 10 days (product use-by date). At storage days 1, 2, 3, 4, 7 and 10, five randomly selected lettuce bags were analysed separately for enumeration of S. Reading and endogenous lactic acid bacteria (LAB). Experiments were performed in triplicate for each storage temperature. In addition, control bags were analysed at day 0 and day 10 of storage for enumeration of total aerobic viable count (AVC) and for presence/absence of Salmonella spp.

Bacterial Analyses
S. Reading was quantified by homogenizing 25 g of inoculated lettuce in 225 mL BPW with a Stomacher (Stomacher 400 Lab Blender, Seward, UK) for 30 s. Tenfold dilution series of the homogenate was prepared in PS and 100 µL of the appropriate dilutions was plated in parallel on Xylose-Lysine-Desoxycholate agar (XLD) (OXOID) and Brilliance™ Salmonella agar (OXOID). The plates were incubated at 37 • C for 24 ± 3 h. Typical colonies were counted, and one typical colony was selected for further confirmation. S. Reading was confirmed by testing presumptive colonies on Triple Sugar Iron agar (Difco, MD, USA) and Urea agar-containing agar base (OXOID) with 40% Urea (Sigma-Aldrich, St. Louis, MO, USA) followed by agglutination with omnivalent Salmonella antiserum (Enteroclon Anti-Salmonella A-67, omnivalent, Sifin, Berlin, Germany). The same homogenate was used for LAB enumeration following Nordic Committee on Food Analysis (NMKL) no. 140 2nd ed., 2007. Plates were incubated anaerobically at 25 • C for five days.
The detection of Salmonella spp. in control bags was conducted in accordance with the NMKL no. 71 5th ed., 1999. Briefly, a homogenate was prepared as described above and incubated at 37 • C for 24 ± 3 h. An aliquot of 100 µL of the enrichment was transferred to 10 mL Rappaport Vassiliadis Soya Peptone broth (RVS) (OXOID) and incubated for 24 ± 3 h at 41.5 • C, followed by plating of a loopful (10 µL) of broth on XLD agar (OXOID) and Brilliance™ Salmonella agar (OXOID) and incubated as described above. Analysis of AVC in control bags was carried out following the method NMKL no. 86, 5th ed., 2013. Plates were incubated under aerobic conditions for three days at 25 • C.

Growth Model
The Baranyi model [23] as defined by Dabadé et al. [24] and Den Besten et al. [25] was employed to estimate the growth kinetic parameters of microorganisms on fresh-cut iceberg lettuce stored at isothermal conditions: where t is the time (h), N(t) is the number of microorganisms (log 10 CFU/g) at time t, N 0 is the initial number of microorganisms (log 10 CFU/g), N max is the maximum number of microorganisms (log 10 CFU/g) and µ max is the maximum specific growth rate (h −1 ). (2):

The adjustment function of A(t) is expressed by Equation
where λ is the lag phase duration (h), and v is the rate of increase of a limiting substrate, assumed to be equal to µ max . The Baranyi model was employed to describe microbiological growth data, and growth parameters were estimated with the NonLinearModel.fit command in the Matlab 8.3.0.532 (R2014a) software (MathWorks Inc., Natick, MA, USA). Determination of starting values in nonlinear regression procedure is a critical step to predict the accurate parameters. Starting values for the parameters N 0 and N max were selected as the minimum and maximum values of experimental data at each temperature, respectively. The starting value of µ max was chosen as the highest value among the first derivatives of the natural logarithm of bacterial count with respect to time. Parameter λ was obtained from the raw data plot as the time point, when exponential growth started. After several successive iterations in the nonlinear procedure, the starting values were converged to optimum values of the parameters.

Survival Kinetic Model
Survival kinetic data of S. Reading on the fresh iceberg lettuce stored at 4 and 8 • C were modelled using Weibull model as described by Mafart et al. [26]: where t is the time (h), N 0 is the concentration of microbial population (log 10 CFU/g) at time t, delta is the scale parameter and p is shape parameter.
Similar to the Baranyi model, the parameters of the Weibull model were calculated with the NonLinearModel.fit command in the Matlab 8.3.0.532 (R2014a) software (MathWorks Inc., Natick, MA, USA).

Goodness of Fit of the Growth and Survival Models
The goodness of fit of the growth and survival models was evaluated by considering the mean squared error (MSE) and adjusted coefficient of determination (adjusted-R 2 ) values using Equations (4) and (5), respectively: where observed is the value obtained in experiments, fitted is the value estimated by the model, n is the number of observations, s is the number of parameters of the model, SSE is the sum of squared errors and SST is the total sum of squares.

Results and Discussion
The obtained growth data of LAB on the fresh iceberg lettuce for the temperatures of 4, 8, and 15 • C are presented in Table 1. The initial concentration of LAB on fresh iceberg lettuce at 4, 8 and 15 • C, were 3.10 ± 0.06, 3.51 ± 0.07 and 3.75 ± 0.35 log 10 CFU/g, respectively. These results are in agreement with the values reported in other studies for the initial contamination of LAB in lettuce [19,[27][28][29]. After 10 days of storage, the LAB concentrations on the fresh iceberg lettuce were 5.62 ± 0.17, 7.16 ± 0.34 and 8.36 ± 0.13 log 10 CFU/g for the temperatures of 4, 8 and 15 • C, respectively (Table 1). In addition, control samples were analysed at day 0 and day 10 of storage for enumeration of AVC and for presence/absence of Salmonella spp. The initial concentration of AVC on iceberg lettuce was 4.48 ± 0.58, 4.45 ± 0.57 and 5.35 ± 0.17 log 10 CFU/g, for the storage temperatures of 4, 8 and 15 • C, respectively. At 10 storage days, the AVC concentration was found to reach 7.57 ± 0.25, 8.85 ± 0.25 and 8.35 ± 0.24 log 10 CFU/g for temperatures of 4, 8 and 15 • C, respectively. Salmonella spp. were not detected in any of the control samples of iceberg lettuce throughout the storage period. LAB presented growth at all tested temperatures. The µ max and λ values of LAB on fresh iceberg lettuce calculated by the Baranyi model for each storage temperature are given in Table 2  The initial S. Reading concentration on the inoculated lettuce bags was around 5 log 10 CFU/g for all storage temperatures (Table 1). At the end of the storage period (10 days), S. Reading concentration was 2.06 ± 0.08 and 2.93 ± 0.25 log 10 CFU/g at 4 and 8 • C, respectively, while at 15 • C the pathogen reached 5.68 ± 0.29 log 10 CFU/g. These results demonstrated that S. Reading exhibited growth on the tested product only at 15 • C. The fitting of the Baranyi model to these data was acceptable as evidenced by the graphical representation in Figure 1   The growth kinetics for different Salmonella serovars in lettuce reported by other studies were in line with the results for S. Reading at 15 °C. However, in several cases, studies demonstrated growth at temperatures of 8 °C or even slightly lower, which was not observed for S. Reading in this study. This is probably due to serovar and strain variations with respect to growth temperatures. For instance, Sant'Ana et al. [19] studied the growth of a cocktail of Salmonella enterica composed of the serovars Typhimurium and Enteritidis in lettuce packaged under modified atmosphere conditions and incubated between 7 and 30 °C. According to experimental data, all Samonella strains were able to grow at 7 °C. Other studies have also shown growth at 5 °C for several Salmonella serovars such as Enteritidis, Typhimurium, Anatum, Newport and Saint Paul [20,30]. In turn, Koseki and Isobe [16] did not find growth at 5 °C for the serovars Typhimurium and Enteritidis. The experiments of the work by Koseki and Isobe [16] were also carried out under the same air conditions as those used in the study by Oliveira et al. [30].
Concerning the prediction capacity of existing microbiology models to estimate S. Reading kinetics, models predicted growth at lower temperatures than 15 °C, which was the only temperature showing growth in our study. Basically, models reflected what was found in the experimental data (i.e., model domain), thus the model from Sant'Ana et al. [19] and Koseki and Isobe [16] predicted growth at 7 and 10 °C, respectively, while the models obtained by Oliveira et al. [30] and Veys et al. [20] predicted growth at a temperature as low as 5 °C. At 15 °C, model results were in agreement on a practical absence of lag time, while μmax ranged from 0.05 and 0.08 h −1 . The best match with the growth rate observed in our work (μmax = 0.056 h −1 ) was found for the models from Sant'Ana et al. [19] and Koseki and Isobe [16], with values of 0.053 and 0.057 h −1 , respectively. It is noteworthy that these models were the ones with a more similar minimum growth temperature to our study. For other The growth kinetics for different Salmonella serovars in lettuce reported by other studies were in line with the results for S. Reading at 15 • C. However, in several cases, studies demonstrated growth at temperatures of 8 • C or even slightly lower, which was not observed for S. Reading in this study. This is probably due to serovar and strain variations with respect to growth temperatures. For instance, Sant'Ana et al. [19] studied the growth of a cocktail of Salmonella enterica composed of the serovars Typhimurium and Enteritidis in lettuce packaged under modified atmosphere conditions and incubated between 7 and 30 • C. According to experimental data, all Samonella strains were able to grow at 7 • C. Other studies have also shown growth at 5 • C for several Salmonella serovars such as Enteritidis, Typhimurium, Anatum, Newport and Saint Paul [20,30]. In turn, Koseki and Isobe [16] did not find growth at 5 • C for the serovars Typhimurium and Enteritidis. The experiments of the work by Koseki and Isobe [16] were also carried out under the same air conditions as those used in the study by Oliveira et al. [30].
Concerning the prediction capacity of existing microbiology models to estimate S. Reading kinetics, models predicted growth at lower temperatures than 15 • C, which was the only temperature showing growth in our study. Basically, models reflected what was found in the experimental data (i.e., model domain), thus the model from Sant'Ana et al. [19] and Koseki and Isobe [16] predicted growth at 7 and 10 • C, respectively, while the models obtained by Oliveira et al. [30] and Veys et al. [20] predicted growth at a temperature as low as 5 • C. At 15 • C, model results were in agreement on a practical absence of lag time, while µ max ranged from 0.05 and 0.08 h −1 . The best match with the growth rate observed in our work (µ max = 0.056 h −1 ) was found for the models from Sant'Ana et al. [19] and Koseki and Isobe [16], with values of 0.053 and 0.057 h −1 , respectively. It is noteworthy that these models were the ones with a more similar minimum growth temperature to our study. For other studies, in which other leafy greens were used, results were much more different. For example, the secondary model developed by Mishra et al. [17] based on data collected from literature for leafy greens including iceberg lettuce predicted a µ max value at 15 • C of 0.096 h −1 , while Yoon et al. [21] studying S. Typhimurium on fresh-cut cabbage reported µ max and λ values for 15 • C of 0.034 h −1 and 10.2 h, respectively. In summing up, the microbiology models developed exclusively with the serovars Enteritidis and Typhimurium in a similar matrix would be the only ones suitable for predicting S. Reading growth at 15 • C, and probably at temperatures around this value but not at temperatures near to the minimum growth temperature given the discrepancies obtained for lower temperatures.
With regard to the possible explanation for these similarities between studies and models, it is likely that the relatively high temperature was a larger determinant than other factors (packaging atmosphere, inoculum method, type of lettuce, etc.). From this, it is apparent that the serovar would not be especially relevant under more optimum growth conditions. Nevertheless, it is remarkable that, in our study, S. Reading was not able to grow at 8 • C, indicating that this strain could have a reduced growth potential under the conditions studied in comparison to other studies. In this sense, we hypothesize that the endogenous microbiota of lettuce could also play an important role as an inhibitory factor on Salmonella, as in our study, LAB was able to grow at all temperatures with a higher growth rate. Nevertheless, more specific research is needed to determine this potential effect from native LAB on Salmonella.
The survival data of S. Reading on the fresh iceberg lettuce were satisfactorily described using the Weibull model as can be observed in Figure 1, with all experimental data within 95% confidence bands, which was also confirmed by the goodness of fit indexes, with MSE < 0.041 and adjusted-R 2 > 0.944 (Table 3). The parameter delta increased from 16.03 to 18.81 at 4 and 8 • C, respectively, while the parameter p slightly decreased from 0.348 to 0.250. Mathematically, the delta value can be interpreted as the time needed to reduce the first decimal reduction of the microbial population. Therefore, its increase observed between 4 and 8 reflects a reduction in the die-off rate of the Salmonella population. Besides, p values were < 1, confirming that the microorganism followed a concave upward curve profile as observed in survival curves (Figure 1). This survival profile is characterized by an asymptotic region, which could involve a resistant and residual population that could remain viable for a longer time despite of the initial die-off rate. To the best of our knowledge, there are no studies reporting the survival kinetic parameters for Salmonella in lettuce during storage, even though some works provided estimates of the logarithmic reductions that could be compared to the ones obtained herein. For instance, the study by Chang and Fang [31], where shredded iceberg lettuce inoculated with S. Typhimurium was incubated under aerobic conditions at 4 • C, obtained a reduction of 0.7 ± 0.4 and 0.8 ± 0.3 log 10 CFU/g for 3 and 7 days, respectively. Whereas, in our study, reductions were higher, with values of 1.7 and 2.3 log 10 CFU/g for 3 and 7 days, respectively. Besides that, Delbeke et al. [32] observed, after 7 days, reductions of around 1 log 10 CFU/g of S. Thompson in lettuce packaged in passive atmosphere and incubated at 7 • C. In this case, values in our work were also higher, with a reduction of 1.8 log 10 CFU /g for the temperature of 8 • C. In the light of these data, apart from the possible differences between serovars, it can be suggested that reduced oxygen content (2-4% O 2 ) in the packaging atmosphere could enhance the inactivation of Salmonella when stored at refrigeration conditions, with log 10 -decreases of 2.8 and 2.0 for 4 and 8 • C, respectively for 10 days of storage. These findings demonstrated that for S. Reading, under the typical commercial conditions for cut lettuce (packaged in MAP), an adequate refrigeration temperature may not only inhibit its growth, but reduce drastically the pathogen contamination in the commercialized product.

Conclusions
The Baranyi model described accurately the growth behaviour of LAB on the fresh iceberg lettuce at different storage temperatures (4, 8 and 15 • C). S. Reading could not grow on the fresh iceberg lettuce stored at 4 and 8 • C. The comparison with other studies demonstrated that the microbiology models for lettuce developed in other Salmonella serovars could be used for predicting S. Reading growth at 15 • C, but not at low temperatures, given the discrepancy in the minimum growth temperatures. Besides that, the survival data of S. Reading suggested a low tolerance of this Salmonella serovar to the storage temperatures of 4 and 8 • C, with higher reduction levels than those reported in literature. The survival kinetics could be properly reflected by using the Weibull model. Results and predictive microbiology models obtained in this study provide a valuable source of information to consider, in quantitative risk assessment and in predictive analysis, the kinetics of a specific Salmonella serovar, S. Reading, on commercial fresh iceberg lettuce.