Exogenous Protein as an Environmental Stimulus of Biofilm Formation in Select Bacterial Strains

A screening of environmental conditions that would elicit robust biofilm in a collection of Serratia marcescens isolated from soil revealed that exogenous milk protein increased biofilm productivity up to ten-fold. A select screening of fish pathogens, freshwater and human isolates identified several other species that responded similarly to exogenous protein. The optimal protein concentration was species specific; S. marcescens at 5% milk protein, Aeromonas sp. at 2% - 3%, Flavobacterium columnare at 1% and Pseudomonas aeruginosa at 0.1% - 0.4%. Media supplemented with milk protein also increased the cell counts in biofilm as well as the protein incorporated into the biofilm matrix. These data suggest that relatively high concentrations of exogenous protein may serve as an environmental trigger for biofilm formation, particularly for pathogenic bacteria exposed to relatively high concentrations of protein in bodily fluids and mucosal surfaces. of 4


Introduction
Since the concept of biofilm was first viewed through the eyes of molecular microbiology three decades ago, our appreciation of its importance in ecology has grown exponentially. We now recognize biofilm as an alternative life strategy for many, if not all species of microbes across three domains. The importance of biofilm in industry [1], public health [2], medicine [3] [4] [5] and the environment [6] [7] has been well documented, leading us to frequently include the analysis of a bacterial strain's ability to form biofilm as part of its species' description.
Growth of a microbe in a biofilm removes it from a pelagic lifestyle that is characterized by mass action events (or close to it). As a pelagic entity, change in location and concomitant change in access to nutrients could happen quickly. Once attached to a surface and embedded in a macromolecular matrix, a very different lifestyle ensues one in which the tempo and mode of life are slowed, the ambient conditions change more gradually (in general) and cell physiology is altered [8].
Biofilm, as part of a life history, is reversible [3] [9] [10]. A pelagic microbial species can attach to a substratum, multiply and then return to the surrounding solution as the biofilm structure becomes more susceptible to turbulence or the bacterium senses suboptimal conditions for sessile life. Environmental signals play an essential role in informing the microbe so that optimal survival strategies are selected for, including attachment and release. Consistent with the substantial phylogenetic and physiological diversity of the microbial world, biofilm, as an eco/evo strategy, has been employed in many different ecosystems in response to a broad range of environmental conditions. Of interest to us are specific environmental signals that elicit behavioral changes leading to the formation of biofilm. There are logical candidates for signals including carbon and energy sources, essential micronutrients and even inhibitors. For example, in Janthino bacterium, both violacein and biofilm production were stimulated by glycerol and inhibited by glucose [11]. The presence of calcium ion has also been found to influence biofilm productivity in Pseudomonas [12], Pseudoalteromonas [13], Xyella [14] and Citrobacter [15]. In addition, CaCl 2 , MgCl 2 , CuSO 4 , sucrose and sodium dodecyl sulfate produced greater biofilm in Yersinia pestis [16]. In some circumstances it is difficult to distinguish between a primary trigger and a secondary adjuvant of biofilm formation.
We report herein on two converging lines of investigation in our laboratory, the identification of environmental signals for biofilm formation in a collection of Serratia isolated from soil, and in several fish bacterial pathogens. Our observations with Serratia clearly indicated that exogenous skim milk protein at relatively high concentrations was sufficient to stimulate biofilm formation ten-fold.
We extended this observation to Flavobacterium columnare, Aeromonas salmonicida, two fish pathogens, as well as a collection of freshwater isolates. We also observed that exogenous protein had little or no effect on some isolates from mammalian hosts, but stimulated others. These data indicate that exogenous protein promotes biofilm production in select strains of bacteria.

Strains and Cultivation
Pseudomonas aeruginosa PA01 was a gift of Dr. M. Bagdasarian (MSU). Hydro-

Phylogenetic Analysis
The rRNA of S. marcescens strains (RL-1-RL-16) specific to this study were sequenced at the MSU genomics facility using the 27F 16S rRNA primer (Sanger chemistry

Standard Biofilm Assay
The protocol for biofilm measurement as described by Merritt

Measuring Cell Growth in Milk Protein Supplemented Media
To test for the effects of milk protein on growth of P. aeruginosa, S. marcescens, Aeromonas sp. and F. columnare, we measured growth in R2B supplemented with 0.1%, 0.2% and 0.4% milk protein in 24 well microtiter plates. We intentionally selected low concentrations at which milk protein is completely soluble. At higher concentrations (>1%) milk protein solutions are colloidal, making it difficult to measure optical density. Growths were performed in 24 well microtiter plates with 4x replication and shaking at 100 rpm on an orbital shaker at 25˚C. Optical density measurements were made at 0, 110 min, 210 min, 300 min 390 min 450 min and 24 hours on a Biotek EPOCH plate reader at 600 nm. At 24 hours the wells were tested for biofilm formation as described above. Uninoculated controls were subtracted from growths at each time point and in biofilm quantitation. Uninoculated controls for each protein concentration were also replicated x4.

Determining Viable Cells within Biofilm
To determine the viable cell count within biofilm formed in supplemented and unsupplemented media, cultures of P. aeruginosa PA01, Aeromonas 060628-1 and S. marcescens RL-5 were established in 24 well plates by inoculating 700 µl of R2B ± milk protein with 100 µl of overnight culture. Control wells contained 800 µl of uninoculated media. Plates were sealed with sterile foil and incubated for 24 hours shaking at 100 rpm and 25˚C. These experiments were set up in duplicate so that biofilm determination with crystal violet (Sigma) and viable cell counts could be performed in parallel. Each plate contained four replicates of each strain and media combination. After 48 hours of growth, one plate was stained with crystal violet as described above for quantitation of biofilm and the duplicate plate was used to determine the viable cells count within biofilms, as follows. The plate was gently rinsed three times in sterile water and the washed biofilm was scrapped off using 600 µl of sterile water and a sterile applicator. The cell slurry was transferred to 1.5 ml eppendorf tubes, vortexed to break up cell aggregates and 10-fold serially diluted for plating onto R2A. Plating was in triplicate and CFUs are reported as the total CFUs per well.

Determination of Protein Content in Biofilm
To determine the protein concentration of biofilms, 24 well plates were used as described. In this experiment we tested P. aeruginosa, Aeromonas 060628-1 and F. columnare. Biofilm of P. aeruginosa and Aeromonas were prepared in R2B and R2B-5%MP while F. columnare was tested in R2B-1%MP. Duplicate plates were inoculated so that both protein concentration and crystal violet staining could be tested in parallel. In each plate, all unique media conditions were replicated four times. After 24 hours of incubation at 25˚C and shaking at 100 rpm, one plate was stained with crystal violet (as described above) and the duplicate plate was used to determine protein content within biofilm as follows.

Confocal Microscopy
A two-well chamber (Lab-TekII, Nalge Nunc International, USA) was inoculated with 700 μL of media and 100 µl of P. aeruginosa or F. columnare. 5% Skim Milk Protein at 1/2x Tryptic Soy Broth (Becton, Dickinson and company, France) medium and 100 μL of F. columnare overnight broth. The chamber was then wrapped in Parafilm (Bemis, USA) to seal it. Next, the sample was incubated for 72 hours at 100 RPM. At 48-hour incubation the media was gently removed and new 700 μL of fresh medium was added into the wells. At 72 hours, the medium was removed and discarded, and the chamber was gently washed 3 times with 1 mL of sterile water. 1 mL of fluorescent solution containing 0.5 mL of 20x Nano Orange dye (Molecular Probes Protein Quantitation Kit N10271) and 0.5 mL of FM4-64 dye (Molecular Probes FM4-64) was added into one well and incubated in the dark for five minutes. The well was washed with 1 mL of sterile water two times. The sample was kept hydrate during microscopy. A confocal microscope (Olympus FluoView FV1000) was used for imaging at 20x and 90x.

Results
During the screening of a variety of environmental conditions designed to stimulate the formation of biofilm by soil isolates of Serratia, we detected a significant increase in biofilm when our standard growth media, R2B, was supplemented with milk protein (MP). Using standard media-grade skim milk protein (Hardy Diagnostics) at 5%, biofilm formation of 16 independently isolated soil Serratia strains increased significantly. In Figure 1 we show the response of five Serratia strains to media supplemented with different concentrations of milk protein (0.05%, 0.5%, 2.5% and 5.0%). All five isolates responded to 5% MP supplementation with a ten-fold increase in biofilm formation. All remaining isolates responded with similar increases (data not shown). In our biofilm assays we routinely use Pseudomonas aeruginosa PAO1 as a positive control. Under all conditions we have tested, PA01 produced a robust biofilm when grown on R2B or TSB but had little response to the presence of exogenous protein when at 0.5% or greater and, in many of our assays, high concentrations of protein in the media slightly inhibited biofilm formation by P. aeruginosa.
To extend these observations to other species we tested the ability of 5% MP to stimulate biofilm production in 200 freshwater bacterial isolates. In Figure 2 we report on 48 randomly picked isolates that were selectively isolated on Pseudomonas isolation agar and were therefore resistant to irgasan, a broad spectrum antimicrobial that targets fatty acid synthesis in bacteria. Among these isolates, Advances in Microbiology  four different concentrations of milk protein in TSB. These data, presented in Figure 3, show that the Aeromonas isolates responded to MP concentrations of 1% -5% while a freshwater Pseudomonas isolate revealed little response until 5% and Rahnella was unresponsive.
To extend this analysis to isolates associated with eukaryotic hosts, we also investigated the relationship between biofilm formation and exogenous protein in five isolates from fish, three from humans and an E. coli strain from bovine. These data are presented in Figure 4. Among the isolates from fish, F. columnare, Hydrogenophaga, Brevundimonas responded strongly and positively to exogenous protein by producing abundant biofilm, but at different optimal protein concentrations ( Figure 4(a)). F. columnare and Hydrogenophaga had greatest biofilm productivity at 1% while Brevundimonas was more productive at 5%. Pseudomonas C22 does not form abundant biofilm in unsupplemented media and productivity increased only modestly at 1% and 5% MP.
Acidovorax was unresponsive to exogenous protein. Figure 4(b) reports on the human and bovine isolates. Aeromonas sp, an unpigmented Serratia and a low Figure 3. Biofilm formation by six freshwater isolates in response to elevated protein concentrations in broth. Six isolates from the biofilm screening described in Figure 2 were tested for biofilm production at four different concentrations of milk protein (0.5%, 1.0%. 2.0% and 5.0%) in TSB. biofilm forming E. coli were isolated from human feces and had varied response to exogenous protein. Both Aeromonas and the Serratia isolates responded with greater biofilm productivity at 1% and 5% but E. coli was unresponsive. Included in this experiment was one of our soil Serratia isolates (RL-4) for comparison. Interestingly, biofilm formation by the E. coli isolate from bovine, identified as a high biofilm forming strain, was inhibited in media supplemented with 1% and 5% MP. Note that this experiment was conducted in LB broth without salt to mimic the conditions used in the initial characterizations of the E. coli strains. Both P. aeruginosa PA01 and our soil Serratia RL-4 had biofilm profiles in supplemented and unsupplemented TSB similar to what we have seen in R2B.
As can be seen from these biofilm assays, in some cases the amount of crystal violet staining material was quite large. In many of the Aeromonas strains tested an opaque disk formed at the bottom of the wells, particularly if the incubation period was extended to 48 hours and the 96 well format was used. An obvious concern was the possibility that crystal violet was staining protein and biofilm matrix atypically and providing a false positive for biofilm formation. To test for this, we ran several analyses in 24-well microtiter plates that prevented the formation of any opaque disk by virtue of the large well diameter. In these experiments, we measured biofilm formation using crystal violet and performed viable plate counts on biofilm from replica plates. These data are presented in Figure 5   and show that, as observed above, both Aeromonas and Serratia responded strongly to exogenous protein, producing at least a 10-fold increase in crystal violet signal while Pseudomonas had little response ( Figure 5(a)). The cell viability from a replica plate revealed a 1.2 -3 order of magnitude increase for Serratia and Aeromonas when grown with 5% exogenous protein, while P. aeruginosa had a robust viable count in the absence of protein and only a modest increase with protein when compared with Aeromonas ( Figure 5(b)).
An obvious question regarding the effect of exogenous protein on the formation of biofilm is whether the biofilm becomes enriched in protein. To test for this, we established biofilm in 24 well plates (4 replicates of each strain on a plate) and replicated the whole plate so that both biofilm formation and the amount of protein within the biofilm matrix could be measured. Because of our interest in fish pathogens we tested Aeromonas and F. columnare with P. aeruginosa as our positive control. The results are presented in Figure 6. As shown previously, both Aeromonas and F. columnare responded strongly to exogenous protein by producing more biofilm while P. aeruginosa PA01 was unresponsive. In this test, we used the optimal protein concentrations of 5% for Aeromonas (and Pseudomonas) and 1% for F. columnare. The biofilm from the replica plate was washed and manually scrapped from the wells and the protein concentration was determined using the Bradford assay, after removing the cells by centrifugation. The amount of protein detected in the biofilm for P. aeruginosa was 8.2 and 5.4 µg/ml for growth without and with protein, respectively. For Aeromonas, the increase in biofilm in response to exogenous protein was accompanied by an increase in matrix protein concentration from 0.55 to 47.3 µg/ml. For F. columnare, the 20-fold increase in biofilm was accompanied by a nearly 20-fold Figure 6. The effect of exogenous protein on the protein concentration within biofilms. These experiments were performed in 24 well plates (4 replicates for each treatment) and each plate was replicated for measuring biofilm (crystal violet) and protein (Bradford assay). P. aeruginosa PA01 and Aeromonas strain 060628-1 were tested at 5% milk protein and F. columnare 090702-1 was tested at 1% milk protein. To determine the effect of exogenous protein on cell growth we incubated P. aeruginosa, S. marcescens, Aeromonas strain 060628-1 and F. columnare strain 090702-1 at three concentrations (0.1%, 0.2% and 0.4%) of milk protein in R2B and monitored growth by optical density at 600 nm. Low concentrations were selected to avoid colloidal solution conditions present at higher concentrations. P. aeruginosa grew well under these experimental conditions but optical density was diminished in a concentration dependent manner when the media was supplemented with protein (Figure 7(a)). In contrast, S. marcescens (Figure 7 grew robustly, regardless of the exogenous protein through 450 min. Statistical differences were detected only after 24 hours of growth when exogenous protein appeared to modestly boost growth. Aeromonas grew slowly (Figure 7(c)) through 450 minutes with no appreciable difference with protein addition. The greatest growth was between 450 min and 24 hours. At 24 hours growth was inhibited at 0.4% exogenous protein. F. columnare grew well in the absence of exogenous protein and poorly, if at all, in its presence (Figure 7(d)). The tendency of this strain to form aggregates in solution accounted for the substantial inter-replicate variability. After 24 hours the plates were processed for biofilm formation ( Figure 8). P. aeruginosa PA01, as mentioned above, is a robust biofilm forming strain. Under conditions of growth in this experiment, enhanced biofilm productivity was detected at all concentrations of exogenous protein.
While S. marcescens grew vigorously, biofilm productivity was quite low at the tested protein concentrations. Aeromonas also lacked biofilm productivity at the lower concentrations of protein but did increase substantially at 0.4% milk protein, in spite of the apparent growth inhibition at this concentration. Biofilm production by F. columnare increased in a concentration dependent manner when the media was supplemented with protein. This robust biofilm production was in contrast to pelagic growth which appeared inhibited by exogenous protein.
In addition to measuring the protein content with a standard Bradford assay we used nano-orange to visualize the biofilm-associated protein. Using the standard microtiter plate protocol, we established biofilm on sterile coverslips with and without exogenous milk protein (1%) using P. aeruginosa and F. columnare as the test strains. After growth, the biofilm was washed with sterile water (x3) and stained with Nano-orange and FM4-64 using the vendors protocol. The biofilm was viewed on an Olympus FluoView FV1000 Confocal Microscope at 20X and 90X magnification (Figure 9). Numerous P. aeruginosa cells were detected at 20X magnification but there was little evidence of a robust contiguous biofilm. Intensely orange spots could be detected suggesting concentrations  of protein spotted the surface. At 90X magnification well isolated cells were seen with little evidence of a protein matrix. In contrast, the biofilm formed by F. columnare showed a thick branched proteinaceous complex at 20X magnification.
Cells were clearly outlined with the lipophilic FM4-64 stain at 90X magnification and showed morphological variation as describe previously [22]. In addition, irregularly shaped orange forms as well as cells decorated with Nano-orange were detected.

Discussion
These investigations began with repeated unsuccessful attempts to form a robust biofilm of S. marcescens isolated from soil. Different temperatures, carbon sources, nutrient availability, osmolarity and substrata were tested without effect on biofilm formation. However, one environment in which S. marcescens can colonize is the human respiratory system and this provided clues to a possible environmental signal initiating biofilm formation in Serratia. Alveolar fluid from human lungs is generally at 5% -13% protein [23]. This environmental feature of the lung led us to test biofilm formation at several concentrations of protein and identify robust biofilm of S. marcescens at 5% milk protein. Moreover, the increase in biofilm productivity was frequently an order of magnitude or greater above that observed in unsupplemented media. Our positive control strain, P. aeruginosa PA01, appeared unresponsive to high concentrations of protein in the medium.
These observations were extended to 48 freshwater isolates, four strains from sturgeon eggs [17], two known fish pathogens, three strains isolated from human gut and one from bovine and the results showed biofilm production that was dependent on two variables, species and protein concentration (a total of 74 strains including the Serratia isolates). Based on these data Serratia isolates from both soil and human gut were highly responsive to 5% exogenous protein, producing 5 to 10 times the amount of biofilm that they made in unsupplemented media. In all cases tested, Serratia required concentrations around 5% and failed to respond to lower concentrations (0.1%, 0.2%, 0.4%, 0.5%, 1% & 2%). Aeromonas strains were also sensitive to exogenous protein in the same manner, increasing biofilm production, although lower concentrations of protein (0.4% -2%) would suffice for some strains. The one strain of F. columnare reported on herein was particularly responsive to exogenous protein with an optimum at 1% protein and evidence of increased biofilm productivity at as low as 0.1% protein.
Additional studies within the Flavobacterium and Chryseobacterium lineages indicated that all isolates of F. columnare tested thus far are responsive to 1% milk protein (Loch & Marsh, unpublished). Biofilm production by P. aeruginosa PA01 was unresponsive to high concentrations of protein (1% -5%) and showed growth inhibition but enhanced biofilm production at low concentrations (0.1%, 0.2% & 0.4%). Those strains that appeared unresponsive at high concentrations included freshwater isolates Kluyvera, Erwinia, nearly all Pseudomonas (11 of 12), all Rahnella aquatilis isolates, 3 of 4 Yersinia isolates and human and bovine E. coli isolates.

Protein as a Surface Conditioning Agent
A number of investigators have reported that soluble protein can serve as a "conditioner" to surfaces that enhance or inhibit the development of biofilm.
Frequently serum is used as a "natural" protein-containing solution to condition surfaces (total protein in serum is typically 60 -80 g/L). For example, Patel et al. [24] showed that initial binding of S. epidermidis cells to hydrophobic polyurethanes was suppressed by serum at 2 hours but enhanced when incubated for 24 hrs. The opposite trend was observed for hydrophilic surfaces where serum inhibited biofilm formation. Similarly, Frade et al. [25] found that serum enhanced biofilm productivity of Candida albicans on metallic and non-metallic surfaces. Finally, using methodologies most similar to our approach, Kipanga et al. [26] demonstrated that polystyrene microtiter plates (Costar) conditioned with foetal calf serum showed reduced biofilm formation by C. albicans. These assays are in general difficult to compare given the diversity of surfaces, strains and complex-

Exogenous Protein-A Trigger or Adjuvant to Biofilm Formation?
As mentioned above, we were particularly interested in identifying environmental triggers of biofilm formation. While our results with exogenous protein are provocative in this regard, we cannot identify milk protein supplement as a trigger as opposed to an adjuvant in biofilm formation. The experiments described in Figure 5 & Figure 6 clearly indicate that the addition of exogenous protein increased cell concentration within the biofilm matrix (and biofilm biomass as measured with crystal violet) as well as the concentration of matrix protein in A. salmonicida and F. columnare. A. salmonicida was particularly efficient at incorporating protein into the matrix, increasing 80-fold over controls lacking milk protein. Interestingly, the primary strains of this study, P. aeruginosa, A. salmonicida, F. columnare and S. marcescens, produce extracellular proteases when grown on R2A or TSA plates with 5% milk protein (data not shown). Other isolates of these strains have a well-documented history of producing extracellular proteases [27]- [33]. Consistent with this was our observations in Figure 6 that when cultivated in microtiter plates for biofilm production, both P. aeruginosa and F. columnare reduced the opacity of exogenous protein in the media, indicating that extracellular proteases were actively degrading milk protein under the conditions of our biofilm test. However, A. salmonicida showed no such activity in broth but did add an abundance of protein to the biofilm matrix, suggesting that exogenous protein was at least a biofilm adjuvant for A. salmonicida. Concluding that exogenous protein is the environmental trigger for F. columnare biofilm formation is consistent with the complete absence of detectable pelagic growth in broth supplemented with the milk protein but with concurrent

Protein, Proteases and Virulence
Some proteases are identified as virulence factors in pathogens including Serratia [29] and Pseudomonas aeruginosa [30] [33]. The simplistic view of these extracellular proteases is that they are foraging for nutrients and clear habitats to occupy as well as impeding host immune responses that are protein based.
While extracellular proteases have been linked to biofilm formation in Enterococcus [34] [35] [36], from our observations it is unclear if extracellular proteases influence the formation of biofilm in P. aeruginosa, F. columnare, S. marcescens and A. salmonicida, under our experimental conditions. With a simple plate assay, we can detect extracellular proteases in these strains but the response to exogenous protein in the production of biofilm is strain specific and Aeromonas does not appear to degrade MP in broth when testing for biofilm.
Whether or not the proteases generate small peptides that are triggers or adjuvants of biofilm production remains to be determined.

The Host-Pathogen Evolutionary Dance
The analogy of an arms race has been used repeatedly for host-pathogen interactions as they evolve over time [37] [43]. Based on the results from our S. marcescens strains we would predict that biofilm would be surfactants that facilitate gas exchange on the mucosal surface [44]. These proteins can represent up to 10% of the dry weight of bronchial lavage fluid [45].
Of particular interest are SP-A and SP-D, now recognized as collectins, that participate in host defense along with their role as surfactants. Both bind bacterial LPS and in addition, SP-D binds peptidoglycan. These proteins have also been implicated in clearance of pathogens, activation of macrophages, modulation of inflammatory response and regulation of innate immunity functions in the lung [44] [45] [46]. We posit that the second virulence strategy of Serratia (and Aeromonas, and F. columnare) is the sequestration of proteins from the environment of their host, into the biofilm matrix. This is consistent with the biofilm matrix as a multifunctional extracellular "organ" of a bacterial consortium [47].
Incorporation of substantial amounts of SP-A into the biofilm as a structural component would locally reduce its concentration in alveolar fluid and mute the host's immune response. Targeting SP-A has been previously documented for P. aeruginosa [48].
In demonstrating the substantial influence of exogenous protein on biofilm productivity we hope that this stimulates further work on this aspect of biofilm formation. Responses to exogenous protein appeared to be strain specific, suggesting that the different environments in which these strains are colonizing may have a range of exogenous protein concentrations to which cognate strains have adapted. Protein and/or peptides in the concentrations ranges where we have detected enhanced biofilm formation would saturate protein binding sites on the cell surface. Some of these sites, as in the case of Enterococcus [34] [49] [50], are linked to two-component regulatory systems, hence exogenous protein may be an environmental trigger for biofilm formation.

Conclusion
We have demonstrated that relatively high concentrations of exogenous protein stimulate robust biofilm formation in S. marcescens, Aeromonas sp. and F. columnare. This attribute is species-specific as P. aeruginosa and several E. coli strains did not respond similarly. Exogenous protein in the growth medium increased both the cell counts and the amount of protein in the biofilm. The formation of biofilm by select strains of bacteria in response to exogenous protein may have survival value, particularly in the microenvironments of hosts where protein concentrations can be high.

Ethical Approval
This article contains no studies in which humans or metazoans were used as subjects.

Conflicts of Interest
The authors declare that they have no conflict of interest.