Engineering

The TRACER protein is composed of BRⅡ, a MMP9-Specific Cleavable Sensitive Peptide, and a Negative Charge Sequence.The design of this section is detailed in Model.

Selection of Sensitive Peptides

To target colorectal cancer, our modeling systematically analyzed the transcriptome data of colorectal cancer and normal tissues, and identified significantly upregulated matrix metalloproteinase (MMP) family and their related pathways, as well as the sensitive peptide sequence PLGLAG that can be cleaved in the middle by MMP-2/9 enzymes.

We selected GSE10950 as the training set and downloaded the supplementary data of the independent external dataset "Transcriptomic and Cellular Content Analysis of Colorectal Cancer by Combining Multiple Independent Cohorts" as the validation set, and used UltraEdit and Excel for data processing. With the help of the Sangerbox platform, we performed differential expression gene screening, gene set enrichment analysis, PPI network and validation set analysis, and obtained the following important guiding conclusions:

1. There are significant differences in gene expression patterns between colorectal cancer tissues and normal tissues.

Fig.1 Differential Expression Heatmap

2. The expression of MMP-9 enzyme in colorectal cancer tissues is significantly upregulated compared with normal tissues, while the difference in MMP-2 enzyme does not reach statistical significance.

Fig.2 Boxplot of Gene Enrichment Analysis for MMP9 vs. MMP2 (MMP9 on the left and MMP2 on the right)

3. MMP-9 enzyme can serve as a molecular marker for colorectal cancer.

Therefore, we chose to use PLGLAG as the CRC-targeting sensitive peptide, which constitutes a part of TRACER.

Molecular Docking and Dynamics Simulation

The molecular weight of the entire fusion protein TRACER-IL24 is 23.8 kDa. Considering that the relatively large molecular weight may affect the cleavage process of the sensitive peptide, we selected the human MMP9 crystal structure (PDB: 5I12) and the fusion protein TRACER-IL24 (Alphafold Predicted) to simulate and analyze the MMP-9 enzyme cleavage process through molecular docking and molecular dynamics. GROMACS was used to evaluate whether the catalytic geometry is accessible for a long time and whether the interface interaction is stable.

Fig.3 Docking Conformation of MMP9 and TRACER-IL24

Fig.4 Molecular Dynamics Simulation Results

Key Findings:

1. The number of hydrogen bonds and atomic contacts at the binding interface between MMP-9 enzyme and the sensitive peptide increases with simulation time and tends to stabilize (a-b).
2. The deeply buried interface of the enzymatic cleavage complex continues to increase and stabilizes, indicating that the enzymatic cleavage active interface has been formed (c).
3. The RMSD rises rapidly and then converges to a plateau, suggesting that the system has completed structural adjustment and remains stable (d).

Structural prediction of the entire TRACER fused with IL24 shows that MMP-9 enzyme can still recognize and stably bind to the sensitive peptide segment PLGLAG, providing solid theoretical support for our TRACER design.

Design of Negative Charged Shielding Peptide

To enable the designed negative charge sequence to achieve the shielding function, we collaborated with the modeling team to conduct predictions from two aspects: structure and shielding ability.

Structural Design

Three candidate schemes were proposed:

Type	Sequence	Figure Reference
Repeated Negative Charge Sequence	E×n	Fig.5(a)
Helical Hydrophobic	EXXEEXXE	Fig.5(b)
Flexible	EEG	Fig.5(c)

After analysis, we chose to add glycine (Gly) to the repeated sequence to enhance flexibility. As shown in Fig. 5(c), the negative charge sequence (Glu-3, Glu-5, Glu-8) was found to effectively bind to the positively charged residues of BRⅡ (ARG-25, ARG-28, ARG-32).

Fig.5 Visualization of Negative Charge Structure Design

Prediction of Shielding Ability

We proposed a transmembrane ability evaluation paradigm termed TRF (Transmembrane Readiness Framework):

Fig.6 TRF Hierarchical Indicator Diagram

The evaluation integrates 8 indicators across three levels: Geometry & Interaction, Thermodynamics, and Kinetics. Results showed that the cleaved state (Cut) exhibited more compact structure, closer proximity to membrane, deeper insertion, favorable orientation, smaller energy barrier, and faster rearrangement compared to the uncleaved state (Total).

TRF Level	Indicator	Total（Uncleaved）	Cut（Cleaved）	Conclusion
Geometry & Interaction	RMSD	Stable (0.02–0.025 nm)	Stable	Both are stable.
	Rg	1.94 nm	1.77 nm	Cut is more compact (facilitates membrane approach).
	Minimum Distance Between Centroid and Membrane	~2.0–2.2 nm	~1.6 nm	Cut is closer to the head group layer.
	Insertion Depth (Smaller value = Deeper insertion)	~6.3 nm	~5.7 nm	Cut inserts deeper.
	Tilt Angle Distribution/Temporal Profile	~99°（horizontal lying）	~155°（oblique insertion/near-vertical）	Cut orientation is favorable for transmembrane penetration.
Thermodynamics	SMD Force-Displacement Slope/Plateau	Steep slope, no plateau	Low slope, obvious segmented plateau	Cut has a smaller effective energy barrier.
	SMD Force-Time/Displacement-Time	Early saturation, limited displacement	Continuous progression, linear increase in displacement	Cut is more likely to overcome the energy barrier.
Kinetics	Orientation-Related Time Constant (τ)	~0.49 ns	~0.12 ns	Cut shows faster rearrangement and stronger adaptability.

Initial Genetic Circuit Design

Initially, we planned to use Escherichia coli (E. coli) as the pharmaceutical production host for the production of the fusion protein, and combine the fusion protein with our self-designed microneedle system for drug delivery.

Fig.7 Genetic Circuit Diagram of TRACER-IL24

In our design, the TRACER protein consists of a negatively charged sequence (red), a cleavage-sensitive peptide (green), and the cell-penetrating peptide BRⅡ (blue). Under normal conditions, the negatively charged sequence reduces the positive charge on the protein surface, rendering the protein unable to enter cells via macropinocytosis. In the colorectal cancer (CRC) tumor microenvironment (TME), the highly expressed matrix metalloproteinase MMP-9 will cleave the cleavage-sensitive peptide, leading to the detachment of the negatively charged sequence. Subsequently, the cell-penetrating peptide delivers the therapeutic agent IL-24 into cancer cells to exert its therapeutic effect.

Genetic Circuit: All fusion protein genes were entrusted to Thermo Fisher for synthesis in accordance with the Biobrick standard. Subsequently, the gene fragments were amplified via PCR, followed by restriction enzyme digestion of the PCR-amplified products. Thereafter, the obtained fragments were analyzed using nucleic acid gel electrophoresis, and the electrophoretic products were purified and recovered by the gel extraction method. Finally, the purified genetic circuit fragments and the empty pET-28a(+) vector were subjected to DNA ligation in a 37°C incubator, with incubation for 24 hours.

Fig.8 Verification Results of Nucleic Acid Electrophoresis (pET-28a(+))

Fig.9 Verification Results of Nucleic Acid Electrophoresis (pUC19)

After performing Polymerase Chain Reaction (PCR) on the target gene fragment, a restriction endonuclease digestion assay was conducted between the PCR product and the full-circuit cloning vector, and the digested fragments were analyzed based on agarose gel electrophoresis.

Fig.8 Results: Lane 1 is the standard nucleic acid molecular Marker Ⅲ; Lane 2 is the pET-28a(+) plasmid vector with a theoretical length of 5369 bp; Lanes 3-4 are the pET-28a(+) plasmid fragments after double digestion with a theoretical length of 5350 bp; Lanes 5-6 are the target fragments obtained after PCR with a theoretical length of 697 bp, and all lengths are consistent with the expected values.

Fig.9 Results: Lane 1 is Marker Ⅲ; Lanes 2-3 are the pUC19 plasmid fragments after double digestion with a theoretical length of 2635 bp; Lanes 4-5 are the target fragments obtained via PCR with a theoretical length of 697 bp; Lane 6 is the pUC19 plasmid vector with a theoretical length of 2686 bp, and all lengths are consistent with the expected values.

The successfully constructed plasmids were transformed into BL21(DE3) competent cells, which were then cultured on LB solid medium containing kanamycin sulfate, and finally single colonies harboring the recombinant plasmids were obtained.

Fig.10 Screening Results of Recombinant Plasmid Transformation into BL21(DE3)

The bacterial culture and kanamycin were added to the LB medium at a ratio of 1:100 and 1:1000 respectively, and incubated at 37°C until reaching the logarithmic growth phase with an OD₆₀₀ value of 0.4-0.6. After that, isopropyl β-D-1-thiogalactopyranoside (IPTG) was added to a final concentration of 0.5 mM for overnight induction. Subsequently, the bacterial cells were sonicated for lysis, followed by centrifugation to collect the supernatant. The protein was purified from the supernatant using affinity chromatography, and the protein concentration was determined by the BCA method. Finally, the molecular weight of the protein was characterized via SDS-PAGE.

Fig.11 Results of Protein Concentration Determination by BCA Assay

Fig.12 Results of Protein Molecular Weight Characterization by SDS-PAGE

The standard curve determined and plotted by the BCA assay is shown in Fig.11. Substituting the measured data, the maximum concentration of the purified samples is approximately 179.4 μg/ml. In SDS-PAGE, no distinct bands were observed except for the Marker shown in Fig.12.

The results of the BCA assay and SDS-PAGE indicated that the extracted protein had a low concentration, failing to show distinct bands in SDS-PAGE and thus unable to meet the requirements of subsequent experiments.

In multiple rounds of construction verification and protein expression experiments, recombinant plasmids and engineered bacteria were successfully obtained; however, proteins with sufficient concentration could not be purified. We speculate that this may be due to the excessively low expression level of the constructed recombinant plasmids. Therefore, we plan to increase the intensity of induced expression and perform concentration after protein extraction to raise the protein concentration.

Based on the results of the previous cycle, where recombinant plasmids and engineered bacteria were successfully constructed but the concentration of the extracted protein was insufficient, we decided, after investigation, to increase the concentration of IPTG induction and adopt three methods—water bath concentration, ammonium sulfate precipitation concentration, and freeze-drying concentration—to enhance the protein concentration.

After obtaining the engineered bacteria, the bacterial culture and kanamycin were added to fresh LB medium at ratios of 1:100 and 1:1000, respectively, for logarithmic growth. Unlike the previous cycle, a final concentration of 1 mM IPTG was then added for overnight induction, followed by disruption of bacterial cells by ultrasonication. The supernatant was collected by centrifugation, and proteins were purified by affinity chromatography.

Three Concentration Methods:

1. Water bath concentration: The obtained purified samples were placed with lids opened and incubated in a 40°C water bath until approximately 1/10 of the liquid remained.
2. Ammonium sulfate concentration (pellet): A saturated ammonium sulfate solution was prepared under ice bath, and slowly added to the purified samples in two portions at a ratio of 3:2, with precipitation formed during the process. The mixture was inverted to mix while adding. After 30 minutes of ice bath, centrifugation was performed to remove the supernatant. The precipitate was resuspended again with saturated ammonium sulfate solution, followed by another centrifugation to remove the supernatant, and then resuspended with 1-3 mL TBS (20 mM Tris, 0.137 M NaCl, pH 7.6±0.1 (25°C)).
3. Freeze-drying concentration: The obtained purified samples were placed in a lyophilizer and lyophilized at -40°C and 500 Pa for 12 hours.

After concentration, the protein content in the purified sample was determined using the BCA method:

Fig.13 Results of Protein Concentration Determination by BCA Assay

The standard curve determined and plotted by the BCA method is shown in Fig.13. By substituting the measured data, it was found that among the three concentration methods, freeze-drying concentration showed better concentration efficiency. However, the maximum concentration of the concentrated and purified sample was approximately 114.7 μg/ml, which still failed to meet our requirements.

The results of the BCA method indicated that after enhanced induction and the application of the three concentration methods, the protein concentration did not show a significant increase.

After multiple rounds of repeated experiments, we found that no high-concentration protein was detected in the supernatant even after concentration. At this point, the results from the modeling group indicated that the fusion protein has low solubility and may be present in the precipitate. Therefore, we plan to separately extract the protein from the supernatant and the precipitate after sonication.

Fig.14 Modeling Analysis of TRACER-IL24 Protein Solubility

In the previous cycle, the experimental results and data from the modeling group enlightened us that the fusion protein may exist in the precipitate after sonication due to its properties. Therefore, we plan to extract proteins from both the supernatant and the precipitate to obtain the fusion protein.

After sonication, we extracted proteins from the supernatant. Subsequently, the pellet was resuspended in lysis buffer and subjected to sonication again. After centrifugation, the pellet was collected, resuspended in buffer, and then mixed with protein denaturation mixture followed by another round of sonication. After centrifugation, the supernatant was collected. This process was repeated twice to obtain the purified product.

The protein content in the purified sample was determined using the BCA method:

First experimental result

Second experimental result

Fig.15 Results of Protein Concentration Determination by BCA Assay

The standard curves determined and plotted by the BCA assay are shown in Figs.15. Substituting the measured data, it was found that the maximum concentration of the purified samples increased, with the highest concentration observed in the supernatant (approximately 247.6 μg/ml), which still failed to meet the requirements of subsequent experiments. Results of the BCA assay indicated that neither the supernatant nor the precipitate exhibited a high protein concentration, thus failing to satisfy the demands of subsequent experiments.

Experiments and learnings from the previous cycles have shown that although plasmids were successfully constructed, transformed, and engineered strains meeting the requirements were screened, no high-concentration protein was extracted from either the supernatant or pellet of ultrasonic disruption. This suggests that there may be issues in gene expression or protein extraction steps.

Regarding the above phenomenon, combined with the work of the modeling team and the results of our literature research, we propose the following possibilities:

1. The solubility of the fusion protein is low, leading to excessive production of inclusion bodies during expression. The protein expression process needs optimization.
2. There is a lack of library construction and experimental screening for combinations between promoter strength, RBS, and the expressed gene.
3. The molecular weight of the selected IL-24 protein may be excessively large.

In the future, we plan to make the following modifications in the biological route:

1. Add fusion-promoting tags to improve solubility and facilitate protein production;
2. Construct a screening library of promoter-RBS-expressed gene for high-throughput screening.

However, due to time constraints, we did not conduct subsequent experiments and optimizations.

Since we failed to obtain the fusion protein with sufficient concentration in the previous cycles, we entrusted Beijing Xinxinsantatech Co. Ltd. to help us synthesize the fusion protein. We treated cancer cells with the obtained protein and detected the cytotoxicity of the fusion protein to cancer cells by evaluating cell viability.

Due to sterilization issues in the incubator, we were unable to conduct the experiment before the wiki deadline. It will be carried out before the Jamboree Presentation.

After conducting the aforementioned experiments, the protein expression process we currently have is insufficient to produce a sufficient quantity of protein for microneedle drug loading. Therefore, we plan to engineer Escherichia coli to serve as a carrier for oral drug delivery, enabling it to both fulfill the production function and release the drug. We referenced the four-module design of the oscillatory lysis circuit described in the study "Synchronized cycles of bacterial lysis for in vivo delivery". Our aim is to avoid attempting colonization of the engineered bacteria; instead, we will achieve effective drug concentrations through regular administration. The engineered bacteria will reach the intestinal mucosa via enteric-coated capsules, grow there, and be shed along with the natural renewal of the mucosa. This design, combined with the microneedles in the hardware component, forms an "In and Out" delivery toolbox, which can significantly improve patient compliance.

To align with the therapeutic target of colorectal cancer, our research revealed that the intestinal tract of colorectal cancer patients exhibits extensive infiltration of inflammatory cells, leading to a significant increase in oxidative damage levels. Consequently, we selected an oxidative damage-activated promoter as the starting point for our activation module and employed the SRRz gene—which enables faster lysis—to trigger bacterial lysis.

Fig.18 Genetic circuit of the deliver module

In this design, E. coli will directly enter the human body for therapeutic purposes. P_j23100 can constitutively express the fusion protein, while upon entering colorectal cancer tissues, E. coli will be induced to lyse, releasing the fusion protein for treatment.

Fig.19 Deliver Circuit Schematic Diagram

Due to time constraints, we failed to complete the subsequent experiments before the Wiki deadline. We will present our subsequent experimental results at the Jamboree presentation. However, based on the Agent-Based Modeling (ABM) framework, we used NetLogo to simulate the lysis conditions under different inoculation doses.

Fig.20 Effects of different initial inoculum sizes on biomass accumulation and lysis cycle

Fig.21 Spatiotemporal visualization results of different inoculum sizes simulated by NetLogo
a. Num=2000; b. Num=5000; c. Num=8000; d. Num=10000
The top left shows curves of total biomass and products; the top right shows the spatial distribution map (red represents regions with high concentration of lysis proteins, and green represents regions of viable cells).

NetLogo simulations revealed the quantitative relationship between "inoculation dose--lysis cycle" at the tissue level: the initial inoculation dose exhibits a negative correlation with the lysis cycle. Within a certain range of inoculation doses, the higher the initial bacterial load, the more rapidly quorum sensing signals accumulate, and the shorter the lysis cycle; when the initial bacterial load is too low, the initiation of lysis is delayed, leading to a longer lysis cycle.

In this cycle, we only redesigned and modified the gene circuit. However, combining the results from previous cycles and modeling, E. coli can effectively express the fusion protein and lyse and die after a certain period. Through constitutive expression combined with induced lysis for drug delivery, we believe it will be possible to produce sufficient drugs in the human body and achieve targeted therapy.

Meanwhile, the NetLogo simulation results provide a theoretical basis for applying engineered bacteria in practical therapeutic settings: by adjusting the initial dosage, precise regulation of the lysis rhythm can be achieved, thereby controlling the lysis cycle and obtaining the optimal therapeutic effect.