PHIOGEN, Baylor and LillixBio Use AI to Explore Immune Protection After Phage Therapy
PHIOGEN, Baylor College of Medicine and LillixBio have launched a new collaboration investigating an unusual possibility for bacteriophage therapy: whether killing bacteria with phages could be combined with deliberate stimulation of the patient's immune system to provide protection against future infections. Announced on October 8, 2026, the programme will combine artificial intelligence-based antigen prediction with experimental phage biology and immunological validation, beginning with Escherichia coli and PHIOGEN's PHI-UI-01 programme for recurrent urinary tract infections.
The concept extends phage therapy beyond its conventional objective of eliminating bacteria during an active infection. PHIOGEN is investigating what it describes as “immunizing therapeutics,” in which bacteriophages would provide immediate antibacterial activity while the bacterial material released during treatment could potentially be used to generate a more durable host immune response. The new collaboration is intended to identify which immune responses arise after phage-mediated bacterial destruction and whether those responses can be strengthened, prolonged or directed toward antigens associated with protective immunity.
This represents a considerably different development strategy from treating phages simply as highly specific antibiotics. In conventional lytic phage therapy, a bacteriophage recognizes a susceptible bacterium, injects its genome, replicates and ultimately destroys the bacterial cell. The therapeutic objective is primarily reduction of the pathogen population, while interactions between the administered phages and the immune system have often been studied because they can influence phage circulation, clearance, inflammation and treatment efficacy.
PHIOGEN's hypothesis introduces another dimension. When a phage destroys a bacterium, bacterial proteins and other cellular components are released into an immunologically active environment. Those microbial antigens may be captured and processed by antigen-presenting cells, potentially allowing components of the adaptive immune system to recognize the pathogen. If the right antigens and immune conditions can be identified, the company wants to determine whether an acute phage treatment could contribute to protection against subsequent exposure to the same or related bacterial pathogen.
The new collaboration does not establish that this occurs reliably in patients. Its purpose is precisely to determine which antigenic and immunological mechanisms could support such an effect and how they might be manipulated. Baylor describes the programme as an effort to map the immune response following phage therapy and determine how potential protection could be strengthened, extended and broadened.
LillixBio will provide the computational component of the programme. Its artificial intelligence platform has generated what the collaboration describes as a comprehensive predicted major histocompatibility complex antigen-presentation map for E. coli. The analysis considers microbial proteins across a broad range of immune diversity and is intended to identify bacterial antigens that could be presented effectively to the adaptive immune system.
Major histocompatibility complex molecules are central to adaptive immunity because they display short protein fragments, or peptides, at the surface of cells where they can be examined by T lymphocytes. In humans, these molecules are encoded by the highly variable HLA system. Different individuals carry different HLA alleles and can therefore present different sets of peptides derived from the same microorganism.
This genetic diversity represents an important challenge for any approach intended to generate broad immune protection. A bacterial protein that produces a strong immune response in one individual may be presented much less efficiently in another. Computational mapping can therefore help identify antigens with predicted presentation across multiple HLA backgrounds rather than selecting candidates according to a single immune genotype.
The scale of the approach is also important. Instead of beginning with a handful of previously characterized E. coli antigens, LillixBio's system is intended to evaluate the bacterial proteome more systematically and predict which peptides could become visible to the immune system. PHIOGEN and Baylor can then prioritize a smaller number of candidates for experimental testing.
Prediction alone cannot establish protective immunity. A peptide may bind an MHC molecule efficiently without generating a useful T-cell response, and an antigen that stimulates immune cells does not automatically prevent a future bacterial infection. The biological validation stage will therefore determine whether the computationally identified candidates actually alter immune responses in relevant experimental systems.
That work will take place in collaboration with Baylor's TAILΦR programme, the Tailored Antimicrobials and Innovative Laboratories for Phage Research. TAILΦR has established capabilities in phage isolation, characterization and therapeutic development, while the new project will draw particularly on its work involving phage adjuvantation, serology and immunological profiling in mucosal and systemic infection models.
This combination allows the programme to connect three levels of investigation. Computational analysis can predict potentially important bacterial antigens and immunomodulatory candidates, laboratory experiments can determine whether those predictions produce measurable immune responses, and PHIOGEN can evaluate how the resulting information could be incorporated into its therapeutic development platform.
The first pathogen being studied is E. coli, reflecting the focus of PHIOGEN's PHI-UI-01 programme on recurrent urinary tract infections. Uropathogenic E. coli is the dominant bacterial cause of many UTIs and is particularly important in recurrent disease, where patients may experience repeated episodes despite apparently successful treatment of previous infections.
Recurrence is biologically complex and cannot be explained by antibiotic resistance alone. Uropathogenic E. coli can interact closely with the urothelium, form intracellular bacterial populations in experimental models and exploit several mechanisms that help it persist or evade host defences. At the same time, immune responses generated during a first UTI do not always produce sufficiently effective adaptive protection to prevent subsequent infection.
Reviews of urinary tract immunity describe a response involving the urothelial barrier, neutrophils, macrophages, dendritic cells, antimicrobial molecules and adaptive lymphocytes. However, immunity in the bladder must simultaneously eliminate pathogens while avoiding excessive tissue damage and maintaining tolerance toward non-pathogenic microorganisms. This creates an environment in which strong, durable adaptive immunity against recurrent infection is not guaranteed.
That makes recurrent UTI an interesting setting in which to test PHIOGEN's hypothesis. A treatment capable of rapidly reducing E. coli burden while also promoting a more useful adaptive response could theoretically address two distinct aspects of recurrent disease: the current bacterial population and susceptibility to future infection.
PHI-UI-01 is being developed specifically around this dual-action concept. PHIOGEN uses directed evolution and screening technologies to select bacteriophages with desired antibacterial characteristics, including activity against resistant bacteria. The company is now attempting to integrate immunological properties into this development strategy rather than treating them as secondary consequences of phage administration.
PHIOGEN emerged from bacteriophage research at Baylor College of Medicine and was founded in 2023 to commercialize technologies developed through TAILΦR. Its broader platform combines phage discovery, genomic characterization, physiological screening, directed evolution and formulation, with the objective of creating standardized products capable of treating larger patient populations rather than relying exclusively on individualized phage selection.
The immunological strategy is not limited to PHI-UI-01. PHIOGEN previously received $1.1 million from CARB-X in 2025 to evaluate PHI-BI-01, a separate programme targeting extraintestinal pathogenic E. coli bloodstream infections. That project was also described as combining immediate bacterial killing with the possibility of longer-term immune protection against reinfection, indicating that immunomodulation has become a broader component of the company's development strategy rather than an isolated experiment in recurrent UTI.
The new Baylor–LillixBio collaboration is particularly relevant because it attempts to provide a mechanistic foundation for that strategy. Demonstrating that patients or experimental animals show fewer recurrent infections after phage treatment would not by itself explain why protection occurred. A systematic antigen-presentation map combined with serology and immune profiling could potentially reveal which bacterial proteins are being recognized, which immune pathways are activated and whether specific immune responses correlate with protection.
This distinction will be essential if the approach is eventually developed as a pharmaceutical product. A reproducible immunizing therapy requires more than an observation that some treated subjects subsequently resist infection. Researchers would need to determine how consistently the immune effect occurs, how long it persists, whether it protects against genetically diverse E. coli strains and whether the immune stimulation remains safe.
The proposed use of immunomodulatory candidates adds another layer. LillixBio's platform will identify candidates predicted to amplify the immune response generated during phage-mediated bacterial lysis. Baylor's experimental systems can then investigate whether those interventions improve the magnitude, quality or duration of the response.
In principle, this could make the phage treatment function partly like an infection-specific immunization event, but there are major differences from conventional vaccination. Vaccines contain selected antigens delivered under controlled conditions before or sometimes after exposure to a pathogen. Phage therapy instead destroys actively infecting bacteria inside a patient, generating bacterial material directly at an infection site with an inflammatory environment that can vary widely between individuals.
The amount and composition of bacterial material released during phage treatment could also depend on bacterial density, phage replication, bacterial strain, site of infection and concurrent antibiotic treatment. Determining whether that variable process can produce a consistent and controllable immune response is therefore one of the scientific challenges underlying the collaboration.
The relationship between phages and immunity is itself complicated. Immune responses directed against bacteriophages can sometimes reduce therapeutic activity by neutralizing or clearing administered viruses. PHIOGEN's approach should not be confused with simply trying to increase immunity against the phage particles themselves. The central objective described by Baylor is to understand and exploit immune consequences associated with phage-mediated bacterial lysis, particularly responses to bacterial antigens that could contribute to protection against recurrent infection.
A successful dual-action therapy would therefore need to balance several processes simultaneously. Phages must remain active long enough to reach and destroy their bacterial targets, bacterial lysis must expose appropriate antigens, the host immune system must process those antigens effectively, and the resulting adaptive response must recognize future pathogenic bacteria without creating harmful inflammation.
AI can accelerate the first stage of identifying candidate antigens, but it cannot resolve those biological questions by itself. Predicted peptide presentation must be followed by biochemical, cellular and animal-model validation, and eventual claims of durable prevention would require prospective human clinical evidence.
This limitation is particularly important when interpreting PHIOGEN's description of PHI-UI-01. The programme is still preclinical, and the October 2026 collaboration does not report results from a randomized clinical trial demonstrating long-term prevention of recurrent UTI. No funding amount for the Baylor–PHIOGEN–LillixBio collaboration was disclosed in the announcement.
The programme should therefore be viewed as a mechanistic and translational research effort rather than evidence that phage treatment currently provides vaccine-like protection in patients. Its significance lies in formally investigating a biological effect that conventional phage development has often treated as secondary: the possibility that destroying bacteria could also reshape subsequent host immunity.
If the collaboration identifies reproducible antigenic signatures associated with protection, it could influence how future phage products are selected. Phages might eventually be evaluated not only according to host range, killing kinetics, resistance suppression and pharmacological characteristics, but also according to the immune context they generate after bacterial lysis.
Such an approach would represent a substantial expansion of how bacteriophages are considered therapeutically. Instead of functioning exclusively as self-replicating antimicrobials, selected phages could potentially become part of treatments designed to coordinate pathogen killing with adaptive immune training.
The strategy could be especially relevant to diseases characterized by repeated infections. Recurrent UTIs are an obvious starting point because many patients experience repeated episodes involving E. coli, but the same conceptual framework could eventually be investigated in other recurrent or persistent bacterial diseases if the underlying mechanism proves robust.
The immediate challenge is much narrower and more scientific: establish whether the immune response following phage-mediated E. coli lysis contains reproducible signals that can be predicted, measured and deliberately improved. The collaboration between PHIOGEN, Baylor and LillixBio brings together computational antigen mapping and direct experimental validation to answer that question.
If the hypothesis survives those tests, it would add a new dimension to the therapeutic potential of bacteriophages. The objective would no longer be limited to finding a virus capable of killing today's bacterial infection, but to understand whether that same intervention can help the immune system respond more effectively when the pathogen attempts to return.
Sources :
Baylor College of Medicine — “PHIOGEN, Baylor College of Medicine and LillixBio launch AI-driven collaboration to map phage-induced immunomodulation and long-term prevention,” October 8, 2026. The announcement describes the AI antigen-mapping programme, PHI-UI-01, E. coli, validation at TAILΦR and the objective of strengthening immune responses following phage-mediated bacterial lysis.
https://www.bcm.edu/news/phiogen-baylor-college-of-medicine-and-lillixbio-launch-ai-driven-collaboration-to-map-phage-induced-immunomodulation-and-long-term-prevention
PHIOGEN — Technology platform and pipeline. Baylor confirms that PHIOGEN was founded in 2023 to commercialize technology developed through TAILΦR, while PHIOGEN describes its phage discovery, directed-evolution and preclinical development platform.
https://www.phiogenpharma.com/
CARB-X — “CARB-X to support PHIOGEN’s phage-based project targeting E. coli infections,” August 26, 2025. CARB-X awarded PHIOGEN $1.1 million for PHI-BI-01, a separate ExPEC bloodstream-infection programme designed to investigate immediate bacterial killing together with longer-term immune protection.
https://carb-x.org/carb-x-news/carb-x-to-support-phiogens-phage-based-project-targeting-e-coli-infections/
World Intellectual Property Organization — Global Awards 2026, PHIOGEN. WIPO's profile identifies PHI-UI-01 as PHIOGEN's programme for recurrent urinary tract infections and describes the company's proposed dual antibacterial and immunological mechanism. These are programme claims rather than independent clinical validation.
https://www.wipo.int/en/web/awards/global/2026/
Hou Y, Lv Z, Hu Q, Zhu A & Niu H — “The immune mechanisms of the urinary tract against infections.” Frontiers in Cellular and Infection Microbiology 15 (2025). The review summarizes innate and adaptive immune mechanisms involved in UTI and highlights the complexity of generating effective protection against recurrent disease.
https://doi.org/10.3389/fcimb.2025.1540149
Bowyer GC et al. — “Tissue Immunity in the Bladder.” Annual Review of Immunology (2022). The review discusses bladder immune architecture and why adaptive responses can be insufficient to prevent recurrent infection.
https://pubmed.ncbi.nlm.nih.gov/35471839/
.png){kind=logo}
Comments
Post a Comment