Ribosomal mutations enable a switch between high fitness and high stress resistance in Listeria monocytogenes.

Multiple stress resistant variants of Listeria monocytogenes with mutations in rpsU encoding ribosomal protein RpsU have previously been isolated after a single exposure to acid stress. These variants, including L. monocytogenes LO28 variant V14 with a complete deletion of the rpsU gene, showed upregulation of the general stress sigma factor Sigma B-mediated stress resistance genes and had a lower maximum specific growth rate than the LO28 WT, signifying a trade-off between stress resistance and fitness. In the current work V14 has been subjected to an experimental evolution regime, selecting for higher fitness in two parallel evolving cultures. This resulted in two evolved variants with WT-like fitness: 14EV1 and 14EV2. Comparative analysis of growth performance, acid and heat stress resistance, in combination with proteomics and RNA-sequencing, indicated that in both lines reversion to WT-like fitness also resulted in WT-like stress sensitivity, due to lack of Sigma B-activated stress defense. Notably, genotyping of 14EV1 and 14EV2 provided evidence for unique point-mutations in the ribosomal rpsB gene causing amino acid substitutions at the same position in RpsB, resulting in RpsB22Arg-His and RpsB22Arg-Ser, respectively. Combined with data obtained with constructed RpsB22Arg-His and RpsB22Arg-Ser mutants in the V14 background, we provide evidence that loss of function of RpsU resulting in the multiple stress resistant and reduced fitness phenotype, can be reversed by single point mutations in rpsB leading to arginine substitutions in RpsB at position 22 into histidine or serine, resulting in a WT-like high fitness and low stress resistance phenotype. This demonstrates the impact of genetic changes in L. monocytogenes' ribosomes on fitness and stress resistance.

A single point mutation in the Listeria monocytogenes ribosomal gene rpsU enables SigB activation independently of the stressosome and the anti-sigma factor antagonist RsbV.

Microbial population heterogeneity leads to different stress responses and growth behavior of individual cells in a population. Previously, a point mutation in the rpsU gene (rpsUG50C) encoding ribosomal protein S21 was identified in a Listeria monocytogenes LO28 variant, which leads to increased multi-stress resistance and a reduced maximum specific growth rate. However, the underlying mechanisms of these phenotypic changes remain unknown. In L. monocytogenes, the alternative sigma factor SigB regulates the general stress response, with its activation controlled by a series of Rsb proteins, including RsbR1 and anti-sigma factor RsbW and its antagonist RsbV. We combined a phenotype and proteomics approach to investigate the acid and heat stress resistance, growth rate, and SigB activation of L. monocytogenes EGDe wild type and the ΔsigB, ΔrsbV, and ΔrsbR1 mutant strains. While the introduction of rpsUG50C in the ΔsigB mutant did not induce a SigB-mediated increase in robustness, the presence of rpsUG50C in the ΔrsbV and the ΔrsbR1 mutants led to SigB activation and concomitant increased robustness, indicating an alternative signaling pathway for the SigB activation in rpsUG50C mutants. Interestingly, all these rpsUG50C mutants exhibited reduced maximum specific growth rates, independent of SigB activation, possibly attributed to compromised ribosomal functioning. In summary, the increased stress resistance in the L. monocytogenes EGDe rpsUG50C mutant results from SigB activation through an unknown mechanism distinct from the classical stressosome and RsbV/RsbW partner switching model. Moreover, the reduced maximum specific growth rate of the EGDe rpsUG50C mutant is likely unrelated to SigB activation and potentially linked to impaired ribosomal function.

Pulsed electric field treatment for preservation of Chlorella suspensions and retention of gelling capacity.

Pulsed electric field (PEF) processing has emerged as an alternative to thermal pasteurization for the shelf-life extension of heat-sensitive liquids at industrial scale. It offers the advantage of minimal alteration in physicochemical characteristics and functional properties. In this study, a pilot-scale continuous PEF processing (Toutlet < 55 °C) was applied to microalgae Chlorella vulgaris (Cv) suspensions (pH = 6.5), which was proposed as a functional ingredient for plant-based foods. Cv suspensions were inoculated with three distinct food spoilage microorganisms (Pseudomonas guariconensis, Enterobacter soli and Lactococcus lactis), isolated from the Cv biomass. PEF treatments were applied with varying electric field strength Eel of 16 to 28 kV/cm, pulse repetition rate f of 100 to 140 Hz, with a pulse width τ of 20 µs and an inlet product temperature Tin of 30 °C. The aim was to evaluate the PEF-induced microbial reduction and monitor the microbial outgrowth during a 10-day cold storage period (10 °C). Maximum inactivation of 4.1, 3.7 and 3.6 logs was achieved (28 kV/cm and 120 Hz) for the investigated isolates, respectively. Under these conditions, the critical electric field strengths Ecrit, above which inactivation was observed, ranged from 22.6 to 24.6 kV/cm. Moreover, repeated PEF treatment resulted in similar inactivation efficiency, indicating its potential to enhance shelf-life further.

Stress resistant rpsU variants of Listeria monocytogenes can become underrepresented due to enrichment bias.

Population heterogeneity is an important component of the survival mechanism of Listeria monocytogenes, leading to cells in a population with diverse stress resistance levels. We previously demonstrated that several ribosomal gene rpsU mutations enhanced the stress resistance of L. monocytogenes and lowered the growth rate at 30 °C and lower temperatures. This study investigated whether these switches in phenotypes could result in a bias in strain detection when standard enrichment-based procedures are applied to a variety of strains. Detailed growth kinetics analysis of L. monocytogenes strains were performed, including the LO28 wild type (WT) and rpsU variants V14 and V15, during two commonly used enrichment-based procedures described in the ISO 11290-1:2017 and the U.S. Food and Drug Administration Bacteriological Analytical Manual (BAM). WT had a higher growth rate than the variants during the enrichment processes. Co-culture growth kinetics predictions for WT and rpsU variants showed that the detection chances of the rpsU mutants were reduced from ∼52 % to less than ∼13 % and âˆ¼ 3 % during ISO and BAM enrichment, respectively, which were further validated through subsequent qPCR experiments. Higher heat stress resistance of rpsU variants did not lead to faster recovery during enrichment after heat treatment, and different pre-culturing temperatures before heat treatment did not significantly affect the growth kinetics of the WT and rpsU variants. Additionally, post-enrichment isolation procedures involving streaking on selective agar plates did not show preferences for isolating WT or rpsU variants nor affect the detection chance of rpsU variants. The difference in detection chance suggests that the selective enrichment procedures inadequately represent the genotypic diversity present in a sample. Hence, the enrichment bias during the L. monocytogenes isolation procedure may contribute to the observed underrepresentation of the rpsU mutation among L. monocytogenes isolates deposited in publicly available genome databases. The underrepresentation of rpsU mutants in our findings suggests that biases introduced by standard isolation and enrichment procedures could inadvertently skew our understanding of genetic diversity when relying on public databases.

Quantitative assessment of food safety interventions for Campylobacter spp. and Salmonella spp. along the chicken meat supply chain in Burkina Faso and Ethiopia.

Rural and small-scale chicken farming is a major source of income in most African countries, and chicken meat is an important source of nutrients. However, chicken meat can be contaminated with Campylobacter spp. and Salmonella spp., pathogens with a high reported burden of foodborne illnesses. Therefore, it is essential to control these pathogens in chicken meat. Quantitative microbial risk assessments (QMRA) can aid the development of effective food safety control measures and are currently lacking in chicken meat supply chains in the African context. In this study, we developed stochastic QMRA models for Salmonella spp. and Campylobacter spp. in the chicken meat supply chain in Burkina Faso and Ethiopia employing the modular process risk model in @Risk software. The study scope covered chicken farming, transport, slaughtering, consumer handling, and consumption. Effectiveness of candidate interventions was assessed against baseline models' outputs, which showed that the mean annual Campylobacter spp. risk estimates were 6482 cases of illness per 100,000 persons and 164 disability adjusted life years (DALYs) per 100,000 persons in Burkina Faso, and 12,145 cases and 272 DALYs per 100,000 persons in Ethiopia. For Salmonella spp., mean annual estimates were 2713 cases and 1212 DALYs per 100,000 persons in Burkina Faso, and 4745 cases and 432 DALYs per 100,000 persons in Ethiopia. Combining interventions (improved hand washing plus designated kitchen utensils plus improved cooking) resulted in 75 % risk reduction in Burkina Faso at restaurants and 93 to 94 % in Ethiopia at homes for both Salmonella spp. and Campylobacter spp. For Burkina Faso, adding good hygienic slaughter practices at the market to these combined interventions led to over 91 % microbial risk reduction. Interventions that involved multiple food safety actions in a particular step of the supply chain or combining different interventions from different steps of the supply chain resulted in more risk reduction than individual action interventions. Overall, this study demonstrates how diverse and scanty food supply chain information can be applied in QMRA to provide estimates that can be used to stimulate risk-based food safety action in African countries.

A web-based microbiological hazard identification tool for infant foods.

An integrated approach to identify and assess Microbiological Hazards (MHs) and mitigate risks in infant food chains is crucial to ensure safe foods for infants and young children. A systematic procedure was developed to identify MHs in specific infant foods. This includes five major steps: 1) relevant hazard-food pairing, 2) process inactivation efficiency, 3) recontamination possibility after processing, 4) MHs growth opportunity, and 5) MHs-food association level. These steps were integrated into an online tool called the Microbiological Hazards IDentification (MiID) decision support system (DSS), targeting food companies, governmental agencies and academia users, and is accessible at https://foodmicrobiologywur.shinyapps.io/Microbial_hazards_ID/. The MiID DSS was validated in four case studies, focussing on infant formula, fruit puree, cereal-based meals, and fresh fruits, each representing distinct products and processing characteristics. The results obtained through the application of the MiID DSS, compared with identification by food safety experts, consistently identified the top MHs in these food products. This process affirms its effectiveness in systematic hazard identification. The introduction of the MiID DSS helps to structure the first steps in HACCP (hazard analysis) and in risk assessment (hazard identification) to follow a structured and well-documented procedure, balancing the risk of overlooking relevant MHs or including too many irrelevant MHs. It is a valuable addition to risk analysis/assessment in infant food chains and has the potential for future extension. This includes the incorporation of newly acquired data related to infant foods via a semi-publicly hosted platform, or it can be adapted for hazard identification in general food products using a similar framework.

Temperature status of domestic refrigerators and its effect on the risk of listeriosis from ready-to-eat (RTE) cooked meat products.

Inadequate domestic refrigeration is frequently cited as a factor that contributes to foodborne poisoning and infection, and consumer behaviour in this regard can vary largely. This study provides insight into the temperature profiles of domestic refrigerators in the Netherlands and the impact on the number of listeriosis cases related to ready-to-eat (RTE) cooked meat products. A survey was conducted among Dutch consumers (n = 1020) to assess their knowledge and behaviour related to refrigerators. Out of these participants, 534 measured their refrigerator's temperature, revealing an average temperature of 5.7 °C (standard deviation (SD) of 2.2 °C) with a maximum of 17 °C. Elderly people (65 years and older) had refrigerators with temperatures that were on average 0.6 °C higher than those of younger people (35 years or younger). The 24-hour temperature profiles of an additional set of actively surveyed refrigerators (n = 50) showed that the temperature measured on the upper shelf was significantly higher (mean 7.7 °C, SD 2.7 °C) than the temperature measured on the bottom shelf (5.7 °C, SD 2.1 °C). Quantitative Microbiological Risk Assessment (QMRA) predicted that the primary factors contributing to the risk of listeriosis were the initial concentration and the time and temperature during household storage. Scenario analysis revealed that storing opened RTE cooked meat products at home for either <7 days or at temperatures <7 °C resulted in a significant reduction of over 80 % in predicted illness cases. Among all illness cases, the elderly represented nearly 90 %. When assessing the impact of the disease in terms of Years of Life Lost (YLL), the contribution of the elderly was 59 %. Targeted communication, particularly directed towards the elderly, on the importance of storing RTE cooked meat products at the recommended temperature on the bottom or middle shelf as well as consuming within two to three days after opening, holds the potential to significantly reduce the number of cases.

Insight in lag phase of Listeria monocytogenes during enrichment through proteomic and transcriptomic responses.

The dynamics of the enrichment-based detection procedure of the foodborne pathogen Listeria monocytogenes from food still remains poorly understood. This enrichment is crucial in the reliable detection of this pathogen and more insight into the recovery mechanism during this step is important to advance our understanding of lag phase behaviour during enrichment. In this study we combined transcriptomic and proteomic analyses to better understand the physiological processes within the lag phase of L. monocytogenes during enrichment. Upon transfer of BHI-cultured stationary phase L. monocytogenes cells to half-Fraser enrichment broth (HFB), motility-associated genes and proteins were downregulated, while expression of metal uptake transporters, resuscitation-promoting factors that stimulate growth from dormancy, antibiotic efflux pumps and oxidative stress proteins were upregulated. Next to this, when cells with a heat stress history were cultured in enrichment broth, proteins necessary for recovery were upregulated with functions in DNA-damage repair, protein refolding, cell-wall repair, and zinc transport. Proteomic results pointed to possible factors that support shortening the lag duration, including the addition of 10 µM zinc and the addition of spent HFB containing presumed concentrations of resuscitation-promoting factors. However, these interventions did not lead to biologically relevant reduction of lag phase. Also, when cells were enriched in spent HFB, final cell concentrations were similar to enrichments in fresh HFB, indicating that the enrichment broth seems not to lack critical substrates. Concludingly, this study gives insight into the proteomic changes in the lag phase during enrichment and shows that supplementation of HFB is not the best strategy to optimize the current enrichment method.

Shelf life estimation of refrigerated vacuum packed beef accounting for uncertainty.

This study estimates the shelf life of vacuum packed beef meat (three muscles: striploin (longissimus thoracis et lumborum, LTL), tenderloin (psoas major, PM) and outside chuck (trapezius thoracis, TT)) at refrigeration temperatures (0 °C-10 °C) based on modelling the growth of two relevant groups of spoilage microorganisms: lactic acid bacteria (LAB) and Enterobacteriaceae. The growth models were developed combining a two-step and a one-step approach. The primary modelling was used to identify the parameters affecting the growth kinetics, guiding the definition of secondary growth models. For LAB, the secondary model included the effect of temperature and initial pH on the specific growth rate. On the other hand, the model for Enterobacteriaceae incorporated the effect of temperature on the specific growth rate and the lag phase; as well as the effect of the initial pH on the specific growth rate, the lag phase and the initial microbial count. We did not observe any significant effect of the type of muscle on the growth kinetics. Once the equations were defined, the models were fitted to the complete dataset using a one-step approach. Model validation was carried out by cross-validation, mitigating the impact of an arbitrary division between training and validation sets. The models were used to estimate the shelf life of the product, based on the maximum admissible microbial concentration (7 log CFU/g for LAB, 5 log CFU/g for Enterobacteriaceae). Although LAB was the dominant microbiota, in several cases, both LAB and Enterobacteriaceae reached the critical concentration practically at the same time. Furthermore, in some scenarios, the end of shelf life would be determined by Enterobacteriaceae, pointing at the potential importance of non-dominant microorganisms for product spoilage. These results can aid in the implementation of effective control measures in the meat processing industry.

The importance of what we cannot observe: Experimental limitations as a source of bias for meta-regression models in predictive microbiology.

Meta-regression models have gained in popularity during the last years as a way to create more generic models for Microbial Risk Assessments that also include variability. However, as with most meta-analyses and empirical models, systematic biases in the data can result in inaccurate models. In this article, we define experimental bias as a type of selection bias due to the practical limitations of microbial inactivation experiments. Conditions with extremely high D-values (i.e. slow inactivation) need very long experimental runs to cause significant reductions. On the other hand, when the D-value is extremely low, not enough data points can be gathered before the microbial population is below the detection limit. Consequently, experimental designs favour conditions within a practical experimental range, introducing a selection bias in the D-values. We demonstrate the impact of experimental bias in meta-regression models using numerical simulations. Models fitted to data with experimental bias overestimated the z-value and underestimated variability. We propose a rapid heuristic method to identify experimental bias in datasets, and we propose truncated regression to mitigate its impact in meta-regression models. Both methods were validated using simulated data. Thereafter the procedures were tested by building a meta-regression model for actual data for the inactivation of Bacillus cereus spores. We concluded that the dataset included experimental bias, and that it would cause an overestimation of the microbial resistance at high temperatures (>120 °C) for classical meta-regression models. This effect was mitigated when the model was built using truncated regression. In conclusion, we demonstrate that experimental bias could potentially result in inaccurate models for predictive microbiology. Therefore, checking for experimental bias should be a routine step in meta-regression modelling, and be included in guidelines on data analysis for meta-regression.

Critical comparison of statistical methods for quantifying variability and uncertainty of microbial responses from experimental data.

Variability and uncertainty are important factors for quantitative microbiological risk assessment (QMRA). In this context, variability refers to inherent sources of variation, whereas uncertainty refers to imprecise knowledge or lack of it. In this work we compare three statistical methods to estimate variability in the kinetic parameters of microbial populations: mixed-effect models, multilevel Bayesian models, and a simplified algebraic method previously suggested. We use two case studies that analyse the influence of three levels of variability: (1) between-strain variability (different strains of the same species), (2) within-strain variability (biologically independent reproductions of the same strain) and, at the most nested level, (3) experimental variability (species independent technical lab variability resulting in uncertainty about the population characteristic of interest) on the growth and inactivation of Listeria monocytogenes. We demonstrate that the algebraic method, although relatively easy to use, overestimates the contribution of between-strain and within-strain variability due to the propagation of experimental variability in the nested experimental design. The magnitude of the bias is proportional to the variance of the lower levels and inversely proportional to the number of repetitions. This bias was very relevant in the case study related to growth, whereas for the case study on inactivation the resulting insights in variability were practically independent of the method used. The mixed-effects model and the multilevel Bayesian models calculate unbiased estimates for all levels of variability in all the cases tested. Consequently, we recommend using the algebraic method for initial screenings due to its simplicity. However, to obtain parameter estimates for QMRA, the more complex methods should generally be used to obtain unbiased estimates.

The Most Probable Curve method - A robust approach to estimate kinetic models from low plate count data resulting in reduced uncertainty.

A novel method is proposed for fitting microbial inactivation models to data on liquid media: the Most Probable Curve (MPC) method. It is a multilevel model that makes a separation between the "true" microbial concentration according to the model, the "actual" concentration in the media considering chance, and the actual counts on the plate. It is based on the assumptions that stress resistance is homogeneous within a microbial population, and that there is no aggregation of microbial cells. Under these assumptions, the number of colonies in/on a plate follows a Poisson distribution with expected value depending on the proposed kinetic model, the number of dilutions and the plated volume. The novel method is compared against (non)linear regression based on a normal likelihood distribution (traditional method), Poisson regression and gamma-Poisson regression using data on the inactivation of Listeria monocytogenes. The conclusion is that the traditional method has limitations when the data includes plates with low (or zero) cell counts, which can be mitigated using more complex (discrete) likelihoods. However, Poisson regression uses an unrealistic likelihood function, making it unsuitable for survivor curves with several log-reductions. Gamma-Poisson regression uses a more realistic likelihood function, even though it is based mostly on empirical hypotheses. We conclude that the MPC method can be used reliably, especially when the data includes plates with low or zero counts. Furthermore, it generates a more realistic description of uncertainty, integrating the contribution of the plating error and reducing the uncertainty of the primary model parameters. Consequently, although it increases modelling complexity, the MPC method can be of great interest in predictive microbiology, especially in studies focused on variability analysis.

Intraspecific variability in heat resistance of fungal conidia.

Microbial species are inherently variable, which is reflected in intraspecies genotypic and phenotypic differences. Strain-to-strain variation gives rise to variability in stress resistance and plays a crucial role in food safety and food quality. Here, strain variability in heat resistance of asexual spores (conidia) of the fungal species Aspergillus niger, Penicillium roqueforti and Paecilomyces variotii was quantified and compared to bacterial variability found in the literature. After heat treatment, a 5.4- to 8.6-fold difference in inactivation rate was found between individual strains within each species, while the strain variability of the three fungal species was not statistically different. We evaluated whether the degree of intraspecies variability is uniform, not only within the fungal kingdom, but also amongst different bacterial species. Comparison with three spore-forming bacteria and two non-spore-forming bacteria revealed that the variability of the different species was indeed in the same order of magnitude, which hints to a microbial signature of variation that exceeds kingdom boundaries.

Predicting B. cereus growth and cereulide production in dairy mix.

This study aims to quantify growth and cereulide production by Bacillus cereus and their potential correlation in an intermediate dairy wet-mix. Systematic experiments were carried out using the emetic reference strain F4810/72 in the suboptimal range of temperature of 12 °C to 20 °C. Growth and cereulide kinetic parameters were estimated and the three parameters (i) time to first cereulide quantification (tcer), (ii) maximum specific growth rates (µmax) and (iii) cereulide production rates (k) were modelled as a function of temperature. As temperature increased, growth lag time and tcer were shorter while microbial increase and cereulide production happened earlier, and at higher rates. Maximum concentration of cells and maximum cereulide concentration proved to be temperature-independent, reaching the average values of 7.9 ± 0.3 log10(CFU/mL) and 2.6 ± 0.2 log10(ng.g-1) respectively. Moreover, the time to reach the widely used threshold of 5 log10CFU/mL (t5log) was tested against tcer, and this suggested that this threshold can be used with increased confidence at lower temperatures to assure toxin is not quantified in this matrix. The average tcer were equal to 314 h, 118 h, 73 h and 45 h for 12 °C, 15 °C, 18 °C and 20 °C respectively. A validation study was performed using independent data sets obtained with the same strain in other dairy matrices. The microbial growth models presented good predictive power even when extrapolated beyond the temperature range of construction. Nevertheless, the models proposed for prediction of toxin production over time presented limitations, especially for food matrices that deviate significantly from the original matrix for which the model was developed, making cereulide predictions less accurate. Our findings suggest that similar modelling approaches can be used to predict growth, time to first cereulide quantification as well as cereulide formation over time for a specific matrix, but that matrix-extrapolations are more suitable for growth than for cereulide.

Heterogeneity in single-cell outgrowth of Listeria monocytogenes in half Fraser enrichment broth is affected by strain variability and physiological state.

The behaviour of pathogens at the single-cell level can be highly variable and can thus affect the detection efficacy of enrichment-based detection methods. The outgrowth of single cells of three Listeria monocytogenes strains was monitored after fluorescence-activated single-cell sorting in non-selective brain heart infusion (BHI) broth and selective half Fraser enrichment broth (HFB) to quantify outgrowth heterogeneity and its effect on the detection probability. Single-cell heterogeneity was higher in HFB compared to non-selective BHI and heterogeneity increased further when cells were heat-stressed. The increase in heterogeneity was also strain-dependent because the fast-recovering strain Scott A showed less outgrowth heterogeneity than the slower-recovering strains EGDe and H7962. Modelling of the outgrowth kinetics during the primary enrichment demonstrated that starting at low cell concentrations could fail detection of L. monocytogenes at least partly due to cell heterogeneity. This highlights that it is important to take single-cell heterogeneity into account when optimizing enrichment formulations and procedures when L. monocytogenes contamination levels are low.

Processing environment monitoring in low moisture food production facilities: Are we looking for the right microorganisms?

Processing environment monitoring is gaining increasing importance in the context of food safety management plans/HACCP programs, since past outbreaks have shown the relevance of the environment as contamination pathway, therefore requiring to ensure the safety of products. However, there are still many open questions and a lack of clarity on how to set up a meaningful program, which would provide early warnings of potential product contamination. Therefore, the current paper aims to summarize and evaluate existing scientific information on outbreaks, relevant pathogens in low moisture foods, and knowledge on indicators, including their contribution to a "clean" environment capable of limiting the spread of pathogens in dry production environments. This paper also outlines the essential elements of a processing environment monitoring program thereby supporting the design and implementation of better programs focusing on the relevant microorganisms. This guidance document is intended to help industry and regulators focus and set up targeted processing environment monitoring programs depending on their purpose, and therefore provide the essential elements needed to improve food safety.

Amino acid substitutions in ribosomal protein RpsU enable switching between high fitness and multiple-stress resistance in Listeria monocytogenes.

Microbial population heterogeneity contributes to differences in stress response between individual cells in a population, and can lead to the selection of genetically stable variants with increased stress resistance. We previously provided evidence that the multiple-stress resistant Listeria monocytogenes LO28 variant 15, carries a point mutation in the rpsU gene, resulting in an arginine-proline substitution in ribosomal protein RpsU (RpsU17Arg-Pro). Here, we investigated the trade-off between general stress sigma factor SigB-mediated stress resistance and fitness in variant 15 using experimental evolution. By selecting for higher fitness in two parallel evolving cultures, we identified two evolved variants: 15EV1 and 15EV2. Whole genome sequencing and SNP analysis showed that both parallel lines mutated in the same codon in rpsU as the original mutation resulting in RpsU17Pro-His (15EV1) and RpsU17Pro-Thr (15EV2). Using a combined phenotyping and proteomics approach, we assessed the resistance of the evolved variants to both heat and acid stress, and found that in both lines reversion to WT-like fitness also resulted in WT-like stress sensitivity. Proteome analysis of L. monocytogenes LO28 WT, variant 15, 15EV1, and 15EV2 revealed high level expression of SigB regulon members only in variant 15, whereas protein profiles of both evolved variants were highly similar to that of the LO28 WT. Experiments with constructed RpsU17Arg-Pro mutants in L. monocytogenes LO28 and EGDe, and RpsU17Arg-His and RpsU17Arg-Thr in LO28, confirmed that single amino acid substitutions in RpsU enable switching between multiple-stress resistant and high fitness states in L. monocytogenes.

Incorporating strain variability in the design of heat treatments: A stochastic approach and a kinetic approach.

For the design of thermal processes, the decimal reduction times (D-values) of target organisms can be used. However, many factors influence the D-value, like inherent organism's characteristics (strain variability), the effect of the history of the cells, as well as product factors and process factors. Strain variability is a very large contributor to the overall variation of the D-value. Hence, the overall reduction of microbial contaminants by a heat treatment is a combination of the occurrence of a strain with a certain heat resistance and its reduction given the prevailing conditions. This reduction can be determined using two approaches: a kinetic analysis based on integral equations or a stochastic approach based on Monte Carlo analysis. In this article, these two approaches are compared using as case studies the inactivation of two microorganisms: Listeria monocytogenes in a pasteurization process and the sporeformer Geobacillus stearothermophilus in a UHT process. Both approaches resulted in similar conclusions, highlighting that the strains with the highest heat resistance are determinant for the overall inactivation, even if the probability of cells having such extreme heat resistance is very low.

Variability in lag duration of Listeria monocytogenes strains in half Fraser enrichment broth after stress affects the detection efficacy using the ISO 11290-1 method.

A collection of 23 Listeria monocytogenes strains of clinical and food origin was tested for their ability to recover and grow out in half Fraser enrichment broth following the ISO 11290-1:2017 protocol. Recovery of sub-lethally heat-injured cells in half Fraser broth was compared to reference cells with no stress pre-treatment. The enrichments were followed over time by plate counts and the growth parameters were estimated with the 3-phase model which described the data best. The reference cells without stress pre-treatment showed a short lag duration, which ranged from 1.4 to 2.7 h. However, significant variation in the ability to recover after 60 °C heat stress was observed among the tested strains and resulted in a lag duration from 4.7 to 15.8 h. A subset of strains was also exposed to low-temperature acid stress, and the lag duration showed to be also stress dependent. Scenario analyses and Monte Carlo simulations were carried out using the growth parameters obtained in the enrichments. This demonstrated that when starting with one cell, the detection threshold for efficient transfer of at least one cell to the secondary enrichment step, i.e. 2 log10 CFU/ml, was not reached by 11 of 23 strains tested (48%) after exposure to 60 °C heat stress. Increasing the incubation time from 24 to 26 h and the transfer volume from 0.1 to 1.0 ml can increase the average probability to transfer at least one cell to the secondary enrichment step from 79.9% to 99.0%. When optimizing enrichment procedures, it is crucial to take strain variability into account as this can have a significant impact on the detection efficacy.

Multilevel modelling as a tool to include variability and uncertainty in quantitative microbiology and risk assessment. Thermal inactivation of Listeria monocytogenes as proof of concept.

Variability is inherent in biology and also substantial for microbial populations. In the context of food safety risk assessment, it refers to differences in the response of different bacterial strains (between-strain variability) and different cells (within-strain variability) to the same condition (e.g. inactivation treatment). However, its quantification based on empirical observations and its incorporation in predictive models is a challenge for both experimental design and (statistical) analysis. In this article we propose the use of multilevel models to quantify (different levels of) variability and uncertainty and include them in the predictions. As proof of concept, we analyse the microbial inactivation of Listeria monocytogenes to thermal treatments including different levels of variability (between-strain and within-strain) and uncertainty. The relationship between the microbial count and time was expressed using a (non-linear) Weibullian model. Moreover, we defined stochastic hypotheses to describe the different types of variation at the level of the kinetic parameters, as well as in the observations (microbial counts). The model parameters (kinetic parameters and variances) are estimated using Bayesian statistics. The multilevel approach was compared against an analogous, single-level model. The multilevel methodology shrinks extreme parameter estimates towards the mean according to uncertainty, thus mitigating overfitting. In addition, this approach enables to easily incorporate different levels of variation (between-strain and/or within-strain variability and/or uncertainty) in the predictions. On the other hand, multilevel (Bayesian) models are more complex to define, implement, analyse and communicate than single-level models. Nevertheless, their ability to incorporate different sources of variability in predictions make them very suitable for Quantitative Microbial Risk Assessment.