Results

The CTX demonstrated strong antifungal activity, completely inhibiting Penicillium digitatum at 25 µM and Geotrichum candidum at 50 µM within 72 hours. After 120 hours, 50 µM was sufficient to block both pathogens, validating CTX as a powerful antimicrobial candidate to address the major post-harvest fungi responsible for significant losses in orange production.

It is important to note that this was an initial assessment of CTX’s potential against two key post-harvest pathogens. Interestingly, the commercial fungicide Imazalil showed no effect against Geotrichum candidum, whereas CTX displayed clear inhibitory activity, revealing a promising activity. This resistance pattern of Geotrichum to conventional fungicides such as Imazalil aligns with insights gathered during our discussions with Alpha Citrus, reinforcing the relevance of developing alternative solutions.

For real-world application within the citrus production chain, further validation will be necessary, including experiments involving the direct application of CTX to fruits under storage conditions. At this stage, however, the limited quantity of purified peptide available constrained this type of testing.

Finally, CTX was officially registered as a new antimicrobial peptide part in the iGEM Registry under the code BBa_25ONI4BC, contributing to the open repository of synthetic biology tools for future research.

Greening, or Huanglongbing (HLB), is recognized as the most devastating citrus disease worldwide. Aware of the urgency to address this challenge, we sought innovative ways to evaluate the potential of our solution. However, working directly with the causal agent, Candidatus Liberibacter asiaticus (CLas), poses significant challenges, as the bacterium cannot be cultured under standard laboratory conditions. Moreover, plant-based assays are complex, time-intensive, and demand large quantities of peptide, as further discussed in more detail in our modeling page.

To overcome these limitations, we established a collaboration with the Centro de Citricultura de Cordeirópolis (IAC), where we developed an indirect yet rapid assay using infected citrus leaves. This test allows us to assess CTX’s potential under conditions that closely mimic natural infection while maintaining experimental feasibility.

The experimental design involved collecting leaves from infected trees and excising approximately 15 mg of tissue from the petiole tip for DNA extraction (baseline sample). The remaining leaf tissue was placed in a 50 mL Falcon tube containing 2 mL of treatment solution and incubated at room temperature for 7 days. After incubation, the same amount of tissue (15 mg) was excised again from the petiole, and DNA was extracted. Both DNA samples (before and after treatment) were sent to a specialized diagnostic clinic for qPCR quantification of CLas, allowing us to compare bacterial loads and evaluate treatment effectiveness. To address this question, we set up a pilot assay in which infected leaves were incubated for 7 days in Falcon tubes containing either water or culture medium. DNA was extracted before and after incubation and analyzed by qPCR. The results confirmed that the bacterium remained detectable after 7 days, with only a slight reduction in concentration. From this cycle, we learned that CLas was able to persist in excised leaf tissue for the entire incubation period, providing the foundation to proceed to the next DBTL cycles.

After confirming that Candidatus Liberibacter asiaticus was detectable in excised leaf tissue, we wanted to understand whether a treatment applied to the incubation solution could reduce the bacterial load inside the leaf in a way that was detectable by qPCR. From the literature, we identified that tetracycline is effective against Liberibacter. For this purpose, excised citrus leaves were incubated for 3 days with CTX (100 mM) or tetracycline (100 µg/mL, positive control). DNA extraction confirmed the presence of Candidatus Liberibacter asiaticus (CLas) by conventional PCR.

Bacterial quantification was carried out by qPCR. This experiment aimed to directly evaluate the potential of CTX as an antimicrobial agent against the citrus greening pathogen under controlled conditions. For this purpose, eight samples were analyzed, separated according to treatments, and collected before and after the experimental period. The results indicated that there was no significant reduction in the bacterial population in the samples after treatment (table). However, this lack of reduction was also observed in the positive control, which should have shown a decrease, suggesting that the exposure time used in the experiment may have been insufficient. Therefore, we suggest that in future assays the treatment period be extended in order to more accurately assess the efficiency of the procedure. In addition to this modification, we also considered the application of CTX directly to induced roots of cuttings derived from HLB-infected plants, since the root system has greater efficiency in absorbing molecules than the petiole and roots obtained from cuttings would present a more homogeneous bacterial load. This approach has not yet been carried out due to the time required for root induction, but it represents a promising alternative to test the effectiveness of different possible treatments. Thus, this test remains a highly promising alternative to address the challenge posed by the inability to culture CLas under conventional laboratory conditions, offering a rapid and versatile method to evaluate different types of treatments against CLas.

In order to guide the design of our biotechnological and expression strategy for CTX production, we assessed the peptide not only against citrus pathogens but also against our intended production hosts (biofactories). Therefore, We tested against three potential production hosts: Escherichia coli, Saccharomyces cerevisiae, and Aspergillus oryzae.

The tests revealed that CTX displayed strong inhibitory effects on E. coli and A. oryzae. The yeast showed good resistance at the tested concentrations, but growth was still slightly affected. These insights confirmed the need for a protective expression strategy—leading to the decision to fuse CTX to carrier proteins for safe production and later cleavage to obtain the peptide.

To enable CTX production in cell factories, our team developed a coupling-based strategy, linking the peptide to a highly expressed protein, thereby inactivating its antimicrobial activity during expression.

To test this hypothesis, we selected sfGFP (BBa_I746916) as the carrier protein because it is highly soluble, easily detectable, and well characterized as a reporter. The K7 construct included a flexible 3×GS linker (BBa_J18921) to provide spatial separation from sfGFP; dual TEV protease sites (BBa_K20750037) flanking the CTX to enable precise release; and an 8×His tag (BBa_K4422007) to allow purification by Immobilized Metal Affinity Chromatography (IMAC). After purification, TEV cleavage yielded a small, tag-free CTX ready for downstream use. Considering CTX’s small size (~2.3 kDa), we planned its separation from the larger carrier fragments (~30 kDa) via ultrafiltration using a 10 kDa MWCO membrane, allowing CTX to pass through while retaining the carrier (size-based cleanup).

We selected E. coli and Aspergillus oryzae to test this hypothesis and express the sfGFP–CTX construct. We successfully engineered both microorganisms and confirmed expression of the fusion protein. In E. coli, expression, purification, and size-based separation yielded approximately 2 mg of CTX per liter of culture.

However, we were not able to clearly visualize the peptide on SDS-PAGE gels, most likely due to its low concentration and small molecular weight. This highlighted the need for more sensitive analytical techniques. In future cycles, HPLC will be applied to accurately detect and quantify CTX at such low concentrations, ensuring reliable characterization of the peptide.

In the filamentous fungus A. oryzae, we successfully clone the sfGFP-CTX into the genome using CRISRPR/CAS9 by the DIVERSIFY system. The expression was validated by SDS-PAGE analysis and fluorescence detection of secretome of 3 candidates cultivated in expression medium.

A faint band at approximately 35 kDa (the expected size of sfGFP–CTX) was observed in all candidate transformants but not in the wild-type A. oryzae control, indicating successful integration and expression of the construct. To verify the presence of sfGFP-CTX expression in our candidates, we measured the fluorescence of the secretome from A. oryzae strain ory7 and the transformants C1, C2, and C3. Fluorescence values were recorded as relative fluorescence units (RFU), which provide a direct comparison of GFP signals across the different secretomes.

Based on protein quantification, we estimated a secretion of ~500 mg/L of total protein, of which approximately 20% corresponded to sfGFP–CTX (~100 mg/L). Since CTX accounts for ~10% of the sfGFP–CTX mass, this corresponds to ~10 mg/L of CTX. sfGFP–CTX mass, this corresponds to ~10 mg/L of CTX.

Although we did not purify the protein from the A. oryzae secretome, our estimates indicated a production level approximately five times higher than that obtained in E. coli. An additional advantage of the fungal system is that the protein is secreted directly into the culture medium, eliminating the need for cell lysis during downstream processing.

With these results, we demonstrated that coupling CTX to sfGFP enables its production in two distinct cell factories—bacteria and filamentous fungi—as well as its successful recovery through a purification strategy.

To explore the possibility of producing CTX in yeast, we designed a strategy that combines yeast surface display with a sequence-specific chemical cleavage system (SNAC-tag) to enable cost-effective production. In this system, CTX is genetically fused to the yeast cell wall protein Aga1p, which is naturally anchored to the yeast surface via a GPI-anchor. To allow peptide release, we inserted a SNAC-tag between the linker–V5 epitope–Aga1p fusion and CTX. The SNAC-tag enables highly specific cleavage in the presence of Ni²⁺ ions, releasing CTX with a precise N-terminal sequence. This strategy eliminates the need for proteolytic enzymes such as TEV protease, which are often costly and introduce additional steps in downstream processing.

We successfully expressed the CTX–Aga1p fusion in yeast and confirmed it by fluorescence assay using an antibody against the V5 epitope.

Following confirmation of expression, we cultivated the CTX–Aga1p yeast strain alongside a control (lacking expression) in YNB medium for 48 hours to generate biomass. This biomass (OD600 ~ 2.o) was subsequently incubated with a nickel cleavage buffer for 18 hours, enabling the release of CTX. The released peptide was quantified using a NanoDrop spectrophotometer.

The fusion and cleavage strategy can successfully release CTX from yeast biomass, reaching measurable yields. Although the peptide represented ~1.5% of the treated biomass, about 12 mg/L, this result confirmed that yeast can act as a feasible production host. However, we also observed that the nickel buffer may interfere with absorbance readings in the Nanodrop, since some signal was detected even in the negative control without CTX expression. This indicates that our current quantification could be overestimated. To address this, in future cycles we plan to apply HPLC-based methods for accurate characterization and quantification of the produced peptide.

Aspergillus oryzae is well known for its exceptional ability to secrete large amounts of proteins, particularly amylases, whose main role is the degradation of polymers such as starch and maltose [5]. Among the secreted proteins, amylases are consistently among the most abundant. The A. oryzae strain used in contains three copies of amylase genes (AmyA, AmyB, and AmyC) in its genome, which ensures their strong expression. Previous studies have shown that the heterologous expression of proteins in filamentous fungi is often very low, usually in the milligram range, whereas the expression of native homologous proteins such as amylases can reach gram levels [6]. To take advantage of this natural secretion capacity, our team designed a strategy to couple CTX to the three amylases of A. oryzae (AmyABC). By fusing CTX to these highly expressed proteins, we aimed to leverage the fungus’s powerful secretion system and thereby achieve high-yield production of the peptide.

The amylases are present in three copies in the genome of A. oryzae. These AmyABC genes are homologous, sharing the same promoter sequence, while their terminators differ: AmyB and AmyC share the same terminator, whereas AmyA has its own unique terminator. Our strategy was to fuse K7–CTX at the C-terminus of each amylase, inserting it immediately before the stop codon. This design would allow the fungus to continue producing its natural amylases, while simultaneously secreting CTX coupled to them. To achieve this, we designed a CRISPR/Cas9 editing strategy with two guide RNAs (gRNAs) to target all three amylase loci simultaneously. As donor DNA, we used synthetic fragments (obtained through IDT sponsorship) containing the fusion sequence, enabling homology-directed repair to insert the K7–CTX module as illustrated in the scheme below.

The donor DNA was received from IDT as a linear gene fragment. To facilitate its use, it was designed to be cloned into the multiple cloning site (MCS) of pRS426 upon arrival. The donor fragment contained 750 bp of upstream and downstream homology arms to enable efficient homologous recombination in A. oryzae. In addition, to prevent potential re-cleavage by one of the guide RNAs, a codon modification (W369F) was introduced into the sequence.

From each transformation, six transformants were isolated and screened by PCR to confirm K7-CTX insertion at the target locus.. The positive candidates should have 530 pb, while the negative 370 pb. As we can see below, all candidates were negative, indicating the absence of cassette integration.

We performed spore PCR to confirm the integration of CTX into the three amylase genes (amyA, amyB, and amyC). However, all candidates tested negative, indicating that the insertion was not successful. Due to time limitations, we were not able to complete the full construction of the amyABC–CTX system during this cycle. Although no positive transformants were obtained in this first attempt, the strategy remains promising. We raised some hypotheses to explain the negative results, such as the possibility that the gRNAs were not efficiently cleaving the desired sites or that transformation efficiency needs to be optimized. Despite these challenges, our team plans to continue developing and refining this approach in future cycles. Coupling CTX to highly secreted amylases could provide a powerful strategy to increase peptide yields and enhance the overall feasibility of the production system and open the doors to solid fermentation production.

Every year, the Brazilian orange juice industry generates millions of tons of residues during juice extraction. These by-products are traditionally destined for a variety of secondary uses, but their potential remains largely underexplored. Our team saw in this challenge an opportunity: to turn citrus waste into a resource for sustainable peptide production. By harnessing the natural ability of Aspergillus oryzae to grow on complex substrates and secrete large amounts of amylase enzymes [7], we aimed to establish a low-cost and eco-friendly system for CTX production through our AmyABC–CTX coupling strategy. To achieve this, we explored the use of solid-state fermentation (SSF) with orange peel residues as the growth substrate.

We inoculated 1 mL of 10^6 spores/mL of Aspergillus oryzae in 30 mL of the ammonium sulfate (5 g/L) and 2.4 g of orange peel substrate (1.68–0.35 mm particles) for 3 and 5 days at 30°C.

The secretome was filtered to remove biomass and protein quantification was carried out by the Lowry method. The day 0 control (no fungus added) showed a relatively high protein signal, which we attribute to compounds released from the substrate.

The SDS-PAGE confirmed that A. oryzae grew on orange peel biomass and secreted ~1100 mg/L of protein, with ~60% identified as amylases (~660 mg/L). Given the CTX–amylase fusion, this corresponds to ~40 mg/L of CTX, or ~2.0 mg peptide from 2.4 g of residue. Extrapolated, this equals ~830 mg of CTX per kilogram of orange peel residue under solid-state fermentation, using the amyABC-CTX strategy

Although the full construction of the CTX–amylase fusion (Part 6) was not completed, this approach demonstrated the highest potential titer among all strategies tested, as detailed on the Entrepreneurship page. Importantly, it also utilizes a low-cost by-product of the orange juice industry, reinforcing our commitment to circular economy and sustainability. These results highlight the strong potential of coupling CTX to highly secreted amylases as a scalable and cost-effective production strategy, providing clear guidance for the next steps in optimizing this approach.

Given the major challenges of performing plant-based assays (such as long growth cycles, high variability, and biosafety requirements) and the fact that CLas cannot be cultured under laboratory conditions, we developed a computational approach to guide and validate our therapeutic strategy. This "virtual laboratory" simulates the cell-by-cell interactions between the CLas pathogen and host defenses. Our modeling efforts followed three key stages: validating the model against real world data, using it to simulate the efficacy of our CTX peptide, and performing a sensitivity analysis to understand the system's core dynamics.

First, we validated our model to ensure its biological relevance. Our simulations reproduced key features of the disease observed in scientific literature, including the spatial displacement of the plant's callose defense as seen in microscopy studies by Bernardini et al., 2022, the long term temporal progression of the infection matching data from Coletta-Filho et al., 2010, and Vasconcelos et al., 2020, and the emergence of sparse, irregular infection patterns consistent with the known non-uniform distribution of CLas in trees. This multi level validation confirmed that our model is a reliable representation of the disease.

Figure 1: The model's simulated infection curve (solid line) shows strong agreement with experimental data points from field studies, validating the model's accuracy.

With the model validated, we used it to compare the efficacy of CTX peptide against a conventional bacteriostatic treatment (Tetracycline). The simulations revealed a stark contrast in outcomes:

1. Tetracycline (Suppressive Action): Offers only temporary control. The simulation shows that while the treatment initially reduces the bacterial load, the disease rebounds to an even more severe state once the drug's effect wanes.
2. CTX Peptide (Curative Action): Provided a permanent solution. Its bactericidal action led to the complete eradication of the pathogen from the system, even at a simulated dose ten times lower than tetracycline.

To visualize this difference, we generated animations of the final tissue state, providing a clear visual proof of a curative outcome.

Figure 3: Final spatial state of the simulated tissue. (A) Untreated, (B) Tetracycline treatment, showing residual infection, (C) CTX treatment, showing a completely cleared grid.

Finally, a sensitivity analysis identified the most critical factors governing the disease's dynamics. We discovered that the final severity of the infection is determined primarily by an "race" between the bacterial growth rate ($r$) and the strength of the plant's defense ($\alpha_C$). Furthermore, the analysis showed that the speed of the outbreak is mainly controlled by the propagation rate ($\beta$), while the final spatial pattern of the infection is dictated by the defense signaling radius ($S_r$). This understanding provides insights into which factors could be targeted for future disease control strategies.

Early and accurate diagnosis is essential for controlling citrus greening. However, traditional molecular approaches such as qPCR, while precise, are time-consuming, costly, and require laboratory infrastructure. To overcome these limitations, we designed a machine learning–based diagnostic Hardware capable of recognizing infection patterns directly from the volatile organic compound (VOC) signatures captured by our device.

To achieve this, we collected and processed measurements from healthy and infected citrus leaves. Each measurement consisted of a multidimensional dataset combining signals from polymer-based interdigitated electrodes (IDEs), MOS gas sensors, and environmental sensors (BME680, SHT31). The goal was to translate these complex chemical and electrical variations into a clear and reliable diagnostic output.

After testing multiple combinations of models and configurations, the most promising approach was found to be Linear Discriminant Analysis (LDA). LDA serves both as a classification method and a dimensionality reduction technique. It projects the data into a lower-dimensional space, maximizing the separation between classes while minimizing the spread within each class. The sensor data showed significant variance, making it suitable for building a discriminative classification model.

The training results are encouraging: the model can successfully distinguish between healthy and infected leaves. However, the system remains sensitive to environmental variations. This is mainly due to the limited size of the dataset and the lack of generalization across different environmental conditions. Measurements taken on different days, for instance, can affect the model’s performance.

Importantly, this sensitivity is not necessarily a result of overfitting. The models show strong performance on the test sets, confirming that the dataset split into training, validation, and test subsets provides a reliable evaluation. The confusion matrix offers a clear visual representation of the model’s ability to classify the leaves correctly and highlights areas where misclassifications occur.

A confusion matrix helps you evaluate a classification model's performance. The rows typically represent the actual (true) labels, while the columns show the predicted labels. The main diagonal (from top-left to bottom-right) contains the correctly classified instances, meaning the predicted label is the same as the actual label. All other cells off the diagonal represent errors, showing where the model confused one class for another (misclassifications).

This table is a classification report that summarizes the excellent performance of a predictive model for the "negative" and "positive" classes. The Precision scores (0.97 and 0.98) indicate that when the model predicts a class, it is highly accurate, leading to a very low false positive rate. The Recall scores (0.98 and 0.97) show that the model is extremely effective at identifying the vast majority of actual instances for each class, meaning it has a low false negative rate. Consequently, the F1-Score, which is the harmonic mean of precision and recall, is an outstanding 0.98 for both classes, confirming a robust and well-balanced performance. The Support column simply indicates that this evaluation was based on 121 actual instances of the 'negative' class and 131 of the 'positive' class.

The high accuracy achieved by the LDA model confirms that the Sniffer prototype is capable of detecting clear chemical distinctions between healthy and infected citrus leaves. These results validate the functionality of the hardware, the effectiveness of the sensor array, and the consistency of the signal acquisition process. However, the system’s sensitivity to environmental variations emphasizes the need for broader datasets and improved signal robustness under field conditions.

Building on this foundation, future versions of the Sniffer will focus on enhancing stability, portability, and user accessibility. Planned improvements include shielding the device within a metal enclosure to reduce noise, integrating a rechargeable battery for autonomous operation in the field, and miniaturizing the electronics onto a compact PCB. Expanding the dataset with samples collected across different orchards, seasons, and tree ages will strengthen model generalization. Additionally, integrating optimized airflow control and a user-friendly software interface will ensure consistent VOC delivery and provide straightforward diagnostic outputs, such as “Healthy” or “Infected,” directly to farmers.

In conclusion, the Sniffer is more than a proof of concept — it is a flexible platform ready for expansion into real-world agricultural applications. With future iterations focused on portability, robustness, and usability, the Sniffer holds strong promise as a universal VOC-based detection system for plant health monitoring and beyond.