Inactivation of Multi-Drug Resistant Non-Typhoidal Salmonella and Wild-Type Escherichia coli STEC Using Organic Acids: A Potential Alternative to the Food Industry

Salmonella and Escherichia coli are the main bacterial species involved in food outbreaks worldwide. Recent reports showed that chemical sanitizers commonly used to control these pathogens could induce antibiotic resistance. Therefore, this study aimed to describe the efficiency of chemical sanitizers and organic acids when inactivating wild and clinical strains of Salmonella and E. coli, targeting a 4-log reduction. To achieve this goal, three methods were applied. (i) Disk-diffusion challenge for organic acids. (ii) Determination of MIC for two acids (acetic and lactic), as well as two sanitizers (quaternary compound and sodium hypochlorite). (iii) The development of inactivation models from the previously defined concentrations. In disk-diffusion, the results indicated that wild strains have higher resistance potential when compared to clinical strains. Regarding the models, quaternary ammonium and lactic acid showed a linear pattern of inactivation, while sodium hypochlorite had a linear pattern with tail dispersion, and acetic acid has Weibull dispersion to E. coli. The concentration to 4-log reduction differed from Salmonella and E. coli in acetic acid and sodium hypochlorite. The use of organic acids is an alternative method for antimicrobial control. Our study indicates the levels of organic acids and sanitizers to be used in the inactivation of emerging foodborne pathogens.


Introduction
Salmonella and Escherichia coli present in food are a direct risk to human health [1][2][3]. Studies related to food contamination by these pathogens are frequent in the literature [4,5]. In Brazil, as a consequence of food contamination, groups of Shiga-toxin producing Escherichia coli (STEC) and non-typhoidal Salmonella have been detected in several foodborne cases [6,7]. Moreover, these bacteria groups

Organic Acid in DiskxDiffusion Tests
The results of disk-diffusion tests are described in Figure 1. Overall, the acetic and lactic acid had a higher antimicrobial effect in both pathogen strains, while the citric acid has proved to be less effective for both bacteria (p < 0.05). When we analyzed E. coli (Figure 1a), all wild-type strains were significantly different from the ATCC strain for both, the lactic and acetic acid. However, for the citric acid, only E113-4 and E26-2 strains were different from the ATCC strain (p < 0.05). For Salmonella, the lactic and acetic acids also had higher inactivation values when compared to citric acid (Figure 1b). For lactic acid, the only difference between strains was found between S3509 and S6130 (p < 0.05). For acetic acid, the ATCC and S3509 strains were more susceptible to inactivation (p < 0.05). Finally, for citric acid, the same inactivation (p > 0.05) was achieved for all strains. In general, the Salmonella group (average of all strains) presented the same inactivation pattern than the E. coli group for all tested acids (Figure 1c). When evaluating the results of E. coli versus Salmonella (Figure 1c), it is possible to identify that Salmonella is more sensitive to acetic and lactic acid than E. coli. However, in citric acid, the same inactivation values (p > 0.05) were obtained for both groups.

Organic Acid in DiskxDiffusion Tests
The results of disk-diffusion tests are described in Figure 1. Overall, the acetic and lactic acid had a higher antimicrobial effect in both pathogen strains, while the citric acid has proved to be less effective for both bacteria (p < 0.05). When we analyzed E. coli (Figure 1a), all wild-type strains were significantly different from the ATCC strain for both, the lactic and acetic acid. However, for the citric acid, only E113-4 and E26-2 strains were different from the ATCC strain (p < 0.05). For Salmonella, the lactic and acetic acids also had higher inactivation values when compared to citric acid (Figure 1b). For lactic acid, the only difference between strains was found between S3509 and S6130 (p < 0.05). For acetic acid, the ATCC and S3509 strains were more susceptible to inactivation (p < 0.05). Finally, for citric acid, the same inactivation (p > 0.05) was achieved for all strains. In general, the Salmonella group (average of all strains) presented the same inactivation pattern than the E. coli group for all tested acids (Figure 1c). When evaluating the results of E. coli versus Salmonella (Figure 1c), it is possible to identify that Salmonella is more sensitive to acetic and lactic acid than E. coli. However, in citric acid, the same inactivation values (p > 0.05) were obtained for both groups. Legend: Average and standard deviation of the Escherichia coli STEC (a); inhibition and nontyphoidal Salmonella strains (b). Capital letters indicate the statistical difference (p < 0.05) in the same acid in different strains. Lowercase letters indicate the statistical difference (p < 0.05) between acids in the same strain. A comparison of E. coli and Salmonella group (average of all strains) inhibition means (c). Capital letters indicate a statistical difference between acids in the same strain, and lowercase letters indicate a statistical difference between strains in the same acid.
For subsequent analysis of inactivation kinetics, we selected the E. coli strain E26-2 once it showed a difference from the ATCC in the three studied organic acids. Besides, the Salmonella strain S6130 was selected because of its difference from ATCC in acetic acid and higher drug-resistant profile among the other Salmonella strains [5]. Citric acid presents the lowest inactivation values Capital letters indicate the statistical difference (p < 0.05) in the same acid in different strains. Lowercase letters indicate the statistical difference (p < 0.05) between acids in the same strain. A comparison of E. coli and Salmonella group (average of all strains) inhibition means (c). Capital letters indicate a statistical difference between acids in the same strain, and lowercase letters indicate a statistical difference between strains in the same acid.
For subsequent analysis of inactivation kinetics, we selected the E. coli strain E26-2 once it showed a difference from the ATCC in the three studied organic acids. Besides, the Salmonella strain S6130 was selected because of its difference from ATCC in acetic acid and higher drug-resistant profile among the other Salmonella strains [5]. Citric acid presents the lowest inactivation values between acids in both analyzed groups. For that reason, it was not used in the subsequent inactivation modeling.

Inactivation Modeling Using Organic Acid and Sanitizer Treatments
The inactivation kinetics of Escherichia coli and Salmonella cells were determined by survival models fitted to the survival data (Table 1). Legend: Salmonella used was (S45-1), and Escherichia coli was (S26-2). The model was selected according to the better adjustment of data. 4D reduction: concentration compounds required to 4-log reduction of the bacterial load. R 2 = adjusted determination coefficient: indicates the goodness of fit. Kmax: rate of population inactivation before the tailing effect. Delta: time required for the first log reduction. MSE: mean square error.
The behavior of Salmonella and Escherichia coli inactivation is described in Figure 2. The pattern of inactivation was linear for quaternary ammonium, and the rate of inactivation (Kmax) resembles in both strains, affirming that the strains are susceptible to this compound. For sodium hypochlorite, the Salmonella inactivation model indicates a linear pattern. However, for the E. coli strain, a log-linear decrease with a tail in higher concentrations of the sanitizer was found. The Kmax of sodium hypochlorite indicates a higher susceptibility in E. coli in the initial contact. However, it is possible to verify a resistance in the last three levels used in E. coli strains that evidenced the tail effect (an almost constant survival rate) ( Figure 2). Concerning organic acid, all models had a linear dispersion for both pathogens and close to the Kmax value, with the exception of E. coli in acetic acid, where the best data adjust was for the Weibull model, and the δ (delta) parameter (concentration required for first log reduction) was obtained. The parameters of each model are described in Table 1.
Results regarding a 4-log reduction of each bacterial load are described in Figure 3. Escherichia coli showed a higher (p < 0.05) resistance to acetic acid exposure when compared to Salmonella ( Figure 3). However, the sodium hypochlorite inactivation of the strains showed a contrary behavior, where E. coli presented a higher inactivation than Salmonella (p < 0.05). The quaternary ammonium and lactic acid did not differ among the strains used (p < 0.05). Results regarding a 4-log reduction of each bacterial load are described in Figure 3. Escherichia coli showed a higher (p < 0.05) resistance to acetic acid exposure when compared to Salmonella ( Figure  3). However, the sodium hypochlorite inactivation of the strains showed a contrary behavior, where E. coli presented a higher inactivation than Salmonella (p < 0.05). The quaternary ammonium and lactic acid did not differ among the strains used (p < 0.05).  Results regarding a 4-log reduction of each bacterial load are described in Figure 3. Escherichia coli showed a higher (p < 0.05) resistance to acetic acid exposure when compared to Salmonella ( Figure  3). However, the sodium hypochlorite inactivation of the strains showed a contrary behavior, where E. coli presented a higher inactivation than Salmonella (p < 0.05). The quaternary ammonium and lactic acid did not differ among the strains used (p < 0.05). Legend: Non-typhoidal Salmonella used was S45-1, and E. coli STEC used was S26-2. The capital letter indicates a statistical difference in concentration compounds required to 4-log reduction of each substance (column). Adjusted means (LS means) were separated by the Student's t-test. Concentration values are presented as average ± standard deviation. Legend: Non-typhoidal Salmonella used was S45-1, and E. coli STEC used was S26-2. The capital letter indicates a statistical difference in concentration compounds required to 4-log reduction of each substance (column). Adjusted means (LS means) were separated by the Student's t-test. Concentration values are presented as average ± standard deviation.

Discussion
Quaternary ammonium and sodium hypochlorite are commonly applied in the food industry. In the present study, wild Salmonella and E. coli strains showed a linear pattern of inactivation when exposed to quaternary ammonium. The obtained results indicate an efficacy on the inactivation of Salmonella and E. coli. This finding is relevant, indicating that the use of this substance remains a good alternative to the inactivation of microbial pathogens. However, the use of quaternary ammonium in farm facilities destined to livestock is still banned in some countries due to the capacity of induction of multi-drug resistance in bacteria [35]. This fact is highlighted for generic efflux pump mechanisms and cassette collectors of resistance. Both are responsible for resistance in some antimicrobial substances too [36]. For example, Deng et al. [37], evaluated Salmonella isolates from foods of animal origin at retail, and the results indicated that the use of disinfectants was related to MDR strains by selective pressure and the mechanisms described above. Moreover, other studies have also evidenced that quaternary ammonium compounds induce antimicrobial resistance [27].
Concerning sodium hypochlorite, the results indicate an inactivation tail for the wild E. coli strain. This substance is commonly applied in the facilities and directly in products of vegetal origin. The constant use of this substance and, consequently, resistance due to selective pressure, as well as its volatile characteristic, is an additional challenge to be used in the food industry [38]. Our hypothesis for the inactivation tail is based on the saturation of efficacy of the sodium hypochlorite due to the lower availability of free-chlorine or volatilization of this compound. Furthermore, the wild genetic profiles can have an influence because metabolic transcriptions are generally more active [3].
Another point is that sodium hypochlorite is applied to biofilm control in the food industry, which may trigger resistance to this substance since the biofilm populations are dense and the inner layer can survive the use of sanitizer [39]. Our results show the need to evaluate alternative substances for biofilm control, since a population of resistant bacteria may be responsible for several cases of contamination in food [40,41]. It is essential to notice that Salmonella had a linear inactivation pattern, which indicates a susceptibility of this wild strain to sodium hypochlorite. According to the study performed by Köhler et al. [42], an efficient reduction in gram-negative bacteria with sodium hypochlorite was found. However, the presence of organic matter and MDR strains was highlighted as limiters of the sanitizer efficiency. For this reason, we performed the assays in Brain Heart Infusion Broth (BHI) broth to simulate the organic matter in the food industry and the time of exposure based on the general protocol used in the facilities.
Organic acids are regarded as an alternative compound for microbiology control in the food industry due to their secure handling, low cost, and quick action [43]. Lactic acid had a linear inactivation pattern in both strains. This finding may be related to the prohibition of the use of lactic acid in the meat production system in Brazil, where the lack of exposure of wild strains to lactic acid in food processing may have led to bacterial strains without previously developed resistance against the acid. Moreover, our results on E. coli inactivation kinetics are similar to other studies, pointing out that lactic acid is more effective than acetic acid, although the use of lactic acid in carcass for decontamination in the USA is permitted [44]. Another advantage of lactic acid compared to acetic acid is its volatile characteristics, where lactic acid is odorless, while acetic acid has a strong characteristic odor. Besides, this characteristic of acetic acid can irritate the skin or eyes.
In contrast to lactic acid, a different behavior was found using acetic acid to inactivate E. coli strains. Linear inactivation was verified in Salmonella, while for E. coli, the model of inactivation fitted was non-linear (Weibull). This behavior may be associated with intrinsic resistance of low pH in E. coli [45,46]. Moreover, the study performed by Hamdallah et al. [47] shows that an experimental evolution was performed to estimate the adaptation and growth capacity of E. coli at adverse pH. The results indicated that pH is the key to transcriptional regulators for acid resistance, and together with selective pressure, it directs the evolution of the strains towards higher resistance profiles. Besides, Chapman and Ross [48] suggested that Salmonella and E. coli protect themselves against acetic acid by mechanisms that retard acidification of the bacterial cytoplasm. In accordance, our results indicate that E. coli may have triggered this previously mentioned mechanism.
On the other hand, for Salmonella, these effects were either not elicited or insufficient. It is important to note that the acetic acid is employed in the practice of nutritional supplementation in animals as a growth promoter tool (usually in poultry reared) [49]. This fact may be associated with the resistance to the wild E. coli strain in the present study.
Another factor corroborating the discrepancy in acetic acid is the screening step used in the present study. The wild-type strains showed higher resistance to organic acids when compared to the ATCC strains, and E. coli was more resistant than Salmonella in lactic and acetic acids (Figure 1). A hypothesis is related to the mechanism of the acid tolerance response (ATR) that remains more active in wild strains subjected to abiotic stresses than in clinical strains stored for long periods [3]. The extensive presence of bacteria resistant to several drugs has been related to Brazilian meat and poultry production [5,50,51]. However, the lack of research associated with the use of sanitizers against bacteria with multi-drug resistance in Brazil and the possible exposure to antimicrobials in the food chain emphasizes the need to use wild strains in studies for the inactivation of pathogens. We evaluated different models of inactivation in wild strains that had virulence genotype and occasional multi-drug resistance. The fitted models for organic acid and sanitizers showed suitable adequacy when explaining the microorganism inactivation kinetics. The statistical parameters of adequacy and adjustment were satisfactory, reinforcing the excellent fit to the data (Table 1).
Moreover, the legislation behind the use of disinfectants is still an essential topic of discussion. The direct use or presence of the quaternary of ammonium in food products is prohibited by the ANVISA (Brazil) [52]. The use of sodium hypochlorite for fruit disinfection is the exception [53]. On the other hand, the direct use of organic acids in the processing of beef is allowed by the European Union and the United States of America to decrease the total count of microorganisms [44,54]. However, the use of organic acids is not permitted in Brazil for beef production [43]. The use of organic acid in the animal production system in Brazil is only approved for chicken processing during the sanitization step and the acetic acid as a growth promoter. In this regard, our study encourages the use of organic acid in the food industry, directly or indirectly, as a sanitization process.
Moreover, the use of organic acids is a potential alternative to overcome bacteria antimicrobial resistance [33,55,56]. Antimicrobial resistance has increased the global incidence of infectious diseases, and thus, organic acids, due to the penetration and disintegrating capacity of the outer membrane of gram-negative cells, represent a potential tool to combat this issue [33,55,56]. Besides, it is worth pointing out that the organic acids can be directly applied in food or on food processing surfaces once they are safe for human intake and do not have a daily limit established.

Sample Collection and Preparation
A total of 12 bacterial strains (six E. coli and six Salmonella) were used and are described in Table 2. The wild-type non-typhoidal Salmonella strains used in the study were previously isolated from chicken meat, as described by Cunha-Neto et al. [5], and E. coli STEC strains were isolated during the processing of beef by Santos et al. [57]. Besides, the ATTC culture of non-typhoidal Salmonella (ATCC-23564) and Escherichia coli STEC (ATCC-2196) was used as a reference to address the comparison with wild-type strains. The criteria for inclusion of the strains in the present study were according to the relevance of the serotype (involvement in food outbreaks) and the resistance to one or more classes of antibiotics. The strains were stored at −80 • C in Brain Heart Infusion Broth (BHI; Kasvi ® , São Paulo, Brazil), medium with 20% glycerol as stock cultures. Posteriorly, the reactivation of the strains was performed. Briefly, an aliquot of 0.1 mL of the stored culture was collected, inoculated into 9 mL of BHI and incubated at 37 • C for 24 h. Subsequently, a second reactivation round was carried out to maximize the cellular metabolic process.

Selection of Resistant Strains in Organic Acid Using Disk Diffusion Method
To select the strains with the highest resistant profile to be used in the inactivation modeling, a disk diffusion assay for organic acids was performed. Acetic, lactic, and citric acids were standardized for 4096 µg/mL concentrations. Each strain was transferred to Müller-Hinton 2 broth (MH; Himedia ® , Mumbai, India) and incubated between 2 and 4 h up to 0.5 MacFarland scale [58]. The assay was carried out according to the Kirk Bauer disk-diffusion test [59]. Briefly, the strains were streaked on Müller-Hinton 2 agar (MH; Himedia ® , India) and diffusion disks (LB; Laborclin ® , São Paulo, Brazil) with 10 µL of each organic acid were included in the diffusion disks. After the incubation period of 24 h at 37 ± 0.1 • C, the halos were measured.

Organic Acid and Sanitizer Treatments and Enumeration of Survival Cells
According to disk-diffusion results, two strains were selected for the study of the inactivation kinetics Salmonella (S45-1) and Escherichia coli (E26-2). To determinate the working concentrations, a minimum inhibitory concentration (MIC) was determined for both pathogens. Therefore, MIC was performed by a microdilution test on a 96-well plate. Briefly, 0.2 mL of BHI with Salmonella strain at 10 5 CFU/mL was included in each well. The concentration of the tested organic acids was calculated according to the volume of the total solution. Besides, the doses with a minimum inhibitory concentration for each pathogen were identified, and the doses required for the model were determined. The dose for total inactivation (DTI) was used as the highest working concentration to study the inactivation efficiency of each compound.
Ten concentrations lower than DTI were utilized to study the bacterial inactivation kinetics. The ranges used for acetic and lactic acids were: 4.00-7.50% (v/v), to sodium hypochlorite: 29.00-70.00% (v/v) and quaternary ammonium: 0.45-0.68% (v/v). The analyses were performed following the microdilution method with an exposure time of fifteen-minutes per substance. Briefly, 100 µL of BHI with their respective inoculated bacteria were distributed in 8-wells into a 96-well plate. The percentages of sanitizers and organic acids were calculated according to each point and included in the BHI broth.
After fifteen minutes, the aliquot of 0.1 mL was transferred to 0.9 mL of saline peptone water (for neutralization of the substances). Posterior dilutions were performed, and an aliquot of 0.1 mL was plated on plate count agar (PCA; Kasvi ® , São Paulo, Brazil). The plates were incubated at 37 • C for 24 h and counted on the electronic plate counter (Eddy-jet-IUL, Barcelona, Spain).

Statistical Analyses and Mathematical Modeling
To evaluate the effects of different organic acids on different strains of E. coli and Salmonella, as well as to choose the highest organic acid-resistant strains, the data of disk-diffusion assay were analyzed using ANOVA with Tukey's test. In order to compare Salmonella vs. E. coli in disk-diffusion inhibition and dose required for 4-log reduction (obtained by inactivation modeling) of each sanitizer and organic acid, a Student's t-test was performed. Models of inactivation were constructed using the software Gina FIT version 1.6 (Katholieke Universiteit Leuven, BEL, Leuven, Belgium). The following models were fitted to the survival data: Log-Linear Bigelow [60], Geeraerd-tail model [61], and the Weibull model [62]. The model evaluation and performance were assessed through the adjusted coefficient of determination (R 2 adj ) and mean square error (MSE) [63]. The significance level used was 0.05.

Conclusions
The use of some chemical compounds is being related to the induction of antimicrobial resistance [17], and organic acids are gaining popularity as an alternative strategy for antimicrobial control [33,55,56]. Properties such as low cost, easy handling, fast application, and a non-limited daily intake to consumers pointed to the use of the organic acids. Moreover, organic acids, mainly lactic acid, can be directly used on the beef surface, in the water employed to sanitizations, or in industrial facilities for inactivation of wild strains with resistance to several antimicrobials. Our study indicates the levels of organic acids and sanitizers to be used in the inactivation of emerging foodborne pathogens while using wild-type strains of E. coli STEC and Salmonella, with a multi-drug resistance profile for the construction of such models that incorporate higher reliability with the expected reduction since they are based on strains with higher resistance profiles.

Conflicts of Interest:
The authors declare no conflict of interest.