Engineering Success

Engineering Success: Cycles in Narirutin Production

For the first time in the world, we have succeeded in biosynthesizing Narirutin, a flavonoid compound that is found mainly in Jabara, by reconstructing its biosynthetic pathway in yeast using enzymes from not Jabara itself, but from Antirrhinum majus and Citrus sinensis. This was achieved by following engineering design approaches through five DBTL cycles. We detailed each cycle and the key design approaches that were especially fundamental for the success of this project.

Cycle 1: Narirutin extraction from Citrus jabara

Citrus jabara fruit, which originated from the Wakayama Prefecture in Japan, is the greatest source of narirutin. Jabara is a hybrid citrus of yuzu and mandarin, and 99% of the flavonoid in the peel of Jabara is known to be narirutin^[1]. Jabara obtains a patent in which narirutin from the peels of Jabara is reputed to have anti-allergic and antihistamine properties^[1][2]. In the first DBTL cycle, our design goal was to confirm the presence of narirutin in Jabara and to establish the extraction and analysis method as the control for our project.

Design 1 & Build 1

We obtained Citrus jabara fruit, sourced from Wakayama (please also refer to the human practice page in which team members visited Wakayama for further investigation), the primary region for Jabara, and designed a simple extraction protocol. The plan was to freeze-dry the peels that we peeled off from the fruit and pulverize the peels, and soak them in ethanol to extract flavonoids since narirutin is soluble in ethanol. Even though the fruit that we tested was not fresh, since we tested during September while Jabara's seasonal period is in December, we hypothesized that narirutin would still be extractable from how adequate it is inside the peel. Moreover, it is also informed that narirutin is chemically more stable than its aglycone precursor naringenin, and studies report how it remains stable during freeze-drying, ethanol extraction, and short-term storage at room temperature^[3].

Test 1

The ethanolic extract of Jabara turned out to be a yellow solution that contained the fruit's compounds. We analyze the extract using Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS), and a prominent peak corresponding to narirutin’s atomic mass was detected from the Jabara peel extract, matching the standard. This confirmed that nariruin is indeed present in the peels of Jabara, even though the fruit we tested was not ripe and fresh.

Learn 1

From the extraction of narirutin from Jabara, we learned that narirutin extraction is effective even from non-fresh peels, and the glycoside is robust. This gave us confidence in our analytical pipeline for the extraction of metabolites from the recombinant yeast and analyzing them using LC-MS/MS. Being able to confirm Jabara's narirutin content also reinforced our project’s premise that Jabara peels are a great source of narirutin, and reproducing this biosynthesis in a lab chassis would definitely be worthwhile.

Cycle 2: Pathway design and Plasmid construction

With the data that Jabara indeed contains nariruin, Cycle focused on designing the biosynthetic pathway in a microbial host, which we struggled most with. It took us more than 2 months to discuss which genes and enzymes we should select and what chassis or microbial host would be best suited for the biosynthesis of narirutin. Narirutin is naringenin-7-O-rutinoside, the flavonone naringenin decorated with a disaccharide attached at the 7-hydroxyl group^[2]. This means that two enzymatic steps are needed: first attaching glucose, forming a 7-O-glucoside, and then adding the rhamnose to that glucose, which leads to the formation of the rutinoside disaccharide. Additionally, the host must supply the activated sugar donors UDP-glucose and UDP-rhamnose for the transfer reactions.

Design 2

We considered two approaches for obtaining the necessary glycosyltransferase genes: 1. cloning the necessary genes from Jabara itself via cDNA amplification, or 2. using known genes from other organisms and citrus plants. We initially explored the former approach, but we were hit with some obstacles. First of all, the genome of C. jabara has not been sequenced, and designing the degenerate primers based on related citrus enzymes proved difficult. Moreover, glycosyltransferases like rhamnosyltransferases are known to have a highly conserved active-site motifs, but very divergent N- and C-terminus sequences, which means that the primers matching other citrus genes would unlikely amplify the Jabara gene. Therefore, we lacked sufficient sequence information, which was required to clone Jabara’s own rutinosyltransferase. In addition to the time constraints, we decided on the latter approach of using characterized enzymes from the literature that likely would perform similar reactions.

Key enzymes that we chose

1. Flavanone 7-O-glucosyltransferase (F7GT)

This is the enzyme that transfers glucose from UDP-glucose to the 7-OH of the naringenin aglycone^[2]. We selected the F7GT from Antirrhinum majus (snapdragon), which is known to produce glucosylate flavonoids at the 7 position. Using F7GT from snapdragon was a risky step to take, not only because this wasn't a citrus plant but also because this enzyme had never been tested in a yeast host or even in a microbial system. Yet, this enzyme selection was the only and foremost enzyme that we could find for our engineering success.

2. Flanonene-7-O-glucoside rhamnosyltransferase (RhaT)

This is an enzyme that attaches rhamnose from the UDP-rhamnose to the 7-O-glucoside ^[2]. Importantly, to make narirutin, we needed an enzyme that forms a 1,6 linkage of a rutinoside rather than a 1,2 linkage that forms a neohesperidoside^[4]. In Citrus sinensis (sweet orange), a 1,6-rhamnosyltransferase is responsible for converting flavonone-7-O-glucosides into a rutinoside form, just like narirutin. Therefore, we chose a rhamnosyltransferase gene from sweet orange for the second step, which we refer to as 16RT.

These choices were decided by prior studies showing that the conversion from naringenin to nariruin requires a two-enzyme system. By utilizing the known enzymes from snapdragon and orange, we avoided the uncertainties of cloning unknown jabara genes.

(In addition, we were lucky to have a collaborator at AIST who had a cloned UDP-rhamnose biosynthesis gene available, which made it possible for us to fight against the time constraints and ensure that our host produces UDP-rhamnose to supply the sugar donor.)

Build 2 & Test 2

We codon optimized the gene sequences for expression in yeast and assembled them into the plasmids. Our design included strong promoters and terminators for each gene - our constructed plasmids carried: 1. the snapdragon F7GT gene, 2. the sweet orange 16RT gene, 3. the UDP-rhamnose synthesis gene to convert UDP-glucose to UDP-rhamnose within the host. These constructs were verified by methods of restriction digestion, colony PCR, and agarose gel electrophoresis. To ensure that no errors had occurred during cloning, we submitted the plasmids for Sanger sequencing. These plasmids were eventually confirmed, and by the end of cycle 2, we had a complete genetic toolkit that could potentially biosynthesize narirutin: the two-step glycosylation pathway genes and the sugar-supplying gene.

Learn 2

From this cycle, we learned the importance of the rationale of enzyme selection throughout the pathway design. Since it took an immense amount of time to design the pathway, we lost a lot of time in the actual experiment. However, our design’s originality lies in this strategy of combining F7GT from a flower (snapdragon) with a 16RT from sweet orange in a yeast chassis. This discussion and reading of literature was one of the most critical steps of the whole project. By the end of Cycle 2, we were ready to test our engineered pathway.

Cycle 3: Testing Pathway functionality in yeast

With the plasmids that we have in hand, in this cycle, we moved to the testing phase of expressing the pathway in the yeast chassis and checking for narirutin production. We initially considered using Escherichia coli as our chassis, but came to the conclusion to select Saccharomyces cerevisiae as our host. Primary reasons are that yeast has its eukaryotic protein-folding ability and also the ability to handle multiple plasmids ^[5]. We transformed yeast with our pathway plasmids and observed how they could produce narirutin from an added precursor.

Design 3

Our pathway begins with naringenin, and since yeast does not naturally produce naringenin, we planned to feed naringenin externally to the cultures, taking a biotransformation approach. We set three conditions to gauge the pathway’s function: 0nM, 0.1nM, 1.0nM of naringenin feed. These conditions were chosen by balancing the solubility of naringenin in water and the hypothesis that too much naringenin could damage the yeast chassis. It is clearly stated that 100nM of naringenin would not dissolve in water, which shows how it has a limited and low solubility. We took into consideration that around 10nM would be the practical upper limit, but were concerned that the yeast chassis would be damaged; therefore, we decided to have 1.0nM as the highest concentration of naringenin to add externally

Build 3 & Test 3

We cultured the yeast in a minimal medium so that the background signals when analyzing using LC-MS/MS would be minimized, and added naringenin at the start of cultivation for the 0.1 and 1nM groups. We monitored cell growth by its intensity and observed an interesting phenomenon in which the 1.0nM naringenin-fed culture reached a slightly higher density than the other groups. Perhaps naringenin or the glycosides might have had some beneficial effects, but we fully do not understand the reason behind this.

Expressing Protein	10:00	13:00	16:00
5, RHM2 (0.1 mM)	1.38	1.32	1.36
4, UGT (0.1 mM)	1.5	1.6	1.44
6, 16RT (0.1 mM)	1.84	1.9	1.48
1 ,RHM2, UGT, 16RT (0 mM)	1.3	1.24	1.24
2, RHM2, UGT, 16RT (0.1 mM)	1.46	1.36	1.28
3, RHM2, UGT, 16RT (1 mM)	1.76	2.12	3.16

Learn 3

This cycle showed data on how we feed naringenin has a profound impact on both cell growth and product formation. As seen in the table above, there is a clear difference in cell culture growth depending on the concentration of naringenin added externally. Moreover, the 1.0mM feed led to the highest cell density that exceeded 3.0 by 16 hours, while the 0mM no feed and 0,1mM conditions had a significantly lower rate of growth over the same period of time.

Cycle 4: Extraction of Narirutin and Analysis

After cultivation, we harvested both the cells and the supernatant, prepared for extraction and analysis. Here, we performed the solvent extraction on the cell pellets and supernatants separately because the product could be inside the cells or might be secreted. Cycle 4 focuses on the characterization of the product more than anything else. Our project goal is to identify whether the pathway functions; therefore, rather than the quantity of the final product, we prioritized whether Narirutin was produced.

Design 4

We wanted to ensure that the compound extracted was truly narirutin. We designed a workflow to improve confidence in the identification of the target product by utilizing LC-MS/MS, which offers significantly higher specificity and sensitivity than LC-MS. While LC-MS uses only one mass spectrometer for basic mass to charge ratio analysis, LC-MS/MS adds a fragmentation step to filter out the interfering compounds and to confirm the identity of narirutin with greater precision.

Build 4 & Test 4

Using LC-MS/MS, we were able to obtain a high-resolution mass spectrum of the narirutin peak from the 1mM naringenin-fed extract. The measured mass matched narirutin’s expected molecular ion within a few milliDaltons, confirming the molecular formula. Moreover, it is also convincing how the fragmentation produced a major fragment at m/z 273, which corresponds to naringenin and indicates how the molecule contains the naringenin core and has successfully transformed into narirutin through the enzymatic processes (refer to fig. 3). Taken together, this analysis proves that the engineered pathway produced the target molecule narirutin!

To share the results, as shown in fig. 4: in 0nM, no narirutin was found as expected; in 0.1nM, there was a very small peak corresponding to narirutin’s mass; in the 1nM feed, we found a distinct narirutin peak in the cell extract! This was definitely a eureka moment for us and quantitatively confirmed how our yeast cells successfully biosynthesized narirutin from naringenin. The retention time also matched that of the narirutin extract from the Jabara peel in cycle 1, which was definite proof for the product’s identity. From fig. 5, we also observed a much larger peak on the intermediate, naringenin-7-O-glucoside, which demonstrated how the first enzyme of snapdragon worked robustly, although the second step of adding rhamnose was partially limiting. Notably, most of the narirutin was found inside the cells rather than in the supernatant, which makes sense as narirutin is a charged glycoside that cannot be transported out of yeast so easily.

Learn 4

This cycle showed significant data. From this cycle, we learned how substrate concentration is crucial because of the dramatic difference in the narirutin production between 0.1nM and 1.0nM. Secondly, we figured out that there is possibly a rate-limiting step of the rhamnosyltransferase in which the enzyme had a lower activity, or the UDP-rhamnose supply was limited. Finally, we were also able to observe that the product mostly stays intracellular, hinting that exporting might be a challenge. These learnings provide new insights and ideas to develop a better narirutin biosynthesis platform for future experiments. Nonetheless, being able to observe a new-to-nature narirutin in yeast was a tremendous success, and to our knowledge, this is the first time narirutin has successfully been biosynthesized in a microbe.

Beyond Cycle 5: Design and Future Improvements

Being able to achieve the baseline success of producing narirutin in the designed pathway was extremely beneficial and significant, based on the fact that Jabara is an expensive plant, and being able to biosynthesize narirutin in our pathway could be more cost-efficient than consuming the peels of the actual plant. Our final Cycle would be more about designing the next steps to increase nariruin yield and essentially turning our proof-of-concept into a more efficient and practical system.

Design 5

1. High-Precursor Feeding

Our future plan would be to increase the concentration of naringenin to feed the yeast microbe. Since providing more naringenin changed the outcome of narirutin dramatically, we plan to extend this further and see the limitations in which the yeast would be able to handle. One idea is to feed naringenin from 5-10nM to the culture.

2. Enhance Enzyme Performance

We might screen or engineer for a more efficient 1,6rhamnosyltransferase, and one idea that we currently have is to go back to our initial idea and attempt to clone the 1,6RT, now that we know that F7GT from snapdragon worked efficiently despite not being a citrus plant, but just a flower. We could also test alternative rhamnosyltransferases from other citrus plants, although this will take time and is not guaranteed.

3. Better host and conditions

Since we used a minimal medium to remove noise for the analysis, this likely limited cell growth and the expression of the enzymes. Therefore, we consider using a richer medium that could possibly boost the titer to maximize the production of nairutin. Another factor to consider is the host itself, and our current Saccharomyces cerevisiae could be improved by using a yeast with higher tolerance.

4. Exporting the Product

We figured out that narirutin stays mostly inside the cells, so extracting it requires cell lysis. In the future, we aim to engineer a solution for the host microbe to secrete narirutin into the medium. One idea we currently have is to express a yeast transporter protein that could pump glucosides out of the cells. There are already known transporters for flavonoid glycosides, so inserting one could help the yeast export narirutin as it is made and would allow the project to develop into a factory from biosynthesis to easy extraction.

Fusion Peptide

Abstract

Allergic rhinitis involves a complex cascade of immune responses, from early-phase Th2 cytokine signaling to late-phase eosinophil recruitment and immune tolerance. To address this, we designed a modular, protease-responsive fusion peptide that targets multiple stages of the allergic response in a single construct. The final optimized sequence incorporates a cell-penetrating peptide (CPP) for mucosal and intracellular delivery, two functional domains for early- and late-phase immunomodulation, a stabilizing EEEE segment, and a CTLA-4 mimic to promote regulatory T-cell activation. Enzyme-cleavable linkers were strategically placed to enable temporally controlled domain release in inflamed tissues. Over 40 candidate sequences were generated and evaluated through computational Build-Test-Learn cycles, using HDOCK, CamSol, DeepSTABp, AlphaFold2, ProsperousPlus, and ChimeraX to assess folding, stability, solubility, receptor binding, and linker functionality. Confidence scores were generally low, reflecting the flexible and intrinsically disordered nature of linkers and short peptides, but structural and functional predictions indicated that the design would retain accessibility and activity. The IL-5Rα domain was removed during optimization due to redundancy and stability concerns and replaced with the EEEE segment to enhance solubility and structural balance. Future work includes peptide assays, such as protease cleavage and chromophore-binding tests, to validate in silico predictions, as well as plasmid-based expression for repeated functional studies. Overall, this project demonstrates a rational, multi-phase design strategy for a targeted anti-allergic peptide, combining computational modeling, protease-responsive release, and delivery optimization to maximize therapeutic potential while minimizing off-target effects.

Design

Our fusion peptide was designed to modulate key stages of the allergic cascade through a modular and protease-responsive structure. Each domain targets a different pathway involved in allergic rhinitis, from early-phase cytokine signaling to immune tolerance induction. The final optimized sequence is: GRKKRRQRRRPPQKPVPLGLAGCKGERFGFLGEEEEIEPDMYPPPY. This design ensures targeted activity, mucosal permeability, and structural stability, addressing both early and late allergic responses in a single therapeutic construct.

Peptide Selection

Initial Design Concept

Originally, our initial construct contained four functional domains:
1. KPV – IL-4Rα/IL-13Rα1 antagonist (reducing Th2 cytokine signaling)
2. CKGERF – CCR3 blocker (preventing eosinophil recruitment)
3. VDECWRIIASHTWFCAEE – IL-5Rα antagonist (inhibiting eosinophil activation and survival)
4. MYPPPY – CTLA-4 mimic (promoting regulatory T-cell activation and immune tolerance)

This structure was intended to sequentially suppress the allergic cascade, from cytokine signaling inhibition (early phase) to inflammation resolution (late phase). However, during design optimization, the domain 3 peptide, VDECWRIIASHTWFCAEE (IL-5Rα antagonist), caused sequence instability and displayed functional overlap with the immune-tolerance domain (MYPPPY). Since both targeted eosinophil regulation and inflammation resolution, the domain was removed to simplify and stabilize the peptide without compromising therapeutic scope.

Linker Selection

To achieve controlled and localized release, enzyme-cleavable linkers were introduced between domains

• Linker 1 – PLGLAG

An MMP-9 sensitive peptide was introduced between domains 1 and 2 to enable early release of domain 1 at the nasal epithelium, where MMP-9 levels are elevated during inflammation.

• Linker 2 – GFLG

A cathepsin B-cleavable peptide was introduced between domains 2 and 3 to allow domain 2 to act in inflamed submucosal regions rich in immune-cell infiltration.

• Linker 3 – IEPD

A granzyme B-responsive peptide was introduced between domains 3 and 4 to facilitate the release of Domains 3 and 4 in chronically inflamed tissues with active cytotoxic T cells.

This modular, protease-responsive design ensures activation only in inflamed tissues, minimizing systemic exposure and maximizing therapeutic precision.

Build

In the dry lab Build phase, our goal was to design a multi-domain fusion peptide with precise temporal and spatial activity against allergic inflammation. For each functional domain, we selected 2–3 candidate sequences based on a literature review of biologically active peptides. Candidate domains included an IL-4Rα/IL-13Rα1 antagonist to reduce Th2 cytokine signaling for domain 1, a CCR3 blocker to prevent eosinophil recruitment for domain 2, an IL-5Rα antagonist to inhibit eosinophil activation and survival for domain 3, and a CTLA-4 mimic to promote regulatory T-cell activation and immune tolerance for domain 4. Using these candidates, we assembled over 40 fusion peptide sequences in silico, connecting domains with enzyme-responsive linkers designed for temporally controlled release. The first linker, sensitive to MMP-9, was placed between the first and second domains to allow early release at the nasal epithelium during inflammation. The second linker, cleavable by cathepsin B, separated the second and third domains, enabling activity in immune-cell–rich submucosal regions. The third linker, responsive to granzyme B, linked the later domains, facilitating release in chronically inflamed tissue with active cytotoxic T cells.

During testing, the IL-5Rα domain was found to reduce overall peptide stability, and its functional role was largely redundant with the CTLA-4 mimic domain. To address this, the IL-5Rα domain was removed and replaced with a flexible, acidic EEEE segment, which acted as a stabilizing linker while maintaining proper spacing and folding between domains. This adjustment improved predicted solubility, structural stability, and overall feasibility for expression.

This dry lab Build phase produced a final optimized fusion peptide sequence, ready for computational testing using HDOCK, CamSol, DeepSTABp, AlphaFold2, ProsperousPlus, and ChimeraX, enabling evaluation of stability, folding, and functional accessibility, and guiding the selection of the most promising candidate for wet lab validation.

Software Testing

To validate the stability, solubility, and interaction potential of our fusion peptide, we conducted a comprehensive series of in silico analyses using multiple bioinformatics tools. Each software was selected for its ability to evaluate a specific aspect of peptide design, from structure prediction to solubility and receptor binding. This computational testing helped refine our sequence before experimental synthesis, ensuring that our construct was both functionally effective and structurally feasible.

Molecular Docking – HDOCK

Purpose: To assess the binding affinity of each domain to its respective target receptor and confirm that fusion did not disrupt key interaction sites.

Method: Performed molecular docking simulations using HDOCK, a hybrid docking algorithm that combines template-based modeling with ab initio docking. Each peptide domain was docked separately against its corresponding receptor.

• KPV peptide with IL-4Rα/IL-13Rα1
• CKGERF with CCR3
• MYPPY with CD80/CD86 (CTLA-4 binding region)

Findings: Docking results showed that all domains retained favorable binding conformations with negative binding free energies, indicating strong receptor affinity. Most combinations had a 0.85~0.99 confidence score, while any score above 0.7 is considered a very good binding possibility. Importantly, the fused structure maintained accessibility of each domain’s binding interface, confirming that the linkers did not interfere with receptor recognition.

Solubility Prediction – CamSol

Purpose: To evaluate peptide solubility and aggregation tendency.

Method: Utilized the CamSol intrinsic solubility prediction tool, which calculates solubility profiles based on amino acid composition and hydrophobicity patterns.

Findings: The final fusion peptide displayed an overall positive solubility score, supporting its suitability for aqueous formulation and nasal delivery.

Stability Prediction – DeepSTABp

Purpose: To estimate the structural stability of the final peptide construct under physiological conditions, including variations in pH.

Method: Applied DeepSTABp, a deep-learning–based predictor trained on large-scale proteome stability datasets. The tool evaluates thermal and conformational stability of peptides and proteins, taking into account physiological pH ranges (pH 5–8), which are relevant for nasal mucosa environments.

Findings: The model predicted high overall stability for the finalized sequence, particularly in the TAT-KPV and MYPPY regions. Stability remained robust across pH 5–8, suggesting the peptide will maintain folding and function under both slightly acidic (inflamed nasal tissue) and neutral conditions. Overall, the fusion peptide is predicted to be thermally and pH-stable, supporting its practical feasibility for nasal delivery and ensuring that functional domains remain active in variable tissue environments.

Structure Prediction – AlphaFold2

Purpose: To generate an accurate 3D structure of the complete fusion peptide and visualize domain folding and linker flexibility.

Method: Used AlphaFold2, one of the most advanced structure prediction systems, to model the tertiary structure based on the final amino acid sequence.

Findings: The predicted model showed that both the domains and the linkers exhibited lower confidence scores (pLDDT 50–70; yellow/orange). Such lower confidence is common in short peptides or intrinsically disordered regions and does not indicate poor design. Rather, it reflects the inherent flexibility of these regions, which is beneficial for independent domain movement and proper linker function.

Protease Cleavage Site Prediction – ProsperousPlus

Purpose: To confirm that the selected linkers were specifically cleavable by their target proteases and would not be degraded by unrelated enzymes.

Method: Used ProsperousPlus, a protease substrate prediction tool, to simulate protease-peptide interactions for MMP-9 (targeting linker 1), Cathepsin B (targeting linker 2), and Granzyme B (targeting linker 3).

Findings: ProsperousPlus accurately identified the chosen linkers as high-probability substrates for their respective enzymes, with minimal predicted cross-reactivity. This confirmed that each domain would be released selectively under inflammation-specific protease conditions, validating the peptide’s spatiotemporal activation design.

Structural Visualization – ChimeraX

Purpose: To visualize inter-domain spatial arrangements, linker orientation, and solvent-accessible surface area for presentation and further analysis.

Method: Modeled structures from AlphaFold2 were visualized using UCSF ChimeraX, allowing for high-resolution inspection of domain interfaces and electrostatic surfaces.

Findings: Visualization confirmed that functional motifs were surface-exposed and spatially separated, ensuring accessibility for receptor binding and enzymatic cleavage. Electrostatic mapping showed complementary charge distributions between peptide domains and their targets, further supporting effective molecular recognition.

Toxicity Prediction – ToxinPred

Purpose: To assess the potential toxicity of the fusion peptide and ensure that it is safe for therapeutic applications, especially given its modular design and inclusion of cell-penetrating and immune-modulating domains.

Method: The candidate sequences were analyzed using ToxinPred, which employs a combination of machine learning algorithms and motif-based approaches to predict peptide toxicity. Both the hybrid and amino acid composition models were used to provide a comprehensive prediction.

Findings: ToxinPred predicted the fusion peptide to be non-toxic with high confidence. The model evaluated both sequence motifs and physicochemical properties associated with known toxic peptides and found no motifs indicative of toxicity. These results suggest that the fusion peptide is unlikely to elicit harmful effects when used in biological assays or therapeutic contexts.

Parts

Parts Code	Parts Name	Type I	Type II
BBa_2576SBJC	TAT cell-penetrating peptide (HIV-1 derived)	Basic Part	Cell-Penetrating Peptide (CPP)
BBa_25IEL30I	KPV peptide – Th2 cytokine pathway anthagonist	Basic Part	Peptide/Coding Sequence
BBa_25LCPBD4	CKGERF peptide – CCR3 antagonist for inhibition of eosinophil recruitment	Basic Part	Peptide/Coding Sequence
BBa_25PZD52G	EEEE acidic spacer – structural stabilizer and anti-aggregation domain	Basic Part	Peptide/Coding Sequence
BBa_25HD5H3J	MYPPY peptide – CTLA-4 mimic for Treg activation and immune tolerance	Basic Part	Peptide/Coding Sequence
BBa_25T2O2F9	PLGLAG peptide – MMP-9 sensitive linker	Basic Part	Peptide/Coding Sequence
BBa_25GQT08B	GFLG peptide – cathepsin B sensitive linker	Basic Part	Peptide/Coding Sequence
BBa_25A49EXK	IEPD peptide – granzyme B sensitive linker	Basic Part	Coding Sequence
BBa_25A49EXK	IEPD peptide – granzyme B sensitive linker	Basic Part	Coding Sequence
BBa_25ZP60RK	TAT-KPV functional domain – cell-penetrating and Th2 cytokine-modulating domain	composite part	Plasmid/Peptide
BBa_25OW5BWL	TTAT-KPV-PLGLAG-CKGERF-GFLG-EEEE-IEPD-MYPPY fusion peptide – multi-functional anti-allergy therapeutic	composite part	Plasmid/Peptide

Learn

We analyzed the results from our computational testing to determine which fusion peptide design would perform best in terms of stability, folding, and functional accessibility. By comparing over 40 candidate sequences using AlphaFold2, ChimeraX, and ProsperousPlus, we were able to identify the most promising combination of domains and linkers. The analyses showed that removing the IL-5Rα domain and replacing it with the EEEE segment significantly improved overall peptide stability and solubility without compromising functional activity. Confidence scores from AlphaFold2 were generally low throughout the peptide, which is common for short peptides and intrinsically disordered regions. This low confidence does not indicate design flaws; rather, it reflects the natural flexibility and dynamic behavior of the linkers and some functional domains, which is desirable for proper temporal release and functional accessibility. Importantly, this phase also highlighted the necessity of including a cell-penetrating peptide (CPP) to facilitate intracellular delivery of the KPV domain and to enable mucosal penetration, ensuring that the early-phase IL-4Rα/IL-13Rα1 antagonistic effects could be realized in target tissues. Overall, this phase enabled us to identify the most effective domain combinations, linker placements, stabilizing modifications, and delivery strategies, providing clear guidance for designing the final fusion peptide sequence for wet lab testing and potential therapeutic applications.

Final Design

Domain 1 – TAT-KPV segment

The KPV peptide antagonizes IL-4Rα/IL-13Rα1, blocking Th2 cytokine signaling responsible for mucus hypersecretion and chronic inflammation in the early phase of hay fever. The addition of a cell-penetrating peptide (CPP) enhances mucosal permeability, allowing intracellular delivery since KPV acts within epithelial cells.

Domain 2 – CKGERF

This peptide binds to CCR3, inhibiting eotaxin-mediated eosinophil recruitment during the late-phase inflammatory response.

Domain 3 – EEEE

This acidic spacer serves as a structural stabilizer and mild electrostatic repulsion domain to prevent aggregation. It replaced the unstable IL-5Rα-binding domain and contributes to maintaining the peptide’s solubility and conformational balance.

Domain 4 – MYPPY

This CTLA-4 mimic promotes TGF-β1–mediated regulatory T-cell activation, facilitating immune tolerance and long-term suppression of allergic relapse.

Together, these domains offer a multi-phase therapeutic strategy, complementing narirutin’s early-phase stabilizing effects while ensuring safe and efficient delivery through the nasal mucosa.

Future Plans

Although the fusion peptide has been successfully synthesized and purchased, the necessary reagents for functional testing did not arrive on time, delaying our planned experiments. Once the reagents are available, we aim to perform peptide assays to evaluate key properties, including solubility, stability, and bioactivity. Specifically, we plan to conduct protease cleavage tests to confirm that the enzyme-responsive linkers function as designed, releasing each domain in response to the target proteases at the appropriate simulated tissue environments. Additionally, we intend to perform chromophore-binding assays to investigate whether the fusion peptide can effectively interact with its target receptors or binding partners, which will provide direct evidence of functional accessibility and domain activity.

In parallel, we plan to order an expression plasmid containing the optimized fusion peptide sequence to enable in-house peptide production. This approach will allow us to generate sufficient quantities for repeated assays, as well as to explore mucosal delivery and intracellular delivery via the CPP. By completing both the purchased peptide assays and plasmid-based expression, we expect to validate our computational predictions, confirm the functional performance of the fusion peptide, and gather essential data for potential therapeutic applications.

Citations

[1] Zhang, Z., et al. (2020). Study on the solubility of naringenin and its derivatives in different solvents Trends in Immunotherapy, 4(1), 1–8.
https://systems.enpress-publisher.com/index.php/ti/article/view/844/412

[2] Ozdal, T., Caba, Z.T., Cavdar, H., Karaca, A.C., Capanoglu, E., Tomas, M. (2023). Narirutin: Advances on Resources, Biosynthesis Pathway, Bioavailability, Bioactivity, and Pharmacology. In: Xiao, J. (eds) Handbook of Dietary Flavonoids. Springer, Cham. https://doi.org/10.1007/978-3-030-94753-8_32-1

[3]M. Igual, E. García-Martínez, M.M. Camacho, N. Martínez-Navarrete, Changes in flavonoid content of grapefruit juice caused by thermal treatment and storage, Innovative Food Science & Emerging Technologies, https://www.sciencedirect.com/science/article/abs/pii/S1466856411000026

[4]Chaudhary, P. R., Bang, H., Jayaprakasha, G. K., & Patil, B. S. (2016). Variation in Key Flavonoid Biosynthetic Enzymes and Phytochemicals in 'Rio Red' Grapefruit (Citrus paradisi Macf.) during Fruit Development. Journal of agricultural and food chemistry, 64(47), 9022–9032. https://doi.org/10.1021/acs.jafc.6b02975

[5] Partow, S., Siewers, V., Bjørn, S., Nielsen, J., & Maury, J. (2010). Characterization of different promoters for designing a new expression vector in Saccharomyces cerevisiae. Yeast (Chichester, England), 27(11), 955–964. https://doi.org/10.1002/yea.1806

[6] Toho University Faculty of Science, Department of Biology, "Cell-Penetrating Peptides," https://www.toho-u.ac.jp/sci/bio/column/0819.html

[7] Revolution Health & Wellness, "KPV Peptide: A Breakthrough for Inflammation, Immunity, and Gut Health," https://revolutionhealth.org/blogs/news/peptide-therapy-kpv

[8] M. Houimel, M. Ben Salah, M. Mlayah, M. Ben Hamida, M. Ben Ali, A. Boudawara, A. Ben Ammar, "Chemokine CCR3 ligands-binding peptides derived from a natural protein," Peptides, https://www.sciencedirect.com/science/article/pii/S016524781200243X

[9] B. Chen, Z. Kang, E. Zheng, Y. Liu, J.W. Gauld, Q. Wang, "Hydrolysis Mechanism of the Linkers by Matrix Metalloproteinase-9 Using QM/MM Calculations," Journal of Chemical Information and Modeling, https://pubmed.ncbi.nlm.nih.gov/34649435/

[10] C. Liesche, P. Sauer, I. Prager, D. Urlaub, M. Claus, R. Eils, J. Beaudouin, C. Watzl, "Single-Fluorescent Protein Reporters Allow Parallel Quantification of Natural Killer Cell-Mediated Granzyme and Caspase Activities in Single Target Cells," Frontiers in Immunology, https://pmc.ncbi.nlm.nih.gov/articles/PMC6092488/

[11] Y. Zhang, J. Zhou, Y. Wang, Y. Wu, Y. Li, B. Wang, G. Liu, Q. Gong, K. Luo, J. Jing, "Stimuli-responsive polymer-dasatinib prodrug to reprogram cancer-associated fibroblasts for boosted immunotherapy," Journal of Controlled Release, https://pubmed.ncbi.nlm.nih.gov/40054628/