Overview

Our project rigorously adheres to the “Design-Build-Test-Learn” (DBTL) engineering cycle, a framework that guides all aspects of our work, from wet and dry lab research to integrated Human Practices.

In the Design phase, we design not only biological systems but also our public engagement and education strategies, while identifying key knowns and unknowns in both technical and societal contexts.

During the Build phase, we assemble biological parts in the lab while also constructing platforms for dialogue with communities, experts, and stakeholders.

The Test phase involves validating the function of biological components and gauging public understanding and acceptance of our technology through interactive activities.

Finally, in the Learn phase, we synthesize insights from both experimental data and societal feedback to close the loop.

This comprehensive and iterative engineering approach ensures our project advances responsibly on both scientific and social fronts, embodying the “middle-out” philosophy essential to synthetic biology.

Wet Lab

Overview

We follow a step-by-step development process for each section, progressively constructing a programmable phage-like particle platform. We designed and validated the synthesis and DNA delivery capabilities of phage-like particles; Building upon this foundation, we engineered a stable and controllable chassis organism using genome integration techniques. Finally, through modular tail-reprogramming, we achieved specific targeting of novel hosts by the phage-like particles. Layered validation confirmed their specific recognition and infection of target pathogens. Each iteration deepened our understanding of the system’ s controllability and functional programmability, driving continuous refinement of the platform.

Verification of the Synthesis and Infectivity of Phage-like particles

To address the uncontrollable self-replication of natural phages, we plan to replace the natural genome within the head coat of the wild-type T7 phage with DNA concatemer formed by rolling-circle replication of the virulent plasmid. The resulting phage-like particles retain the ability to infect target bacteria and deliver toxin genes, leading to host cell death. Since the DNA fragment lacks genes for phage coat synthesis, these particles cannot produce progeny, thereby enabling controlled bactericidal activity. To visualize the synthesis and infectivity of the phage-like particles during validation, we replaced the toxin gene MazF on the plasmid with bjGFP (a gene encoding green fluorescent protein), and conducted the following experiments.

Figure 1: The model of packaging visualization plasmid

1st Iteration

Design

We aim to infect E. coli MG1655 containing pUC19-packaging-P_TetA-bjGFP with WT T7 to obtain T7 Mix containing both WT T7 and phage-like particles. Subsequently, we infected WT E. coli MG1655 (untransformed with the plasmid) using the T7 Mix. Ultimately, we observed bjGFP expression in the bacteria, with colonies emitting green fluorescence.

Build

We cultured E. coli MG1655 containing the pUC19-packaging-P_TetA-bjGFP plasmid to an OD₆₀₀ of 0.7, added 4×10¹⁰ PFU/mL of the WT T7, and incubated at 37 °C with shaking until most bacteria lysed. Add 0.2 mL chloroform, vortex vigorously, centrifuge, and collect the supernatant. Filter through a 0.22 μm membrane to obtain the T7 Mix.

Subsequently, we cultured uninoculated WT E. coli MG1655 to an OD₆₀₀ of 0.5 and placed it on ice. We performed a 10⁻¹³ serial dilution of the T7 Mix, with the final well containing LB as a blank control. Take 20 μl of each dilution and mix with 20 μl of recipient bacterial suspension or LB (cell-free control). Incubate at 37 °C with shaking for 1 hour. Plate 10 μl of each mixture onto the LB plates containing 50 μg/mL Amp and incubate at 37 °C overnight.

Test

In all the dilution gradient experimental groups, no colonies were observed growing on LB plates containing Amp. Simultaneously, no colony growth was observed in the cell-free control group, ruling out the possibility of medium contamination.

Learn

Upon careful literature review, we discovered that the successful construction of similar packaging systems referenced in our work utilized the BW25113 ΔtrxA strain. The thioredoxin encoded by the trxA gene serves as a host factor essential for the functional activity of the T7 phage DNA polymerase. In the ΔtrxA strains, T7 DNA polymerase cannot be effectively “activated,” resulting in extremely low replication efficiency. Consequently, wild-type T7 phage cannot complete effective genomic replication in this host, with its life cycle blocked at the DNA replication stage, significantly inhibiting its ability to lyse bacteria.

In contrast, the WT E. coli MG1655 strain used in our experiments possesses an intact trxA gene. Consequently, when T7 Mix infects WT E. coli MG1655, the numerically dominant wild-type T7 phages with full replication capacity rapidly complete the entire cycle of adsorption, injection, replication, assembly, and lysis.

We hypothesize that the rapid and efficient lysis by WT T7 phage eliminates the majority of host cells before they can stably take up and express the DNA injected by phage-like particles. This results in the absence of any surviving colonies on plates containing ampicillin.

2nd Iteration

Design

To minimize interference from the WT T7 lysing WT E. coli MG1655 during the final assessment of phage-like particle infectivity, we attempted to modify the final detection method. After obtaining the T7 Mix, we mixed WT E. coli MG1655 with the T7 Mix to create the double-layer plate, hoping to observe colonies emitting green fluorescence on the upper layer.

Build

We obtained the T7 Mix using the same method as the previous iteration. Then, 100 μl of OD₆₀₀=0.7 WT E. coli MG1655 was mixed with 100 μl of the T7 Mix. After incubating at room temperature for 10 min, we added 4 ml of LB containing 0.7% agar and 50 μg/ml Amp, poured it onto a 1.5% agar plate, and cultured it at 37 °C for 12 hours to observe the results.

Test

In the WT E. coli MG1655 + T7 Mix group, we observed colony growth but no fluorescence. The control group (LB + T7 Mix) showed no colony growth, ruling out the possibility of contaminating bacteria. We picked colonies from the G1 plate and cultured them overnight in liquid LB containing 50 μg/mL Amp, hoping to observe the phenotypic expression of bjGFP. Unfortunately, after overnight incubation, the liquid remained completely clear with no bacterial growth.

Figure 2: Left (G1): WT E. coli MG1655+T7 Mix, Right (G5): LB+T7 Mix

Learn

Based on the above results, we speculate that the bacterial growth in liquid medium is inhibited because WT T7 inevitably contaminates the plate during colony picking. WT T7 proliferates extensively and lyses the host, preventing the host bacteria from growing in the liquid medium. Therefore, we are unable to verify the synthesis and infectivity of phage-like particles using the double-layer plate method.

3rd Iteration

Design

To more convincingly validate the synthesis and infectivity of the bacteriophage-like particles, we attempted a novel detection method. After obtaining the T7 Mix, we co-cultured the WT E. coli MG1655 with the T7 Mix in the liquid medium for a period of time, expecting the bacteriophage-like particles to inject the DNA within their capsids into the WT E. coli MG1655 during this period. Subsequently, we resuspended the bacteria in LB medium, washing away as much T7 Mix as possible. The cells were then spread onto LB plates containing 50 μg/mL Amp and incubated at 37 °C overnight. We anticipated observing colonies that would grow and emit green fluorescence.

Build

We prepared the T7 Mix using the same method as the previous iteration. Then, we mixed OD₆₀₀=0.6 WT E. coli MG1655 with the T7 Mix at the three different ratios, cultured the mixture in liquid LB for 1.5 hours, collected the cells, and resuspended them three times in LB to wash away as much T7 Mix as possible. Finally, the cells were spread onto LB plates containing 50 μg/ml Amp and incubated overnight at 37 °C.

Test

In all three experimental groups with different mixing ratios, no colony formation was observed. The plates remained clear, inconsistent with the expected growth of green fluorescent colonies.

Learn

Based on these results, we hypothesize that the absence of colonies on LB plates containing 50 μg/ml Amp is due to the high proportion of WT T7 in the T7 Mix. During co-culture with WT E. coli MG1655, the majority of WT E. coli MG1655 was infected by WT T7. We were unable to sufficiently reduce the interference of WT T7 in the validation infection experiment, thus failing to observe the desired phenotype.

4th Iteration

Design

Due to the unfavorable progress of previous infection experiments, we adopted the stratified validation approach to verify whether the plasmid rolling circle replication-formed DNA concatemer was successfully packaged within the head capsid of the phage-like particles in the T7 Mix, thereby confirming the successful synthesis of the phage-like particles. Simultaneously, in experiments verifying the infectivity of the phage-like particles, we attempted to incorporate a phage inhibitor to mitigate the influence of wild-type T7 during co-culture with WT E. coli MG1655. This phage inhibitor demonstrates varying degrees of suppression against a broad spectrum of phages while allowing normal bacterial growth and protein expression.

Build

For verifying phage-like particle formation, we obtained the T7 Mix using the same method as previous iterations. A portion of T7 Mix underwent heat lysis to release DNA from the particle capsids. The M13-F and M13-R primers were then selected for PCR amplification to detect specific bands.

For verifying phage-like particle infectivity, we cultured WT E. coli MG1655 with phage inhibitor added at a 1:500 ratio. When OD₆₀₀ reached 0.7, 8 ml of OD₆₀₀=0.7 WT E. coli MG1655 was mixed with 300 μl T7 Mix. The mixture was cultured in liquid LB for 1.5 hours, then the cells were harvested and resuspended three times in LB to wash away as much T7 Mix as possible. Finally, the cells were spread onto LB plates containing 50 μg/ml Amp and incubated at 37 °C overnight.

Test

In the PCR reaction, our target specific band is 2028 bp. Lane 1 exhibited three bands, with the longest band matching the expected length. Sequencing results confirmed the target band’ s sequence is entirely correct.

Figure 3: The results of PCR performed on T7 Mix after thermal cracking

M: star marker D5000 ladder; 1: PCR products from the T7 Mix after thermal cracking

Figure 4: The design of detection primer and the sequencing results

After overnight incubation at 37°C, we observed the growth of target colonies emitting green fluorescence in all three replicate groups under the identical conditions across the three experimental groups, consistent with our expected results.

Figure 5: Phage inhibitor-treated E. coli MG1655 co-cultured with T7 Mix

Learn

On LB plates containing 50 μg/ml Amp, we obtained the expected results, confirming that DNA from the phage-like particles was successfully injected into WT E. coli MG1655 and that the reporter gene bjGFP was successfully expressed. However, in the electrophoresis gel, we observed two non-specific bands in addition to the target band, suggesting possible interference from impurity DNA in our T7 Mix. We analyzed the composition of the previously used T7 Mix and the entire heat-lysate process, realizing that the free target plasmid DNA might be present in the T7 Mix. This DNA cannot be removed using filter membranes or the heat-lysate process. Therefore, we need to further eliminate the interference to more robustly demonstrate the presence of target DNA within the phage-like particles.

5th Iteration

Design

After obtaining the T7 Mix, we thoroughly treated it with the RNase and the DNase to digest any potential plasmid DNA and other impurities, yielding T7 Mix 2.0. Subsequently, we performed PCR reactions using the same parameters as the previous iteration to detect the specific bands, verifying the DNA within the phage-like particles. We then re-conducted experiments related to the infectivity of the phage-like particles using T7 Mix 2.0.

Build

After obtaining the T7 Mix, we added 20 μl RNase and 5 μl DNase, mixed thoroughly, and incubated at 37°C for 30 minutes. We then added the phage pellet, centrifuged, aspirated the supernatant, resuspended in LB, and collected the T7 Mix 2.0. Subsequently, we used T7 Mix 2.0 to perform the same experiments as before to validate the synthesis and infectivity of the phage-like particles.

Test

After ruling out interference from free plasmid DNA in the T7 Mix, we repeated the PCR reaction and obtained the results shown in Figure 6.

In the PCR reaction, we observed the target specific band, and sequencing results confirmed the target band’ s sequence was entirely correct. Simultaneously, PCR reactions performed on the control group (WT T7 lysate) did not yield any relevant specific bands.

Figure 6: The results of PCR performed on T7 Mix 2.0 after thermal cracking

M: star marker D5000 ladder; 1: The PCR products from thermally lysed T7 Mix 2.0; 2: The PCR products from thermally lysed WT T7 phage

After overnight incubation at 37 °C, the experimental group observed the growth of target colonies emitting green fluorescence, consistent with our expected results. The control group showed no colony growth, ruling out interference from extraneous microorganisms.

Figure 7: Phage inhibitor-treated E. coli MG1655 co-cultured with T7 Mix 2.0

Learn

After thorough RNase and DNase treatment of the T7 Mix, we eliminated interference from the free plasmid DNA and other contaminating DNA. Following heat-lysate PCR, we confirmed the successful synthesis of the phage-like particles. On the LB plates containing 50 μg/ml Amp, we observed target colonies emitting green fluorescence, confirming that the DNA within the phage-like particles was successfully injected into WT E. coli MG1655 and that the reporter gene bjGFP was successfully expressed.

Reconstructing the Phage Genome

1st Iteration

Design

The process of segmentally reassembling phage genomes onto E. coli involves three steps: first, obtaining phage genomic fragments; second, seamlessly cloning these fragments into the pKD46-SacB vector; and finally, performing double-exchange screening to integrate the genomic fragments into the E. coli genome.

For the first step - amplifying T7 phage genomic fragments - we propose treating the phage lysate supernatant with the thermal lysis method. This process denatures the phage capsid through thermal disruption, releasing the encapsulated DNA. This approach offers a straightforward and efficient method for obtaining phage genomic fragments. The treated liquid is directly used as a template to amplify T7g-1-1, T7g-1-2, T7g-2, T7g-3, T7g-4-1, and T7g-4-2. Homologous arms are added to the 5’ ends of the upstream and downstream primers to enable homologous recombination with the vector.

Build

After inoculating with phages, the bacterial lysate obtained from overnight cultures was filtered through a membrane to yield lysate supernatant. Following treatment at 100°C for 4 minutes, the lysate supernatant was directly used as template for PCR to amplify phage genomic fragments. We aimed to obtain PCR amplification templates through simple and rapid procedures and successfully acquire amplified genomic fragments.

Test

Gel electrophoresis analysis of our PCR products revealed that only T7g-1-1 was successfully amplified. Sanger sequencing of this fragment confirmed the accuracy of the product.

After confirming the partial feasibility of the thermal lysis approach, we repeated the experiment and attempted to incorporate a centrifugation step to precipitate proteins and other impurities. However, the results remained unsatisfactory: longer genomic fragments still failed to amplify successfully, and T7g-2 exhibited significant non-specific amplification.

Figure 8: Results of PCR amplification of genomic fragments using the pyrolysis-treated supernatant as template

M: Star Marker 1kb Ladder; 1: T7g-1-1; 2: T7g-1-2; 3: T7g-2; 4: T7g-3; 5: T7g-4-1

Learn

We suspect this may be due to excessive impurities in the pyrolysis-treated supernatant interfering with PCR, and possibly due to excessive thermal treatment causing fragmentation of the phage genome. Both factors could potentially lead to the PCR amplification failure.

2nd Iteration

Design

To eliminate the influence of impurities in the lysate supernatant and obtain the complete T7 phage genome, we envisioned purifying the phage genome and then using it as a PCR template to attempt amplification of the phage genome fragment again.

Build

We extracted the T7 phage genome from the lysate supernatant using the phage genome extraction kit. After diluting it to 10 ng/μL, we used it as a PCR template to amplify genomic fragments.

Test

We performed the PCR amplification using the extracted phage genome as a template, amplifying the T7g-1-1, T7g-1-2, T7g-2, T7g-3, T7g-4-1, and T7g-4-2 genomic fragments. The PCR products were analyzed by electrophoresis and Sanger sequencing, both confirming correct results. All phage genomic fragments were successfully obtained.

Figure 9: Agarose gel electrophoresis of individual genomic fragments amplified via phage genome PCR

M1: Star Marker 1kb Ladder; 1: T7g-1-1; 2: T7g-1-2; 3: T7g-2; 4: T7g-2; 5: T7g-4-1; 6: T7g-4-2; M2: Star Marker D5000 Ladder

Learn

Our method for amplifying the extracted T7 phage genomic fragments proved viable, successfully incorporating the homologous arms required for seamless cloning at both ends of the fragments. The next step involves constructing the homologous recombination vector and verifying the proper expression of the gene fragments.

Tail fiber replacement: Phage infection

Selection of replacement site

1st Iteration

Design

Taking tomato bacterial spot disease as an example, to enhance the T7 phage’ s ability to specifically recognize and infect the pathogenic bacterium Pseudomonas syringae, we designed a strategy to replace the tail fibers of the T7 phage.

We plan to employ homologous recombination technology to replace the key tail filament protein gene gp17 in the T7 phage genome with the tail filament protein gene HOQ98_gp43 from the Pseudomonas syringae phage phi psa 17. This will completely replace the gp17 protein in its tail filament with the VO98_215 protein, thereby improving its binding efficiency to the pathogen’ s outer membrane receptor.

Build

We selected 50 bp sequences upstream and downstream of the T7 phage gp17 gene as the homologous arms for genomic homologous recombination. The HOQ98_gp43 gene was placed between these arms and inserted into the high-copy vector pUC19, constructing the recombinant vector pUC19-VO98_215. Subsequently, wild-type T7 phage was used to infect E. coli MG1655 harboring the pUC19-VO98_215 plasmid. Following infection, the T7 genome was injected into the host as linear double-stranded DNA and underwent homologous recombination with the fragment on the host’ s pUC19_VO98_215 plasmid. The gp17 gene in its genome was replaced with the HOQ98_gp43 gene, releasing recombinant progeny with altered tails and tail filaments.

Finally, we infected the target bacterium Pseudomonas syringae DC3000 with the obtained recombinant phage.

Figure 10: The model of Genome - Plasmid pUC19-VO98_215 Homologous Recombination

Test

Following the infection experiment, we observed no bacteriophages on the agar plates containing Pseudomonas syringae DC3000. Furthermore, the titer of the recombinant bacteriophage was lower than that of the wild-type T7 bacteriophage in parallel experiments.

Figure 11: Cycle1.1 Double-Layer Plating

Learn

We conducted an in-depth analysis of the functional structure of the tail filament protein and reviewed extensive literature. Studies indicate that the N-terminal domain of the T7 bacteriophage’ s gp17 protein is responsible for connecting the tail filaments to the viral particle, while the C-terminal domain mediates binding to host receptors.

Following the complete replacement of the gp17 protein in the tail filament with the VO98_215 protein, the tail filament may fail to attach to the viral particle due to structural alterations.

2nd Iteration

Design

We identified that conventional protein sequence alignment methods were insufficient for determining substitution sites and strategies within the required timeframe, which led us to seek assistance from the dry lab team. They designed Alphage, a software tool for the automated prediction of tail fiber protein substitution strategies. This software pinpointed the optimal insertion site for the heterologous tail substitution in the gp17 protein to be within the amino acid range of 140-150.

Following this, we consulted the literature and utilized Protein-Protein Docking Model provided by the dry lab team to validate this series of sites. We ultimately selected the 149th residue as the final substitution site, based on its superior availability and post-substitution stability. The C-terminal region of the modified heterologous tail fiber specifically recognizes Pseudomonas syringae, while the N-terminal T7 fragment ensures stable attachment to the virus particle.

Figure 12: Sequence Alignment of Protein gp17 and VO98_215 (ESPript 3.0)

Build

We selected the 50-bp region upstream of the codon for amino acid 149 in the T7 phage gp17 gene as the upstream homologous arm for genomic homologous recombination substitution, while the downstream homologous arm was identical to that used for full-length replacement. The segment of the HOQ98_gp43 gene spanning from the codon for amino acid 149 to the stop codon was positioned between the homologous arms and inserted into the high-copy vector pUC19, yielding the recombinant vector pUC19-VO98_215(C-). Subsequent steps mirrored the full-length replacement protocol. Following T7 infection of the host bacterium, the downstream portion of the gp17 gene in the genome was replaced with the HOQ98_gp43 gene, releasing genetically altered, heterozygous tail protein recombinant progeny.

Finally, we infected the target bacterium Pseudomonas dongfangensis DC3000 with the obtained recombinant phage.

Figure 13: The model of Genome - Plasmid pUC19-VO98_215(C-) Homologous Recombination

Test

Following infection experiments, no plaques were observed on the Pseudomonas syringae DC3000 colony-containing plates. Moreover, the titer of the recombinant phage was lower than that of the wild-type T7 phage in parallel experiments.

Figure 14: Cycle1.2 Double-Layer Plating

Learn

After careful analysis, we identified several potential causes:

• Failure of homologous recombination

• Inability to express the tail filament protein

• Inability of the recombinant phage to adsorb, infect, or effectively lyse the target bacteria.

All of these factors could potentially explain the absence of plaques observed on the Pseudomonas syringae DC3000 bacterial lawn plates.

We recognize that more detailed experiments are required to systematically validate whether the tail replacement was successful and whether the resulting phage-like particles can effectively infect the target bacterium DC3000.

Tiered verification

Design

To address whether tail replacement was successful and whether the resulting phage-like particles could effectively infect the target bacterium DC3000, we designed experiments at the DNA and protein levels to validate tail replacement and phenotypic phage adsorption and infection of the target bacterium, respectively.

<DNA Level>

Specific primers were designed to validate, via PCR and sequencing, the presence of phages with successful genomic recombination and gene replacement in the recombinant progeny produced after T7 infection of host bacteria (Escherichia coli MG1655 (pUC19_VO98_215) and Escherichia coli MG1655 (pUC19_VO98_215(C-))).

<Protein Level>

Add a tag to the recombinant tail protein and use Western Blot to separate and verify the expression of the target protein.

<Adsorption Level>

Use transmission electron microscopy to capture the adsorption of recombinant phages onto DC3000.

<Infection Level>

Since phage plaques cannot be directly observed, other screening markers are used to indirectly verify the infection of DC3000 by recombinant phages.

Build

<DNA Level>

Following T7 phage infection of host bacteria (Escherichia coli MG1655 (pUC19-VO98_215) and Escherichia coli MG1655 (pUC19-VO98_215(149-))), a liquid mixture containing recombinant progeny was produced. After filtration and sterilization, the phage genome was extracted. Specific primers rr1 and rr2 were designed for genomic PCR, with preliminary validation via electrophoresis. Subsequently, genomic PCR was performed using specific staggered primers VO98_215/(150-)-up and VO98_215/(150-)-down. After electrophoretic verification, sequencing and alignment were conducted.

Figure 15: Binding Status of Specific Primers

<Protein Level>

In plasmids pUC19-VO98_215 and pUC19-VO98_215(C-), 6xCAT was inserted between the HOQ98_gp43 gene and downstream homologous arms, yielding two His-tagged plasmids: pUC19-VO98_215(C-his) and pUC19-VO98_215(C-)(C-his). A 6xHis tag was added to the successfully expressed tail filament protein. Following magnetic bead-based preliminary separation, the proteins were ultrafiltration-concentrated and finally separated and verified via Western Blot to assess target protein expression.

Figure 16: The model of His - Tag Plasmid

<Adsorption Level>

Transmission electron microscopy (TEM) imaging was performed to visualize the adsorption of recombinant phage onto DC3000. A mixture of recombinant phage and DC3000 was incubated at a fixed ratio and used as the TEM sample. A control group featuring adsorption of wild-type T7 phage onto DC3000 at the same ratio was included to validate changes in the adsorption capacity of the recombinant phage onto DC3000, thereby confirming the functionality of the replaced tail filament.

<Infection Level>

Green fluorescence is used as the forward screening marker. The packaging plasmid pUC19-pac_J23119 containing the fluorescent protein bjGFP expression cassette was introduced into Escherichia coli MG1655 (pUC19-VO98_215) and Escherichia coli MG1655 (pUC19-VO98_215(C-)) to construct dual-plasmid host strains: Escherichia coli MG1655 (pUC19-VO98_215, pUC19-pac-J23119-GFP) and Escherichia coli MG1655 (pUC19-VO98_215(C-), pUC19-pac-J23119-GFP). Infection of the dual-plasmid hosts with wild-type T7 phage yielded phage particles with capsids containing only the pUC19-pac-J23119-GFP plasmid, where the tail protein had been replaced. DC3000 liquid culture was inoculated with the phage particles, mixed, and spread onto the resistance plates to indirectly verify the infectivity of the recombinant phage against DC3000.

Figure 17: The model of packaging plasmid pUC19-pac-J23119-GFP

Test

<DNA Level>

Genomic PCR results for all 4 specific primer pairs matched expectations:

• Primer VO98_215-up_test showed a band only in T7 ∆gp17::VO98_215.

• Other primers produced bands in all strains except wild-type T7, with correct band sizes.

Sequencing results confirmed the presence of phages with successful genomic recombination and gene replacement in the recombinant progeny.

Figure 18: The results of phage genome PCR.

For primers VO98_215-up_test and VO98_215-down_test, the expected positive result band size is approximately 200bp.

For primer VO98_215/(C-)-up, the expected positive result band size is approximately 700bp.

For primer VO98_215/(C-)-up, the expected positive result band size is approximately 1400bp.

<Protein Level>

Wild-type T7 supernatant showed no bands, while T7 ∆gp17::VO98_215 and T7 ∆C-gp17::C-VO98_215 exhibited bands, confirming successful expression of the new tail filament protein following gene replacement.

Figure 19: Western Blot of phage WT-T7, T7 ∆gp17::VO98_215 and T7 ∆C-gp17::C-VO98_215.

The expected band size of the target protein is approximately 64 kDa, and it is present only in the phage T7 ∆gp17::VO98_215 and T7 ∆C-gp17::C-VO98_215.

<Adsorption Level>

Wild-type T7 phages only approached DC3000 cells and exhibited negligible adsorption. Recombinant phages T7 ∆gp17::VO98_215 and T7 ∆C-gp17::C-VO98_215 showed varying degrees of adsorption to DC3000 cells.

Figure 20: Negative staining Transmission Electron Microscope detects the ability of phage with recombinant tail fiber to infect Pseudomonas syringae pv. Tomato DC3000.

A: Dividing Pseudomonas syringae pv. Tomato DC3000; B: Pseudomonas syringae pv. Tomato DC3000; C: T7 bacteriophage; D: WT-T7 can approach but cannot infect Pseudomonas syringae pv. Tomato DC3000; E: T7 ∆gp17:::VO98_215 infecting Pseudomonas syringae pv. Tomato DC3000; F: T7 ∆C-gp17:: C-VO98_215 infecting Pseudomonas syringae pv. Tomato DC3000.

<Infection Level>

Neither the recombinant phage nor the wild-type T7 phage could lyse DC3000, thus phage plaques were not directly observable. However, phage-like particles with genomes replaced by the packaging plasmid and altered tail filaments injected the packaging plasmid into DC3000 cells upon infection. After plating, these particles produced kanamycin-resistant DC3000 colonies that exhibited green fluorescence.

Figure 21: LB plates spread with DC3000 after infection with PLPs

Learn

Experiments confirm successful tail replacement and demonstrate that tail-modified phage-like particles effectively infect target bacteria. These particles, which successfully encapsulate the packaging plasmid with altered tails, successfully delivered the plasmid from E. coli to DC3000. This confirms that the phage-like particles we obtained can deliver DNA from the chassis organism to the target organism.

3.3 Construction of pUC19_VO98_215（C-）

1st Iteration

Design

A recombinant fragment containing the gene replacement homologous arm and the vector homologous arm was obtained via nested PCR from the pUC19_VO98_215 vector. After linearizing the pUC19 vector with the restriction enzyme EcoRI, the recombinant fragment was seamlessly cloned into the vector.

Build

Using vector pUC19-VO98_215 as a template, the recombinant fragment was obtained through two rounds of PCR. After the first round, the fragment contained the complete downstream homologous arm (gp17 down arm) of the gene replacement and a partial upstream homologous arm (gp17(149) up arm). The second round of PCR completed the upstream homologous arm and added vector homologous arms to both ends of the fragment. The pUC19 vector was digested with the restriction enzyme EcoRI and verified by electrophoresis.

Test

The seamless cloning product was transformed into MG1655 and plated. However, multiple colony PCRs failed to yield the expected positive clones. Notably, the negative control plate showed numerous colonies, indicating a high blank transformation rate.

Figure 22: The results of colony PCR. Positive colonies were expected at around 2kb and there are no colonies in the figure that match the expectation.

Figure 23: Negative Control Plate.

Learn

Upon careful analysis, we concluded that the root cause lay in the challenging control of the ratio between the recombinant fragment and the linearized pUC19 vector. Additionally, the small size of the pUC19 vector led to a high rate of vector self-ligation and linear fragment transfer during transformation. Furthermore, EcoRI digestion could not guarantee complete linearization of all pUC19 vectors, contributing to the failure to obtain positive clones.

2nd Iteration

Design

To address the high blank rate, we decided to obtain the linearized vector using reverse PCR followed by DpnI treatment combined with gel recovery.

Build

High-purity linearized vector was obtained via reverse PCR, DpnI digestion, and gel recovery, with all other procedures unchanged.

Test

Colony PCR confirmed the desired positive clones.

Figure 24: The results of colony PCR. Positive colonies were expected at around 2kb, which two picked colonies in the lane 3-4 turned out to be.

Learn

Blank transformations remain a significant challenge in seamless cloning vector construction. Employing reverse PCR to generate linearized vectors followed by DpnI digestion proves an effective strategy for enhancing positive clone yield.

Dry Lab

Overview

Learn how we developed Alphage — an intelligent tool for precisely modifying T7 phage tail-fiber proteins — through iterative engineering cycles, and see how we systematically enhanced its scientific rigor and user-friendliness.

Software

1st Iteration

Design

To predict phage tail fiber modification schemes, we developed Alphage, a tool designed to assist the Wet Lab in identifying replacement sites for tail fiber replacement. We relied on homology to identify replacement sites and defined the UI logic.

Build

We first performed pairwise global alignment of the two tail-fiber sequences to compute homology and assess each site’ s replaceability. Concurrently, a simple UI was designed for testing purposes.

Figure 25: Picture of the initial software interface

Test

Preliminary testing revealed that the software generated only a limited number of valid replacement sites. After discussion, we concluded that the simplistic sequence alignment logic would miss numerous potential homologous fragments.

Figure 26: Picture of the initial result graph

Learn

We need to adopt a more comprehensive sequence alignment tool with in-depth computational capabilities to ensure full-sequence alignment as thoroughly as possible.

2nd Iteration

Design

In this cycle, we selected DeepBLAST — a sequence alignment tool based on convolutional neural networks (CNNs), which demonstrates high accuracy in aligning distantly related sequences.

Build

We deployed DeepBLAST on a local server and tail-fiber amino-acid sequences from the initial database were aligned against the T7 tail-fiber reference.

Test

DeepBLAST’ s full-sequence alignment function is robust, adequately meeting our needs for assessing the homology of tail fiber protein sequences. However, the existing software lacked the ability to evaluate the feasibility of successful replacement at replacement sites.

Learn

Next, we need to explore alternative approaches to assess the reliability of replacement sites.

3rd Iteration

Design

In this cycle, through discussions and consultations with instructors and professors, we selected ESMFold for structure-based stability estimation and PHIEmbed for phage–host interaction prediction.

Build

We integrated PHIEmbed into the software to predict the efficiency of phage-host infection after replacing the T7 phage tail fiber protein at the replacement site.

In parallel, we consulted Prof. Yue Feng (Beijing University of Chemical Technology), who recommended incorporating stability assessment. Consequently, ESMFold was integrated to generate pLDDT scores as a proxy for structural integrity.

Test

Currently, the three models in the software work well together and generate biologically meaningful parameters during computation. However, these parameters lack interconnections to support the overall evaluation of tail fiber protein replacement schemes.

Learn

Based on the existing indicator scores, we need to achieve objective integration of multi-dimensional scores and construct a more scientifically robust comprehensive scoring model.

4th Iteration

Design

To achieve objective integration of multi-dimensional scores, we used the “paired ranking transformation + linear regression optimized by binary cross-entropy” algorithm to solve for the optimal weight coefficients among the scores.

Burges, C., Shaked, T., Renshaw, E., Lazier, A., Deeds, M., Hamilton, N., & Hullender, G. (2005). Learning to rank using gradient descent. In Proceedings of the 22nd International Conference on Machine Learning (pp. 89–96). ACM.

Build

We collected the currently available scoring datasets and constructed a robust machine learning model. Using the “paired ranking transformation + linear regression optimized by binary cross-entropy” algorithm, we solved for the optimal weight coefficients among the scores, ensuring the weight assignment is statistically valid and supported by data.

Test

The model provided default weights for the evaluation parameters: 62% for “homology”, 29% for “Δhomology”, and 9% for “infection”. Based on these weights, comprehensive scores were computed, with schemes ranked in descending order of confidence.

Figure 27: Parameters setting interface

Learn

So far, the software’ s computational logic is essentially complete; what we still need is an appealing UI to deliver a more intuitive and comfortable user experience.

For more information of our final results, please visit our wiki page.

5th Iteration

Design

To deliver a better user experience, we have redesigned the UI based on the software’ s computational logic.

Build

We added an interactive interface allowing users to adjust parameter weights. On the results page, the computed replacement schemes are presented through line graphs paired with concise textual descriptions.

Test

We conducted tests using the project’ s replacement schemes and obtained the following results:

Figure 28: Result display interface

Learn

Currently, we have successfully developed Alphage, a robust tool for predicting phage tail fiber protein replacement. In the future, we will further expand the existing dataset and conduct comprehensive testing of the software’ s functions.

Human Practice

Overview

To enhance the effectiveness of synthetic biology education, we iteratively developed a story-based curriculum for primary students and a card game using the “Design-Build-Test-Learn” cycle. The CHUN GENG Program built a logic-driven knowledge system using metaphors and character guidance, while SynBio Halli Galli simplified component recognition with color-coded pathways.

Both projects evolved through user feedback, transitioning from concept delivery to systemic understanding, demonstrating the flexibility and impact of engineering thinking in science communication.

SynBio Halli Galli Engineering

1st Iteration

Design

In the initial phase, we set the popular science objective as enabling players to memorize synthetic biology components and understand the construction of genetic pathways.

Based on the characteristics of this objective, we drew inspiration from the game Halli Galli and adopted its following advantages:

Build

We defined three types of cards:

• Component Cards: These include four basic synthetic biology components.

• Contamination Cards: These simulate laboratory biosafety scenarios such as phage outbreaks, culture medium contamination, and mutation accumulation, with an emphasis on biosafety.

• Function Cards: These introduce commonly used tools like CRISPR-Cas9 and DNA ligase.

Additionally, we retained the basic rules of Halli Galli including reward and penalty mechanisms as well as the method for determining victory or defeat.

Test

We contacted team members who worked on the previous card game project, and we jointly discussed the playability and rationality of our cards. Their feedback was: a single pathway might bore players, adding diverse promoters and designing pathways with different functions would help increase the diversity of the cards.

Learn

We incorporated their suggestions, leading to several key changes in our card design:

1. We redesigned three types of promoters: a promoter that responds extensively to low pH, an anaerobically activated promoter, and a heat-induced promoter.

2. Inspired by these multiple promoters, we optimized the target genes for the corresponding conditions: fluorescent protein genes, drug resistance genes, and protease genes.

2nd Iteration

Design

Based on the feedback from the first iteration, we began designing the component patterns to be displayed on the cards.

Build

We aimed for the component patterns to intuitively represent the conditions for pathway expression while being both aesthetically pleasing and concise. With this in mind, we created designs for each component card, contamination card, and tool card.

Test

Before officially producing the card designs, we hand-drew 90 cards using pencils and cut paper to make physical prototypes, then invited other team members to test-play them.

From the test-play experience, we identified a problem in the design: when components with overly detailed drawings were arranged in quantities ranging from 1 to 4 on a 6×8 cm card, the detailed patterns became too small.

This significantly increased the game difficulty, prolonged reaction times (which slowed down the game pace), and failed to effectively stimulate players’ reaction abilities.

Learn

Therefore, we began to consider: how to maximize simplification while retaining the key features of the components. As a result, we chose to keep only the symbols and letter abbreviations of the components.

For contamination cards and function cards, which could have designs covering the entire card surface, we retained the detailed drawings.

3rd Iteration

Design

We used different solid colors to fill each type of component, aiming to reduce the game difficulty for players without a background in synthetic biology. When constructing genetic pathways, players could use the color guidance to gain a clearer understanding of the complete pathway structure, thereby achieving our popular science goal.

Build

In accordance with our design plan, we added colors to the original hand-drawn patterns and launched the second round of test-playing.

Test

With the experience from the first test-play, we became much more proficient in the second round. However, after several rounds, we gradually noticed that the construction of a single pathway was somewhat monotonous.

Learn

Thus, based on biological knowledge, we distinguished between eukaryotic and prokaryotic genetic pathways, using solid and hollow drawing styles respectively to differentiate them. At this point, our card game was finalized.

We finalized the card designs, commissioned the production of a certain number of physical cards, and released the card instruction manual via a promotional post.

The cards became a key element in scenarios such as popular science education, team communication, and project demonstrations.

CHUN GENG Program Engineering

1st Iteration

Design

We planned to expand the audience of synthetic biology popular science to primary school students, creating an immersive learning experience by combining games and stories.

Considering the natural appeal of picture books to children, we decided to create a character named “Chocolate” (based on a real teddy bear dog named Chocolate). Through Chocolate’ s perspective to connect scattered knowledge points, we would guide children into the world of synthetic biology.

During the popular science classes, we would intersperse game activity scenarios as one form of immersive popular science, creating an atmosphere where the character and the students explore synthetic biology together, and realize the game through character interaction.

Build

We built a story framework: the dog Chocolate travels to the world of synthetic biology and explores the mysteries of this world with children. We created a 5-minute PPT story, anthropomorphizing cells and endowing them with different “job functions”, and adapted previous iGEM projects into plotlines.

In the story, DNA is compared to a “cell instruction manual”, bases to “letters”, biological components to “words”, biological devices to “sentences”, and genetic pathways to “instructions”.

Test

We conducted a trial lecture for students with no background in synthetic biology.

The participants had a good understanding of the metaphors in the story and particularly resonated with the character “Chocolate”. They were able to retell the metaphors in the story and initially understand “how cells work”, which confirmed the feasibility of story-based popular science.

Learn

We verified that the “storytelling + character guidance” form of popular science is effective for primary school students. On this basis, we decided to further expand the content, adopting the “language system” analogy method to gradually build a complete knowledge system.

Based on the story framework, we continued to expand the popular science content, using the language system familiar to children (DNA - cell instruction manual, bases - letters, biological components - words, biological devices - sentences, genetic pathways - instructions guiding cell work) to analogize synthetic biology concepts, and organized the teaching content in accordance with the bottom-up thinking of engineering.

2nd Iteration

Design

Building on the success of the first iteration, we expanded the 5-minute story into a complete 90-minute course, covering the entire process from basic concepts to logical structures.

Build

The story still starts with specific synthetic biology examples.

Under the framework of the language system, it introduces the material basis (compared to writing materials) from the most basic level, the central dogma (grammar), and the previous synthetic biology concepts. Moreover, the concepts of feedforward, positive feedback, and negative feedback are introduced into genetic circuits, which are analogized through the protagonist “Chocolate’ s” three twists and turns of experiences in wanting to eat ice cream, to compare the connections and differences between the three circuits.

For example:

• Feedforward structure: Finish homework first before making a request to increase the success rate;

• Positive feedback: Moderate behavior leads to long-term trust;

• Negative feedback: Excessive behavior leads to restrictions.

Test

Before the official class on the next day, other members of the CHUN GENG Program attended a trial class, fully simulating the teaching scenario of the next day. Regarding the complete story Chocolate’ s Journey into Synthetic Biology, some students pointed out that it was somewhat redundant to explain the concepts in plain language first and then present the concepts themselves; it would be better to directly replace the concepts with the metaphorical terms.

Learn

This suggestion led to a major adjustment in the focus of our class.

We realized that for primary school students, understanding logic is more important than memorizing concepts. We weighed the pros and cons of presenting concepts and weakening concept indoctrination.

First, too many concepts might make it too difficult for primary school children to understand; on the other hand, weakening concept indoctrination might affect children’ s accurate perception of synthetic biology. We spent a long time weighing and making a decision.

3rd Iteration

Design

Based on the second iteration, we finally completely replaced professional concepts with metaphorical language.

We clarified that the focus of our popular science class lies in whether children can fully understand the logic and principles behind synthetic biology. The memorization of concepts, based on understanding, will become easy, and the timing (sooner or later) does not matter.

Build

As a result, we reduced concept indoctrination and made the following replacements:

• Use “cell instruction manual” instead of “DNA”;

• Use “letters” instead of “bases”;

• Use “words” instead of “biological components”;

• Use “sentences” instead of “biological devices”;

• Use “instructions” instead of “genetic pathways”.

Test

In the official class, considering the difficulty of the course, we designed a rich reward mechanism in the class to encourage children: small stars were given out based on the number of correct answers, and different rewards were given at different levels according to the total number of stars accumulated; in addition, after every 2-3 knowledge points, we set up small question challenges, and the questions were completely designed based on the previous knowledge points.

Learn

The class received a positive response.

Based on the children’ s performance in the question challenges and the accuracy of their answers, the children had a good grasp of the knowledge. In the final activity of “designing a life instruction manual”, they were able to independently combine “words” and “sentences” to construct simple genetic circuits. This activity not only improved the children’ s thinking ability and broadened their cognitive horizons but also made them feel the fun and magic of learning synthetic biology.

Through three iterations, the CHUN GENG Program gradually moved from “story guidance” to “logical understanding”, and from “terminology indoctrination” to “metaphor construction”. With this, we attempted to build a synthetic biology popular science system suitable for primary school students, with characters as the main line and the language system as the framework. The entire process realized the effectiveness and flexibility of the “Design - Build - Test - Learn” cycle.

References

[1] Hamamsy, T., Morton, J.T., Blackwell, R. et al. Protein remote homology detection and structural alignment using deep learning. Nat Biotechnol 42, 975–985 (2024). https://doi.org/10.1038/s41587-023-01917-2

[2] Zeming Lin et al., Evolutionary-scale prediction of atomic-level protein structure with a language model. Science 379, 1123–1130 (2023). DOI: 10.1126/science.ade2574

[3] Gonzales MEM, Ureta JC, Shrestha AMS. Protein embeddings improve phage-host interaction prediction. PLoS One. 2023 Jul 24;18(7):e0289030. doi: 10.1371/journal.pone.0289030. PMID: 37486915; PMCID: PMC10365317