Pneumonia is the second most common hospital-acquired infection that ranks first among intensive care units (ICUs) patients. Hospital-acquired pneumonia (HAP) is often caused by drug-resistant bacteria and cause death in 20–70% of cases (Koulenti et al., 2017). In pneumonia, infection of the lung parenchyma is observed with acute inflammation and tissue consolidation. Therefore, the filling of bronchioles and alveoli with inflammatory secretions reduces gas exchange and causes respiratory symptoms. On the other hand, occurrence of severe inflammatory phase by immune cells in the lung is one of the important factors of tissue damage in pneumonia. Although this process is controlled by reducing the pathogenic agent under the process of immune homeostasis. However, infectious and some therapeutic agents, by interfering with homeostasis, intensify the inflammatory phase and cause tissue damage and disease progression (Gopallawa et al., 2023).

Gram-negative bacteria (GNB), especially Enterobacteria including Klebsiella pneumoniae, followed by Acinetobacter baumannii, and Pseudomonas aeruginosa are the most common bacterial causes of HAP (Torres et al., 2017). These bacteria are also part of the ESKAPE group (including Enterococcus faecium, Staphylococcus aureus, K. pneumoniae, A. baumannii, P. aeruginosa, and Enterobacter spp), and are on the priority list of the World Health Organization (WHO) to find new therapeutic methods due to their high drug resistance (Tacconelli et al., 2018). MDR (Multi-Drug Resistant) and XDR (Extensively-Drug Resistant) clinical isolates of A. baumannii constitute the dominant population of hospital-acquired infectious agents, and they cause high mortality, especially in systemic infections. Also, A. baumannii causes wide range of nosocomial infections such as pneumonia, bacteremia, meningitis, soft tissue infection, and urinary tract infection (UTI), which are seen especially in immunocompromised patients. Presence of different resistance mechanisms such as drug-degrading enzymes, efflux pumps and changes in the cell envelope have left limited antibiotic options for treatment of drug-resistant strains of A. baumannii (Ayoub Moubareck and Hammoudi Halat, 2020). Colistin is one of the last treatment choices for these strains, however, its toxicity and noticeable side effects limit its use (Zheng et al., 2020).

The presence of efflux pumps and their increased expression are among the most important mechanisms by which A. baumannii acquires resistance to antibiotics in particular to colistin. As a result of efflux pumps functions not being restricted to specific substrates, they participate in pumping out multiple antibiotics from the cell and are associated with the development of MDR strains. Five families of efflux pumps are responsible for drug resistance in A. baumannii. The first group is the resistance nodulation division (RND) superfamily (AdeABC, AdeIKJ, AdeFGH), which associates with developing resistance to aminoglycosides, fluoroquinolones, tetracyclines and carbapenems (Shi et al., 2024). The second family is the ATP-binding cassette (ABC) superfamily (AdeABC, MsbA) which contributes to resistance against carbapenems and various disinfectants. The third family is the small multidrug resistance (SMR) protein subfamily (AbeS), which is involved in resistance development to fluoroquinolones and macrolides. The fourth family is the multidrug and toxic compound extrusion (MATE) superfamily (AbeM), which causes resistance to sulfonamides, aminoglycosides and chloramphenicol by affecting H⁺ exchange. The last family is the major facilitator (MFS) superfamily (TetA, TetY, CraA, AmvA), which is involved in resistance development to tetracyclines, carbapenems and aminoglycosides (Chang et al., 2016).

Several methods for drug-resistant strains of A. baumannii infection control have been investigated, such as antibiotic combination therapy, herbal derivatives and phage therapy (Karakonstantis et al., 2021). Six decades after the discovery of the first predatory bacterium, Bdellovibrio bacteriovorus, studies on the use of predators as a live antibiotic in the control of drug-resistant pathogens continue. Predators are driven to prey-rich areas by chemotaxis, and then they get attached to the prey. Despite details of prey tropism still being vague, the role of type IV pili in initial attachment, rose-like structures, MIDAS adhesin family protein, and fiber-like proteins (mosaic adhesion proteins or MAT) on the surface of Bdellovibrio for prey attachment and subsequent entry into the periplasm are confirmed (Caulton et al., 2024; Tyson et al., 2024). Rose-like structures are only seen on the surface of predator cells during the attack phase (Huwiler et al., 2023). Also, MATs are a superfamily with diverse binding domains which let them bind to different cell envelopes of prey bacteria, and make predation possible on a wide range of preys (Caulton et al., 2024).

Bedlovibrio reproduces by entering the periplasm of the bacterial prey and then exits by lysing the host cell. The number of predators produced varies depending on the size of the prey cells and the amount of nutrients available for Bdellovibrio (Summers and Kreft, 2022). Effect on a wide range of gram-negative bacteria, nontoxic to mammalian cells (Gupta et al., 2016), and isolated B. bacteriovorus from the human gut (Iebba et al., 2013) are the reasons that make predatory bacteria an attractive option for combating pathogens. Limited in vivo studies have been performed with predatory bacteria. The aim of these studies has been to control local infections, and their results confirm the ability of predatory bacteria to reduce the pathogen’s load (Romanowski et al., 2016, 2023; Cui et al., 2021; Liu et al., 2023; Tajabadi et al., 2023). Furthermore, a significant systemic pathogen inhibition by B. bacterivorus has been demonstrated only in a few studies, i.e., oral administration to chickens (Atterbury et al., 2011), intranasal administration to rats (Shatzkes et al., 2016), and intravenous administration to rats (Shatzkes et al., 2017) and mice (Findlay et al., 2019).

A successful treatment should have limited and controllable side effects in addition to inhibiting the pathogen. Although the presence of modified lipid A in the LPS of B. bacteriovorus has been confirmed to weakly induce inflammatory cytokines (Needham and Trent, 2013), an important question is how long the continuous release of endotoxin from the prey pathogen can be tolerated by the host in a systemic infection? Perhaps the perpetual cycles of predation could be controlled using engineered predator bacteria privileged with specific prey preferences. Thus far, a single study has reported the effects of nasal administration of predatory bacteria to control K. pneumoniae in an experimental respiratory infection of rats (Shatzkes et al., 2016).

In the present study, we developed a rat pneumonia model with MDR clinical isolate of A. baumannii and used intratracheal administration of B. bacteriovorus HD 100 for treatment. Tissue tropism of the predatory bacteria, the time of their elimination from the host body, and the induction of inflammatory and destructive responses are important questions to be answered upon systemic use of a live bacteria as an alternative to antibiotics. To elucidate these questions, we studied the systemic release of predatory and pathogenic bacteria in the rat body and the response of two pro-inflammatory cytokines to the pathogen, predator and their interactions.

The increase of antibiotic resistance in A. baumannii, which is one of the common causes of HAP, shows the necessity of finding an alternative to deal with this pathogen (Ayoub Moubareck and Hammoudi Halat, 2020). Therefore, in this study, we established an acute pneumonia model with an MDR strain of A. baumannii and evaluated the efficiency of the B. bacteriovorus HD100 as a novel treatment agent. Because A. baumannii is an opportunistic pathogen, acute pneumonia was induced in Wistar rats by inducing transient neutropenia, inoculation with a high dose, and using a clinical A. baumannii isolate. A. baumannii spread throughout the animal’s lung within 30 min after intratracheal administration (Figure 3), and the pathogen load in lung increased up to 6 h, and then showed a continuous decrease. Similar rapid systemic spread of A. baumannii has been reported in mice and has been attributed to the escape of the pathogen from the host’s innate immune system (Harris et al., 2019). Also, similar studies showed that the peak of the microbial counts of A. baumannii in the lung occurs between 4 h to 6 h after inoculation and then decreases. Furthermore, a higher pathogen load in lungs than other organs is indicative of a persistent lung infection (Russo et al., 2008; Harris et al., 2019).

According to the antibiogram results for clinical A. baumannii isolate, colistin was used for antibiotic treatment (Mobarak-Qamsari et al., 2023). CMS is administered intravenously or by inhalation and is degraded in body to active colistin, which has a half-life of approximately 4 h. Colistin efficacy and penetration are related to tissue characteristics such as tissue vascularity, lipid composition, extracellular matrix composition, tissue inflammation and drug binding affinity. Greater vascular muscle tissues lead to more facilitated colistin penetration compared with less vascular adipose tissues. Conversely, more fat in the tissue is associated with reduced drug penetration (Sanabria et al., 2021). Although colistin has a high electrostatic binding affinity for the lipid membranes of the liver, kidney, lung, brain, and heart; the drug penetration into the pleural cavity, lung tissue, cerebrospinal fluid, and bones is poor, and the concentrations are not considered antibacterial. In addition, increased inflammatory responses in the tissue are associated with reduced drug penetration and efficacy. Just like the increased-inflammation in patients with severe respiratory infections, inhaled administration of CMS has been recommended to achieve higher drug concentrations and reduce systemic side effects in these patients (Mondal et al., 2024). In addition, studies have confirmed that in the pulmonary administration, the plasma concentration of the drug remains high for a longer period of time (Yapa et al., 2013). Also, for combination therapy of predatory bacteria with colistin, we determined MIC to measure the sensitivity of B. bacteriovorus HD100 to colistin (8 μg/mL). Because MIC values in our study were not determined under the CLSI standard conditions, it cannot be compared with other bacteria. Also, for the first time, to our knowledge, B. bacteriovorus HD100 suspension for intratracheal administration was purified from endotoxins by the solvent-extraction method (Szermer-Olearnik and Boratyński, 2015).

The most effective treatment was observed in the antibiotic treated group and followed by combined treatment group. High survival rate and the lowest septic score during 10 days, the lowest lung bacterial counts up to 24 h, and the lung mild inflammatory response at 24 h and later were the benefits of the antibiotic treatment. However, a more intense infiltration of neutrophils at 6 h post-inoculation was observed for the antibiotic group than for the combined treatments. It is probably due to the infection of a large number of A. baumannii by the predator, and as a result, the antibiotic did not affect the pathogen. Therefore, less endotoxin was released and these groups also had a lower septic score and TNF-α serum level. On the other hand, the most severe pulmonary inflammatory reactions occurred in the B. bacteriovorus HD100 treated group, that continued up to 48 h. Studies that have used pulmonary administration of LPS to induce lung injury have reported similar pulmonary and systemic inflammatory responses to our observations. But these effects disappeared after 24 h, because they had used a single dose of endotoxin (Chuang et al., 2013; Fu et al., 2017).

Studies have shown that the most important early defense against A. baumannii respiratory infection is involved with macrophages and neutrophils cooperation. The release of LPS and its interaction with Toll like receptor-4 (TLR-4) activates M1 macrophages in the lung, which ultimately leads to the production of inflammatory cytokines and chemokines such as macrophage neutrophil chemokine-2 (MIP-2), IL- 1β, and TNF-α. Moreover, bacterial OMVs also play a role in activating the MyD88 pathway in the recruitment of immune cells by activating TLR-2. Therefore, large numbers of neutrophils are driven to the lung, which are primarily responsible for phagocytosis and killing of the pathogen and contribute to the recruitment of other immune cells through the TLR4-NFκB pathway (Marion et al., 2019). In the presence of large numbers of A. baumannii, more neutrophils are recruited to the lung, causing more inflammation and further damage to the airways (Harris et al., 2019). Since in lung histology, cell damage caused by the presence of the predator alone or together with the host pathogen has not been observed, the persistent inflammatory reactions in our study are probably due to the continuous release of endotoxin due to the continuous degradation of pathogens in the predation cycle. On the other hand, the three groups that received the predator showed continuous high levels of TNF-α after 6 h (except for the combined treatment at 24 h).

Even in the B. bacteriovorus HD100 control group, pulmonary inflammatory response and serum inflammatory cytokine levels were increased during the study period. So, the group showed the highest septic score and mortality rate on day 10. These observations may also suggest that the predatory bacteria may influence the microbiota modification. Since B. bacteriovorus HD100 is an obligate predator, it must have found a host in the body that can be dissected by day 10. In a study that examined the lung microbiome of cystic fibrosis (CF) patients for 1 year, the presence of predatory bacteria Bdellovibrio and Vampirovibrio was confirmed in the lungs. In addition, minor changes in the composition of the microbiome were observed, which was associated with the reduction of Parcobacterium. According to the curves presented in this manuscript, more predatory bacteria were isolated with higher disease severity (de Dios Caballero et al., 2017).

The lower systemic spread of A. baumannii can be considered as the only advantage of the treatment with predatory bacteria. This was also observed for combination therapy and is probably related to pathogen involvement in the lung in the predation cycle. A major problem in intratracheal inoculation of predatory bacteria was the rapid spread to the brain of the animals, where they persisted until day 10 of examination. Im et al. showed that B. bacteriovorus HD100 is resistant to serum complement system, which is likely related to the neutral charge of the predator LPS. Thus, as we observed in this study (Appendix Figures 2B–E), systemic spread of the predator is facilitated (Im et al., 2017). Similarly, Findlay et al. studied systemic administration of B. bacteriovorus HD100 to treat plague in mice. Although the predatory bacteria saved the mice from death and reduced the amount of the pathogen, they remained in the animal’s body for a long time. Also, imaging of the labeled predator revealed that B. bacteriovorus HD100 has tropism to fat-based tissues. The presented images confirmed the high intensity of the predator’s presence in the brain of the animals (Findlay et al., 2019).

In vivo systemic studies on predatory bacteria are very limited. In a study, Shatzkes et al. reported that intravenous administration of B. bacteriovorus has no effect on reducing the systemic burden of K. pneumoniae in rats. It also took 20 days for the predator to be removed from the animal’s body (Shatzkes et al., 2017). In another study by the same researchers, intranasal administration of B. bacteriovorus and Micavibrio aeruginosavorus was evaluated to reduce the microbial load of K. pneumoniae in the lung. The results showed that the lung pathogen load was reduced by more than 3.0 log10 in the animals, but no histological complications in the lungs or continuous increase of inflammatory cytokines were reported (16). These results may be related to use of immunocompetent rats and because K. pneumoniae is not a rat pathogen and is easily cleared by an efficient immune system. In addition, intranasal administration helps eliminate a number of pathogens and predatory bacteria before reaching the lungs. However, an acute inflammatory response in the lungs at 24 h after B. bacteriovorus HD100 inoculation was reported by these authors (Shatzkes et al., 2016).

It seems that the complications observed in our research are related to the continuous predatory cycle due to the wide range of prey, as well as, the wide systemic circulation of the predators. Therefore, the first step to use predatory bacteria as living antibiotics is to focus on the predation control through genetic modification. Despite the obscures life cycle of predatory bacteria and their interactions with the host, relying on new findings is promising. Studies have shown that about 40% of B. bacteriovorus genes, called the predatosome, are related to predation. Duncan et al. divided this set into four functional gene classes, of which class III is involved in bdelloplast turnover and predator evasion. Although, B. bacteriovorus causes prey cell death within 10–20 min after entry, it requires 3–4 h to complete the predation cycle. Genes involved in badloplast turnover and predator evasion are suitable candidates for gene manipulation to limit predation cycles (Duncan et al., 2019). Therefore, it is possible to engineer predatory bacteria that can enter the periplasm of the prey and destroy it while not being able to complete their cycle.

Another option is to alter proteins involved in prey binding. One example is the induction of mutations in the MAT complex, which can create predators with limited host range. However, due to the large number of MAT proteins, single-gene mutations can only delay predation (Caulton et al., 2024). Another suggestion is the induction of mutations in Bd0875, a protein of the MIDAS adhesin family, which was investigated in the study by Tyson et al. Mutations in this protein cause the formation of round and dead badloplasts (Tyson et al., 2024). It is also possible to generate engineered strains in which gene expression is controlled by riboswitches, as Dwidar and Yokobayashi did by generating theophylline-responsive strains that affected flagellar production and hunting ability in B. bacteriovorus (Dwidar and Yokobayashi, 2017).

Determining the role of predatory bacteria on the microbiome and the consequences of these changes is an important question that needs to be answered. Furthermore, predator–prey interactions and their effects on immune responses in an in vivo setting should be evaluated. More research needs to determine when predatory bacteria are eliminated from the host’s body. Are predator metabolites responsible for inflammatory responses, or endotoxin, which is released from gram-negative prey once after each predator cycle, plays a role in inflammatory complications. Finally, considering the complications observed by predatory bacteria in our study, it can be said that there is still a long way to introduce these microbes as a live antibiotic. On the other hand, the results of this research can suggest a study on the possibility of predators in acute lung diseases. Therefore, controlling the population of predators in the lung microbiome of pneumonia patients may be possible to help them survive.