Periodic Reporting for period 1 - PHAGECOUNTER (Discovery of counter defense strategies of bacteriophages against bacterial immunity)
Reporting period: 2023-04-01 to 2025-03-31
Our planet is truly microbial, with microbes driving essential functions across all ecosystems. But what impacts microbes most profoundly are their viruses—bacteriophages, or simply phages. When phages infect bacteria, the microbial hosts fight back using highly complex immune systems. The field of bacterial immunity has recently witnessed a surge of groundbreaking discoveries, with more than 200 different defense system types discovered over the last 5 years. It is now evident that bacterial viruses carry an arsenal of anti-defense strategies to neutralize bacterial defense systems, with the greatest diversity in their direct inhibitors. For a long time, research focused almost exclusively on anti-restriction-modification (anti-RM) and anti-CRISPR proteins, and despite recent discoveries expanding the known repertoire of inhibitors, these have so far been identified for only about 10% of all defense system types.
The diversity, distribution, and genomic organization of anti-defense systems remain largely unexplored. Because these proteins are small and fast-evolving, they often escape detection by traditional bioinformatics. To address this gap, my project developed a machine learning–based approach to systematically discover anti-defense proteins. This strategy enabled me to pursue three key objectives: (1) to identify a broad diversity of anti-defense proteins of phages, (2) validate their anti-defense properties, and (3) characterize a novel feature of anti-defense proteins – their broad specificity.
This work provides the first systematic framework for uncovering the hidden diversity of anti-defense systems and reveals new molecular strategies used by phages to overcome bacterial immunity. By identifying inhibitors with broad specificity, it also opens the door to understanding how phages adapt to diverse bacterial hosts and immune landscapes. These insights have broad implications for microbial ecology, phage therapy, and biotechnology, where modulating bacterial immunity is of growing interest. This research holds great promise for improving phage therapy, a promising alternative to antibiotic resistance.
The diversity, distribution, and genomic organization of anti-defense systems remain largely unexplored. Because these proteins are small and fast-evolving, they often escape detection by traditional bioinformatics. To address this gap, my project developed a machine learning–based approach to systematically discover anti-defense proteins. This strategy enabled me to pursue three key objectives: (1) to identify a broad diversity of anti-defense proteins of phages, (2) validate their anti-defense properties, and (3) characterize a novel feature of anti-defense proteins – their broad specificity.
This work provides the first systematic framework for uncovering the hidden diversity of anti-defense systems and reveals new molecular strategies used by phages to overcome bacterial immunity. By identifying inhibitors with broad specificity, it also opens the door to understanding how phages adapt to diverse bacterial hosts and immune landscapes. These insights have broad implications for microbial ecology, phage therapy, and biotechnology, where modulating bacterial immunity is of growing interest. This research holds great promise for improving phage therapy, a promising alternative to antibiotic resistance.
The overarching goal of this project was to identify and characterize phage-encoded anti-defense proteins that inhibit bacterial immune systems.
Aim 1: The first objective was to identify anti-defense proteins in phages and pair them with corresponding bacterial defense systems using a comparative genomics approach. We analyzed closely related strains of Vibrio bacteria and their phages of the Nahant collection, integrating genomic data with known infection outcomes from cross-infection matrix. However, we found that comparative genomics alone was insufficient to accurately pair defense systems with their phage-encoded inhibitors. To overcome this, I initiated a collaboration with a mathematician, applying a convolutional neural network model he developed and applied it to the cross-infection matrix of the Nahant collection. This approach successfully paired bacterial defense systems with candidate inhibitors in phage genomes, representing a major methodological improvement over the original plan.
Aim 2: To validate the predicted anti-defense proteins, I built a library of plasmid-cloned Vibrio defense systems and expressed them in a defense-less host strain, which we constructed for the purpose of screening without interference of genomically encoded defense genes. This clone library was tested for defense activity against an array of phages. For each defense system that showed a clear defense phenotype, I cloned the predicted anti-defense gene and assessed whether its expression suppressed the defense. 90% of all tested combinations confirmed the computational predictions, providing strong experimental validation.
Aim 3: The initial goal was to determine the timing of anti-defense gene expression during infection. However, during the course of the project, we discovered that several anti-defense proteins could inhibit multiple, non-homologous defense systems. This unexpected finding—rarely described and never systematically explored—prompted a shift in focus. The revised aim centered on confirming additional cases of broad specificity and initiating exploratory work into their molecular mechanisms of action. This adjustment was scientifically motivated, as such broadly acting anti-defense proteins could reveal common vulnerabilities across diverse bacterial immune systems and offer new conceptual entry points for the field.
Aim 1: The first objective was to identify anti-defense proteins in phages and pair them with corresponding bacterial defense systems using a comparative genomics approach. We analyzed closely related strains of Vibrio bacteria and their phages of the Nahant collection, integrating genomic data with known infection outcomes from cross-infection matrix. However, we found that comparative genomics alone was insufficient to accurately pair defense systems with their phage-encoded inhibitors. To overcome this, I initiated a collaboration with a mathematician, applying a convolutional neural network model he developed and applied it to the cross-infection matrix of the Nahant collection. This approach successfully paired bacterial defense systems with candidate inhibitors in phage genomes, representing a major methodological improvement over the original plan.
Aim 2: To validate the predicted anti-defense proteins, I built a library of plasmid-cloned Vibrio defense systems and expressed them in a defense-less host strain, which we constructed for the purpose of screening without interference of genomically encoded defense genes. This clone library was tested for defense activity against an array of phages. For each defense system that showed a clear defense phenotype, I cloned the predicted anti-defense gene and assessed whether its expression suppressed the defense. 90% of all tested combinations confirmed the computational predictions, providing strong experimental validation.
Aim 3: The initial goal was to determine the timing of anti-defense gene expression during infection. However, during the course of the project, we discovered that several anti-defense proteins could inhibit multiple, non-homologous defense systems. This unexpected finding—rarely described and never systematically explored—prompted a shift in focus. The revised aim centered on confirming additional cases of broad specificity and initiating exploratory work into their molecular mechanisms of action. This adjustment was scientifically motivated, as such broadly acting anti-defense proteins could reveal common vulnerabilities across diverse bacterial immune systems and offer new conceptual entry points for the field.
A major and unexpected outcome of this project was the discovery that some phage-encoded anti-defense proteins can inhibit multiple, unrelated bacterial immune systems. While individual cases of such broadly specific anti-defenses have been sporadically reported, this is the first systematic demonstration that broad-spectrum anti-defense mechanisms exist and may be more widespread than previously recognized. This finding significantly advances the field by revealing a new layer of complexity in the phage–bacteria arms race and challenges the long-standing assumption that anti-defense proteins are narrowly specific. The next critical step is to elucidate the molecular mechanisms underlying this broad activity, which will provide fundamental insights into how phages evolve to overcome diverse bacterial immune strategies.
In addition to its basic scientific importance, this discovery has strong applied potential. Broadly acting anti-defense proteins may inform the design of improved phage therapies by enhancing a phage’s ability to infect and kill multi-drug resistant bacteria. Moreover, because these proteins modulate immune activity, they may serve as leads for developing novel immunomodulatory tools relevant beyond microbiology. To fully exploit these findings, further research is needed to: (i) biochemically characterize the mechanisms of broad specificity of anti-defense proteins, (ii) assess their prevalence across phage populations and host species, and (iii) evaluate their utility in engineered phage cocktails to use in phage therapy. The work served as the basis for my ERC Starting Grant (StG) application, and as of the time of writing, I have progressed to the interview stage. Support for continued research, potential intellectual property (IP) protection, and partnerships with translational or biotech entities will be crucial in moving these discoveries toward application.
In addition to its basic scientific importance, this discovery has strong applied potential. Broadly acting anti-defense proteins may inform the design of improved phage therapies by enhancing a phage’s ability to infect and kill multi-drug resistant bacteria. Moreover, because these proteins modulate immune activity, they may serve as leads for developing novel immunomodulatory tools relevant beyond microbiology. To fully exploit these findings, further research is needed to: (i) biochemically characterize the mechanisms of broad specificity of anti-defense proteins, (ii) assess their prevalence across phage populations and host species, and (iii) evaluate their utility in engineered phage cocktails to use in phage therapy. The work served as the basis for my ERC Starting Grant (StG) application, and as of the time of writing, I have progressed to the interview stage. Support for continued research, potential intellectual property (IP) protection, and partnerships with translational or biotech entities will be crucial in moving these discoveries toward application.