Combined Application of Bacterial Predation and Violacein to Kill Polymicrobial Pathogenic Communities

Violacein is a bisindole antibiotic that is effective against Gram-positive bacteria while the bacterial predator, Bdellovibrio bacteriovorus HD100, predates on Gram-negative strains. In this study, we evaluated the use of both together against multidrug resistant pathogens. The two antibacterial agents did not antagonize the activity of the other. For example, treatment of Staphylococcus aureus with violacein reduced its viability by more than 2,000-fold with or without B. bacteriovorus addition. Likewise, predation of Acinetobacter baumannii reduced the viability of this pathogen by more than 13,000-fold, regardless if violacein was present or not. When used individually against mixed bacterial cultures containing both Gram-positive and Gram-negative strains, violacein and B. bacteriovorus HD100 were effective against only their respective strains. The combined application of both violacein and B. bacteriovorus HD100, however, reduced the total pathogen numbers by as much as 84,500-fold. Their combined effectiveness was also demonstrated using a 4-species culture containing S. aureus, A. baumannii, Bacillus cereus and Klebsiella pneumoniae. When used alone, violacein and bacterial predation reduced the total population by only 19% and 68%, respectively. In conjunction with each other, the pathogen viability was reduced by 2,965-fold (99.98%), illustrating the prospective use of these two antimicrobials together against mixed species populations.

Microscopic imaging. Microscopic images were obtained using a Zeiss LSM 780 NLO microscope. To visualize the bacterial strains, dyes were used. Prior to treatment with either violacein or B. bacteriovorus HD100, each of the bacterial cultures were mixed with 6 µM Syto-9 (Invitrogen, USA). This is a live stain and all viable bacterial cells were fluorescently green afterwards. After washing the cells with HEPES to remove any extra dye, they were exposed to either the predatory cells (PPR of 0.1) or 20 mg/ml of violacein in HEPES. After one hour, propidium iodide (Invitrogen, USA) was added to the bacterial cultures to a final concentration of 30 µM. This dye is a dead stain and labels any non-viable bacterial cells red. After 30 minutes at room temperature, the cells were pelleted (16,000 × g, 5 min), washed and resuspended in HEPES before being imaged.

Antibiotic resistance determination.
To determine the antibiotic resistant nature of the bacterial strains, we used the same protocol as described previously 18 . Strain resistance was determined using the latest breakpoint tables available at the European Committee on Antimicrobial Susceptibility Testing (EUCAST) website (http:// www.eucast.org/clinical_breakpoints/). Spot viability assay. To compare the activity of conventional antibiotics and the dual treatment protocol used in this study, we tested representative bacteriostatic and bactericidal antibiotics, i.e., chloramphenicol and gentamycin, respectively. For the assays, a concentrated dosage that was 10-fold higher than the typical dose used in labs was employed. As such, the bacterial community was exposed to either 350 mg/l chloramphenicol, 500 mg/l gentamicin or both. After treatment, the cells viabilities were evaluated using a spot viability plate. For this, the exposed cultures were serially diluted into LB media and 10 µl was spotted on LB agar within a square petridish (SPL, Korea) and incubated for 24 hour at 37 °C. The extent of growth was used as an indication of viability.
Statistical analysis. In this research, all assays were accomplished at least three replicates and the standard deviations among the samples are indicated with error bars on the graphs. Statistical analysis was done using Student's t-test to compare two sets of results and the statistical significance was marked on the graphs using the marks *, **, *** for p-values of less than 0.05, 0.01, and 0.001 respectively. For comparing three or more data sets, analysis of variance (ANOVA) tests were performed followed by the Tukey post-hoc test. Statistically, significantly different groups using a p-value < 0.05 are shown on the graphs using letters (a, b, c and d).

Activity Spectrum of Violacein and Bacterial Predation.
Violacein is quite active against S. aureus, killing more than 99% of the culture when added at a concentration of 20 mg/L or greater 18,21,23 . Using this concentration, we tested the activity of violacein against six different strains (Tables 1 and 2), including three Grampositive and three Gram-negative species. As shown in Fig. 1a, only the Gram-positive cultures were sensitive to violacein, with each showing significant viability losses after 24 hours. In contrast, none of the Gram-negative cultures were negatively impacted by violacein (Fig. 1a). The A. baumannii and K. pneumoniae strains used throughout this study were multidrug resistant clinical isolates ( Table 2). The results with B. bacteriovorus HD100 stood in stark contrast as only the Gram-negative strains experienced a loss in their viabilities (Fig. 1b). These results clearly illustrate the different activity spectrums for both antimicrobials and highlight the limitations of each. This dichotomy is further demonstrated in Fig. 1c where S. aureus or E. coli were imaged after being exposed to either. In agreement with Fig. 1a

Violacein and B. bacteriovorus Do Not Adversely Impact Each
Other's Activity. The above data shows violacein was not active against the three Gram-negative strains. Given B. bacteriovorus HD100 is also Gram-negative, we speculated that violacein would also not be harmful towards this predator or its activity. Figure 2a shows that this was the case when A. baumannii was used as the prey strain. In both cases, the A. baumannii populations were similarly reduced by more than 4-log due to the activity of B. bacteriovorus HD100. Similarly, the activity of violacein was not thwarted by the presence of B. bacteriovorus HD100, as illustrated in Fig. 2b. This figure shows reduction of the S. aureus populations was comparable whether or not the predator was added.
The Activities of Violacein and B. bacteriovorus are Compatible with Each Other. We next tested the possibility of using both treatments together against cultures containing both a Gram-positive and a Gram-negative strain. For these tests, we selected S. aureus (average initial population: 4.2 × 10 8 CFU/ml) and either A. baumannii or K. pneumoniae (average initial population: 7.9 × 10 8 CFU/ml).
As shown in Fig. 3a, both antimicrobials were active against their respective pathogen: treatment with violacein alone killed only S. aureus while the use of B. bacteriovorus HD100 by itself only reduced the A. baumannii population. As a result, the total pathogen number did not decrease very much, i.e., only 51% with violacein and 69% with B. bacteriovorus HD100. When violacein and B. bacteriovorus HD100 were used together, however, both pathogens were significantly killed and the total pathogen numbers were reduced by 99.8% (620-fold). Similar results were obtained when S. aureus and K. pneumoniae were cultured together (Fig. 3b) as individual treatments with violacein and B. bacteriovorus HD100 led to a 41% and 81% loss in total pathogens, respectively, while a dual treatment caused a 99.999% (84,500-fold) reduction in the pathogen viability.
Dual treatment tests were also performed using either A. baumannii or K. pneumoniae alongside a different Gram-positive bacteria, i.e., S. epidermidis (Fig. 4). In both cultures, the viability of each pathogen was reduced by more than 99.3%.

Application of the Dual Treatment to a Complex Microbial
Community. Subsequently, we tested if a dual treatment is also effective against a complex microbial consortium harboring four different human pathogens, i.e., S. aureus, A. baumannii, B. cereus and K. pneumoniae (Fig. 5). The population of each pathogenic strain  is listed in Table 3, with a total initial population of 5.9 × 10 8 CFU/ml. After treatment with either B. bacteriovorus HD100 or violacein alone, the Gram-negative and Gram-positive strain viabilities decreased, respectively, but we could not reliably measure the impact due to the overwhelming presence of the surviving strains. The use of violacein alone resulted in a 19% reduction of the total number of pathogens while B. bacteriovorus HD100 led to a 68% loss (3.1-fold). As shown in Fig. 5, when both antimicrobials were used, the number pathogens was reduced by 99.96% (2,970-fold).

Dual Treatment Impacts on Multidrug Resistant Bacterial Populations. As a final test using both
predatory bacteria and violacein together, a test culture containing multidrug resistant S. aureus (Table 2) was prepared. The other strain was the same A. baumannii as above, which is also multidrug resistant. Both of these strains are resistant to numerous antibiotics, as listed in Table 2, and the initial population of each pathogen within the cultures was 5.2 × 10 8 and 3.2 × 10 8 CFU/ml, respectively. For comparison, we also treated the mixed pathogen culture with a concentrated blend of gentamicin with chloramphenicol (Fig. 6). The image show the use of B. bacteriovorus HD100 and violacein was much more effective at killing the pathogens than the use of gentamicin and chloramphenicol, even though the concentrations of the two antibiotics were much higher than what is typically used.

Discussion
Due to the universal spread of antibiotic resistance, it is projected that the number of human mortalities each year resulting from antimicrobial resistant pathogens will surpass that caused by cancer by 2050 36 . Two alternative antimicrobials currently being studied by various groups are violacein and predatory bacteria 18,23,25,28,37,38 . In addition to their antibiotic activities, both also show promising toxicology results, a key characteristic for broad spectrum antibiotics 8,39 ; in tests with mice, violacein was non-toxic at concentrations as high as 1 mg/kg 40 while B. bacteriovorus was not toxic towards several different animal hosts [41][42][43] or in human cell cultures [44][45][46] . Treatment of S. aureus with violacein not only led to a significant loss of viability but also caused a substantial degree of aggregation. Similar results were seen in other studies where S. aureus was exposed to either violacein 22 or galangin 47 . In the former study, violacein was shown to disrupt the membrane integrity of S. aureus, which led to a rupturing of the cell membrane and leakage of the cellular components. It was hypothesized by Cushnie et al. (2007) that galangin similarly damages the membrane and exposes the hydrophobic regions located within the phospholipid bilayer, which then interact and cause aggregation 47 . The same process likely occurs with violacein, which would account for the cell aggregates seen here. The activity of violacein also helps to explain the inherent resistance of Gram-negative bacterial strains since the outer membrane would act like a sponge and absorb violacein, protecting the inner membrane. With B. bacteriovorus, however, only E. coli was killed while the S. aureus cells remained viable (Fig. 1c). This is a clear demonstration that S. aureus is not a prey for B. bacteriovorus and reaffirms the long held knowledge that BALOs only predate and consume Gram-negative strains [29][30][31][32] , despite claims in one study that S. aureus is also attacked 48 . Although the mode of activity for violacein has recently been resolved, much still remains unanswered about how predatory bacteria recognize, enter and kill their prey and the processes involved in each of these steps.
The results above illustrate inherent limitations for both violacein and B. bacteriovorus HD100; violacein is largely effective against Gram-positive bacteria while B. bacteriovorus HD100 attacks only Gram-negative strains. Given the different activity spectrums for these two antimicrobials, we proposed they might be complementary and used alongside each other against polymicrobial populations. However, it was not known if predatory bacteria and violacein negatively impact the activity of the other. Here we show that this was not the case -violacein was just as effective against S. aureus regardless of whether B. bacteriovorus was present while predation of A. baumannii proceeded just as successfully in the presence or absence of violacein. This suggested that they can be used together and hinted at their potential use to eradicate polymicrobial populations.
Polymicrobial infections are a serious problem since complex bacterial communities can benefit from the properties of each of the members. An example of this is the protection afforded to a susceptible population by a resistant member of the community, such as through enzymatic inactivation of the antibiotic 5,6 . Another possibility is the "inoculum effect", whereby the antibiotic activity is diluted out by the presence of other bacteria, which act as "sinks" for the antibiotic and limit its effectiveness against susceptible microbe populations 7 . We were concerned that this may be a problem for violacein. Violacein is a hydrophobic compound and attacks cellular membranes, implying the presence of naturally-resistant Gram-negative bacterial cells may act as a "sink" for this antibiotic and dilute its impact on S. aureus or other Gram-positive pathogens. The results, however, suggest that this is not a problem as violacein at the concentration used was as effective at killing S. aureus regardless if A. baumannii or K. pneumoniae were present.
Likewise, although they are not attacked by B. bacteriovorus, Gram-positive strains may act as decoys and slow the predation of susceptible Gram-negative bacterial strains 49 . As B. bacteriovorus does not display a clear chemotaxis towards prey cells 50 , it is generally thought that interactions between B. bacteriovorus HD100 and its prey are random events that occur as the predator swims within the media. By extrapolation, this means B. bacteriovorus (a) Impact of 20 mg/l violacein alone, 1 × 107 PFU/ml B. bacteriovorus HD100 alone or both together against a mixed culture containing S. aureus and A. baumannii. The viabilities were measured after 24 hr. Each of the antimicrobials was effective against their respective pathogen when used alone, but together led to a significant loss in the total pathogen numbers. a, b and c = p < 0.05 (n = 3). (b) Impact of 20 mg/l violacein alone, 1 × 107 PFU/ml B. bacteriovorus HD100 alone or both together against a mixed culture containing S. aureus and K. pneumoniae. The viabilities were measured after 24 hr. Each of the antimicrobials was effective against their respective pathogen when used alone, but together led to a significant loss in the total pathogen numbers. a, b, c and d = p < 0.05 (n = 3).
will also come in contact and interact with non-prey bacterial cells. Such interactions have been observed with both Neisseria gonorrhoeae 51 and Bacillus subtilis 49 , although neither is predated upon. Stemming from this reasoning, Wilkinson (2001) developed the first mathematical model to describe the impact decoys may have on B. bacteriovorus predation rates and efficacies using a double-Monod framework based upon a continuous flow system 52 . Hobley et al. (2006) followed this with experimental data using B. subtilis as a decoy in predation cultures with E. coli S17-1 as the prey, and developed their own model based upon the Lotka-Volterra equations 49 . They found predation rates were slower during the first seven hours when B. subtilis was present as a decoy at a   Table 3). The viabilities of the culture as a whole were measured after 24 hr. Each of the antimicrobials led to mild but significant losses when used alone, but together led to a 99.96% loss in the total pathogen numbers (*p < 0.05; ***p < 0.001) (n = 3).
ratio of approximately 1:2 (decoy:prey). In contrast, the study by Van Essche et al. (2010) found the presence of a decoy, i.e., Gram-positive Actinomyces naeslundii ATCC 12104, at a 14-fold excess had no impact on predation of Aggregatibacter actinomycetemcomitans ATCC 43718 53 . Similar results were reported by Loozen et al. (2014) in their study with a six-member bacterial community 54 . The six species present within the community were all oral bacteria and included four Gram-negative and two Gram-positive species. Along with the Gram-positive bacterial strains, two of the Gram-negative strains, i.e., Porphyromonas gingivalis and Prevotella intermedia, were not predated upon by B. bacteriovorus HD100. However, A. actinomycetemcomitans and Fusobacterium nucleatum are both prey and the presence of the four other strains did not hinder their predation.
Clearly some non-prey bacteria can act as decoys and slow down predation initially 49 , while others have no impact 53,54 . Here, we found the presence of S. aureus did not negatively impact predation. Similar results were seen with S. epidermidis. One clear difference between S. aureus and S. epidermidis, though, was the viabilities after a dual treatment with K. pneumoniae. With S. epidermidis, the viabilities were comparable with those seen after single treatments in Fig. 1. With S. aureus, however, the viabilities of both pathogens were reduced further by the dual treatment (Fig. 3).
The activities of violacein and B. bacteriovorus HD100 were also evaluated using a four-member pathogen culture, with positive results. As with the two-strain cultures, the presence of the four different strains did not significantly deter the activities of either antimicrobial when used alone (Table 3) nor when used together (Fig. 5). In the latter case, the number of viable pathogens was reduced by more than 3-log and stands as a clear demonstration the potential these two antimicrobials have when used together.
Not only are violacein and B. bacteriovorus HD100 effective against their respective classes of bacteria, they are also both active against multidrug resistant pathogens. For violacein, the minimum inhibitory concentrations for a wild-type S. aureus and four other strains, including a clinical isolate that is resistant to seven different antibiotics, were identical 18 . Along the same lines, the antibiotic resistant nature of the prey had no obvious impact on the ability of B. bacteriovorus to attack 28 , while predation reduces the presence of the antibiotic resistance marker by degrading the prey DNA 55 . Consequently, violacein and B. bacteriovorus are both active against multidrug resistant pathogens. This was illustrated here in experiments with a mixed culture containing multidrug resistant strains of S. aureus and A. baumannii. For comparison, we also exposed this culture in parallel to high doses of chloramphenicol, gentamicin or both of these antibiotics. With chloramphenicol and/or gentamicin, the culture viabilities were not significantly impacted, even though the concentration of each antibiotic was 10-fold higher than commonly employed in the lab. In contrast, a dual treatment with violacein and B. bacteriovorus HD100 led to a 2-to 3-log loss, a result that is indistinguishable with that in Fig. 3 with the wild-type S. aureus. Figure 6. Dual treatment of a culture containing multidrug resistant strains of A. baumannii and S. aureus. A mixed culture containing A. baumannii alongside S. aureus CCARM 3090 was treated with either chloramphenicol, gentamicin, these two antibiotics together (Chl + Gen) or the dual treatment (Predation & Violacein). The concentrations of the two antibiotics were 10x higher than commonly used in lab experiments. After 24 hr, the cultures were serially diluted and 10 µl spotted onto an agar plate. The image here shows the resulting growth, illustrating the much higher degree of killing obtained with the dual treatment protocol.  Table 3. Impact of predation or violacein on the survival of the four pathogens within a mixed population. a N.D. -Not detected on the agar plates with a 10 7 dilution.
In conclusion, this study evaluated the activities of violacein and B. bacteriovorus HD100 when used together against polymicrobial cultures. Violacein, being active primarily against Gram-positive strains such as S. aureus, and B. bacteriovorus HD100, which attacks only Gram-negative bacteria, were found to be compatible and not diminish the activity of the other. This was demonstrated in cultures containing two and four different bacterial species, including both Gram-positive and Gram-negative strains. When used alone, the overall viabilities decreased marginally (less than 1-log reduction) but, when violacein and B. bacteriovorus HD100 were used together the number of viable pathogens was reduced by 3-log or greater, even when multidrug resistant pathogens were present.