Cycle 1: Project Ideation and Initial Pivot

Design

Our team initially voted to pursue a project targeting the gut–brain axis, aiming to engineer a bacterial probiotic that could contribute to the treatment or prevention of Alzheimer’s disease. We focused on the kynurenine pathway, exploring strategies to upregulate or downregulate metabolites involved in neuroinflammation and neurodegeneration.

Build

We began gathering literature and mapping potential genetic targets for metabolic regulation within the kynurenine pathway. We also started developing conceptual models to predict the effects of modulating pathway intermediates on neurological outcomes.

Test

As our research deepened, we encountered limited concrete data on feasible genetic targets and measurable outputs. Simulations and literature review revealed that the pathway’s complexity and its dependence on multi-organ interactions would make experimental validation impractical within our scope.

Learn

We realized that a project involving the gut–brain axis and neurological endpoints would require extensive multi-system modeling and neuro-assays beyond our available time and resources. Guided by our mentors and Dr. Brett, we decided to pivot while maintaining our focus within the gut microbiome field. This learning led us to redirect our project toward a gut-specific disorder: Celiac disease.

Cycle 2: Concept Development

Design

After establishing a new focus, we revisited the design phase. Two main branches of thought emerged:

1. Gluten-Degrading Enzyme Production: Express a plant- or fungal-derived enzyme within a probiotic bacterium, allowing it to degrade gluten in the lumen of the small intestine into fragments too small to trigger immune recognition.
   Our research identified AN-PEP, an enzyme previously studied by the 2024 iGEM UT Austin team [1]. Their work provided a valuable starting point, though they faced difficulties achieving functional expression in their probiotic strain. We aimed to build upon their findings and overcome those limitations.
2. Epitope Masking: A team member proposed an alternative strategy—rather than fully degrading gluten, we could mask the regions recognized by the immune system using analog or binding proteins.

Build

We began in silico construction of gene parts for gluten degradation and modeled AN-PEP activity on representative gluten sequences. Simultaneously, we conducted literature-based modeling to explore the feasibility of epitope masking approaches, refining our ideas based on structural information about gluten peptides.

Test

Computational modeling revealed that cleavage alone was not always sufficient. Depending on sequence variability, some cleavage products remained large enough to retain immunogenic epitopes. Similarly, early modeling showed that epitope masking using analog proteins was biologically complex and difficult to achieve without unintended immune consequences.

Learn

We concluded that relying on a single gluten-degrading enzyme might be inadequate for complete detoxification. This led to two refined design directions:

• A multi-enzyme approach combining two or more enzymes with complementary cleavage motifs to produce smaller, less immunogenic fragments.
• A modified epitope-targeting strategy focusing on blocking or degrading the highly immunogenic 33-mer peptide region of gluten, as previously explored by the 2024 iGEM UM–Macau team [2].

Cycle 3: In Silico Evaluation of AN-PEP Efficiency

Design

During the cloning design process, we identified a key question: could AN-PEP degrade gluten efficiently enough to prevent immune detection? If AN-PEP only partially cleaved gluten, the remaining fragments might still trigger immune responses, reducing therapeutic effectiveness.

Build

To address this concern, we conducted a computational analysis using our Software. The cleavage motif of AN-PEP was input into software that scanned sequences from multiple gluten strains (compiled from a meta-analysis [link] ). The predicted fragments were then compared against a database of known gluten IgE epitopes in humans. This allowed us to identify potential overlaps and assess whether AN-PEP could fully eliminate immunogenic regions.

Test

Results were encouraging: under idealized conditions, AN-PEP appeared capable of cleaving several gluten variants into non-immunogenic fragments. However, the software did not incorporate kinetic parameters—such as enzyme concentration, substrate availability, or digestion time—limiting our ability to predict real-world performance. These findings indicated the need for future wet-lab experiments to test enzymatic activity and degradation rates.

Learn

We concluded that while AN-PEP is a strong candidate enzyme, relying solely on it introduces risk. Implementing a multi-enzyme system could improve degradation efficiency by acting on diverse cleavage motifs, generating smaller fragments, and reducing the probability of leaving intact epitopes. This insight shaped the direction of our next design cycle, focusing on enzyme combination strategies.

Cycle 4: Cloning and Wet-Lab Validation of AN-PEP

Design

Following the computational validation of AN-PEP, we proceeded to clone the construct using the Golden Gate assembly method. The plasmid backbone used in this procedure contains a fluorescent green cassette that is excised when successfully digested by BsaI type II restriction enzymes. Therefore, successful clones were expected to appear white, indicating the likely presence of the assembled AN-PEP plasmid.

However, as detailed in the Sequencing section of our Wet Lab Experiments page, sequencing results from several white colonies revealed unexpected gaps of 1–1.5 kb precisely where the AN-PEP gene, promoter, ribosome binding site (RBS), secretion tag, and terminator should have been. This discrepancy suggested either incomplete digestion or errors during synthesis or assembly. To evaluate a larger number of colonies efficiently and cost-effectively, we designed a secondary screening approach to assess insert presence by plasmid size.

Build

We performed plasmid mini-preps on selected colonies and used 1% agarose gel electrophoresis to compare plasmid sizes. To minimize the interference of supercoiled plasmid conformations, which can obscure accurate size estimation, each sample was digested with XbaI prior to loading on the gel. This allowed for clearer visualization and more reliable comparison between expected and observed plasmid sizes.

Test

We analyzed eight colonies from each of the four plates (though we ran a total of 20 colonies). After electrophoresis, none of the samples displayed plasmid sizes consistent with the predicted construct length. This consistent discrepancy across all plates strongly indicated that the AN-PEP insert was not successfully incorporated.

Learn

The absence of the correct-sized plasmid across multiple clones suggested a systematic failure in the cloning process rather than random colony variation. Potential causes include incomplete restriction digestion, ligation inefficiency, or synthesis errors within the AN-PEP fragment itself. This outcome highlighted the need to re-evaluate our assembly protocol and consider alternate cloning strategies, such as Gibson Assembly or re-synthesizing the AN-PEP insert, to ensure successful plasmid construction. To avoid time loss, the full transcription unit was synthesized.

Cycle 1: Streamlining Software Use

Design

There were two possible designs for the software. One combined all functionalities into a single program, the other had one script for each function. We wanted to ensure our program was convenient for the end user. Using one program is more convenient. However, ensuring reasonable run times is also an important quality of life consideration. We needed to determine which setup was best for the end user.

Build

To determine which design was most convenient both designs were created in R using the R-studio console and tested on our database - GlutPro 6.1 [3].

Test

The programs were compared for speed and efficiency of use. Due to the large amount of data intended to be processed the program with the different functions divided into different files performed significantly faster than the combined program.

Learn

The reduced time requirement of the program built with separate scripts was determined to be a better quality of life feature than having all the programs in one file. As such it was this version of the program that was carried through into future versions.

Cycle 2: The Addition of a Positive Control

Design

Through many smaller DBTL cycles the code had been brought to a point where the enzymes could cut the sequences, epitopes could be matched and output was presented in a series of files - in effect the code fully met the requirements of the minimum viable product. What remained unsure was the accuracy of the code.

Build

To determine if the code was performing accurately a positive control was created. Running this code along with the results would provide confidence to the users that their results were accurate.

Test

The positive control functioned as intended and has since its inception returned positive results.

Learn

The inclusion of the positive control demonstrated that our code was accurate and could be used as a model for epitope binding to enzyme fragments.

References

1. UT Austin iGEM 2024. (2024). Igem.wiki. https://2024.igem.wiki/austin-utexas/results
2. UT-Macau. (2024). Igem.wiki. https://2024.igem.wiki/um-macau/results
3. Bromilow, S., Daly, M., Gethings, L. A., Mills, E. N. C., Nitride, C., & Shewry, P. R. (2020). The GluPro suite of curated cereal seed storage prolamin protein sequence datasets [Data set]. figshare. https://doi.org/12613154