Wet Lab Results ​

Melatonin Synthesis ​

The synthesis of melatonin relies on two plasmids: one is pWB980-ori_m-hTPH1_hPCBD1_hQDPR_TDC (Figure 1.1), which mediates the biosynthesis of serotonin from L-tryptophan; the other is pHY300PLK_SNAT_COMT (Figure 1.2), which facilitates the conversion of serotonin to melatonin.

aaa
Figure 1.1
aaa
Figure 1.2

Both plasmids were subjected to comprehensive characterization using three experimental techniques, including Reverse Transcription-Polymerase Chain Reaction (RT-PCR), Western Blotting (WB), and Enzyme-Linked Immunosorbent Assay (ELISA). Additionally, the effect of the second plasmid on cell growth was investigated to evaluate their potential influence on cellular physiological processes during melatonin synthesis.

The effect of the second plasmid on bacteria growth ​

We Expand culture of pHY300PLK-SNAT-COMT glycerol stock and pHY300PLK glycerol stock: 5ml liquid LB medium (Amp: 100ng/ml) + 10μl glycerol bacterial solution (4 replicate experiments).

Then, measure OD₆₀₀ at 0h, 1h, 2h, 3.5h, 4.5h, 6h, 7h, 10h, 14h and 16h during the expansion culture.

Browning Index Time Course
Figure 2(a)
Browning Index Statistical Comparison
Figure 2(b)

Figure 2.

(a): Growth Curve of E. coli DH5α (OD₆₀₀) showing the growth dynamics of E. coli DH5α cultures containing pHY300PLK - SNAT - COMT and pHY300PLK plasmids over time, with OD₆₀₀ measured at different time points during expansion culture.

(b): Data is presented as Mean ± SEM, with N = 4 biological replicates. Bar graph showing OD₆₀₀ values of E. coli DH5α cultures after 6 hours of cultivation, comparing those with pHY300PLK and pHY300PLK-SNAT-COMT plasmids. Statistical analysis via Mann - Whitney test (two-tailed) yielded p = 0.0286 < 0.05, significance denoted by *.

RT-PCR result ​

Agarose gel electrophoresis analysis of RT - PCR products
Figure 3. Agarose gel electrophoresis analysis of RT - PCR products. For each target gene (m - hTPH1, hPCBD1, hQDPR - flag, TDC - his), the samples in the lanes labeled "(-)" were derived from bacteria harboring the pWB980 - ori plasmid, while those in the lanes labeled "(+)" were from bacteria containing the pWB980 - ori_m - hTPH1_hPCBD1_hQDPR - flag_TDC - his plasmid. For each target gene (SNAT-his, COMT-flag), the samples in the lanes labeled "(-)" were derived from bacteria harboring the pHY300PLK plasmid, while those in the lanes labeled "(+)" were from bacteria containing the pHY300PLK_SNAT-his_COMT-flag plasmid. The distinct bands (960 bp for m-hTPH1, 315 bp for hPCBD1, 786 bp for hQDPR-flag, 1584 bp for TDC-his, 669 bp for SNAT-his, 1158 bp for COMT-flag) indicate the successful amplification of corresponding gene fragments, confirming the transcriptional activity of all four target genes.

The control group showed no expression, while the experimental group exhibited expression. Therefore, we infer that the genes of all six proteins were successfully transcribed into RNA.

Western blot ​

We have hQDPR-Flag and TDC-His in plasmid pWB980-ori_m-hTPH1_hPCBD1_hQDPR_TDC, so we used western blot to detect them.

Western blot result of hQDPR-Flag and TDC-His
Figure 4. This graph shows the result of our western blot. pWB_rbs_TPH_rbs_hPCBD_rbs_hQDPR-flag_rbs_TDC-his denote sample from E.coli with plasmid with target genes and pWB denotes that from E.coli with vector backbone pWB980-ori without target genes. There is specific band for TDC-his, but the size is different from 59.1kDa, which is derived by Snapgene. There is no specific band for hQDPR-flag.

We have COMT-Flag and SNAT-His in the plasmid pHY300PLK_SNAT_COMT, so we used western blot to detect them.

Western blot result of COMT-Flag and SNAT-His
Figure 5(a)
Western blot result of COMT-Flag and SNAT-His
Figure 5(b)

Figure 5

pWB-rbs-SNAT-His-+rbs-COMT-flag denotes sample from E.coli with plasmid with target genes on pWB980-ori and pWB denotes that from E.coli with vector backbone pWB980-ori without target genes.

pHY-rbs-SNAT-His-+rbs-COMT-flag denotes sample from E.coli with plasmid with target genes on pHY300PLK and pHY denotes that from E.coli with vector backbone pHY300PLK without target genes.

There is a specific band for SNAT-his. But the size is different from 24.7kDa, which is derived by Snapgene.

Western blot result of COMT-Flag and SNAT-His
Figure 6. pWB-rbs-SNAT-His-+rbs-COMT-flag denotes sample from E.coli with plasmid with target genes on pWB980-ori and pWB denotes that from E.coli with vector backbone pWB980-ori without target genes. pHY-rbs-SNAT-His-+rbs-COMT-flag denotes sample from E.coli with plasmid with target genes on pHY300PLK and pHY denotes that from E.coli with vector backbone pHY300PLK without target genes. There is specific band for COMT-flag. The size is also correct, corresponding to 41.6kDa, which is derived by Snapgene.

Results also show that they are expressed more in vector backbone pWB980-ori. We evaluated band gray-level values and performed one-way ANOVA to analyze the difference in COMT expression between the backbone vectors pWB980-ori and pHY300PLK. As shown in Figure1.2., there was a statistically significant difference between them.

Figure 7. Data is presented as Mean ± SEM, with N = 3 biological replicates. pWB+: the samples are from E.coli DH5α with plasmid with target genes on pWB980-ori. pWB: the sample is from E.coli DH5α with vector pWB980-ori without target genes, serves as negative control. pHY+: the samples are from E.coli DH5α with plasmid with target genes on pHY300PLK. pHY: the sample is from E.coli DH5α with vector pHY300PLK without target genes, serves as negative control. P-values of ordinary one-way ANOVA are listed in the graphs. P ≤ 0.001: very highly significant, denoted by ***. P ≤ 0.05: significant, denoted by *.

ELISA ​

We detected the serotonin content in both the control group and the experimental group (pWB980-ori_m-hTPH1_hPCBD1_hQDPR_TDC) using this kit.

Comparison of Serotonin Concentrations Between Control Group and Experimental Group
Figure 8. Comparison of Serotonin Concentrations Between Control Group and Experimental Group Data are presented as Mean ± SEM, with N = 4 biological replicates for pWB980-ori, and N = 4 biological replicates for pWB980-ori-TPH-hPCBD1-hQDPR-TDC. This figure presents the concentration of serotonin in the control group and the experimental group, with statistical analysis performed using an Unpaired t-test (two-sided).The results show a significant difference (P=0.0008 < 0.05, donated by ***) between the experimental group and the control group. Based on this statistical finding, it is concluded that pWB980-ori-TPH-hPCBD1-hQDPR-TDC can successfully convert L-tryptophan into serotonin.

By using the t-test, we can find that there is a significant difference between the experimental group and the control group. We conclude that pWB980-ori-TPH-hPCBD1-hQDPR-TDC can convert L-tryptophan into serotonin.

Then, we detected melatonin content in both the control group and the experimental group (pHY300PLK_SNAT_COMT) using this kit.

Bacterial Growth - OD600
Figure 9(a)
Melatonin Concentration
Figure 9(b)

Figure 9. Comparison of Bacterial Growth (OD600) and Melatonin Concentration After 16 Hours of Culture

(a) Bacterial Growth - OD600

This bar graph presents the OD600 (optical density at 600 nm) of the bacterial liquid from the control group (pHY300PLK) and the experimental group (pHY300PLK−SNAT−COMT) after 16 hours of culture (data shown in Table 5). The data was analyzed using the Mann-Whitney test. The results indicated no significant difference (ns) in OD600 between the control group and the experimental group (P>0.05). Thus, there is no significant difference in bacterial concentration.

(b) Melatonin Concentration

This bar graph presents the melatonin concentration (pg/mL) in the control and experimental groups as measured by ELISA after 16 hours of culture (data shown in Table 6). The data was analyzed using the Mann-Whitney test. In contrast to the OD600 results, the melatonin concentrations showed significant variations (P=0.0409<0.05, denoted by ∗) between the two groups.

This significant difference confirms that the biosynthetic pathway successfully enabled the experimental strain (pHY300PLK-SNAT-COMT) to produce a significantly higher amount of melatonin compared to the control.

Unfortunately, due to constraints in time and technology, we were unable to obtain more precise results.

However, We can preliminarily speculate that pHY300PLK-SNAT-COMT is capable of converting serotonin into melatonin.

Wax Synthesis ​

We constructed a plasmid designed to regulate wax synthesis via lactose control. In the presence of lactose, bacteria will take up large amounts of carbon sources to synthesize acyl-CoA; in the absence of lactose, bacteria utilize stored acyl-CoA to synthesize wax.

Figure 10. The final plsamid of wax group. G6PD: Glucose-6-phosphate dehydrogenase, AcrB: Acridine resistance protein B, WS/DGAT: Wax ester synthase/acyl-CoA diacylglycerol acyltransferase
Figure 10. The final plasmid of wax group. G6PD: Glucose-6-phosphate dehydrogenase, AcrB: Acridine resistance protein B, WS/DGAT: Wax ester synthase/acyl-CoA diacylglycerol acyltransferase

We planned to construct the plasmid using Gibson Assembly technology, but due to technical and experimental limitations, this goal was not fully achieved. To validate our designed plasmid, we purchased a pre-constructed plasmid directly from a biosynthetic company and conducted a series of experiments including polymerase chain reaction (PCR), Western Blot (WB), absorbance measurement after wax staining and gas chromatography (GC).

Given that our designed metabolic pathway may impose a significant burden on bacteria, we also included experiments to assess the plasmid's impact on bacterial growth.

The effect of our plasmid on the growth of bacteria ​

We plotted bacterial growth curves by measuring the OD600 of bacterial cultures at different time points. Since we had separately synthesized the regulatory sequence and expression sequence when purchasing the plasmid earlier, we set up a total of five experimental groups: empty plasmid (Control), regulatory sequence (with IPTG), regulatory sequence (without IPTG), expression plasmid (with oleic acid), and expression plasmid (without oleic acid).

Growth curve analysis to assess the metabolic burden of engineered circuits
Figure 11. Growth curve analysis to assess the metabolic burden of engineered circuits. The optical density (OD600) of various engineered E. coli strains was monitored over a 12-hour period. The Control group represents the empty plasmid, the Regulate group represents the regulatory plasmid, and IPTG+ indicates the addition of IPTG. Expression denotes the expression plasmid, and SO+ indicates the addition of oleic acid.

However, for the group treated with oleic acid, due to the low solubility of oleic acid, we encountered significant pipetting errors during liquid aspiration. Consequently, the results deviated considerably from the actual situation.

As shown in Figure 10, most groups exhibit similar growth patterns. Due to errors introduced by oleic acid, the data for the oleic acid-treated group display significant fluctuations, yet it still demonstrates that this plasmid has a substantial impact on bacterial growth.

The PCR result ​

Gel Electrophoresis Results for WS/DGAT, G6PD, and AcrB
Figure 12. Gel Electrophoresis Results for WS/DGAT, G6PD, and AcrB. Electrophoresis results for WS/DGAT, G6PD, and AcrB are shown. Target bands for G6PD (500 bp) and AcrB (1405 bp) were successfully amplified from the experimental samples. The target band for WS/DGAT was not detected. WS/DGAT: Wax ester synthase/acyl-CoA diacylglycerol acyltransferase, G6PD: Glucose-6-phosphate dehydrogenase, AcrB: Acridine resistance protein B.
Browning Index Time Course
Figure 13(a)
Browning Index Statistical Comparison
Figure 13(b)

Figure 13. (a)Gel Electrophoresis Results for WS/DGAT using different primer sets. (b)Sequence of different primer sets. In figure (a) primer set 1 and primer set 2 produced bands at the expected 500-bp position, whereas primer set 3 and the negative control did not.

We discovered that the issue might stem from the primers we were using. Therefore, we redesigned the primers for PCR.

We conducted PCR experiments using both new and old primers. Primer set 1&2 are new, while primer set 3 is the previous one. PCR results indicate that our newly designed primer sets 1 and 2 successfully amplified the target band, confirming the successful design of the new primers. Meanwhile, primer set 3 and the negative control failed to amplify any bands, further validating the rigor of the experiment.

Western blot ​

Western blot analysis confirming the expression of WS/DGAT-flag
Figure 14. Western blot analysis confirming the expression of WS/DGAT-flag. Cell lysates from three independent biological replicates (Group 1, 2, and 3) were analyzed by immunoblotting using an anti-FLAG antibody. For each replicate, a distinct band corresponding to WS/DGAT-flag was detected in the experimental sample ('sample') at an approximate molecular weight of 90 kDa. The protein was absent in the negative control lanes ('-'). The positions of molecular weight markers are indicated on the left. WS/DGAT: Wax ester synthase/acyl-CoA diacylglycerol acyltransferase.
Quantification of WS/DGAT protein expression by Western blot analysis
Figure 15. Quantification of WS/DGAT protein expression by Western blot analysis. The relative band intensity was measured for cells containing the pBS-WS/DGAT expression vector versus a negative control. Data are presented as the mean ± SEM of three independent biological replicates (n=3), with individual data points overlaid. Statistical significance was determined by an unpaired Welch's t-test. The double asterisk (**) indicates a statistically significant difference (P = 0.001).

We performed Western blot experiments for proteins involved in the wax synthesis pathway. However, due to technical limitations and experimental constraints, we conducted only one WB experiment. In this experiment, we observed only the WS/DGAT protein producing a band at the correct position, while the other two proteins did not. We also performed a T-test analysis on the grayscale values of the WS/DGAT results, which revealed significant differences, rigorously validating the effectiveness of our WB results.

Absorbance measurement after wax staining ​

Oil Red O is a dye that readily binds to wax components. By measuring the absorbance curve of the bacterial culture after staining, we can determine whether wax components are present in the culture, thereby verifying whether the designed plasmid functions as intended.

Figure 16. Raw absorption spectra of all experimental samples and controls. The absorbance spectra of engineered bacterial cultures were measured alongside the key reagents used in the lipid quantification assay. The distinct peak for Oil Red O (red line) confirms its maximum absorbance at approximately 518 nm. The broad spectra of the bacterial samples represent the combined absorbance from cell components and light scattering before normalization.
Figure 16. Raw absorption spectra of all experimental samples and controls. The absorbance spectra of engineered bacterial cultures were measured alongside the key reagents used in the lipid quantification assay. The distinct peak for Oil Red O (red line) confirms its maximum absorbance at approximately 518 nm. The broad spectra of the bacterial samples represent the combined absorbance from cell components and light scattering before normalization.
Figure 16. Raw absorption spectra of all experimental samples and controls. The absorbance spectra of engineered bacterial cultures were measured alongside the key reagents used in the lipid quantification assay. The distinct peak for Oil Red O (red line) confirms its maximum absorbance at approximately 518 nm. The broad spectra of the bacterial samples represent the combined absorbance from cell components and light scattering before normalization.
Figure 17. Comparison of raw absorption spectra from bacterial cultures. This figure highlights the spectral differences between the target bacteria and control strains after induction. The culture containing the target expression vector (green line) shows a pronounced increase in absorbance between 400-550 nm, indicating the production of wax that is absent in the control strains.
Figure 16. Raw absorption spectra of all experimental samples and controls. The absorbance spectra of engineered bacterial cultures were measured alongside the key reagents used in the lipid quantification assay. The distinct peak for Oil Red O (red line) confirms its maximum absorbance at approximately 518 nm. The broad spectra of the bacterial samples represent the combined absorbance from cell components and light scattering before normalization.
Figure 18. Normalized absorbance spectra of engineered E. coli cultures. After background correction, the raw spectra were normalized using the absorbance value at 700 nm as a reference to eliminate the influence of cell density variations. This data processing method isolates the spectral signature of Oil Red O within the bacteria. The results confirm that the increased absorbance of the target strain (green line) directly originates from the Oil Red O-stained wax, validating the functional efficacy of the engineered metabolic pathway.

The result shows that the experiment group manage to achieve a much higher optial density at 500-600nm while the one at 600/700nm keeps at a lower value, which suggest that it achieve higher light absorption with a lower bacteria density. More importantly, the range is exactly the absorption peak of oil red stain. We could conclude that our bacteria succeeded in synthesizing a high level of neutral lipids,however, the component of the prodcut are still requires more accurate analysis methods like GC-MS.

Gas chromatography ​

To detect the production and accumulation of fatty acids and fatty alcohols, we also performed gas chromatography analysis. Below are some of our test results:

Figure 19. Parts of the result of gas chromatography analysis.The figure shows some of the detected substances and their relative absorbance values.
Figure 19. Parts of the result of gas chromatography analysis.The figure shows some of the detected substances and their relative absorbance values.

The overall results indicate that our bacterial samples contain significant amounts of fatty acids and fatty alcohols. This provides supporting evidence that our fatty acid accumulation component is functional. However, we did not conduct control group testing, which may lead to inaccuracies in our conclusions.

Dry Lab Results ​

Overview ​

Our dry lab efforts produced working software, models, datasets, and protocols that directly supported our wet lab experiments and created generalizable resources for future teams. Below we summarize the concrete results achieved in each subproject.

SynbioMCP – AI Agent Framework ​

Result: We successfully implemented SynbioMCP, an AI-agent software system that integrates large language models with bioinformatics tools. The proof-of-concept includes modules for sequence translation, protein visualization, and DNA optimization.

Validation: Tested across multiple use cases relevant to our project and benchmarked against traditional manual workflows.

Deliverables:

  • A functioning plug-in framework with modular extensibility.
  • Documentation and a demo video published on our wiki and GitHub/GitLab.
  • Example toolchains showing automated workflows (e.g., visualization and optimization pipelines).

Extended impact: Our user experience research, based on grounded theory, demonstrated that offline engagement and real-time feedback loops can guide software usability. This provides a framework future teams can reuse for technology validation and user studies.

BactaGenome – Genome-Scale Modeling ​

Result: We developed a deep learning pipeline for transcriptional regulation modeling, adapted from AlphaGenome. The system was trained and tested on E. coli using the PRECISE RNA-seq dataset and RegulonDB.

Key findings:

  • Models trained with simple MSE loss could fit overall expression trends but lacked gene-specific clarity.
  • Switching to the combined loss function improved interpretability and predictive performance.

Deliverables:

  • Training scripts, model configuration files, and evaluation code openly shared.
  • Comparative results illustrating the difference between loss functions.

Extended impact: Expert feedback (from DeepMind researchers and Prof. Pan) validated the novelty and robustness of our approach, increasing reproducibility for other teams attempting genome-scale modeling.

Molecular Dynamics & Thermodynamic Simulations ​

Result: We established a complete MD simulation workflow for enzyme thermal stability, focusing on lychee-related metabolic enzymes.

Pipeline includes:

  • System setup and preparation with annotated input files.
  • GPU-based MD execution (tested with V100 GPUs), including troubleshooting CUDA/NVML errors and handling maxwarn flags.
  • Post-simulation analysis: Energy profiles, structural stability metrics, and radius of gyration (Rg).

Deliverables:

  • Reusable GROMACS input/output files.
  • Python scripts for smoothing, plotting, and interpreting MD data.

Impact: Provides a ready-to-use workflow that lowers the barrier for future teams interested in computational protein stability analysis.

Lychee Recognition Dataset & Model ​

Result: We collected and curated an imaging dataset for lychee fruit, capturing variation and freshness at multiple post-harvest timepoints.

Model: A machine learning pipeline was trained to classify varieties and predict days-after-harvest.

Deliverables:

  • Two dataset versions (V1 and V2), labeled with metadata including image and post-harvest duration.
  • A model capable of non-destructive freshness evaluation.

Impact: This work demonstrates how agricultural imaging + machine learning can be applied to real-world food preservation problems, and the framework is adaptable to other crops.

Datasets, UI, and Tools ​

Result: Designed and deployed a dataset collection interface that allowed structured input of lychee sample metadata.

Deliverables:

  • Completed Version 1 and Version 2 datasets, covering image capture and freshness labeling.
  • Documented dataset plan with standardized schema to encourage reuse.

Impact: Establishes a reusable data infrastructure for crop freshness studies and can serve as a template for future iGEM data-driven projects.

Conclusion ​

Through the integration of AI agents, deep learning models, molecular simulations, and imaging datasets, our dry lab team delivered tested pipelines and reusable resources. These results not only supported our wet lab work but also contribute tools, data, and insights that can empower future iGEM teams and researchers.