Part 1/15 : Bacteriophages and the Immune System: How Immunity Can Shape the Success or Failure of Phage Therapy

For decades, phage therapy has been described through an apparently straightforward biological relationship. A bacteriophage encounters a susceptible bacterium, recognizes a surface receptor, injects its genome, replicates and eventually destroys its host. When translated into therapeutic language, the reasoning seems almost equally simple: identify a phage capable of lysing the pathogen and administer enough viral particles to control the infection.

The reality inside the human body is considerably less linear.

A therapeutic bacteriophage does not move through an empty vessel containing only its bacterial host. From the moment it is administered, it encounters epithelial barriers, mucus, soluble proteins, circulating antibodies, phagocytic cells and highly organized innate and adaptive immune networks. At the same time, the bacterial population itself is changing under the combined pressure of viral predation and host immunity.

Phage therapy is therefore not fundamentally a two-player interaction between a virus and a bacterium. It is a tripartite biological system in which the bacteriophage, the bacterial population and the mammalian immune system continuously alter one another.

This third player may help the phage. It may eliminate it. It may recognize phage components, remove circulating virions, generate neutralizing antibodies or change the inflammatory environment in which bacterial killing occurs. In some situations, the immune system appears necessary for therapeutic success. In others, particular immune cells can measurably reduce the local concentration of active phages.

The same phage that produces spectacular bacterial clearance in vitro may therefore behave very differently inside a patient.

The limits of the phage kills bacterium model :

Much of experimental phage selection still begins with bacterial susceptibility. A bacterial isolate is exposed to individual phages or phage cocktails and the resulting phenotype is measured using spot tests, efficiency of plating, liquid culture kinetics or related killing assays. These experiments remain indispensable. A phage unable to productively infect the target bacterium is unlikely to become an effective therapeutic agent.

But bacterial permissiveness is only one dimension of in vivo efficacy.

One of the most influential demonstrations of this problem came from the study of acute respiratory infection caused by Pseudomonas aeruginosa. In a murine pneumonia model, Roach and colleagues examined phage treatment in immunocompetent animals and in animals with altered immune compartments. Their results showed that the therapeutic effect could not be explained by phage replication and bacterial lysis alone. Neutrophils were critical to successful infection clearance, particularly when phage-resistant bacteria emerged [1].

The mathematical framework accompanying the experiments suggested that the innate immune response did not necessarily need to eliminate the entire bacterial population. Under the modeled conditions, an immune response capable of removing approximately 20% to 50% of bacteria could move the system into a region where phage therapy became effective [1].

This changes how phage therapy should be imagined.

The phage may not need to eradicate every bacterium. Instead, viral replication can rapidly reduce bacterial density and disrupt the bacterial population sufficiently for innate immunity to regain control. The immune system, in turn, can eliminate bacterial subpopulations that are increasingly difficult for the phage to infect, including newly emerging phage-resistant variants.

This interaction was described as immunophage synergy [1].

The concept is clinically important because bacterial resistance to a therapeutic phage does not necessarily mean immediate treatment failure. In the 2017 pulmonary model, resistant bacteria appeared during the first 24 to 48 hours following treatment, yet the infection could still be cleared when an effective innate immune response was present [1].

A conventional phage susceptibility assay cannot reproduce this ecology. It shows what occurs between a phage and a bacterium under defined laboratory conditions. It does not show what happens when several million neutrophils are recruited into infected tissue, when bacterial density falls below an immune control threshold or when resistant bacterial variants acquire phenotypic costs that change their interaction with host defenses.

In vivo phage activity is therefore partly an ecological process.

The immune system does not simply observe bacterial lysis :

It is tempting to imagine the mammalian immune system as a passive beneficiary of phage therapy. According to this interpretation, phages eliminate bacteria and the immune system merely resolves the remaining infection.

Experimental evidence increasingly challenges that view.

Bacteriophages can interact directly with mammalian cells even though those cells are not permissive hosts for productive phage replication. Phage particles can be internalized, transported through cellular barriers and processed in intracellular compartments.

In 2017, Nguyen and colleagues demonstrated rapid and directional transcytosis of diverse bacteriophages across confluent epithelial cell layers [2]. Using their experimental rates and estimates of phage abundance at mucosal surfaces, the authors calculated that approximately 10^10 phage particles could potentially cross epithelial barriers each day in an average human body [2].

The number is an estimate rather than a direct measurement of daily human translocation, but its scale is biologically provocative. Mammalian tissues may encounter bacteriophages far more frequently than the classical separation between bacterial viruses and animal cells would suggest.

Crossing an epithelial layer also has consequences for phage pharmacology. A virion transported from the luminal or apical environment into underlying tissues can encounter resident immune cells. Conversely, cellular uptake may remove infective phages from the extracellular compartment in which their bacterial targets are located.

Internalization is therefore not synonymous with successful delivery.

This distinction becomes particularly important for inhaled phage therapy, where billions of viral particles may be deposited directly onto a respiratory surface containing mucus, epithelial cells, resident macrophages and recruited leukocytes. The biological fate of those particles cannot be predicted solely from their plaque-forming activity against Pseudomonas aeruginosa on an agar plate.

Human epithelial cells can respond differently to different phages :

The respiratory epithelium is not merely a physical wall separating therapeutic phages from deeper tissues.

In 2024, Zamora and colleagues exposed human airway epithelial cells derived from a person with cystic fibrosis to four lytic Pseudomonas aeruginosa bacteriophages: OMKO1, LPS-5, PSA04 and PSA34 [3]. In several experiments, epithelial cells were exposed to phage concentrations reaching 10^9 plaque-forming units per millilitre [3].

The four phages did not produce a single uniform host response.

The epithelial transcriptional profile changed after phage exposure and the cells secreted antiviral and proinflammatory mediators. Cytokine responses differed according to the phage examined. All four phage treatments were associated with basolateral IFN-β secretion in the experimental system, while other responses, including IFN-λ1, IL-8, TNFSF13B and TNFSF8, varied between phages [3].

This observation has profound implications for therapeutic phage selection.

Two phages may infect the same bacterial species. They may show similar efficiencies of plating and comparable bacterial killing kinetics. Yet they may not be immunologically equivalent when they encounter human tissue.

The molecular basis of these differences remains incompletely understood. Phage capsid architecture, virion morphology, surface chemistry, nucleic acid composition, structural proteins and the physicochemical environment surrounding the epithelial cell could all influence recognition or cellular processing.

The important point is that the word bacteriophage does not describe a single immunological entity.

Phages are extraordinarily diverse biological particles. It would be surprising if thousands of structurally and genetically distinct bacteriophages interacted identically with mammalian cells.

Macrophages reveal the paradox at the centre of phage therapy :

Few immune cells illustrate the complexity of this relationship as clearly as macrophages.

Macrophages are essential components of innate immunity. They recognize microorganisms and damaged tissue, internalize biological particles, secrete cytokines and participate in the coordination of inflammation and tissue repair. In the lung, alveolar macrophages occupy the precise anatomical environment in which inhaled therapeutic phages may be expected to encounter bacterial pathogens.

Their presence should logically help control a bacterial infection.

That is true.

But they may simultaneously interfere with the phage being used to treat it.

A 2025 study investigated this apparent contradiction during Pseudomonas aeruginosa pneumonia treated with phage PAK_P1 [4]. The investigators depleted approximately 98% of alveolar macrophages in their experimental animals before infection [4].

Without phage therapy, macrophage depletion was harmful. Every macrophage-depleted animal reached the humane endpoint within 24 hours, whereas progression in control animals took more than 48 hours [4]. This confirmed that alveolar macrophages had an important protective function during infection.

The situation changed after phage administration.

In control animals, phage treatment increased survival from 0% at 2.5 days to 67% at day eight. In macrophage-depleted animals, treatment increased survival from 0% at 24 hours to 80% at day eight [4].

More strikingly, bacterial quantification 22 hours after infection revealed approximately 10^5 CFU/mL in the lungs of phage-treated control mice. In phage-treated animals depleted of alveolar macrophages, P. aeruginosa fell below the experimental detection limit of 2.5 × 10^2 CFU/mL [4].

The authors reported a difference of at least four orders of magnitude in pulmonary bacterial burden between the two phage-treated groups [4].

Mathematical modeling and experimental measurements provided a possible explanation. Phage decay in vivo was approximately three times faster in immunocompetent animals than in macrophage-depleted animals, and alveolar macrophages were shown to phagocytose the therapeutic bacteriophage [4].

The interpretation is uncomfortable but scientifically important.

A cell that protects the host from bacterial pneumonia can simultaneously remove an antibacterial virus from the infected environment.

Macrophages are neither simply friends nor enemies of phage therapy. Their net effect emerges from competing biological functions. Removing them can improve phage-mediated bacterial killing while worsening the host's capacity to survive infection and manage tissue damage.

Phage therapy therefore cannot be optimized by asking whether macrophages are beneficial.

The correct question is more difficult: how do macrophage-mediated bacterial control, phage clearance, inflammatory signaling and tissue protection interact over time in a specific infected organ?

The immune system can also remember a therapeutic phage :

Innate immunity acts rapidly, but repeated phage exposure introduces a second layer of complexity.

Bacteriophages are protein-rich viral particles. Their capsids, tails, tail fibres, sheaths and other exposed structural components can provide antigens for adaptive immune recognition. Antibody responses against therapeutic phages have been documented for decades, but their clinical importance has often been difficult to interpret.

A 2024 Nature Communications study offered unusually direct experimental evidence that phage-specific adaptive immunity can impair subsequent therapy [5].

Berkson and colleagues constructed a five-phage cocktail targeting vancomycin-resistant Enterococcus. The initial treatment significantly reduced the intestinal bacterial burden in mice. However, two treatment courses induced neutralizing antibodies against the administered phages and accelerated phage clearance from tissues [5].

The immune response was not identical for every phage in the cocktail. Myophages produced a stronger neutralizing antibody response than the siphophages included in the same preparation [5].

When the cocktail was subsequently reused in animals possessing anti-phage immunity, the treatment no longer produced a significant reduction in faecal VRE burden [5].

The experiment raises a fundamental therapeutic question.

Can a patient become immunologically resistant to their own phage therapy?

The answer appears to be yes in at least some biological contexts, although the clinical consequences remain highly variable.

Human evidence is beginning to illustrate that complexity. In 2024, Bernabéu-Gimeno and colleagues described three nebulized monophage treatments for Staphylococcus aureus or Pseudomonas aeruginosa respiratory infections in people with cystic fibrosis [6]. Phages were nebulized for ten days alongside standard antibiotic therapy [6].

In two treatments, bacterial loads fell by between 3 and 6 logarithmic units. Viable phages remained detectable in sputum for as long as 33 days after treatment completion [6].

Yet serum phage-neutralizing antibodies were detected in all three cases between 10 and 42 days after treatment [6].

This was especially important because the phages had been administered by nebulization rather than an invasive systemic route. Local delivery to the respiratory tract did not prevent the development of a measurable systemic anti-phage antibody response.

The mere detection of neutralizing antibodies does not prove therapeutic failure. In the cystic fibrosis cases, substantial bacterial reductions were observed despite the later appearance of anti-phage immunity [6]. Clinical phage therapy has also produced examples in which neutralizing antibodies were detected without clearly preventing successful bacterial control.

Timing may therefore be decisive.

A phage may reduce the pathogen before a mature neutralizing antibody response develops. A short course may succeed where prolonged or repeated exposure creates stronger immunological pressure. Reusing the same phage weeks or months later may not reproduce the pharmacological behaviour observed during the first treatment.

In this sense, a bacteriophage is unlike a conventional small-molecule antibiotic.

Its pharmacology can change because the patient remembers it.

Sometimes a phage can manipulate mammalian antiviral immunity :

The relationship becomes even more unusual when phages produced by bacteria actively influence mammalian immune pathways.

The filamentous Pf bacteriophage of Pseudomonas aeruginosa provides one of the clearest examples.

In 2019, Sweere and colleagues showed that Pf phage could be taken up by mammalian immune cells and stimulate a TLR3-dependent antiviral response through TRIF and type I interferon signaling [7].

This is conceptually remarkable. TLR3 is classically associated with the recognition of viral double-stranded RNA. Pf phage is a bacterial virus, yet phage-associated RNA within internalized particles contributed to activation of an antiviral immune program [7].

The consequence was favourable to P. aeruginosa.

Type I interferon signaling reduced tumour necrosis factor production and impaired bacterial phagocytosis, limiting bacterial clearance [7].

A virus of bacteria had altered mammalian immunity in a manner that benefited its bacterial host.

The conventional phage therapy narrative is almost completely inverted. The phage does not kill the pathogen. Instead, a bacterium-associated phage manipulates host antiviral immunity and creates an immunological environment in which the bacterium becomes more difficult to eliminate.

Pf phage is a filamentous phage associated with P. aeruginosa and should not be treated as a model for every therapeutic lytic bacteriophage. Nevertheless, the study demonstrates what is biologically possible.

Mammalian immune pathways can distinguish and respond to phage-derived material. The resulting immune response can alter bacterial clearance.

Bacteriophages may even influence mucosal inflammation :

The intestinal virome provides another warning against describing phages as immunologically invisible.

Gogokhia and colleagues experimentally increased bacteriophage levels in mouse models and examined the consequences for intestinal immunity [8]. Phage exposure induced both phage-specific and bacteria-specific immune responses. In the experimental colitis model, increasing bacteriophage abundance aggravated intestinal inflammation through mechanisms involving TLR9 and IFN-γ [8].

Human ulcerative colitis-associated bacteriophages also stimulated greater IFN-γ responses in the experimental systems than phages obtained from healthy individuals [8].

Again, these results should not be extrapolated to claim that therapeutic bacteriophages generally cause intestinal inflammation. The study investigated complex phage populations and mucosal immune biology rather than a purified clinical phage product.

Its importance lies elsewhere.

The mammalian immune system is capable of responding to bacteriophages in a phage-dependent and context-dependent manner. The response can extend beyond simple clearance of viral particles and influence tissue-level inflammation.

The patient is part of the phagogram :

Modern therapeutic phage selection is increasingly sophisticated. Candidate phages can be sequenced, screened for undesirable genes, tested against panels of bacterial isolates and characterized for adsorption, latent period, burst size, stability and resistance emergence.

Yet the host immune system is rarely integrated into the earliest stages of phage selection with comparable resolution.

The evidence now suggests that this separation is becoming difficult to defend.

Neutrophils can be essential partners in therapeutic bacterial clearance [1]. Epithelial barriers can transport phages across mammalian cell layers at biologically meaningful rates [2]. Respiratory epithelial cells can produce different cytokine profiles in response to different lytic phages [3]. Alveolar macrophages can accelerate local phage loss and reduce phage-mediated bacterial clearance [4]. Repeated treatment can generate neutralizing antibodies and compromise the reuse of a phage cocktail [5]. Nebulized phages can induce systemic neutralizing antibodies in people with cystic fibrosis within 10 to 42 days [6]. A filamentous Pseudomonas phage can exploit mammalian antiviral signaling to impair bacterial phagocytosis [7]. Intestinal bacteriophage expansion can influence TLR9-dependent mucosal inflammation [8].

These findings do not support a simple conclusion that the immune system is good or bad for phage therapy.

They support a more demanding idea.

The therapeutic unit is not the phage alone.

It is the phage in a bacterial population, inside a particular tissue, exposed to the immune architecture of a particular host at a particular stage of infection.

This may explain why apparently excellent in vitro phages sometimes become pharmacologically disappointing in vivo. It may also explain why phage resistance does not invariably cause treatment failure and why comparatively modest phage-mediated reductions in bacterial density can sometimes precede successful infection clearance.

The future of phage therapy may therefore require an expansion of what we mean by phage susceptibility.

A bacterial isolate can be susceptible to a phage.

But is the phage stable in the patient's biological compartment? Is it rapidly removed by resident phagocytes? Does it stimulate epithelial cytokine release? Are pre-existing antibodies capable of neutralizing it? Has the patient previously encountered a related phage? Will repeated administration accelerate viral clearance? Does bacterial lysis lower the pathogen population sufficiently for neutrophils to take control?

These are not secondary details surrounding the antibacterial mechanism.

They are part of the mechanism.

For a century, phage therapy has largely been narrated as the deliberate use of bacterial viruses against bacterial pathogens. That definition remains correct, but it is becoming biologically incomplete.

Inside the human body, every therapeutic phage enters an immunological ecosystem.

And in that ecosystem, the immune system is not a spectator.

It is the third player.

Sources :

[1] Roach DR, Leung CY, Henry M, et al. Synergy between the Host Immune System and Bacteriophage Is Essential for Successful Phage Therapy against an Acute Respiratory Pathogen. Cell Host & Microbe. 2017;22(1):38–47.e4. https://doi.org/10.1016/j.chom.2017.06.018

[2] Nguyen S, Baker K, Padman BS, et al. Bacteriophage Transcytosis Provides a Mechanism To Cross Epithelial Cell Layers. mBio. 2017;8(6):e01874-17. https://doi.org/10.1128/mBio.01874-17

[3] Zamora PF, Reidy TG, Armbruster CR, et al. Lytic bacteriophages induce the secretion of antiviral and proinflammatory cytokines from human respiratory epithelial cells. PLOS Biology. 2024;22(4):e3002566. https://doi.org/10.1371/journal.pbio.3002566

[4] Zborowsky S, et al. Macrophage-induced reduction of bacteriophage density limits the efficacy of in vivo pulmonary phage therapy. Nature Communications. 2025;16:5725. https://doi.org/10.1038/s41467-025-61268-1

[5] Berkson JD, Wate CE, Allen GB, et al. Phage-specific immunity impairs efficacy of bacteriophage targeting Vancomycin Resistant Enterococcus in a murine model. Nature Communications. 2024;15:2993. https://doi.org/10.1038/s41467-024-47192-w

[6] Bernabéu-Gimeno M, Pardo-Freire M, Chan BK, et al. Neutralizing antibodies after nebulized phage therapy in cystic fibrosis patients. Med. 2024;5(9):1096–1111.e6. https://doi.org/10.1016/j.medj.2024.05.017

[7] Sweere JM, Van Belleghem JD, Ishak H, et al. Bacteriophage trigger antiviral immunity and prevent clearance of bacterial infection. Science. 2019;363(6434):eaat9691. https://doi.org/10.1126/science.aat9691

[8] Gogokhia L, Buhrke K, Bell R, et al. Expansion of Bacteriophages Is Linked to Aggravated Intestinal Inflammation and Colitis. Cell Host & Microbe. 2019;25(2):285–299.e8. https://doi.org/10.1016/j.chom.2019.01.008