Predatory bacteria as living antibiotics – where are we now?

Antimicrobial resistance (AMR) is a global health and economic crisis. With too few antibiotics in development to meet current and anticipated needs, there is a critical need for new therapies to treat Gram-negative infections. One potential approach is the use of living predatory bacteria, such as Bdellovibrio bacteriovorus (small Gram-negative bacteria that naturally invade and kill Gram-negative pathogens of humans, animals and plants). Moving toward the use of Bdellovibrio as a ‘living antibiotic’ demands the investigation and characterization of these bacterial predators in biologically relevant systems. We review the fundamental science supporting the feasibility of predatory bacteria as alternatives to antibiotics.

Antibiotics have revolutionized modern medicine, but their effectiveness is threatened by the spread of multi-drugresistant bacteria for which there is no available treatment [1][2][3].Antimicrobial resistance (AMR), often driven by unnecessary antibiotic use, is a serious, global health and economic threat [4][5][6][7].Of particular concern is the declining antibiotic pipeline with limited antibiotics in development (in both number and diversity) to meet current and anticipated patient demands [8][9][10][11].Moreover, scientific and economic challenges have contributed to many large global pharmaceutical companies discontinuing their antibiotic development programmes [12], making the need for new approaches to infection control ever more exigent [13].
Over the last few years, fundamental scientific research into alternatives to antibiotics has gathered momentum.These include 'living antibiotics'; live agents such as bacteriophage (hereafter referred to as phage) or indeed their components [14], predatory bacteria and probiotics, and the competitive exclusion of pathogens [15,16].Whilst evaluation of all potential novel non-traditional approaches currently under investigation is beyond the scope of this review, their development and associated challenges has been the subject of recent publications and symposia [8,[16][17][18][19][20][21][22][23].Most new technologies, particularly those which are disruptive to an industry, are subject to many technical and regulatory challenges, which in the case of 'living antibiotics' include activity spectrum, pharmacokinetics, immune response, manufacturing issues (including production, stability, storage and delivery), regulatory and quality-control frameworks and market acceptance.Individually, these potential non-traditional antibiotics may never entirely fill the vacuum left by antibiotics, however, they may prove to be invaluable in complimenting conventional chemotherapy, either used separately or in combination with them.As such, it is imperative that the research community debates their use and builds scientific and clinical evidence prior to their requirement, and there are encouraging signs that this is happening [18].
One novel approach to treat infections is the use of living predatory bacteria, such as Bdellovibrio bacteriovorus, small Gram-negative bacteria found ubiquitously in soil and aquatic environments, that naturally invades and kills other Gram-negative bacteria.The feasibility of using Bdellovibrio therapeutically revolves around answering several important questions: Against which pathogens is Bdellovibrio effective?Is Bdellovibrio harmful to the patient either directly or indirectly via products of predation?and Do prey develop resistance to Bdellovibrio?Although predatory bacteria have been researched by the academic community since the 1960s, recent work supported by DARPA's visionary and multidisciplinary pathogen predators programme has driven fundamental scientific research towards experimentally addressing these questions [24].
While there are many different predatory bacteria [25], this review chiefly focuses on B. bacteriovorus and the fundamental science supporting its potential translation.

PREDATORY LIFESTYLE OF B. BACTERIOVORUS
B. bacteriovorus consumes its Gram-negative bacterial prey in a process that typically lasts 3-4 h.This predatory lifecycle is a complex process that is beginning to be understood at the molecular level [26][27][28][29][30][31][32][33].Initially, B. bacteriovorus recognizes, attaches to [34] and enters the prey cell, reinforcing, traversing and re-sealing its entry port [35].Invasion is accompanied by prey cell-rounding, establishment of a stable bdelloplast and, at this point, death of the prey cell [36].Through the sequential release of an arsenal of enzymes [37], B. bacteriovorus digest the prey and use the resulting nutrient pool to grow as a long filament.Synchronous division of this single filament produces an odd or even number of progeny cells, which each develop flagella or gliding engines (depending on conditions), then burst from the dead prey cell and begin to seek new prey.Depending of the size of the prey cell, and hence the nutrients available from inside the prey bacterium, an average of 4-6 B. bacteriovorus progeny are released per cell [38].Although recognized as an obligate predatory bacterium, B. bacteriovorus can switch to a host-independent lifestyle, demonstrating axenic growth on complete media [39].

WHY PREDATORY BACTERIA?
Several features of the B. bacteriovorus lifecycle and genome make it attractive as a potential therapeutic agent against Gram-negative bacterial pathogens.During the predatory lifecycle the prey cell is killed in a short time (<30 min) [40] and therefore the prey would have to express means of defence quickly enough to resist predation, something not yet seen [41].Unlike some antibiotics, which can cause a cascade of events that lead to bacterial autolysis and release of inflammatory molecules [42], B. bacteriovorus predation does not result in the initial lysis of the prey, as the prey contents are consumed from within a stable bdelloplast structure prior to lysis.In addition, there is no single receptor for prey recognition and attachment.After prey invasion there is upregulation, both in number and functional diversity, of prey-destructive enzymes [43] with potential genetic redundancy [44], suggesting that simple prey resistance to predation by B. bacteriovorus is unlikely to occur.

EFFECT OF ADMINISTRATION OF BDELLOVIBRIO ON THE HOST RESPONSE
In order for the therapeutic potential of Bdellovibrio to be realized, predation of Gram-negative pathogens must be fully characterized in biologically relevant systems.Such characterization must address issues such as the host response, toxicity, inflammation, tissue damage or the inhibition of wound healing.Some of these issues have been addressed using in vitro cell culture and in vivo animal models, evaluating the effect of predatory bacteria on both individual components of the immune system as well as the whole host.A number of human cell lines, including corneal-limbal epithelial cells, blood monocytes, macrophages, kidney epithelial cells, liver epithelial cells and spleen monocytes, have been exposed to different predatory bacteria, at a range of m.o.i. and duration (between 2 h and 24 h), and levels of pro-and antiinflammatory cytokines measured [45][46][47][48][49]. Common to these studies were the pro-and anti-inflammatory cytokines IL1B, TNFa, IL6, IL8 and IL10; cytokines known to be stimulated in response to bacterial outer membrane lipopolysaccharide (LPS) [50], and essential for host defence against pathogens.These studies show that while B. bacteriovorus are not silent in terms of their immune response, the levels of inflammatory cytokines produced following exposure were negligible or low; B. bacteriovorus is not as immune-stimulatory as Gramnegative pathogens tested alongside [45,46,49], perhaps in part due to a unique lipid A structure [45], and its possession of sheathed flagellum [51].Additional cell-viability imaging, cytotoxicity measurements and assessment of morphological changes to animal and human cultured cells on exposure to predatory bacteria preliminary indicate that B. bacteriovorus is non-toxic to human cells [46][47][48], although more studies are needed.Extending this for B. bacteriorvorus, Raghunathan and colleagues investigated the uptake, persistence and clearance of live predatory bacteria using a human macrophage cell line (U937 cells) [49].By exposing U937 cells to fluorescently labelled B. bacteriovorus, Raghunathan and colleagues were able to microscopically visualize and count live intracellular bacteria.In parallel experiments, predatory bacteria engulfed by the U937 cells were recovered and enumerated following experimental lysis of the U937 cells.Assessing Bdellovibrio numbers post-administration is challenging, but these experimental approaches demonstrated that high numbers of predatory bacteria were able to survive up to 24 h inside cells (typically 14 bacteria per U937 cell (microscopy)), thereby defining a period of persistence and potential ability to prey on intracellular pathogens.B. bacteriorvorus had no measurable effect on the viability of the eukaryotic cell, supporting and amplifying findings from previous work [46][47][48].In addition, by performing uptake experiments in the presence of pharmacological inhibitors, a role of the host actin cytoskeleton and its rearrangement in the uptake of B. bacteriorvorus was demonstrated.B. bacteriorvorus is ultimately trafficked via the phagolysosomal pathway, as evidenced by their targeting to acidic vacuoles [49].This is supported by similar observations in a murine macrophage cell line [52], and both are important observations to consider for administration of B. bacteriovorus as a therapeutic.These studies illustrate the potential for B. bacteriovorus to target intracellular pathogens, which may be inaccessible to many antibiotics and other biological control agents such as phage; although this latter point is under active investigation [53][54][55].These in vitro cellular experiments, along with animal models described below, have been important steps to investigate efficacy, lack of toxicity and potential predator bioavailability.
Early in vivo host response focussed on the viability of Bdellovibrio sp. in the intestines of endothermic and exothermic vertebrates.Westergaard and Kramer [56] investigated the viability and persistence of B. bacteriovorus strain MS7 experimentally administered in the intestines of catfish (Ictalurus punctatus), leopard frogs (Rana pipiens), mice and rabbits showing little or no recovery of B. bacteriovorus 24-48 h post-inoculation.B. bacteriovorus offered to mice via drinking water for 3 days could not be recovered at all from the intestinal tract.Likewise, either none or very few B. bacteriovorus were recovered from rabbit ileal loops approximately 24 h post-injection, regardless of whether axenic cultures were used, or Escherichia coli prey were co-injected.Importantly, B. bacteriovorus was shown to be non-pathogenic to the animals.This early work of Westergaard and Kramer has been extended by several groups using experimental animal models.Atterbury et al. found that B. bacteriovorus did not produce any noticeable signs of disease when chickens were dosed with axenic cultures of B. bacteriovorus HD100 [57].Administration of B. bacteriovorus was shown to be non-toxic and non-immunogenic to the ocular surface of rabbit eyes [58], and did not interfere with wound healing in a Galleria (wax moth larvae) model [46].These studies have been extended to include assessments of host morbidity, histopathology, levels of pro-and anti-inflammatory cytokines, predatory bacterial dissemination in the body and long-term assessments of general health in rat [59][60][61], mouse [52,62] and zebrafish larval models [63].Overall, in vivo administration of B. bacteriovorus in a range of animal models and via a number of routes of administration has demonstrated that they are not detrimental to the health of these animals.

EXAMINING THE PREY RANGE OF B. BACTERIOVORUS: IN VITRO AND IN VIVO PREDATION
In vitro efficacy of B. bacteriovorus has been demonstrated against a wide range of prey including bacteria associated with gut, oral, wound and ocular infections [46,[64][65][66] and bioterrorism agents [67].B. bacteriovorus successfully reduce pathogen numbers in both laboratory buffer and human serum [68] and against prey in biofilms [69 -71], which are often recalcitrant to antibiotic treatment [72].Importantly, a number of multi-drug-resistant human clinical isolates have been shown to be susceptible to predation by B. bacteriovorus [46,65,68,70,73,74]; this list encompasses a number of the E(S)KAPE pathogens for which new treatments are needed [6] including Gram-negative colistin-resistant isolates expressing mcr-1 [75].Work in vitro has also investigated the ratio of B. bacteriovorus to prey required for successful predation [68]; an important consideration for future in vivo experiments and clinical translation for use as a therapeutic if antibiotics fail to treat the infection.
While experimental in vitro systems, especially laboratory buffer or growth medium-rich environments may demonstrate predatory efficacy, they are an inadequate substitute for the complexities of in vivo therapy, which must take into account the host immune response and pathogen survival and subversion tactics [76][77][78].Animal infection models have been instrumental in assessing both the ability of B. bacteriovorus to prey upon Gram-negative bacteria in vivo, as well as addressing any host response (and therefore safety) to such administration (see Effect of administration of Bdellovibrio on the host response), adding to the growing evidence of the feasibility of using predatory bacteria for therapeutic applications, at least in animals and in the future extending this to humans.
Atterbury et al. [57] administered B. bacteriovorus HD100 orally to groups of chickens, which had been experimentally infected with Salmonella Enteritidis p125109.In the 3 days following B. bacteriovorus treatment, Salmonella numbers in the caeca of B. bacteriovorus-treated birds were significantly reduced compared with control animals.Moreover, the appearance of the caeca in B. bacteriovorus-treated birds was normal, compared with many of the caeca from control animals, which showed clear evidence of typhilitis consistent with Salmonella infection.Interestingly, B. bacteriovorus treatment in these birds was associated with a higher mean weight (although not statistically significant) than buffer-treated groups and also resulted in changes to native bacterial populations in the caeca.B. bacteriovorus were not isolated from the caeca of treated birds at the end of the trial, suggesting that B. bacteriovorus was a relatively short-lived population in the environment of the chicken gut.This study was also crucial in priming future studies considering any potential change in bacterial composition of the gut microbiome that would be associated with an administration of such a therapeutic.
Further important studies have demonstrated successful in vivo predation of pathogens by B. bacteriovorus.Using a rat model, Shatzkes and colleagues administered B. bacteriovorus 109J in a series of four intranasal doses to treat rats experimentally infected with Klebsiella pneumoniae in the lungs, showing that pathogen burden was significantly reduced in B. bacteriovorus-treated animals compared to control groups [59].Shatzkes et al. [60] further investigated the use of B. bacteriovorus 109J to treat K. pneumoniae infections in rats, but this time with a septicaemic model of disease initiated by injection into the tail veins of rats.B. bacteriovorus were unable to significantly reduce the infection.This was the first administration of B. bacteriovorus directly into the bloodstream, and an important step in assessing the ability of predatory bacteria to clear bloodstream infections.
The Mostowy and Sockett laboratories employed a zebrafish larval localized (hindbrain) infection and predator treatment model [63,79,80].Along with extensive genomic homology to humans [81], the physical characteristics of zebrafish larvaeinclude a well-understood, fully developed immune system and optically transparent nature which lends itself well to innovative live fluorescent microscopy.This means the larval model has great translational potential especially with regard to studying infection and treatment kinetics [80,82].Using live-cell imaging and bacterial enumeration, Willis et al. demonstrated successful in vivo predation of Shigella flexneri by B. bacteriovorus [63].Using fluorescently labelled (GFP) Shigella flexneri and mCherry-labelled B. bacteriovorus, the authors were able to observe the in vivo behaviours of both bacteria over time.In the absence of treatment, zebrafish larvae showed increasing GFP fluorescence over time.Upon treatment with B. bacteriovorus, this pathogen-associated GFP fluorescence was diminished.This was supported by enumeration of both bacterial species and confocal microscopy evidence of live bacterial predation inside zebrafish larvae, demonstrating that injection of B. bacteriovorus reduces otherwise lethal antibiotic-resistant Gram-negative pathogen Shigella flexneri numbers by active predation.This work has now been extended to include two additional antibiotic-resistant E(S)KAPE pathogens and expanded to simultaneously quantify in vivo predation, zebrafish host response and development of a mathematical model of infection and cure (Jess Tyson and Liz Sockett personal communication; manuscript submitted).
In a recent study, Russo and colleagues [83] demonstrated that B. bacteriovorus 109J could be used to significantly reduce Yersinia pestis numbers in the lungs of experimentally infected mice.In a separate study, groups of mice (SKH-1) were pre-treated with B. bacteriovorus via intraperitoneal injection, then infected with a lethal challenge of Y. pestis [84].Following this, mice were treated daily with B. bacteriovorus and the spread of Y. pestis was tracked with in vitro imaging.B. bacteriovorus treatment was associated with significantly lower levels of Y. pestis, measured both by luciferase signal and by viable counts of the bacteria from the spleen at the end of the experiment.Interestingly, no protection was recorded for Balb/c mice infected with Y. pestis then treated with B. bacteriovorus, suggesting that the protective effect may be dependent on the genetic or immune background of the animal.Not all in vivo administration of B. bacteriovorus has been successful in reducing pathogen load.Despite promising results in vitro using a tissue culture model in a prior investigation [85], administration of B. bacteriovorus to treat calves experimentally infected with Moraxella bovis, a causative agent of infectious bovine keratoconjunctivitis, failed to result in significant improvement in corneal ulcer formation compared with untreated control animals [86].Although an important investigation, the authors themselves state that there are limitations to the study, and perhaps treatment failure could be explained by the observation that they were unable to record persistence of B. bacteriovorus in the tears of calves in this experimental model, despite having previously shown B. bacteriovorus persistence on the corneal surface.
As seen from the extensive research above, there is mounting evidence that Bdellovibrio sp.persist non-pathogenically and long enough to be therapeutically active, are not highly immune-stimulatory, have minimal adverse effects on the bacterial microbiota, and are not known to become part of normal host microbiota [56,57,61,63].Both successful and unsuccessful in vivo assays such as these reviewed here are vital for assessing what indications, routes of administration (topical versus intravenous) and sites of infection, a potential predatory bacterial administration would be most effective against.

UNDERSTANDING THE RELATIONSHIP OF BDELLOVIBRIO PREDATION AND HOST RESPONSE
The clinical outcome of an infection in animals and humans following therapy depends on the interaction of factors such as the patient's immune system and general health, the nature of the pathogen and the type and delivery of treatment.In considering in vitro and in vivo bacterial predation as separate entities to host response, the interaction and cooperation between bacterial predation of a Gram-negative pathogen and response of the host is not captured.The zebrafish hindbrain infection model has been instrumental in exploring this [63].This model demonstrated that B. bacteriovorus persisted nonpathogenically for a sufficient duration to effectively prey on Shigella during an experimental infection [63].It was in this work that interactions with cells of the host immune system were observed microscopically.Moreover, through the use of Pu.1 morpholinos to knockdown host macrophages and neutrophils, this study found that the maximal therapeutic benefits of B. bacteriovorus resulted from the synergy of bacterial predation and the host immune system.Postadministration, the predatory bacteria population is ultimately cleared by interaction with the host's immune system [49,63] and indeed this clearance may be of benefit in (self-) limiting the treatment.By employing RNAseq analysis in a zebrafish model of infection and predatory cure, we are beginning to understand this complex relationship (Jess Tyson and Liz Sockett personal communication; manuscript submitted).

COMPARISON TO OTHER MICROBIOLOGICAL ANTIBACTERIAL APPROACHES
In thinking about treatment of infections with predatory bacteria, researchers can look to the phage research and clinical community for motivation in addressing challenges from in vitro to clinical setting.Although in use for years in some countries [87,88] recent high-profile cases have demonstrated the successful administration of phage cocktails in treating drug-resistant infections in a clinical setting [89,90].Superficially, B. bacteriovorus and phage share many characteristics as potential biological control agents.They both exhibit predatory or parasitic lifecycles, they are selfreplicating and self-limiting, persisting only while susceptible hosts or prey are present and, in the case of lytic phage, both have a broadly similar lifecycle.They also appear to have little or no adverse effect when introduced into the bodies of animals and humans.However, there are some notable differences between the two, which may restrict their use in certain indications.Firstly, B. bacteriovorus are unequivocally 'living' and metabolically active, compared with phage, which are inert particles until they establish contact with a host via a specific receptor.This may be both an advantage and disadvantage for B. bacteriovorus in that it may use energy-requiring processes such as the active 'locating' of prey that are unavailable to phages.However, if no prey is found, B. bacteriovorus cannot enter a dormant state and will die.Conversely, phages can remain 'viable' for many years without expending any energy, but are unable to actively pursue their hosts.They must rely on stochastic probability to encounter their hosts, and compensate for this comparative inefficiency by replicating more quickly than B. bacteriovorus and with a much greater burst size.Previous studies with Salmonella typhi and E. coli have shown that phage do not always eliminate the whole population of their hosts [91].Often in these cases, the titres of phage and host increase and decline with characteristic out-of-phase population oscillations [92].This phenomenon has also been described in predatory protozoa [92] and Bdellovibrio [91,93].As such, the replication of both B. bacteriovorus and phage may be subject to a minimum host/prey threshold, below which they cannot sustain their populations indefinitely.However, this may be therapeutically beneficial as elimination of a pathogen is often not necessary to achieve significant reductions or complete resolution of disease symptoms [63,94].The host range of an individual phage is usually restricted to a number of strains within a species or one or two closely related species of bacteria, which makes the maintenance of their populations more challenging than B. bacteriovorus.The host specificity and rapid acquisition of bacterial resistance to phage needs to be considered; the broad bacterial prey range and the lack of simple resistance mechanisms of predatory bacteria give it an advantage in comparison.Interestingly, recent work by Hobley and colleagues has highlighted the combinatorial power of using predatory bacteria alongside phage [95]; the separate kinetics of phage susceptibility and Bdellovibrio predation allowed greater bacterial cell killing of E. coli in certain conditions.Both B. bacteriovorus and phages will encounter challenging conditions when introduced into the body of endothermic animals.Phage are more likely to be resilient when encountering harsh physical and chemical conditions, such as higher temperatures and extremes of pH, compared with B. bacteriovorus.Phage can cross the blood-brain barrier but may not be able to persist intracellularly without deliberate modification.Indeed, previous studies suggest that when they enter the bloodstream they can be quickly sequestered by the reticuloendothelial system [96].This may be countered, at least in some cases, with the serial passage of phage in mammals to re-isolate phage capable of circulating in mammals for long periods of time [96].Phages are known to be capable of transferring DNA between bacteria through generalized and specialized transduction, which is a risk that does not apply to B. bacteriovorus [37].This risk can be mitigated by careful screening of phage genomes to remove particular phage possessing integrases for example, and others which may harbour genes encoding bacterial toxins or other virulence factors.
The therapeutic application of Bdellovibrio and phages require the industrial production of both.This presents different challenges for each, as phages can be produced in much greater numbers and more quickly, but their production may require the co-culturing of large quantities of pathogenic host bacteria.Conversely, non-pathogenic hosts of Bdellovibrio such as E. coli and P. putida can be used, but these would need to be efficiently separated from the final therapeutic preparation.In addition, maintaining the viability of a Bdellovibrio preparation is likely to be more challenging than for phage.It will be interesting to see what lessons can be learnt from the technical challenges of producing therapeutic preparations of these biocontrol agents [97], and also determining whether any synergy can be achieved when using them together as treatments.

TRANSLATING FUNDAMENTAL SCIENCE INTO APPLICATION TO MEET THE CHALLENGE OF AMR
Academic researchers, including most recently some working on the DARPA pathogen predators programme, have transformed the research field and demonstrated the potential of using whole live predatory B. bacteriovorus to kill a broad range of antimicrobial-resistant (AMR) clinical pathogens in vitro and in vivo; supporting further its promise as a therapeutic.To be considered as a credible therapeutic option, we need compelling evidence from future human trials that this treatment is effective, has no (or minimal) negative effect on the patient and has some advantage over existing treatments, at least in specific circumstances.We need to demonstrate clinical value and measure treatment effect to illustrate the benefit to patients.Future economic and regulatory framework discussions for non-traditional approaches need to be fulfilled, and whilst beyond the scope of this review, momentum is gathering in this area [9,12,22] (follow http://amr.solutions/blogindex.html for updates, comprehensive expert analysis and opinion).But do we have all the information we need?and if not, what still needs to be addressed?Informed by in vivo animal models, we need to move towards both safety and efficacy trials in humans.Animal models are crucial, but have limitations, with the protocols to date administering predatory bacteria prior to, or soon after, the pathogen of interest.Investigation of more established infections is needed, with support for defining predatory bacteria dose range (and possibly pathogen numbers), dose number and schedule required to treat the infection.Perhaps surprisingly, knowledge on the actual numbers of pathogens at different human bodily sites during infections and recovery is still very much a developing field.Development of largescale predator growth and purification methods are currently limited [98] and need expanding for larger-scale safety and efficacy trials.Infection treatment is time critical therefore an assessment of stability, long-term storage and delivery of a live, predatorily-active therapeutic needs further work.
What is an appropriate clinical model?What does the treatment endpoint look like?We need to define the indications in which an advantage of using predatory bacteria can be seen.Not every infection may be accessible to predatory bacteria -evidence from animal models to date suggests topical or localized rather that systemic applications would be more amenable.We need to define what constitutes a successful treatment outcome and for this we may need to measure pathogen levels and debate whether it is sufficient to reduce pathogen numbers to a level at which a healthy human immune system will clear the infection itself [99,100].
Further work is needed to evaluate the dissemination of predatory bacteria from the administration site, along with determining any long-term effects of exposure on the host or their resident microbiota.Animal models and the ability to recover and count live predators in cellular models are important milestones in assessing the effect of predatory bacterial treatment on the host, its bioavailability and dissemination.It would be interesting to expand this work to visualize the detailed subcellular location of B. bacteriovorus in the presence of a pathogen [101] to inform the availability to prey on intracellular pathogens.
In conclusion, it may seem counter intuitive to treat a bacterial infection with administration of another bacterium, and this requires a paradigm shift.The potential effects of treating a Gram-negative bacterial infection with administration of a live predatory bacterium must be considered in context with the burden of an infection on the host.Although non-replicating in the absence of prey, an immune response, albeit a low one, to the presence of B. bacteriocarrying out predation, would be expected as part of the treatment, but this may be actually beneficial.However, faced with an otherwise untreatable Gram-negative infection, could this 'cost' be considered acceptable?One final note once the scientific community has addressed the scientific challenges ahead of their requirement, we are faced with a dilemma to use them as a therapeutic or keep them in reserve.Thinking about the striking analogy to fire prevention, and of antibiotics being the fire extinguishers of modern medicine [102] that are developed and hopefully never used, should we preserve and have B. bacteriovorus ready and available should a fire start?