Presence of virulence genes
Our study showed variations in the frequency of virulence genes in S. Minnesota strains. However, the detection of avrA, sodC and invA genes was unanimous. For the invA gene, this pattern was expected by Rowlands in 2014 , when analyzing 237 strains of Salmonella spp. isolated from food in Brazil also reported the presence of the invA gene in all strains. Zou in 2012  in a study conducted in North Carolina (USA), from June 2009 to September 2010, found that 99.3% of S. Enteritidis strains isolated from human outbreaks of salmonellosis had the gene. According to Whang in 2009 , the invA gene seems to be very conserved in Salmonella spp., which justifies its high occurrence.
In addition to invasiveness, all strains demonstrated survival potential under oxidative stress (sodC) and apoptosis induction in infected cells (avrA). The existence of these virulence mechanisms reinforces the pathogenic potential of strains and, despite the absence of reported cases of salmonellosis by this serovar, demonstrates the possibility of causing disease in humans. Therefore, its isolation in environmental samples, chicken meat and its derivatives may indicate a potential source of transmission of this agent to humans.
Borges in 2013  evaluated 84 strains of S. Enteritidis isolated from 1996 to 2010 in the state of Rio Grande do Sul, Brazil, from different avian sources, and observed a similar result to our study, with the presence of the invA and avrA genes in 100% of the isolates and the IpfA and agfA genes were identified in 99% (83/84) and 96% (81/84) of the isolates, respectively. In 2016, Ahmed  also observed similar results in Egypt in 20 isolates of S. Typhimurium from chicken, and in 10 isolates of human origin, with frequency of 100% of invA and avrA genes (30/30). They concluded that a high frequency of these genes is observed in serovars potentially causing salmonellosis in humans.
The absence of the sefA gene in the studied strains may be justified according to the study by Amini in 2010 , in which they observed that this specific gene is not present in all serovars, being restricted to group D Salmonella, such as Enteritidis, Dublin, Moscow and Blegdon serotypes.
Although not present in all strains, the genetic potential to form biofilms, associated with the presence of agfA, lpfA and luxS genes, was observed in 75% (15/20) of the strains. The presence of these genes demonstrate that the studied strains are fully capable of adapting to adverse environmental conditions through the acquisition of sessile life form. But the expression of only one of them already allows the initiation of the process of acquiring sessile form. Similar studies with different Salmonella serotypes have shown the presence of the agfA gene in over 91,4%, the lfpA gene in 80.63% and the three genes together in more than 73.34% of strains isolated from birds [7, 6, 45, 28].
According to Borges in 2013 , the frequency of virulence determinants and the establishment of genetic profiles of isolates can be used to determine more effective control protocols and prevention measures in industries.
Influence of temperature on S. Minnesota biofilms
In addition to genotypic characteristics, it is known that extrinsic factors also directly affect the sessile lifestyle. Among them, the presence of moist, nutrient-poor surfaces and temperature variations, which in poultry production offer the ideal conditions for Salmonella to settle and form biofilms .
The absence of biofilms at 4 °C was defined according to the BFI determination described by Naves in 2008 . These data are in agreement with the work done by Dhakal in 2019 , who found optical density values equal to those of the negative control for six Salmonella serovars tested when kept at refrigeration temperature. However, a study conducted with a standard S. Minnesota strain showed that this serovar has the capacity to form medium intensity biofilms under in vitro conditions using the traditional methodology .
In addition, our study indicates that the BFI classification for this serovar is strain-dependent, and there are probably other factors that determine this variability, such as the presence and expression of genes linked to biofilm formation (agfA, lpfA and luxS), which encodes aggregative fimbriae and the quorum sensing system whose functions are linked to the process of surface fixation and bacterial communication, respectively, facilitating biofilm formation . In our study, all strains had at least one of these genes, indicating the potential for sessile life that was expressed differently and according to external factors and the expression of this potential.
The increase in biofilm intensity was directly related to the increase in temperature. This is consistent with the study by Borges in 2013  in biofilms in S. Heidelberg, S. Typhimurium, S. Agona, S. Infantis, S. Brandenburg and S. Tennessee at temperatures of 3 °C, 12 °C, 28 °C and 37 °C, whose BFIs ranged from nonexistent to moderate.
Considering the presence of biomass identified as statistically different from the negative control, we observed a higher number of strains capable of forming biofilms at different temperatures. Thus, even under thermal stress conditions under low temperatures, S. Minnesota is still capable of forming a biomass that allows its viability, presenting potential contaminant.
The results found in our study indicate that S. Minnesota can become a difficult problem to control during broiler processing in industry. Considering the high prevalence of the microorganism in Brazilian broiler poultry and slaughterhouses , it is possible to suggest its permanence in the industrial environment, even under stress conditions, through the production of biofilm. This information serves as a warning to the poultry industry as an incentive to develop more rigorous and effective control measures for this agent.
The same behavior in relation to the biofilm formation identified at temperatures of 25ºC and 36 °C contradicts the results found in the literature, which states that temperatures below the optimal growth temperature and near the environment intensify the biomass production in Salmonella biofilms [41, 35]. According to Cabarkapa in 2015 , the factors involved in biofilm production have different responses depending on the bacterial strain and according to the incubation temperature.
Significant differences (p < 0.05) in relation to BFI within the same serotype indicate that there is probably influence of intrinsic characteristics, such as the presence of fimbriae, flagella, membrane proteins and others , which even present at the molecular level vary in their expression. Thus, environmental conditions alone are not decisive and limiting in the formation of biofilms. Genotypic diversity and the way in which these factors are expressed and determine the phenotype which, in association, influence the production of sessile biomass by S. Minnesota should also be considered.
The image analysis allowed to confirm the differences identified in the biofilm according to the temperatures (Fig. 5). The ultrastructure of the biofilm formed by S. Minnesota at temperatures of 25 and 36 ° C showed a more stable and mature conformational characteristic compared to the temperature of 4 ° C consistent with that found in different Salmonella serovars .
Chemical agents reduced S. Minnesota biofilm
Peracetic acid, sodium hypochlorite and chlorhexidine are chemical agents widely used in the cleaning of slaughterhouses, including the industry where strains have been isolated. These biocides promoted a reduction in S. Minnesota biofilm counts after exposure for 15 minutes. However, the use of peracetic acid and chlorhexidine demonstrated the same efficiency in significantly reducing counts for resistant strains, differently of what was observed for sodium hypochlorite, which in all strains showed resistance. This is alarming because sodium hypochlorite is one of the most widely used cleaning and disinfecting agents in the industry.
The presence of strains tolerant to different sanitizers suggests that the inappropriate use of these agents in the routine of the industrial environment may result from sublethal exposure to these biocides, representing a real risk to the resistance and adaptation of these bacteria, besides favoring the production of biofilms , since under these conditions the adaptive response mechanisms of stress bacteria are activated, promoting their survival even in harsh environments .
Regarding the high resistance observed to sodium hypochlorite, it is also possible to associate the involvement of molecular factors, such as RpoS, Dps and sodC genes, linked to oxidative stress, being the last one identified in all strains of our study. These genes are actively expressed in S. Enteritidis SE86, resistant to the presence of sodium hypochlorite at 200 ppm . It is also possible that for S. Minnesota there are similar mechanisms at work in this process. In addition, the properties of this sanitizer can be altered according to the pH variation and the presence of organic matter that alternate the availability of hypochlorous acid, reducing its efficiency .
The variation in the counts for the different strains identified after contact with peracetic acid and chlorhexidine demonstrated that the persistence of the presence of these microorganisms may be a strain-dependent characteristic.
The existence of strains with resistance profile to all agents (profile IV) (Table 4) (12/20–60%) indicates that there are probably intrinsic or extrinsic adaptive mechanisms that allow their survival. This fact may characterize a potential risk in industry due to the difficulty in eliminating these microorganisms and the risk of spreading this characteristic to other bacterial strains.
Although the use of chemical compounds brings benefits in the disinfection stage, these agents usually have the limitation of not destroying the residual structures of the bacterial biofilm matrix, which can facilitate the resurgence or even the maintenance of these structures on the surfaces . In our study, we observed that the external matrix was maintained only when the strains were treated with sodium hypochlorite. Therefore, bacterial resistance seems to be related to the limitation of the presence of organic matter and, for this case, special efforts are required for the complete removal of S. Minnesota biofilms adapted to this biocide. Probably, the effectiveness in controlling these microorganisms will be achieved through sanitation plans that combine cleaning measures focused on the elimination of the extrapolymeric matrix combined with the use of different agents and on the periodic rotation of disinfectants, respecting the periods between disinfections.
In a different way, peracetic acid and chlorhexidine proved to be effective in eliminating the external matrix and in disrupting the conformation of mature biofilm. This profile may be associated with the mechanism of action aimed at the denaturation of proteins, cellular enzymes, increased permeability of the bacterial cell and cell lysis .
Similarity between strains
Phylogenetic analysis identified strains with distinct molecular profiles. Only pulsotypes C, K and M grouped strains that allowed an epidemiological evaluation. Pulsotype “C” grouped two strains, M17 and M14, with similarity of 82.8%, coming from the industry B broiler poultry in different periods, 2009 and 2010, with the presence of avrA, sodC, invA, agfA and luxS genes in common. Although they belong to the same industry (B), they are strains of distinct flocks; one from the state of São Paulo and one from Mato Grosso do Sul, indicative of a possible dissemination of this profile, possibly influenced by transport. In addition, the strains behavior in biofilm tests was identical with resistance to the three chemical agents evaluated in sessile form.
The pulsotype "K" grouped four strains, two (M04 and M03) clones isolated from the slaughterhouse of industry A, located in the state of Minas Gerais. Cross-contamination may be the reason for the occurrence of this profile, since they were isolated in very close periods, both dated in November 2014. The presence of avrA, sodC, invA, agfA and lpfA genes was common in both.
Other strains of the K (M20 and M16) pulsotypes were isolated in 2010; both in the aviary and in the industry B, cutting room in Mato Grosso do Sul. The avrA, sodC, invA and luxS genes and resistance to sodium hypochlorite and peracetic acid in biofilms were common among these strains. These data indicates that the maintenance of these microorganisms in sessile form and contamination of the final product may be caused by cross contamination between utensils and the slaughtering environment itself .
In addition to cross contamination, there are also signs of neglect of biosecurity standards due to the presence of the agent in clean and dirty areas of processing. According to Moura in 2014 , this oversight is decisive for the maintenance of the microorganism in the environment. At the same time, the pulsotype K indicates that there was a spread of this profile, since there are similar strains in both Mato Grosso do Sul and Minas Gerais, in different slaughter units.
The “M” pulsotype grouped three strains (M08; M07 and M05) with similarity equal to 83,6% of 2014, coming from industry A slaughterhouse in the state of Minas Gerais. The strains presented the avrA, sodC, invA and agfA genes in common.
In addition to cross contamination, it is possible to suggest that strains with homology greater than 80% persist in both the poultry and slaughterhouse environment, probably due to the presence of biofilms, which, even if identified at low intensities in our assays, demonstrated the viability of the strains sessile bacteria at high counts, even in the presence of biocidal agents.
The sessile life form probably caused this serovar to be isolated at different times, as well as facilitating its spread among the different industry environments.