Escherichia coli and Salmonella Enteritidis dual-species biofilms: interspecies interactions and antibiofilm efficacy of phages

Escherichia coli and Salmonella Enteritidis are foodborne pathogens forming challenging biofilms that contribute to their virulence, antimicrobial resistance, and survival on surfaces. Interspecies interactions occur between species in mixed biofilms promoting different outcomes to each species. Here we describe the interactions between E. coli and S. Enteritidis strains, and their control using specific phages. Single-species biofilms presented more cells compared to dual-species biofilms. The spatial organization of strains, observed by confocal microscopy, revealed similar arrangements in both single- and dual-species biofilms. The EPS matrix composition, assessed by Fourier-transform infrared spectroscopy, disclosed that the spectra extracted from the different dual-species biofilms can either be a combination of both species EPS matrix components or that the EPS matrix of one species predominates. Phages damaged more the single-species biofilms than the mixed biofilms, showing also that the killing efficacy was greatly dependent on the phage growth characteristics, bacterial growth parameters, and bacterial spatial distribution in biofilms. This combination of methodologies provides new knowledge of species-species and phage-host interactions in biofilms of these two major foodborne pathogens.

www.nature.com/scientificreports www.nature.com/scientificreports/ broad-spectrum disinfectant/virucide, sanitizer, with an application on a hard, non-porous surface. Resistance to QAC was due to the EPS produced by Salmonella, which conferred protection to both species 4 . The interaction between these two species in biofilms formed on HEp-2 cells showed that Salmonella biofilms outgrew and displaced pre-formed E. coli biofilms 13 . In another study, a specific strain of Salmonella unable to form biofilms utilized the curli proteins cross-seeded by E. coli to enhance its adherence to dual-species biofilms 14 .
Phages, the natural predators of bacteria, have been extensively employed in the control of single-species biofilms of common bacteria present in the food industry [15][16][17][18][19][20] . However, the application of phages to mixed bacterial populations is scarce but the results, reported to date, have been encouraging. For instance, phages successfully controlled dual-species biofilms of Staphylococcus lentus and Pseudomonas fluorescens, and further were able to decrease the number of the non-susceptible strain, when its specific phage was not added, through their release from the biofilm consortia 21 . Dual-species biofilms of Staphylococcus aureus and Staphylococcus epidermidis challenged with phages phiIPLA-RODI and phiIPLA-C1C in a cocktail also reduced significantly the number of attached cells 22 .
The focus of this work was to assess the influence of E. coli and S. Enteritidis strains on single-and dual-species biofilm formation, and also to investigate biofilm control using phages vB_EcoM_Daica (Daica) and ϕ135, specific for E. coli and S. Enteritidis, respectively.

Results
Characterization of E. coli and S. Enteritidis strains growth parameters. The constructed E. coli and S. Enteritidis strains carrying sfGFP-and mCherry-expression plasmids were characterized to determine their specific growth rates and doubling times ( Table 1). The differences in EC 434 and EC 515 growth, and of SE Ex2 and SE 269 resulted in statistically different µ max values (p < 0.05) ( Table 1). Furthermore, also the µ max between EC 515 and SE 269 was statistically significant (p < 0.05) with EC 515 presenting the slowest doubling time (t d ).
Characterization of single-and dual-species biofilms. The four strains were characterized according to a methodology described previously 23 to find if these were weak or strong adherent bacteria. Species had differences in adherence and consequent biofilm formation and thus for the dual-species biofilms, two sets of combinations that were used consisting of the strong (EC 434 and SE Ex2) and the weak (EC 515 and SE 269) biofilm-formers. Overall, all strains presented higher numbers in single-species biofilms compared to dual-species biofilms (Fig. 1). For instance, biofilms of EC 434 (7.5 × 10 7 CFU/cm 2 ) and SE Ex2 (1.73 × 10 8 CFU/  Table 1. Specific growth rate (µ max ) and doubling time (t d ) of the E. coli and S. Enteritidis strains used Statistical analysis of the results was performed using GraphPad Prism 6 (GraphPad Software, La Jolla, CA, USA). Differences between strains and species parameters were assessed using two-way ANOVA followed by Tukey's multiple comparison statistical test Asterisks (*) indicate significant differences between strains (*p < 0.05) and daggers ( † ) indicate significant differences between species (p < 0.05). www.nature.com/scientificreports www.nature.com/scientificreports/ cm 2 ) dropped to approximately 2.5 × 10 6 CFU/cm 2 when added together (Fig. 1A). Also, EC 515 and SE 269 biofilms showed 1 log less viable cells in dual-species biofilms compared to single-species biofilms (Fig. 1B).

Figure 1. Biofilm formation of (A) EC 434 and SE
To better understand the differences between the E. coli and S. Enteritidis strains and their biofilm-forming abilities, the Competitive Index (CI) and the Relative Increase Ratio (RIR) indexes were calculated (Fig. 2). While in CI the growth curves of the two species in mixed biofilms are compared, the RIR compares the growth curves of both species in single-species biofilms. A negative CI was detected throughout all time points ( Fig. 2A). However, only at 24 and 28 h of biofilm growth of SE Ex2 + EC 434 the CI and RIR were found to be statistically different (p < 0.05), indicating a competitive advantage of SE Ex2 over EC 434. On the other hand, a positive CI was observed at 24 and 28 h of growth for EC 515 + SE 269 biofilms, indicating a competitive advantage for EC 515 (Fig. 2B).
The bacterial distribution within the 48-h-old single-and dual-species biofilms formed by E. coli strains (mCherry fluorophore), and/or S. Enteritidis (sfGFP) was visualized using CLSM (Figs. 3 and 4). EC 434 single-species biofilms presented a heterogeneous spatial distribution, with bacteria accumulated more in some areas than others, reaching a thickness of approximately 4 µm (Fig. 3A, I and II). SE Ex2 biofilms were more evenly spread throughout the polystyrene coupon, presenting a thickness of ≈13 µm (Fig. 3A, III and IV). The difference in thickness between EC 434 and SE Ex2 biofilms was is in agreement with the number of viable cells obtained for each biofilm at 48 h (Fig. 1A). Analysis by CLSM of dual-species biofilm formed by EC 434 + SE Ex2 showed both strains growing on the same areas on the coupon, suggesting an influence of each species on the spatial distribution of the other species (Fig. 3B, I-III). The thickness EC 434 + SE Ex2 biofilms reached approximately 11 µm (Fig. 3B, IV), due to the lower number of bacterial counts obtained (1.55 Log 10 CFU/cm 2 and 1.76 Log 10 CFU/cm 2 for EC 434 and SE Ex2) (Fig. 1A).
The spatial characterization was also assessed for 48-h single-and dual-species biofilms of EC 515 and SE 269 (Fig. 4). EC 515 alone formed clustered biofilm structures, reaching, in some areas of the coupon, 11 µm in thickness (Fig. 4A, I and II). EC 434 strain reached only a 4 µm thickness (Fig. 3A, I and II). SE 269 formed a biofilm with a reticulate appearance equally dispersed through the coupon with a thickness of approximately 24 µm (Fig. 4A, III and IV), while SE Ex2 presented only a 13 µm thickness. EC 515 + SE 269 biofilms, maintained their single-species biofilm spatial arrangement (Fig. 4B, I-III) although reaching only a maximum thickness of ≈15 µm (Fig. 4B, IV).
EPS spectra obtained by FTIR-ATR spectroscopy of E. coli and S. Enteritidis from single-and dual-species biofilms was analyzed (Fig. 5). Globally, the spectra presented a very similar and typical shape containing the absorption bands of lipids (3000-2800 cm −1 ), proteins/amides I and II (1700-1500 cm −1 ), a mixed region of phospholipids and nucleic acids (1500-1185 cm −1 ), polysaccharides (1185-900 cm −1 ), and the fingerprint region (900-600 cm −1 ) 24,25 . Regarding the EC 434 + SE Ex2 mixed biofilm EPS matrix, the spectrum is similar to EC 434 or to SE Ex2 spectra, depending on the considered spectral region (Fig. 5A), pointing to the absence of a predominant EPS matrix species. Additionally, the PCA model revealed an undefined clusterization with all spectra being quite dispersed in the whole scores map (Fig. 6A). Concerning the EC 515 + SE 269 EPS matrix, this spectrum is more similar to the one of EC 515. Indeed, the infrared spectra of the dual and single-species biofilms are almost totally superimposable with the EC 515 EPS matrix, being quite different from the SE 269 spectrum (Fig. 5B). The corresponding scores map of the PCA model also reflects this similarity, showing two defined clusters, one containing the EPS spectra from S. Enteritidis biofilm and the other containing the EPS matrix spectra from both single E. coli and the dual-species biofilms (Fig. 6B).

Phage characterization. Transmission electron microscopy (TEM) visualization showed that E. coli phage
Daica is a myovirus, having a long contractile tail (Fig. 7A), while the S. Enteritidis phage ϕ135 is a siphovirus with a long non-contractile tail (Fig. 7B). . Differences between single-and dual-species biofilm formation abilities were assessed using two-way ANOVA followed by Tukey's multiple comparison statistical test. Asterisks (*) indicate a significant difference (p < 0.05) between CI and RIR.
E. coli and S. Enteritidis dual-species biofilm control with E. coli phage Daica and S. Enteritidis phage ϕ135 was investigated (Fig. 11). Dual-species biofilms of EC 434 + SE Ex2 using the phage cocktail, resulted in a maximum viable cell decrease at 8 h, with EC 434 being reduced by 1.15 Log 10 and SE Ex2 by 0.88 Log 10 (Fig. 11A). EC 515 + SE 269 dual-species biofilms reached the lowest numbers of viable cells at 4 h of treatment having the phages reduced EC 515 by 1.07 Log 10 and SE 269 at 8 h by 2.42 Log 10 (Fig. 11B).

Discussion
E. coli and Salmonella represent an important public health concern 26,27 , being responsible for numerous foodborne outbreaks, and often isolated from contaminated meats 28,29 , vegetables 30,31 , as well as food processing surfaces. Their presence in food processing surfaces leads often to food products cross-contamination events 32 . E. coli and Salmonella's ability to form biofilms make them less susceptible to common disinfectants, physical removal, and other elimination processes 33,34 . In this work, we assessed the efficacy of two phages alone or in a cocktail, for the control of single-and dual-species biofilms, respectively. Furthermore, structural and compositional characteristics of E. coli and S. Enteritidis single-species biofilms were analysed, and dual-species biofilm interactions studied 35 . www.nature.com/scientificreports www.nature.com/scientificreports/ The E. coli strains tested belong to serotypes O6 (EC 434) and O1:K1:H7 (EC 515). Mono-species biofilm formation by E. coli serotypes has been shown to be serotype-dependent 36 but also influenced by the levels of curli expression, and presence/absence of virulence factors, such as Shiga toxins 1 and 2 (stx1 and stx2), intimin (eaeA), enterohemolysin A (hlyA), fimbriae, and autotransporters (e.g. EhaB and EspP) [37][38][39] . For instance, a study with 39 Shiga toxin-producing E. coli isolates (seropathotypes A, B, and C), showed that seropathotype A, that includes the serotypes responsible for the highest incidence of foodborne outbreaks (O157:H7 and O157:NM), had greater ability to form biofilms than isolates from seropathotypes B (eg O26:H11, O45:H2, and O103:H2) and C (O157:H26) 38 . According to our results, significant discrepancies between the biofilm-forming ability by the E. coli serotypes O1 and O6 were perceived, with higher counts (1.5 Log 10 ) in biofilms of EC 515 compared to EC 434 (Fig. 1). SE Ex2 and SE 269 isolates belong to S. Enteritidis serovar however their genotype is unknown. Thus assumptions equivalent to those described for E. coli cannot be established. The adherence ability after 24 h of SE Ex2 and SE 269 showed to be different, as also confirmed by the biofilm thickness using CLSM (Figs. 3 and 4). The CFU/cm 2 determinations showed a 1.2 Log 10 difference with SE Ex2 reaching higher numbers than SE 269 (Fig. 1). Single-species biofilms of E. coli treated with Daica resulted in equivalent antibiofilm efficacy against both strains used (Fig. 10). This result is possibly due to a combination of three different factors: the latent period of Daica in the tested strains and the burst size reached per infected bacterium (Fig. 9A); and also the strain growth parameters. Significant differences of µ max were detected for both E. coli (EC 434 and EC 515) and S. Enteritidis (SE Ex2 and SE 269) ( Table 1). Daica has a longer latent period and a lower burst size on EC 434 cell, compared to EC 515. Additionally, EC 434 strain has a higher doubling time which could result in less new cells to be infected by progeny phages. In EC 515, Daica produced around 7.3 times more progeny phage particles per infected cell, however, in this strain, the phage latent period was longer. All these factors influenced in the treatment outcome, having EC 515 treatment with Daica reached similar reductions (see Figs. 9A and 10A, and also Table 1). Salmonella phage ϕ135 showed a higher antibiofilm outcome in SE 269, compared to SE Ex2 biofilms. This is greatly due to a higher burst size in the first strain (Fig. 9). Also, the growth rate of SE 269 was lower and therefore the non-infected biofilm cells proliferated at a slower rate than the SE Ex2 non-infected cells, consequently resulting in a smaller number of descendent cells present at the sampling time points assessed (Figs. 9B and 10B, and also Table 1). www.nature.com/scientificreports www.nature.com/scientificreports/ The strain combinations for the dual-species biofilm formation studies were: EC 434 + SE Ex2 (strong biofilm-formers), and EC 515 + SE 269 (weak biofilm-formers). The competitive ability of strains in mixed biofilms has been linked to the relative density of each bacterium added to form biofilms 40 , which in our study was exactly the same (200 µL/well at 1 × 10 7 CFU/mL). Therefore, this factor cannot explain the higher/lower surface coverage by the different dual-species biofilms. The number of viable biofilm cells per area analysed in dual-species biofilms was always lower than in single-species biofilms (Fig. 1). Antagonism between biofilms formed with E. coli O157:H7 strain USDA 5 and Salmonella strain 457-88 was hypothesized to be due to species competition for adherence to the surface, leaving a smaller area available for adhesion and consequent biofilm formation by the less competitive species involved 41 . Mixed biofilm populations can also have a dominance of  www.nature.com/scientificreports www.nature.com/scientificreports/ one species over the other. For instance, biofilms formed by Staphylococcus aureus and E. coli showed a dominance of the latter species, due to the shorter generation time of E. coli compared to S. aureus 42 . It seems that faster-growing bacteria control the environment, and favor their growth over other slower-growing strains resulting in a long-term prevalence in the surface area 43,44 . Growth characteristics between the two sets of species used for dual-species biofilms showed that only the set of strains EC 515 and SE 269 was significantly different (p < 0.05). This may explain why EC 515, presenting a faster growth rate, also covered a larger biofilm surface area when formed along with SE 269 (Fig. 1B). The EC 515 covered the surface area more in height forming preferentially 3D structures, while SE 269 cells were more scattered in the polystyrene surface (Fig. 4B). The same spatial distribution found in single-species biofilms was maintained by SE 269 and EC 515 when grown together.
The other set of strains used in dual-species biofilms, EC 434 and SE Ex2, had similar µ max and t d values (p > 0.05) being present in the biofilms in fairly similar numbers (Fig. 1A) and equally distributed in the same areas of the coupons (Fig. 3B).  www.nature.com/scientificreports www.nature.com/scientificreports/ FTIR-ATR spectroscopy coupled with chemometrics, a reliable and alternative method to accurately discriminate bacteria at different taxonomic levels, including E. coli clones 45 , and Salmonella serogroups and serotypes 46 , and has also been used in biofilm contexts. For instance, FTIR-ATR has been used to monitor biofilm in situ, in real-time, and under fully hydrated conditions 47 , the early stages of biofilm formation, and the EPS matrix from biofilms at mid-development and mature phases 47 . In the latter work, FTIR-ATR spectroscopy showed that mature biofilms presented substantially higher protein and carbohydrate peaks, and a low level of lipids that was, nonetheless, higher than in mid-development phase biofilms 47 . In the present work, FTIR-ATR was used to compare the EPS matrix composition of single-and dual-species biofilms (Fig. 5). The FTIR-ATR analysis showed few spectral differences among the samples with a lower absorption band of lipids in all EPS samples analyzed than the absorption bands of proteins or carbohydrates. Also, it seemed that the ratio between proteins/carbohydrates  www.nature.com/scientificreports www.nature.com/scientificreports/ was higher for EC 515 and SE 269 (Fig. 5B). A previous study has reported a competitive advantage of EPS producer strains in mixed biofilm populations and also reported that this EPS secretion helps progeny cell movement up and out of the focal cell layer, providing better oxygen conditions to their descendants 48 . Taking into account the EPS matrix spectral profiles obtained by FTIR-ATR, the mixed biofilms of EC 434 + SE Ex2 resembled the EPS spectra obtained with EC 434 in terms of proteins/amides I and II, and the phospholipids and nucleic acids absorption peaks, and with the polysaccharide absorption band of SE Ex2 (Fig. 5A). So, in general, in this set of strains, none seemed to be a better EPS producer than the other. Although some differences in spectral band intensities were observed, FTIR-ATR spectroscopy is a semi-quantitative method and therefore quantitative relations about biofilm matrix components should be taken carefully. The polysaccharide component of the EPS matrix offers diverse benefits to the cells in biofilms, sustaining surface and cell-to-cell adhesions, providing protection, and allowing cell growth in 3D structures 49 . If we take into account only the polysaccharides absorption peak (peak 4), this clearly resembles the one obtained by SE Ex2 which suggests that this strain secretes more polysaccharides than EC 434. As described above, this can confer this strain a competitive advantage for surface adhesion and predominance in the mixed consortia (Fig. 5A) which is in agreement with the higher levels of SE Ex2 cells in the surface compared to EC 434 (see Fig. 1A and negative CI index in Fig. 2A). The other set of strains, EC 515 + SE 269, showed a FTIR-ATR spectra superimposing the spectrum of EC 515 observed in single-species biofilms (Fig. 5B). Therefore, it is clear that there is a predominance of the EC 515 EPS matrix in these dual-species biofilms but that also EC 515 predominated due to its µ max ( Table 1). Dominance of a surface by one specific bacterium in mixed biofilms can be growth rate-dependent or polysaccharide-dependent. However, this latter observation needs to be further validated, since only the FTIR-ATR absorbance peak between 1185 and 900 cm −1 was taken into consideration for this assumption.
The use of both phages as antibiofilm agents showed that short treatment periods are more effective towards both single-and dual-species biofilms (Figs. 10 and 11). The antibiofilm effect using the phage cocktail comprising Daica and ϕ135 phages, on dual-species biofilms formed by EC 434 and SE Ex2 resulted in similar Log 10 reductions for both species (Fig. 11A). Both strains have non-statistically different µ max and, therefore, in theory, the phage producing a higher burst size should reduce more biofilm cells. This was not, however, the situation observed after phage treatment of these biofilms. Taking into consideration that EC 434 and SE Ex2 biofilms also presented fairly similar viable cell numbers throughout the phage treatment experiments, and being ϕ135 able to produce 30.7 times more new phage particles per infected SE Ex2 cell compared to Daica on EC 434, this indeed should have resulted in greater SE Ex2 reductions. CLSM analysis showed that these strains are scattered throughout the coupon surface, and this can hinder phages from reaching their hosts (Fig. 3B). Furthermore, FTIR-ATR analysis of the EPS matrix also showed that the spectral peaks of these strains were more similar, in some peaks, to SE Ex2 (polysaccharide absorption peak, peak 4), and in other peaks more related to EC 434 (Fig. 5A). If the assumption made before, that SE Ex2 secretes more polysaccharides, this SE Ex2 EPS matrix can be masking the phage receptors present in the host cell surfaces of EC 434 explaining in this case, why a higher phage burst size did not result in an enhanced antibiofilm activity against EC 434. Nevertheless, further experiments need to be performed to understand this behavior, possibly resorting to time-lapse microscopy imaging, flow cytometry or even spinning disk confocal microscopy.
Phage challenge of the EC 515 and SE 269 dual-species consortia resulted in higher reductions of SE 269 biofilm cells than of EC 515 (Fig. 11B). In this case, this result is a reflection of ϕ135 having a better burst size than Daica, and also due to the slower growth rate of SE 269 ( Fig. 10B and Table 1). Furthermore, CLSM showed that EC 515 preferentially formed 3D clusters, which can impair phage diffusion to reach cells that are in deeper biofilm layers (Fig. 4B). According to FTIR-ATR spectroscopy analysis, the EC 515 + SE 269 EPS spectrum obtained was remarkably similar to the spectrum of EC 515 EPS. These results indicate that, even though EC 515's EPS predominates in the dual-species biofilm samples, it is possibly confined only to the clustered areas, and therefore does not constitute an obstacle for phage ϕ135 to reach SE 269 cells.
In conclusion, this work gives new insights into the many factors influencing dual-species biofilm development. Moreover, it sheds light on the interaction of single and cocktail phages for the control of single-and dual-species biofilms. A better understanding of species-species and phage-host interactions was reached through analysis of bacterial and phage growth parameters, viable cell determination, before and after phage treatment, CLSM imaging of the biofilm structures, and also EPS matrix analyses by FTIR-ATR spectroscopy. Even though some of the strains showed to be more competitive regarding their surface attachment, reaching higher viable cell numbers than the other species involved, this did not have a direct influence on the action of phages. Infectivity of the phage was more associated with their growth characteristics, in particular, the latent period duration and burst size, and also to the difficulty of phages to infect one particular strain, EC 515, which formed 3D biofilms structures rather than being uniformly dispersed over the surface, possibly impairing phage diffusion. Furthermore, the EPS matrix of one species can constitute a problem to the other one, and the phage killing ability the latter species, by possibly covering the phage receptors necessary for the adsorption step to take place.
Construction of sfGFP and mCherry strains. S. Enteritidis strains were transformed with the sfGFP-pBAD plasmid (Addgene plasmid #54519) which encodes for the superfolder green fluorescent protein (sfGFP), and the E. coli strains with plasmid pNUT086, which encodes for the mCherry fluorophore. Electrocompetent S. Enteritidis and E. coli cells (100 µL) were mixed with the respective plasmids, placed in a 1 mm gap electroporation cuvette, and a pulse of 25 μF, 200 Ω, 1.8 kV was applied. After, 900 µL of liquid LB medium were added to the mixture, cells incubated at 37 °C (120 rpm, 1 h), and plated on solid LB supplemented either with ampicillin (100 µg/mL) or with kanamycin (50 µg/mL), for S. Enteritidis cells and E. coli cells, respectively.
Characterization of the growth parameters of E. coli and S. enteritidis strains. Specific growth rates and doubling times of all strains were determined according to previously described methods 52 . Experiments were performed in triplicate.
Biofilm formation assessment by Crystal Violet staining. The method described by Stepanović et al. (2000) was used to categorize the biofilm-forming capacity of bacteria into: 1) non-biofilm-formers (OD ≤ ODc); 2) weak (OD c < OD ≤ 2 × OD c ); 3) moderate (2 × OD c < OD ≤ 4 × OD c ); and 4) strong biofilm-formers (OD > 4 × OD c ). OD c is the cut-off OD (ODc = OD average of sterile LB + 3 × Standard Deviation (SD) of the ODs of negative controls). Sterile LB served as negative control giving an average value of 0.089 ± 0.005 and an ODc of 0.104. Experiments were performed in triplicate.
Biofilm formation. The ability of the different strains to form biofilms was determined according to a method using the microtiter plate biofilm model previously described 23 , with some modifications. Briefly, an overnight culture was grown (37 °C, 120 rpm) in LB supplemented with kanamycin or ampicillin, according to the plasmid used. After, all cultures were diluted to 1 × 10 7 CFU/mL, and 200 μL/well of culture transferred to polystyrene microtiter plates (96-well flat-bottom, Sarstedt, Inc., Germany) and incubated at 37 °C (48 h, 120 rpm). After, biofilms were washed twice with saline and detached using an ultrasonic bath (50 kHz, 6 min, Sonicor SC-52, Sonicor Instruments, UK) 53 . Bacterial suspensions were homogenized, 10-fold serially diluted, and plated onto solid LB supplemented with either kanamycin or ampicillin, and incubated overnight at 37 °C. Experiments were performed in triplicate.

Determination of the competitive index (CI) and the relative increase ratio (RIR).
In dual-species biofilms, the Competitive Index (CI) was established as the EC 434/SE Ex2 or EC 515/SE 269 ratios within the output sample divided by the corresponding ratio in the inoculum (input): CI = (EC 434/SE Ex2 or EC 515/ SE 269) output/(EC 434/SE Ex2 or EC 515/SE 269) input, where output and input are the counts of viable cells [Log 10 (CFU/cm 2 )] obtained at defined time points or the inoculum (t = 0), respectively 12 . CI values were interpreted as follows: a CI value equal to 0 indicates equal competition of the two species; a positive CI value indicates a competitive advantage for the species on the numerator; a negative CI value indicates a competitive advantage for the species on the denominator. The Relative Increase Ration (RIR) was calculated based on the counts of viable cells [Log 10 (CFU/cm 2 )] obtained from single-species biofilms of each strain 12 .

Confocal laser scanning microscopy (CLSM).
Single-and dual-species biofilms were formed on polystyrene coupons (NuncThermanox, Thermo Scientific, MA, USA) placed on 24-well polystyrene microtiter plates as described above, with slight modifications. The volume of bacterial suspension was adjusted to 1 mL/well instead of 200 μL/well. Expression of mCherry fluorophore and sfGFP was induced by adding 1 mM IPTG and 0.2% (v/v) of L-arabinose 54 Fourier transform infrared-attenuated total reflectance (FTIR-ATR) spectroscopy of EPS. EPS of biofilms was lyophilized and analyzed by FTIR-ATR spectroscopy. Samples were transferred to the ATR crystal and a pressure of 150 N/cm 2 was applied. Infrared spectra were acquired (PerkinElmer Spectrum BX FTIR System spectrophotometer, PerkinElmer, Waltham, USA) with a PIKE Technologies Gladi ATR accessory (PIKE Technologies, Inc., Madison, USA) from 4000 to 600 cm −1 with a resolution of 4 cm −1 and 32 scans co-additions. For each sample, three instrumental replicates (obtained in the same day) and two biological replicates (obtained www.nature.com/scientificreports www.nature.com/scientificreports/ in two different days from two independent bacterial growths) were obtained, corresponding to a total of six spectra for each extracted EPS. Between each EPS measurement, the background was acquired. Spectra corresponding to the instrumental replicates were averaged prior to the analysis.
Phage propagation and titration. SE Ex2 and EC 434 were used for the production and titration of phages ϕ135 and Daica. Phages were amplified using the plate lysis and elution method 58 . Titration of the phage was performed according to a previously described protocol 59 .
Transmission electron microscopy (TEM) analysis of phages. The morphology of phage particles was observed by TEM as previously described 60 . Phage DNA extraction, genome sequencing, and annotation. Phage DNA was extracted as described before 60 . Phage DNA was prepared for sequencing using the Illumina Nextera XT library preparation kit generating 250 bp paired-end sequencing reads and sequenced in the Illumina HiSeq platform. Reads were trimmed to remove adapters, contaminations, or low-quality sequences, and assembled with the CLC Genomics Workbench version 7 (CLC Bio, Aarhus, Denmark). Structural annotation was conducted with MyRAST 61 and also checked manually (Geneious 9.1.4, Biomatters, NJ, USA). Potential frameshifts were checked with BLASTX 62 . Gene functions were predicted using BLASTP programs 63 (E-value ≤ 10 −5 ) and HHPRED server 64 , consulted between June and July 2018, TMHMM 65 and Phobius 66 . The search of tRNA encoding genes was performed using tRNAscan-SE 67 . Putative promoters were searched using PhiSITE 68 , and putative regions were manually verified. ARNold 69 was used to predict Rho-independent terminators and the energy was calculated using Mfold 70  Phage one-step growth characteristics. The one-step growth curves of phages Daica and ϕ135 on the two tested E. coli and S. Enteritidis strains were performed as described previously 51 using a multiplicity of infection (MOI) of 0.001. Experiments were performed in triplicate.
Biofilm treatment with phages. Biofilms formed on 96-well microtiter plates for 24 h, as described above, were washed once with saline, and exposed to 100 μL of LB and 100 μL of phage at a multiplicity of infection (MOI) of 1. Single-species biofilms of S. Enteritidis were exposed to ϕ135, and E. coli to Daica. Dual-species biofilms were exposed to both phages. Microtiter plates were incubated at 37 °C, 120 rpm, and samples were taken after 4, 8 and 24 h. Control experiments were performed using 100 μL of LB and 100 μL of SM buffer (5.8 g/L NaCl, 2 g/L MgSO 4 .7 H 2 O, 50 mL 1 M Tris, pH 7.5). The number of viable biofilm cells (CFU/cm 2 ) was determined before and after phage infection, with serial dilutions performed in saline-FAS solution [0.9% (w/v) NaCl, 2 mM of ferric ammonium sulfate (FAS)] to destroy non-infecting phages 74 . Experiments were performed in triplicate.
Infrared spectra data analysis. Two principal component analysis (PCA) models 75 were developed including EPS spectra from the two E. coli and two S. Enteritidis strains, in single-and dual-species biofilms. Prior modeling, spectra were pre-processed with standard normal variate (SNV) followed by the application of a Savitzky-Golay filter (15 smoothing points, 2 nd order polynomial, and 1 st derivative) 76,77 and mean-centered. All data analyses were performed in Matlab version 7.9 (Mathworks, USA) and the PLS Toolbox version 5.5.1 for Matlab (Eigenvector Research, USA). Statistical analysis. Statistical analysis of the results was performed using GraphPad Prism 6 (GraphPad Software, CA, USA). The independent experiments and the results are presented as mean ± standard deviation (SD). Differences between single-and dual-species biofilm formation abilities were assessed using two-way ANOVA followed by Tukey's multiple comparison statistical test. CI and RIR were compared using unpaired Student's t-test, with significant differences being indicative of a probable competition between the species 12 . Differences were considered statistically significant if p < 0.05 (95% confidence interval).