The citrus industry, a cornerstone of the agricultural economy supporting countless farmers and consumers, is facing severe challenges from the brown citrus aphid (Toxoptera citricida). These aphids adeptly conceal themselves on the undersides of citrus leaves, stealthily sucking plant sap, making early detection and effective control particularly difficult².

Moreover, T. citricida reproduces via parthenogenesis, where females can produce dozens of offspring daily without mating, leading to exponential population growth. Even more concerning is their adaptive survival strategy: when food resources become scarce, aphids develop into winged morphs that can fly away from orchards to establish new colonies, initiating fresh infection cycles. They also lay dormant eggs that overwinter and hatch in spring, perpetuating the infestation year after year³.

These behavioral traits render conventional control methods insufficient. Even if a small number of individuals survive treatment, they can rapidly rebuild the population, making effective long-term management exceptionally challenging.

Fig 1. Cunning aphids often hide on the underside of citrus leaves, making them difficult to detect.

Facing the severe threat posed by aphids, the SZU-China team utilized knowledge from synthetic biology to construct an intelligent citrus aphid management project named Aphigo. This project decomposes the complex task of pest control into three independent, yet collaborative, BioBrick modules: (1) an early detection system for real-time monitoring of aphid density; (2) a trunk-injected RNAi system, where efficient RNAi molecules are protected and delivered via functionalized VLPs, establishing long-lasting systemic protection within the plant; and (3) an engineered Metarhizium control system, a targeted therapeutic agent designed for the rapid emergency eradication of outbreak populations.

Traditional pest control strategies often rely on single-method approaches⁴, which struggle to adapt to dynamic field conditions. In contrast, Aphigo's two core intervention modules—the trunk-injected RNAi system and the engineered Metarhizium control system⁵—complement each other to form an integrated defense framework combining prevention and treatment.

In this system, the RNAi molecules encapsulated within MS2 particles are directly introduced into the plant's vascular system via trunk injection. They then circulate with the sap, achieving stable presence throughout the plant and enabling long-lasting, systemic, and proactive prevention. This endogenous control strategy can cover the hidden parts where aphids conceal themselves during their latent period. The engineered Metarhizium control system, on the other hand, serves as a targeted therapeutic agent, specifically designed to address aphid population outbreaks. Triggered by alerts from the detection system, it utilizes the dual lethal capabilities of Metarhizium to rapidly eradicate outbreak populations, achieving highly efficient emergency intervention.

In summary, Aphigo integrates biosensing (Module 1), sustained endogenous prevention (Module 2), and rapid targeted treatment (Module 3) to establish a digital pest management model characterized by real-time monitoring and categorized intervention. This approach transforms conventional reactive strategies and offers a precise, sustainable new paradigm for the citrus industry.

Below, we will delve into the core design of the project, providing a detailed explanation of each module’s working principles and technological innovations.

Fig 2. The three major modules of APHiGO, a comprehensive defense system integrating prevention and treatment.

The early detection system is built upon a genetically modified strain of Bacillus subtilis. The selection of B. subtilis as the chassis organism offers multiple advantages. First, it is a extensively studied model organism with a well-defined genetic background and high amenability to genetic engineering⁷. Second, B. subtilis is approved by the U.S. Environmental Protection Agency (EPA) as a biocontrol agent and is already used as an active ingredient in multiple biopesticides⁸. Most importantly, B. subtilis utilizes the sucrose phosphotransferase system to hydrolyze sucrose into glucose and fructose-6-phosphate. These metabolic products serve as key substrates for the subsequent generation of the detection signal—methyl salicylate (MeSA). As a common plant commensal bacterium, B. subtilis can colonize leaf surfaces, providing a natural delivery advantage for its use as a biosensor directly on citrus trees.

Fig 3. Three key advantages of selecting Bacillus subtilis as the chassis for the detection system

The core principle of this module leverages the biological characteristics of aphid feeding on citrus trees. Aphids typically gather on the undersides of leaves, extracting sap from the plant's vascular tissues. During feeding, they excrete sucrose-rich honeydew, which drips onto the lower leaf surfaces. This sucrose serves as a direct signal input for the biosensor. Our engineered bacteria respond to this signal through an integrated sucrose-inducible promoter, converting it into a volatile chemical—methyl salicylate (MeSA)—detectable by hardware sensors. This enables real-time, non-invasive monitoring of aphid density.

The workflow of the module follows a refined metabolic pathway:

• 1.Sucrose Utilization: Bacillus subtilis metabolizes sucrose and breaks it down into monosaccharides such as glucose. These sugars subsequently enter the native shikimate pathway, ultimately yielding the key intermediate, chorismate⁹.
• 2. Conversion of Chorismate to Salicylic Acid: To transform non-toxic chorismate into a signaling molecule, we introduced the pchBA gene cluster from Pseudomonas aeruginosa. The enzymes encoded—isochorismate synthase and isochorismate pyruvate lyase—efficiently catalyze the conversion of chorismate to salicylic acid (SA).
  (Reaction formula or schematic diagram)
• 3. Conversion of SA to Volatile MeSA: Finally, to release the signal from the microbial cells into the environment, we incorporated the BSMT gene from Petunia nyctaginiflora. This gene encodes a salicylic acid methyltransferase that catalyzes the methylation of SA, converting it into volatile methyl salicylate (MeSA)¹⁰.
  (Reaction formula or schematic diagram)

Fig 4. Metabolic process of engineered Bacillus subtilis producing MeSA, where the green dashed box indicates the original metabolic pathway of the chassis cell

Once synthesized, MeSA—as a volatile compound—rapidly diffuses into the air and is detected by sensors¹¹. This process achieves signal amplification and externalization, transforming a localized biological event on the leaf surface into a gaseous signal that can be monitored remotely.

The selection of MeSA as the final signaling molecule is based on both ecological and technical considerations. As a natural plant volatile, MeSA is released in large quantities when plants suffer from pest infestation, serving as a "distress signal" to attract natural enemies of the pests. This allows our system not only to monitor aphid presence but also to leverage natural ecosystems for synergistic control. More importantly, as a gaseous molecule, MeSA can be captured and quantified by highly sensitive gas sensors.

In field conditions, aphids often cluster unevenly on young shoots, but the honeydew they secrete can be dispersed by wind and deposited on other leaves. By deploying gas sensors to monitor MeSA concentrations, our system overcomes the uncertainties associated with direct aphid detection. This design enables timely alerts for farmers, shifting pest management from a reactive approach to a proactive strategy.

When the detection system triggers an alert, a precise, efficient, and durable control method is required. Conventional foliar spraying relies on direct contact of pesticides with plant surfaces and absorption through the leaves. However, this approach faces multiple challenges in citrus trees. The waxy layer on citrus leaves forms a natural barrier that impedes the penetration of external substances. Additionally, plants exhibit low absorption efficiency for foreign compounds, making it difficult for agents to traverse the cuticle and cell wall. Even when pesticides reach the leaf surface, environmental factors such as wind and rain, as well as plant surface structures, lead to uneven distribution and incomplete coverage. Crucially, aphids often inhabit the undersides of leaves—particularly concealed areas such as veins and young shoots—making it difficult for foliar sprays to reach them effectively. This results in low pesticide utilization and significantly reduced control efficacy.

To address these limitations, we developed an RNAi-based trunk injection system. By directly introducing functionalized RNAi agents into the plant’s vascular system, this method bypasses leaf barriers and establishes systemic protection throughout the plant.

To achieve both highly efficient and stable systemic prevention within the trunk injection system, we adopted a strategy of "parallel multi-target design and phased iterative optimization" by concurrently developing multiple RNAi candidates.

Fig 5. Process diagram of Multi-generational Design of RNAi Molecules

First-Generation Design — Single-Target dsRNA

In the initial phase of RNAi design, we conducted extensive literature research to identify key survival genes in aphids, including chitin synthase (CHS), cytochrome P450, and cuticle protein CP19, which play critical roles in insect development, resistance, and xenobiotic metabolism.

Chitin Synthase CHS (KR611528): Chitin is a major component of the insect exoskeleton and peritrophic matrix, accounting for approximately 40% of the insect's dry weight. During the molting process, the chitin-rich structures in the exoskeleton and peritrophic membrane need to be reconstructed to accommodate the insect's increasing body size. CHS plays a critical role in this process. Aphids rely on molting for development and typically undergo four molts to become adults. Silencing the CHS gene leads to molting disorders, severely affecting insect growth and development, and ultimately resulting in death.

Cytochrome P450 (KX224318.1): CYP450 enzymes are key to metabolizing foreign compounds in insects. They aid in detoxifying pesticides, plant toxins, and other environmental stressors. Silencing CYP450 may compromise detoxification capacity, insecticide resistance, and ecological adaptability.

Cuticle Protein CP19 (MF374532.1): CP19 belongs to the insect cuticular protein family, helping reduce water loss and supporting stress adaptation. It also plays a structural role in exoskeleton formation and maintenance. Silencing this gene effectively inhibits aphid growth and survival.

In addition to the above genes, we also analyzed some key growth-related genes, such as the protein synthesis and cell function genes EF-1a and 18S ribosomal RNA (18S rRNA), which are involved in protein synthesis and cell division. The RyR endoplasmic reticulum calcium channel receptor gene plays a central role in insect molting, muscle function, and body development; the glyceraldehyde-3-phosphate dehydrogenase gene (GAPDH) helps insects cope with environmental toxins and provides energy support; acetylcholinesterase (AChE) and the insulin receptor (AcInR),regulate nerve conduction and motor function.

To ensure RNAi fragment specificity, we performed BLAST homology analysis in the NCBI database to evaluate sequence similarity with non-target organisms such as humans, citrus, and honeybees. A homology threshold of 20% was set, and only genes below this threshold were selected as candidates. Based on this criterion, CHS, CYP450, and CP19 were chosen as RNAi targets to minimize off-target risks.

We then used MEGA software for multiple sequence alignment of homologous genes across related species, identifying conserved regions in each target gene and recording their start and end positions to guide the design of long dsRNA fragments¹².

Furthermore, leveraging multiple RNAi prediction databases, we screened siRNA target regions with high silencing efficiency¹³. Integrating conservation and interference efficiency analyses, we designed three specific dsRNA molecules: dsCHS, dsCYP450, and dsCP. All final dsRNA sequences were re-validated via BLAST to ensure high specificity to Toxoptera citricida ¹with controlled off-target risks.

Second-Generation Design — Fused Multi-Target dsRNA

In our first-generation design, we successfully constructed dsRNA molecules targeting individual key genes (CHS, CYP450, CP19). To achieve more effective pest control, through in-depth discussions with Dr. Ruoyu Chen, we decided to adopt a more advanced strategy in the second-generation design: constructing fused multi-target dsRNA. By concatenating multiple short dsRNA fragments—each targeting different genes or efficiently targeting specific regions of a target gene—into a long, fused dsRNA molecule, we can simultaneously silence multiple functionally related genes. This approach enables coordinated silencing of multiple functionally related genes, significantly enhancing RNAi-mediated lethality while reducing the risk of pest resistance.

From the validated first-generation dsRNA fragments, we selected short sequences (60–100 bp) with high targeting efficiency, strong specificity, and evolutionary conservation. These optimized fragments were directly linked in tandem to form fused constructs. During the assembly process, we utilized RNAfold to analyze the secondary structure of the dsRNA molecules¹⁴, paying particular attention to avoiding complex structures or repetitive sequences at fusion sites. This ensures that the long fused dsRNA can be efficiently recognized and uniformly cleaved by the insect Dicer enzyme, generating functional siRNA molecules.

We ultimately constructed two types of fused molecules:A dual-target fused dsRNA (dsF2), targeting both CYP450 and CHS genes; A triple-target fused dsRNA (dsF3), simultaneously targeting CHS, CYP450, and CP19.

To evaluate biosafety, we employed the Off-target function of the dsRNAEngineer platform to compare the designed fused dsRNA sequences against an integrated transcriptome database of non-target organisms—including pollinators, aquatic species, humans, and crops. This step ensures minimal potential interference with beneficial insects such as ladybugs and other non-target organisms.

Through this iterative advancement, we have successfully upgraded our RNAi design from a "single-target strike" to a "multi-pathway coordination" strategy. We anticipate that the second-generation fused dsRNA molecules will more effectively disrupt aphid growth and development, providing a more robust tool for our integrated pest management system¹⁵.

Third-Generation Design — Multi-Tandem Short Hairpin RNA(tri-shRNA)

While second-generation fused dsRNA enables coordinated multi-gene interference, its long dsRNA sequences are susceptible to nuclease degradation, intracellular immune recognition, and elevated off-target risks. Building on this foundation, we developed a third-generation RNAi design—tandem short hairpin RNA (tri-shRNA)—to enhance stability and reduce off-target effects through more precise sequence architecture.

shRNA is composed of a short hairpin structure, whose stem sequence can be processed into siRNA with interference activity. After entering the cytoplasm, shRNA is recognized and cleaved by the Dicer enzyme, generating double-stranded siRNA approximately 21–23 nt in length. Subsequently, the siRNA is loaded into the RNA-induced silencing complex (RISC). After unwinding, a single guide strand is retained. This guide strand locates the target mRNA through base complementarity, and the Argonaute protein within RISC ultimately executes the cleavage, leading to the degradation of the target mRNA and thereby achieving silencing of the target gene. The hairpin structure of shRNA effectively enhances its stability and significantly reduces the risk of degradation by RNases.

More importantly, shRNA can be designed to incorporate multiple targeting sequences in tandem, enabling simultaneous targeting of multiple genes and enhancing interference efficiency. In this generation's design, three distinct shRNA units targeting CHS, CP19, and CYP450 respectively are tandemly arranged within a single transcript. Each unit is constructed based on previously screened high-efficiency siRNA fragments and incorporates the universal hairpin loop sequence TTCAAGAGA. The tri-shRNA is processed by Dicer enzyme within the cell, generating three distinct siRNAs that target CHS, CP19, and CYP450 mRNAs respectively. These siRNAs are subsequently loaded into the RNA-induced silencing complex (RISC), leading to simultaneous silencing of the three target genes and enhancing the synergistic lethal effect.

During the design process, we used RNAfold to simulate the secondary structure of the entire sequence to ensure that each shRNA unit maintains structural independence, avoiding mutual interference and ensuring effective recognition and cleavage by Dicer. Additionally, through the off-target analysis of the dsRNAEngineer platform, we confirmed that this tandem shRNA sequence poses no potential silencing risk to non-target organisms. This multi-tandem short hairpin RNA structure maintains high stability while achieving coordinated interference of multiple genes, and is expected to significantly enhance the control efficiency against aphids.

Fourth-Generation Design — Tandem Artificial miRNA(bi-amiRNA)

The fixed sequence of RNA may not be efficiently processed by the endogenous RNAi-related proteins of certain organisms. To address this issue, we introduced a multi-tandem artificial miRNA (bi-amiRNA) structure in the fourth-generation RNAi system, further utilizing the endogenous microRNA (miRNA) processing mechanism of eukaryotes to achieve efficient silencing of target genes.

During the design process, we integrated the previously screened CYP450 siRNA sequence into the api-mir-71 scaffold, and the CHS siRNA sequence into the api-mir-3017a scaffold¹⁶. We used a universal linker (AGGCAT) to connect these two miRNA units, forming a double-tandem structure. We then used the RNAfold tool to simulate the secondary structure of the overall sequence, ensuring that each amiRNA unit maintains structural independence and avoids mutual interference. Each amiRNA unit is separately cleaved by Dicer enzyme in the cell, generating their respective siRNAs that target and silence CYP450 and CHS genes¹⁷.

By introducing the amiRNA structure, we have not only solved the processing problems that may be caused by fixed sequences in traditional shRNA design, but also fully utilized the advantages of the miRNA processing mechanism, enhancing the effect of multi-gene targeted silencing. We expect that the multi-tandem amiRNA design can significantly improve control efficiency against aphids and provide more efficient lethal effects.

Design of Protective Delivery System

To ensure that the designed RNAi molecules remain stable in the complex plant-insect interaction environment, we selected MS2 virus-like particles (VLPs) as the protective delivery vector. MS2 VLPs can achieve self-assembly into structurally stable complexes through the specific recognition between the capsid protein and the target RNA, thereby maintaining the integrity and function of the RNAi molecules after application¹⁸.

MS2 is naturally a positive single-stranded RNA bacteriovirus that specifically infects male E. coli carrying the F-pilus. Its viral particle consists of a genome RNA of approximately 3.6 kb and an icosahedral protein capsid composed of 90 capsid protein dimers. In the natural MS2 genome, there is a specific stem-loop structure—the "pac site"—which serves as the initiation signal for capsid protein recognition and assembly. After the capsid protein dimer binds to the pac site with high affinity, it triggers local RNA conformational changes, exposing more protein-binding interfaces. This drives the subsequent coordinated assembly of capsid proteins through protein-protein interactions, gradually encapsulating the RNA strand and ultimately forming hollow spherical particles with a diameter of approximately 27 nm¹⁹. This process completely encapsulates the RNA, effectively shielding it from degradation by environmental nucleases²⁰.

Based on the natural assembly mechanism of the MS2 virus, we developed a recombinant MS2 VLP delivery system. This system retains the self-assembly and pac site recognition functions of the capsid protein but completely removes the viral genome to ensure no replication or infection capability. We introduced the pac site into the C-terminus of the target RNAi molecule, enabling its specific recognition by the capsid protein²¹. Using an in vivo assembly strategy, MS2 capsid protein dimers and RNAi molecules containing the pac site were co-expressed in E. coli BL21(DE3), allowing direct intracellular self-assembly into complete VLP particles. Finally, structurally uniform VLP-RNAi complexes with fully encapsulated RNA were obtained through purification.

Targeting Enhancement Strategies

The VLP-RNAi complexes we constructed can effectively prevent degradation by environmental nucleases. However, previous studies have shown that nucleic acid molecules encapsulated by MS2 have poor endosomal escape ability after cellular uptake, and unmodified VLP-RNAi complexes are randomly distributed in aphids, making it difficult for most RNAi molecules to be effectively absorbed and function. Therefore, improving the uptake and delivery efficiency of the complexes in the aphid gut is a core challenge we face²² To enhance the uptake and delivery efficiency of the complexes in the aphid gut, we systematically fused two functional peptides to the surface of the MS2 capsid protein:

TAT Penetrating Peptide: To overcome the bottleneck of endosomal entrapment and lysosomal degradation, we displayed the HIV-TAT cell-penetrating peptide on the particle surface. TAT, as a classical cell-penetrating peptide, is rich in basic amino acids. Residues such as histidine protonate in the acidic endosomal environment, inducing a massive influx of protons, which elevates osmotic pressure and ultimately leads to endosomal swelling and rupture. Additionally, its arginine-rich region can electrostatically interact with endosomal membrane phospholipids, directly disrupting membrane stability. These two mechanisms can effectively deliver the cargo into the cytoplasm.

GBP3.1 Aphid Gut-Targeting Peptide²³: This peptide can specifically recognize surface proteins of aphid gut cells, achieving tissue-specific targeting. We fused it between the monomers of the capsid protein dimer, maintaining the stability of the dimer structure while endowing the complex with the ability to directionally accumulate in the aphid gut.

Through the synergistic action of these two functional peptides, we can significantly improve the targeted delivery efficiency of VLP-RNAi complexes in aphids, ensuring that RNAi molecules efficiently enter aphid gut cells and exert lethal effects.

MS2 VLPs are essentially non-replicative, non-infectious hollow protein shells. Their design completely removes the original viral genome, retaining only the capsid protein and its self-assembly function²⁴. Through this modification, MS2 VLPs can effectively protect RNAi molecules from environmental nuclease degradation, ensuring their stability and function. Additionally, to address the targeting of VLP-RNAi complexes in aphids, we further enhanced the directional delivery capability of VLP-RNAi complexes in aphids by fusing TAT penetrating peptide and GBP3.1 aphid gut-targeting peptide. The overall design ensures the stability, targeting, and efficient delivery of the delivery system.

This module employs trunk injection as the core delivery strategy and has developed four generations of RNAi molecules through multiple rounds of iteration: from single-target dsRNA, to fused multi-target dsRNA, then to tandem shRNA, and finally upgraded to biomimetic artificial miRNA (amiRNA). These molecules are further encapsulated and functionalized using MS2 virus-like particles, enabling stable transport in complex environments and efficient targeting to the aphid gut.

Technology is primarily applied during the early growth season of citrus trees. The agent directly enters the plant's internal vascular tissues, effectively reaching concealed areas where aphids hide during the latent period—such as the undersides of leaves and young shoots—overcoming the challenge of uneven coverage associated with traditional spraying. The system also establishes long-lasting protective layers within the plant, reducing farmers' exposure to pesticides while avoiding the significant water waste caused by spraying, achieving safer and more water-efficient green pest control.

After establishing the trunk injection-based prevention system (Module 2), we designed Module 3—a targeted therapeutic agent based on engineered Metarhizium—to address potential localized outbreaks of aphid populations. This module not only infects pests externally like natural Metarhizium, but also functions as a living bio-factory, continuously producing and delivering our precisely designed triple-fusion RNAi molecules (dsF3) within the aphids. This achieves a "dual strike" against the pests, enabling rapid eradication of outbreak aphid populations in the environment.

We selected Metarhizium anisopliae as the chassis organism. As a natural entomopathogenic fungus, its spores can germinate upon contact with the cuticle of the brown citrus aphid, differentiate to form specialized infection structures—appressoria—and then penetrate the insect cuticle, invade the hemolymph, and proliferate extensively. During this process, Metarhizium secretes various metabolites that destroy host tissue structures, ultimately leading to aphid death²⁵.

Simultaneously, Metarhizium has been registered as an environmentally friendly biopesticide in many countries and is widely used in agricultural production of various crops such as citrus and rice. It is safe for non-target organisms including mammals, birds, bees, and ladybugs. This strain has a clear genetic background and a mature genetic manipulation system, providing the feasibility for us to subsequently engineer it for stable expression of targeted RNAi molecules²⁶.

To modify Metarhizium for expressing targeted RNAi molecules, we employed Agrobacterium-mediated genetic transformation. This method is based on the ability of Agrobacterium to transfer the T-DNA region from its Ti plasmid into host cells, thereby achieving integration of the target gene. During the transformation process, Agrobacterium delivers T-DNA into host cells through its secretion system and integrates the target gene into the Metarhizium genome, ensuring stable expression of the target gene in the fungus.

We first inserted the target gene—the triple-fusion RNAi molecule—into the T-DNA region of the Ti plasmid, and then introduced it into Metarhizium via Agrobacterium-mediated transformation. After transformation, the target gene integrates into the Metarhizium genome, enabling stable and continuous expression and effective function²⁷.

Considering the potential risks of genetically modified microorganisms (GMOs) in application, we referred to previous designs from the NYMU_Taipei team and the Hangzhou-SDG team, and combined them with our requirements to design a safety switch system to ensure the safety of the engineered fungus. The core components of this system include:

Promoter (PMcl1)²⁸: A promoter that is specifically activated within the insect hemocoel, which only becomes active after Metarhizium successfully invades the aphid's hemocoel.

Toxin (SuperNova photosensitive lethal protein): SuperNova is a genetically encoded photosensitive protein. Under natural light exposure, SuperNova transitions from its ground state to an excited state and effectively transfers energy to surrounding oxygen molecules, generating reactive oxygen species (ROS). ROS can damage intracellular lipids, proteins, and nucleic acids, ultimately leading to loss of cell function and fungal death.

Targeting Enhancement (NLS): A nuclear localization sequence (SV40) fused to one end of the SuperNova protein ensures the toxin is directed into the fungal nucleus, thereby enhancing its lethal effect.

Upstream Insulator (TtrpC)²⁹: Blocks potential unintended activation by spurious promoters, preventing non-target expression.

Unlike previous protoplast-mediated transformation methods, this system uses Agrobacterium-mediated transformation, whose random integration might lead to accidental activation by strong endogenous promoters, causing leaky expression of the toxin protein in non-target environments. To mitigate this risk, we introduced the TtrpC terminator as a transcriptional insulator upstream of the safety switch. Ultimately, the entire safety system is constructed as a complete genetic circuit:

TtrpC Terminator——PMcl1——SuperNova−NLS——TtrpC Terminator.

Through this design, we achieve control from application to elimination. When the engineered Metarhizium spores are applied to the environment, the PMcl1 promoter remains silent in non-host conditions, suppressing SuperNova expression and ensuring the spores remain dormant and stable. Only when the Metarhizium spores successfully invade the aphid hemocoel is the PMcl1 promoter activated, initiating the expression of the SuperNova-NLS fusion protein. Subsequently, the nuclear localization sequence (NLS) directs the SuperNova protein into the fungal nucleus.

It is important to note that the aphid hemocoel is a dark environment lacking activating light signals, which keeps SuperNova inert within the nucleus, allowing the fungus to proliferate normally and complete its pest-killing task. After Metarhizium kills the aphid and grows hyphae out of the carcass to release spores, SuperNova becomes activated by natural light, thereby killing the escaped fungus.

This module centers on engineered Metarhizium to establish a targeted therapeutic system that integrates dual mechanisms of fungal infection and RNAi-mediated gene silencing. It features the following key characteristics:

Dual-Mode Action for Enhanced Control Efficacy

While wild-type Metarhizium is already a well-established biocontrol agent capable of infecting aphids through cuticle penetration, our engineered strain introduces a powerful second mode of attack. It continuously produces target-specific dsRNA inside the host. This internal RNAi production acts synergistically with the fungal infection process, creating a combined assault that significantly improves lethality against the brown citrus aphid.

A Self-Replicating and Disseminating RNAi Factory

A key innovation of this design lies in the engineered fungus functioning as a living, self-propagating bio-factory. Once applied, it can persist in the environment, replicate, and spread within the aphid population. This enables continuous, in-situ production and delivery of dsRNA molecules directly inside the target pests, establishing a persistent and self-amplifying control system that reduces the need for repeated applications.

When deployed in response to outbreak alerts from the detection system, this module delivers rapid emergency control. Combined with the trunk-injected RNAi preventive system, it further refines an integrated "prevention–treatment" framework for sustainable citrus aphid management. This integrated strategy not only enhances immediate control efficacy but also helps delay the development of pest resistance due to its dual-mode mechanism, aligning with our goal of developing ecologically balanced and durable pest management solutions.

The stable, economical, and large-scale supply of dsRNA is a core bottleneck for the commercial application of RNAi technology. Traditional unidirectional transcription systems suffer from low efficiency and a tendency to produce excessive non-target single-stranded RNA, severely hindering scaled-up preparation. To meet the project's continuous demand for large quantities of high-purity dsRNA and with an eye toward future industrial prospects, we designed and optimized a highly efficient and economical in vitro transcription (IVT) production system. By introducing bidirectional T7 promoters and multiple terminator arrays, we achieved efficient and high-quality synthesis of dsRNA.

Optimization Point 1: Bidirectional T7 Promoter Design for One-Step dsRNA Synthesis

We introduced highly active T7 promoters at both ends of the target sequence. This ingenious "bidirectional T7" design allows the DNA template to be transcribed simultaneously from both the sense and antisense strands by T7 RNA polymerase. The two transcription products are naturally complementary and can rapidly anneal within the reaction system to form complete dsRNA molecules. This design eliminates the cumbersome steps required in traditional methods—transcribing two single strands separately and then annealing them—greatly improving synthesis efficiency and ensuring the synthesized product is free from excessive impurities³⁰.

Optimization Point 2: Introduction of Multiple Terminator Arrays to Ensure Precise Transcription Termination

To prevent transcriptional "read-through" that produces heterogeneous byproducts of varying lengths, and to ensure the yield and purity of dsRNA, we designed a triple transcription terminator array downstream of each T7 promoter. This array sequentially comprises the T7UUCG terminator (a synthetic class I terminator), the rrnB T1 terminator (an endogenous ρ-independent terminator from E. coli), and the T7 terminator (a canonical class II terminator).

This design establishes a multi-layered, synergistic termination safeguard mechanism, whose core principles are as follows:

The multiple terminator array ensures product length homogeneity, significantly improves the yield and purity of the target dsRNA, and avoids the production of long-chain RNAs with potential off-target effects due to transcriptional read-through. It also reduces the metabolic burden on cells, thereby achieving a balance of high yield and high purity.

Optimization Point 3: Determination of the Optimal Configuration Through Systematic Screening of Time Gradients and Terminator Arrays

Building on the successful construction of the multiple terminator array, we further adopted a modular design strategy. By pre-introducing specific restriction enzyme sites between adjacent terminators, we constructed plasmids with single terminators (1T), double terminators (2T), and triple terminators (3T) to comprehensively evaluate the impact of different arrays on dsRNA yield and purity.

To assess transcription efficiency under different reaction conditions, we established time gradients of 4 hours, 6 hours, and 12 hours in small-scale tube reactions to simulate the dynamics of dsRNA synthesis under varying reaction durations. This provides critical process references for subsequent large-scale in vitro transcription production.

Based on the three core modules developed earlier, we have built the "CitrusShield" citrus protection platform, which deeply integrates biotechnology and digital agriculture to provide an intelligent solution for pest and disease management in citrus orchards, with a core focus on early pest detection and timely intervention.

The foundational core of this platform is the pre-deployment of engineered Bacillus subtilis sustained-release formulations in the citrus orchard. These genetically modified microorganisms persistently colonize leaf surfaces. When aphids feed and produce honeydew, the engineered bacteria perceive the sucrose signal and release methyl salicylate (MeSA) as a pest indicator signal. The platform utilizes an intelligent sensor network deployed throughout the orchard to monitor three key ecological indicators in real-time: environmental temperature and humidity, methyl salicylate (MeSA) concentration, and carbon dioxide concentration³².

These data are transmitted in real-time via the Internet of Things to a cloud-based analysis system , providing a reliable basis for pest population assessment.

To facilitate easy operation for farmers, we developed a WeChat mini-program that presents the aforementioned key data indicators through an intuitive and user-friendly interface This lightweight application allows farmers to view real-time and historical data for methyl salicylate (MeSA) and carbon dioxide concentrations via their smartphones. The system can also display trends on daily, weekly, and monthly scales.This lightweight application enables farmers to view real-time and historical data on methyl salicylate (MeSA) and carbon dioxide concentrations via smartphones, ensuring that farmers can promptly monitor changes in orchard environment. (For details, please refer to the CitrusShield page).

Based on dynamic changes in the monitoring data, farmers can make scientific control decisions. When methyl salicylate concentrations remain at low levels, indicating aphids are in the latent period, trunk-injected RNAi formulations are used for preventive control. When methyl salicylate concentrations rise persistently accompanied by abnormal carbon dioxide levels, indicating signs of a population outbreak, Metarhizium formulations are used for emergency eradication.

In the future, as field test data accumulates and algorithm models improve, we plan to integrate an intelligent decision-making function into the mini-program This function will then automatically generate tiered control recommendations based on real-time data analysis. In the early stages of pest occurrence, it will recommend preventive treatment plans and appropriate RNA interference (RNAi) dosages. When pest density increases, it will automatically switch to emergency control mode, guiding farmers to use fast-dissolving Metarhizium formulations.

The APHiGO project achieves a closed-loop, digitalized pest management system, transforming complex biotechnology into simple, intuitive tools. It enables every farmer to become a practitioner of precision agriculture, working together to protect our citrus groves and safeguard that sweet future.