DRY LAB
Does our software provide assistance for experimental design?
Proof of concept:
Our software, Alphage, integrates multiple deep learning models trained on large datasets, ensuring high prediction accuracy and reliable results. By comparing the tail fiber protein sequences of the target phage and T7, and evaluating homology, structural stability, and host infection efficiency, Alphage comprehensively ranks and selects the optimal replacement strategy. This provides an a priori prediction for the experimental design of replacing T7 tail fiber protein with that of phage phi PSA 17, reduces trial-and-error costs, and accelerates project development.
Future steps:
Although we have developed a user-friendly and efficient system, unfortunately, time constraints have limited our ability to expand the underlying dataset. In the future, we plan to optimize the software for faster search and execution, incorporate additional phage data, and perform comprehensive functional testing.
Does our software offer user-friendly operation?
Proof of concept:
All source code and dependencies have been uploaded to GitLab, and a detailed user manual provides comprehensive guidance on installation and usage. We have also designed a simple and efficient graphical user interface (GUI) to streamline user interaction. To enhance user experience, results are presented through a combination of charts and text, intuitively displaying prediction schemes for users to evaluate and select.
Future steps:
Although the GUI has undergone multiple iterations, user feedback remains limited. Going forward, we aim to reach a broader user base and continue refining the interface and interaction logic based on real-world feedback.
Does our model deliver real-world simulation accuracy?
Proof of concept:
Model parameters and conditions are derived from Wet Lab experiments and literature data, enabling simulation of real-world scenarios under experimental conditions to a reasonable extent. The Stability Model for a Dual-Plasmid System successfully simulated the competition between pUC57 and pUC19, guiding the Wet Lab in optimizing the vector design. At the structural level, the Protein-Protein Docking Model used AlphaFold3 to predict binding affinity and stability after replacing the T7 phage gp17 protein with the tail fiber protein of phage phi PSA 17. This verified the feasibility of the replacement strategy and increased confidence in its experimental implementation.
Future steps:
Although most model data are derived from experiments and literature, exact simulation of real-world processes remains challenging, and some experimental details remain unaccounted for. Going forward, we will work more closely with Wet Lab teams to enable mutual validation and conduct targeted field studies to collect real-world data. This will enhance the model’s realism and predictive accuracy.
WET LAB
Has it been successfully verified that phage-like particles can be successfully produced?
Proof of concept:
To verify that the plasmid was successfully packaged and capable of transducing into bacteria, we constructed a functional validation system, ΦOR-pacB-DTR-pacC-pTetA-bjGFP-Ter. Using the green fluorescent protein bjGFP as a visual reporter gene, this system fully simulates and confirms the entire process from plasmid rolling-circle replication, concatemeric DNA packaging, and bacteriophage-like particle formation to gene delivery and heterologous gene expression can be achieved.
When this class of phage particles infected the host bacteria and injected into their DNA, PTetA-bjGFP was expressed in the host bacteria. Under conditions without TetR deterrence, the TetA promoter drove sustained expression of the bjGFP gene, and we observed green fluorescence from the host colonies on Amp plates. This clear fluorescent phenotype visually demonstrates that the full chain design from plasmid roll-up replication, multiplex DNA packaging, phage-like particle formation to gene delivery and heterologous gene expression is successful and effective.
Future steps:
Due to time constraints, we did not integrate the J23119-TetR-MazE-Ter pathway into benthic bacteria, and therefore did not package the ΦOR-pacB-DTR-pacC-pTetA-MazF-Ter pathway into phage-like particles. We need to replace the reporter gene bjGFP with MazF to verify that phage-like particles with toxin genes can successfully kill the target pathogens in future experiments, and replace the reporter gene bjGFP with other genes to verify the feasibility of phage-like particles as delivery platforms.
Are vectors with dual screening functions successfully constructed?
Proof of concept:
In order to construct strains that could successfully achieve homologous double exchange of the target fragment with the chassis bacterial genome by Amp and sucrose double screening, we inserted the SacB gene, which secretes levansucrase, into pKD46 with a recombinant system. The recombinant vector was constructed using seamless cloning, and the integrity and correctness of the dual-screening functional vector was verified at the molecular level by PCR. The chassis bacteria containing the dual-screening functional vectors were cultured, and the medium for dual screening with antibiotics and sucrose was set up for culture-control experiments to verify that the dual-screening functional vectors had the expected functionality at the cellular level.
Future steps:
We did not quantitatively verify the intensity of levansucrase expression at the transcriptional and translational levels, and therefore could not determine the threshold at which sucrose levels would have an effect on the growth of chassis bacteria expressing the SacB gene, which could lead to false positives in the double screen. We did not set up a control experiment for the growth of chassis bacteria with or without the double-screened vector to determine whether the stress of the vector on the chassis bacteria affected their growth. We also need to quantify the relationship between levansucrase expression and chassis bacteria growth by determining the enzyme activity of levansucrase using a colorimetric assay with 3,5-dinitrosalicylic acid.
Whether the deleted phage genome can be properly expressed in E. coli?
Proof of concept:
To verify whether the deleted phage genome could be expressed in E. coli, we transformed the constructed homologous recombinant vector into E. coli BL21 (DE3) strain. This strain can express T7 RNA polymerase under IPTG induction, which drives protein expression of phage genes carried in recombinant plasmid pKD46-SacB-1, 2, 3 and 4. Comparison of the total protein of the transformants with that of wild-type BL21 by SDS-PAGE can detect whether the target protein is expressed or not.
Future steps:
Subsequent experiments will firstly optimise the expression vector to exclude the influence of copy number on protein expression; subsequently, the protein expression species will be qualitatively detected in E. coli MG1655 integrating the T7 RNA polymerase gene; finally, the plasmid containing the packaging site will be transfected into this engineered bacterium, and it will be observed by electron microscopy whether phage-like particles are successfully assembled, so as to further demonstrate the expression and assembly ability of the deleted genome.
Was the gp17 gene successfully replaced?
Proof of concept:
To modify the host range of phage-like particles, we attempted to replace the gp17 gene in the T7 phage genome with the gene HQO98_gp43 encoding the phage phi PSA 17 tail fibronectin.We verified the gene replacement by both phage genome PCR and sequencing and Western Blot. Phage genome PCR and sequencing demonstrated that the gp17 gene was replaced at the molecular level; Western Blot demonstrated that the protein was altered after gene replacement at the expression level.
Future steps:
When quantifying the phage target proteins obtained we found that the target proteins in the bacteriophage precipitate lysed by the phage were much more than those in the supernatant (which is usually considered to contain only active phage), which suggests that the gene was recombined and expressed efficiently but the amount of engineered phage obtained was extremely low. Unfortunately, however, due to time constraints we were unable to perform colloidal gold electron microscopy scanning of the engineered phage with his tag added to verify the binding of recombinant tail filaments to phage torsos, which needs to be verified and improved in future experiments.
Can an engineered phage with modified tail filaments do what we want it to do?
Proof of concept:
The ultimate goal of altering the tail filament proteins of phage is to allow the engineered phage to effectively target and adsorb the target bacterium and to inject the DNA contained in the head shell into the target bacterium. the results of Western Blot showed that we have obtained the tail filament-altered engineered phage, and we verified the gene substitution by two methods: transmission electron microscopy and packaging plasmid as a screening marker. The results of transmission electron microscopy demonstrated that the engineered phage could effectively target and adsorb the target bacteria, while the screening experiments of packaging plasmid as a screening marker demonstrated that the engineered phage could inject the DNA contained in the head shell into the target bacteria.
Future steps:
At present we have no way of knowing how well the recombinant tail filament binds to the phage torso and how efficiently the plasmid is packaged, i.e., we have not quantified the efficiency of engineered phage production. Nor have we quantified the minimum amount of phage to function and the efficiency of its action. This needs to be measured in future quantitative experiments and modelling.