In recent years, biocontrol as an alternative for chemical disinfection has gained strong interest. Studies have already been performed in primary animal and plant production, food industry and even in hospitals (Zhao et al., 2006; Vandini et al., 2014; Luyckx et al., 2016; Bosmans et al., 2017; Hossain et al., 2017; Zheng et al., 2018). However, the possibility to use biocontrol agents against persistent pathogenic strains in the broiler environment has received only limited attention (Alves et al., 2015). In this research, a realistic in vitro model for biofilm formation on the inside of the DWS was developed and validated. This model was utilized to study interactions between Salmonella Java and Pseudomonas putida strains previously isolated from DWS and to evaluate the potential of P. putida as BCA in this niche. Biofilm formation was evaluated based on bacterial counts. This quantification method proved more repeatable and reproducible compared to OD measurement of crystal violet after resolubilization. The high variation of the OD measurements for biofilm quantification was possibly due to the rinsing step under running tap water to remove the excess stain. During this step, an uncontrollable mechanic force is applied to the coupons whereby pieces of the biofilm can detach. Therefore, further optimisation is required to use OD measurement as a method to quantify biofilm formation in the newly developed in vitro biofilm model. However, bacterial counts are more valuable than OD measurements for mixed cultures as it allows to quantify the strains separately and determine the underlying social interactions.
The difference in monoculture biofilm-forming capacity between the three P. putida strains in the DWS in vitro model confirmed the previous observations concerning biofilm formation by these strains in 96-well MTPs (Maes et al., 2019). Differences in biofilm-forming capacity between strains of the same species are commonly reported and can be due to mutations in biofilm regulating genes (Arevalo-Ferro et al., 2005; Chia et al., 2009; Agarwal et al., 2011; Lianou and Koutsoumanis, 2013; López-Sánchez et al., 2016). Very few literature was found concerning monoculture biofilm-forming capacity of S. Java. Agarwal et al. (2011) screened a multitude of Salmonella serotypes, among which one S. Java strain, for biofilm formation in 96-well MTPs. The S. Java strain was evaluated as a weak biofilm former based on OD measurements. In the current study, where biofilm formation was evaluated under more realistic conditions, the S. Java field strain was evaluated as the best biofilm former based on bacterial counts compared to the other strains that were included (among which another Salmonella serotype i.e. Salmonella Mbandaka). Even at low inoculum densities, which are more realistic for the investigated niche (Berghaus et al., 2013), S. Java was capable to form a significant amount of biofilm. It was already demonstrated for Listeria monocytogenes that persistent strains show increased biofilm formation relative to non-persistent strains (Borucki et al., 2003; Soto et al., 2006; Vestby et al., 2009). The strong biofilm-forming capacity of the S. Java strain in this study could therefore be an explanation for the persistent character of this Salmonella serotype in broiler houses.
A framework based on the cooperation criterion (Mitri and Foster, 2013) and biodiversity effect (consisting of a selection effect and a complementarity effect) (Loreau and Hector, 2001; Parijs and Steenackers, 2018) was applied to characterize the social interactions between S. Java and the three P. putida strains. The study of social interaction provides essential information to identify effective BCAs. Several characteristics should be taken into account. First, it is advantageous if the BCA shows a strong inhibitory effect and a high cell number when co-cultured with the strain to be controlled. If the BCA has a higher number of cells than the unwanted strain this would indicate the BCA has a higher fitness in this niche, leading to gradual enrichment (selection) and higher dominance of the BCA over time. Furthermore, the niche overlap between the BCA and the unwanted strain should be maximal, preferentially leading to high levels of interference competition. Both factors are reflected in a low (preferentially negative) complementarity effect.
All evaluated P. putida strains were able to reduce the attachment and biofilm formation by S. Java, supporting the potential of P. putida as a BCA against S. Java in the DWS of broiler houses. The ability of Pseudomonas strains to inhibit the growth of several pathogenic bacteria, among which Salmonella, was previously attributed to the production of iron-capturing siderophores and the toxic pigment pyocyanin (Oblinger and Kraft, 1970; Gram, 1993; Cheng et al., 1995; Das and Das, 2015; Khare and Tavazoie, 2015). P. putida specifically also produces several biosurfactants that can inhibit biofilm formation and even break down existing biofilms (Kuiper et al, 2004). In addition, in silico genome mining revealed two clusters for biosynthesis of bacteriocins and one cluster for a type I polyketide synthase. P. putida thus has a wide arsenal of weaponry that could inhibit S. Java biofilm formation. Therefore, it could be interesting to preserve P. putida biofilms on the inside of the DWS in broiler houses and not to remove them by chemical disinfection.
All three P. putida strains showed different inhibitory effects on S. Java. When inoculated in equal amounts, strain P2 inhibited S. Java to the highest extent and even increased its own biofilm formation compared to mono-culture. The other P. putida strains also inhibited S. Java significantly, albeit to a lower extent, and they formed less biofilm in mixed- than in mono-culture. The observation that all three P. putida strains engage in competitive interactions with S. Java fits with a growing body of recent theoretical and experimental work indicating that competition, not cooperation, dominates interactions among microbial species (Mitri and Foster, 2013; Ghoul and Mitri, 2016; Parijs and Steenackers, 2018). More specifically, interactions between Pseudomonas aeruginosa and Salmonella Enteritidis and Typhimurium, between Pseudomonas fluorescens and Salmonella Typhimurium, Montevideo and Poona, and between P. putida and Salmonella enterica were also identified as competitive (Leriche and Carpentier, 1995; Chorianopoulos et al., 2008; Pang et al., 2017; Pang and Yuk, 2018; Olanya et al., 2015). In general, interactions between Pseudomonas and other Enterobacteriaceae such as Escherichia coli and Klebsiella pneumoniae are also predominantly competitive, however the species dominating the co-culture and whether mutual inhibition or exploitation occurs is strongly strain- and condition-dependent (Cerqueira et al., 2013; Culotti and Packman, 2014; Gomez et al., 2017; Lopes et al., 2014; Zhao et al., 2019). Moreover, the type of interaction between strains is greatly dependent on environmental conditions, among which the stress gradient plays an important role (Piccardi et al., 2019). Therefore, the presence of medication administered through the DWS can also influence the interaction between P. putida and S. Java in practice.
Although all interactions were competitive in nature, the positive complementarity effects in all strain combinations indicate that the niches between both species do not completely overlap, alleviating the competitive interactions. In addition, despite the inhibitory effect of P. putida, Salmonella remained the dominant species when equal inoculum densities were applied as counts for Salmonella were always higher than counts for Pseudomonas spp. The dominance of Salmonella Typhimurium relative to Pseudomonas aeruginosa in dual-species biofilms was already described by Pang et al. (2017), but in the same study Salmonella Enteritidis was equally distributed to P. aeruginosa. In contrast to our study, coexistence between Pseudomonas and Salmonella Agona enhanced biofilm formation by S. Agona in terms of increased biovolume in the study of Habimana et al. (2010). Overall, this suggests that the behaviour of Salmonella in dual-species biofilms with Pseudomonas is strongly dependent on respectively serotype and strain. In addition, differences in biofilm growth conditions (flow, incubation time, incubation temperature, stress factors, surface type, etc.) could also lead to different interactions between the strains (Dai et al., 2017; Piccardi et al., 2019).
Consistently, we found that changing the inoculum ratio affects the outcome of competition greatly. When the inoculum proportion of P3/S1 was lowered to a more realistic 1:0.001, P3 was able to exploit resources provided by competitor S. Java and dominate the biofilm. This exploitation could for example be due to superior positioning in the biofilm or the consumption of metabolic by-products generated by Salmonella (Pfeiffer et al., 2001; Kim et al., 2014; Scholz and Greenberg, 2015; Parijs and Steenackers, 2018). When P. putida was first allowed to form a biofilm on the surface of the DWS and only afterwards S. Java was applied, a further increase in the competitive effect against S. Java was established, as evident by a stronger percentage reduction in Salmonella cell numbers. Again, P. putida was able to dominate the biofilm and exploit S. Java. One possible explanation for the enhanced inhibitory effect in the sequential set-up is that P. putida covers the abiotic surface and prevents the adhesion of Salmonella in a process called surface blanketing (Rendueles and Ghigo, 2012). However, surface blanketing is unlikely due to the low density of the P. putida biofilms. Indeed, prior research indicates that at similar densities Pseudomonas forms a sparse biofilm that does not cover the complete surface (Furiga et al.; 2015). The above described effects of nutrient and interference competition are therefore likely more important.
Although S. Java was strongly inhibited by the P. putida strains, prevention of Salmonella colonization was not yet complete. Given the strain variations observed in this study, other P. putida strains might be able to reduce S. Java to an even higher extent. It would therefore be interesting to evaluate the biocontrol potential of additional P. putida strains, possibly in combination with other Salmonella-biocontrol species. Moreover, biofilms on the inside of the DWS in broiler houses are composed of a diverse range of microorganisms (Maes et al., 2019), which might also interact with pathogen and BCA. Future biocontrol assays should take this species diversity into account.
Another important factor to consider is that, although co-culturing Salmonella and Pseudomonas can lead to less biofilm formation by Salmonella, different studies reported an increased Salmonella tolerance to disinfectants in these mixed species biofilms (Leriche and Carpentier, 1995; Pang and Yuk, 2018). Parijs and Steenackers (2018) reported this increased tolerance can be a consequence of competitive release in the biofilm upon treatment or of an increase in inherent tolerance due to the presence of competing species. Another downside of biofilms present on the inside of the DWS in broiler houses (independent on the strain composition) is clogging of the pipes and capture of medicine particles, leading to under dosing of the animals and increasing the risk for animal health and the development of drug resistant strains (Roberts et al., 2008; Høiby et al., 2010). Therefore also the combination of biocontrol strategies and chemical (disinfection/drug) treatments should be investigated.