Intra-ocular Predation of Fluoroquinolone-Resistant Pseudomonas aeruginosa and Serratia marcescens by Predatory Bacteria

Endogenous endophthalmitis caused by Gram-negative bacteria is an intra-ocular infection that can rapidly progress to irreversible loss of vision. While most endophthalmitis isolates are susceptible to antibiotic therapy, the emergence of resistant bacteria necessitates alternative approaches to combat intraocular bacterial proliferation. In this study the ability of predatory bacteria to limit intraocular growth of Pseudomonas aeruginosa, Serratia marcescens, and Staphylococcus aureus was evaluated in a New Zealand White rabbit endophthalmitis prevention model. Predatory bacteria Bdellovibrio bacteriovorus and Micavibrio aeruginosavorus were able to reduce proliferation of keratitis isolates of P. aeruginosa and S. marcescens. However, it was not able to significantly reduce S. aureus, which is not a productive prey for these predatory bacteria, suggesting that the inhibitory effect on P. aeruginosa requires active predation rather than an antimicrobial immune response. Similarly, UV-inactivated B. bacteriovorus were unable to prevent proliferation of P. aeruginosa. Together, these data suggest in vivo predation of Gram-negative bacteria within the intra-ocular environment.


Introduction
Alternative approaches to antibiotics have become a major focus of research due to increasing resistance among bacterial pathogens.One avenue of research for this purpose is the use of predatory bacteria as a "living antibiotic" [1][2][3][4].Predatory bacteria such as Bdellovibrio bacteriovorus and Micavibrio aeruginosavorus use a wide range of Gram-negative bacteria as a food source including numerous pathogens [5].It has been demonstrated that predatory bacteria are indifferent to the antibiotic-resistance status of their prey [6,7].Moreover, it has been demonstrated that predatory bacteria are non-toxic to mammalian cells and in animal models and have the ability to attenuate Gram-negative bacteria several in vivo models, including but not limited to airway and oral infections in rodents and central nervous system infection of zebrafish hindbrain ventricles [8][9][10][11][12][13][14][15][16].
Recent studies have evaluated the use of predatory bacteria for treatment of ocular surface infections [17][18][19].These have demonstrated some efficacy against Escherichia coli [17] in a mouse model and Pseudomonas aeruginosa in rabbit models including the prevention of corneal perforations [19].However freeze-dried B. bacteriovorus proved to be ineffective in treating Moraxella bovis in a large animal keratoconjunctivitis model [20].The prior studies have focused on the anterior portion of the eye.In this study the goal was to evaluate the use of predatory bacteria in preventing bacterial proliferation in an endophthalmitis model using New Zealand White rabbits.
Endogenous endophthalmitis is frequently caused by Gram-negative bacteria that travel from the blood stream, through the blood-brain barrier and into the posterior portion of the eye [21,22].There bacteria can rapidly proliferate, induce a damaging immune response, and cause damage to the tissues crucial for vision such as the retina.This can lead to severe surgical interventions such as removal of the eyes content (evisceration) or the entire globe (enucleation).
Endophthalmitis caused by Gram-negative bacteria such as Klebsiella pneumonia is especially prominent in Asia [21][22][23].At our hospital the most frequent Gram-negative bacteria that cause endophthalmitis are Pseudomonas aeruginosa and Serratia marcescens, whereas bacteria from the Staphylococcus and Streptococcus genera are the most prominent overall [24].
In this study, the ability of B. bacteriovorus strain HD100 and M. aeruginosavorus strain ARL-13 were evaluated for the ability to prevent growth of bacterial pathogens within the eye and reduced proliferation of P. aeruginosa was reproducibly observed.Furthermore, whether inhibition of P. aeruginosa was due to active predation or an immune response triggered by predatory bacteria, was tested using ultra violet light inactivated B. bacteriovorus.Data suggests that both tested predatory bacteria prey upon P. aeruginosa within the vitreous humor of the rabbit eye.

Materials and Methods
Strains and bacterial growth conditions.The predatory bacteria Bdellovibrio bacteriovorus HD100 (ATCC 15356) [25,26] and Micavibrio aeruginosavorus ARL-13 [27] were used in the study.Predator lysates (co-cultures) were prepared as previously reported [28].Predators B. bacteriovorus and M. aeruginosavorus were incubated at 30°C with E. coli strain WM3064 [29] (1 × 10 9 CFU/ml) for 24 and 72 hours respectively.The resulting lysates were passed twice through a 0.45-μm pore-size filter (Millipore, Billerica, MA, USA).Predators were washed with phosphate buffered saline (PBS) and concentrated by three sequential 45-minute centrifugations at 29,000 x g.Finally, predator pellets were suspended in PBS to reach a final concentration of 1.7 × 10 10 PFU/ml B. bacteriovorus and 3.5 × 10 9 PFU/ml M. aeruginosavorus.B. bacteriovorus UV inactivated cells were prepared by placing 1 ml of purified predator sample in a well of a 12 well plate and radiating the plate 20 times on the Auto Cross Link setting, while mixing the sample in-between each cross-link (UV Stratalinker 1800; Stratagene, San Diego, CA, USA).Lack of predator viability was confirmed by PFU plating, in which no plaque had developed.
Structural integrity of the predator cells was confirmed by light microscopy (1000x magnification).
Pathogens used for this study were P. aeruginosa strain PaC, which is a fluoroquinolone resistant ocular isolate [30].S. marcescens strain K904 is a keratitis ocular isolate [31].S. aureus strain E277 is an endophthalmitis isolate from The Charles T. Campbell Laboratory deidentified strain collection.These were streaked to single colonies on TSA blood agar plates from stocks stored at -80˚C.Single colonies were grown with aeration in lysogeny broth for 16-18h and then adjusted in PBS to an inoculum of 5-10 x 10 3 CFU in 25 µl for injection into eyes.

Animal experiments
This study was approved by the University of Pittsburgh's Institutional Animal Care and Use Committee (Protocol #15025331 "The Use of Predatory Bacteria to Treat Ocular Infections") and conformed to the ARVO Statement on the Use of Animals in Ophthalmic and Vision Research.
Female New Zealand White rabbits weighing 1.5-2 kg were received from Charles River Laboratories' Oakwood Rabbitry.Following systemic anesthesia with 40 mg/kg of ketamine & 4 mg/kg of xylazine administered intramuscularly, and topical anesthesia with 0.5% proparacaine, the right eyes were inoculated in the vitreous via pars plana injection with 25 μl of the bacterial suspension or PBS depending on the group.Immediately following injections of the bacteria, the same eyes were injected with 0.1 ml (100 μl) of the predatory bacteria or PBS per the experimental groups above.Injection of the predatory bacteria or PBS was performed in a different location than the pars plana injection of bacteria.Rabbits were treated with 1.5 mg/kg ketoprofen, administered intramuscularly, to reduce pain after recovery from anesthesia.At 24 hours after inoculation, the rabbits were examined using a slit lamp and imaged.Rabbits were euthanized with an overdose of intravenous Euthasol solution following systemic anesthesia with ketamine and xylazine administered intramuscularly.Vitreous humor taps were performed on the infected eyes by inserting a 23-gauge needle attached to a 1 cc syringe into the vitreous chamber about 4 mm from the limbus and removing about 0.2 -0.3 ml of fluid.The vitreous humor was transferred to a sterile tube and placed on ice.Standard colony counts determinations were performed on the vitreous samples using 5% sheep blood agar plates and incubated overnight at 37 o C. Cytokines were detected from vitreous humor using commercial kits IL-1β (Sigma-Aldrich), TNFα (Thermo Scientific).Clinical signs of endophthalmitis were determined by a masked reviewer and used a 10-point scale with a 0-2-point score given for discharge, redness of eye, chemosis, anterior eye involvement, and hypopyon.

Statistical analysis.
Data was analyzed using GraphPad Prism software.All experiments were repeated at least twice.Kruskal-Wallis with Dunn's post-test was used to compare medians and ANOVA with Tukey's posttest was used to analyze means.

Results
Predatory bacteria prevent proliferation of P. aeruginosa.P. aeruginosa strain PaC is a fluoroquinolone resistant ocular clinical isolate that was chosen because it is susceptible to predation by predatory bacteria [32] .The ocular vitreous chamber of NZW rabbits were injected with P. aeruginosa (5.0 x 10 3 CFU) followed by B. bacteriovorus (4.3 x 10 8 PFU) or M. aeruginosavorus (8.8 x 10 7 PFU).Controls included injection of vehicle (PBS) or individual microbes.At 24 hours eyes were examined by a slit lamp, imaged, and graded for clinical signs of inflammation.Eyes injected with predatory bacteria had increased inflammatory scores that were not significantly higher than the vehicle only control (PBS + PBS), Figure 1A.Eyes injected with P. aeruginosa only had a notable increase in inflammatory score (p<0.0001), by comparison, ocular inflammation was reduced in eyes injected with both P. aeruginosa and either predatory bacteria.Although, the clinical signs of inflammation were not significantly different between the PaC alone and the PaC with predatory bacteria, the eyes injected with both PaC and B. bacteriovorus HD100 were not significantly worse than the vehicle control eyes (Figure 1A).Representative images of eyes from each group are shown in Figure 1B.
The trend toward reduced inflammation following addition of predatory bacteria to P. aeruginosa treated eyes correlated with a >95% reduction in the P. aeruginosa bacterial burden in eyes injected with both predatory bacteria and P. aeruginosa compared to P. aeruginosa alone, p<0.05 (Figure 1C).Unchecked, P. aeruginosa replicated from an inoculum of 5 x 10 3 to a burden 4.8 x 10 6 CFU in 24h in the eye.This was 17-fold and 25-fold higher than what it achieved with B. bacteriovorus HD100 and M. aeruginosavorus respectively.
Damage associated pro-inflammatory cytokine IL-1β levels followed a similar trend to the clinical scores and were largely unaffected by predatory bacteria alone, elevated with P. aeruginosa alone, and significantly mitigated in eyes with both predatory bacteria and P. aeruginosa (Figure 1D).A matching trend for pro-inflammatory cytokine TNFα was measured, although to a lesser extent where the predatory bacteria did not significantly reduce the cytokine levels compared to the P. aeruginosa alone (Figure 1E).

Predatory bacteria reduce intraocular proliferation of S. marcescens.
Figure 2 shows experimental data in which predatory bacteria were evaluated for their ability to prevent intraocular growth of S. marcescens.Clinical presentation of S. marcescens infected eyes was less severe than that of P. aeruginosa and not notably changed by the addition of predatory Similar to clinical presentation, IL-1β levels were induced less by S. marcescens than P. aeruginosa and were unaffected by co-incubation with predatory bacteria (Figure 2D).By contrast, TNFα levels were higher in S. marcescens infected eyes than P. aeruginosa and were significantly reduced when S. marcescens was co-incubated with M. aeruginosavorus (Figure 2E).

Predatory bacteria do not prevent S. aureus replication in the eye
Although B. bacteriovorus was shown to attach to the Gram-positive bacterium S. aureus, S. aureus does not support B. bacteriovorus or M. aeruginosavorus full predation or growth cycle [25,[33][34][35].Inclusion of this bacteria should therefore give insight into whether the prevention of P. aeruginosa and S. marcescens intraocular growth by predatory bacteria is due to active predation or other mechanism, such as predatory bacteria-induced biosynthesis of antimicrobials or immune response triggered by the presence of the predators.The clinical evaluation showed modest inflammation with S. aureus alone that was not significantly altered with addition of predatory bacteria (Fig 3A).Unchallenged by predatory bacteria, the S. aureus strain was able to grow 87-fold from 1.0 x 10 4 to 8.9 x 10 5 in 24 hours.B. bacteriovorus was associated with a 2.26-fold reduction in S. aureus burden, whereas M. aeruginosavorus reduced less than 2-fold (19%) and neither of these changes were significantly different by ANOVA (Figure 3B).IL-1β levels were not reduced by the presence of predatory bacteria (Figure 3C).

Viable predatory bacteria are required to reduce intraocular proliferation of P. aeruginosa
To further test whether the reduction in P. aeruginosa was mediated by predation, UVinactivated B. bacteriovorus were tested.The UV-inactivated bacteria correlated with modestly increased clinical scores (Figure 3D).Importantly, the P. aeruginosa inhibition by live B. bacteriovorus was absent for eyes using the UV-inactivated predator (Figure 3E).Similarly, the reduction in P. aeruginosa induced IL-1β was reduced by live but not inactivated predatory bacteria (Figure 3F).

Discussion
This proof-of-principle study determined whether predatory bacteria could prey upon bacteria within the eye.This study demonstrated a clear ability of both tested predatory bacteria to reduce P. aeruginosa proliferation within the eye.While the reduction was less with S. marcescens, predation may be masked by the remarkably fast replication of S. marcescens strain K904 in the eye.While two different experiments differed in the extent to which P. aeruginosa caused clinical inflammatory signs (Figure 1A versus Figure 3A), the IL-1β levels were reproducibly reduced when predatory bacteria were added to P. aeruginosa infected eyes.IL-1 is associated with ocular tissue damage that triggers production of pro-inflammatory cytokines and reduces barrier function [36].IL-1β is a marker of bacterial endophthalmitis associated with productive antimicrobial host-responses [37][38][39].Unlike the tested pathogens, predatory bacteria failed to significantly induce intraocular cytokine levels above that of eyes injected with the PBS vehicle.This is in agreement with prior in vitro and in vivo studies [12,13,19,20,28,32,[40][41][42] and, at least for B. bacteriovorus due to its unusual outer membrane composition and membranesheathed flagellum which reduce TLR4 and TLR5 activation [43,44], which are major mediators of inflammation in bacterial endophthalmitis [45,46].
Data from this study support the hypothesis that predatory bacteria actively prey upon P. aeruginosa and possibly S. marcescens in the vitreous chamber of the eye.This is based on two observations.First is that predatory bacteria failed to significantly reduce the proliferation of a bacteria that they are unable to prey upon, which suggests that an antimicrobial host response is not strongly induced by predatory bacteria.The second being that UV-inactivated predatory bacteria were unable to reduce intraocular P. aeruginosa.While innate ocular defense mechanisms such as neutrophils likely play a role in inhibition of pathogen growth, the reduced proliferation of P. aeruginosa and S. marcescens in vitreous chamber of the eye in the eyes coincubated with predatory bacteria is most likely due to active predation.Although our findings do support the hypothesis that active predation is required for the effect seen in Gram-negative pathogens, one cannot rule out that an elevated immune response might be triggered following predation as prey cell debris accumulates.A synergistic effect between the immune system and Bdellovibrio was previously suggested in a study monitoring predation of Shigella using a Zebrafish infection model [16].
In conclusion, this study strongly suggests that predatory bacteria can kill bacteria in the vitreous chamber of the eye.However, on its own would likely be insufficient to treat a clinical endophthalmitis.Further studies to determine whether predatory bacteria coupled with antibiotics would improve antimicrobial outcomes as has been previously tested in vitro [47].
bacteria (Fig 2A).Representative images are shown in Figure 2B.S. marcescens was able to grow 85,800-fold from 5.6 x 10 3 inoculum to 4.8 x 10 8 intraocularly in 24 hours.This was reduced 2.2 and 2.7-fold by B. bacteriovorus and M. aeruginosavorus respectively, and was significantly reduced by B. bacteriovorus (Fig 2C).

Figure 1 .
Figure 1.Effect of predatory bacteria on inflammation and P. aeruginosa (PaC) proliferation in

Figure 2 .
Figure 2. Impact of predatory bacteria on S. marcescens (SM) proliferation and consequent

Figure 3 .
Figure 3. Differentiation between predation and induced antimicrobial host-response.A, D.