As an important pillar of agricultural economy , the citrus industry is being quietly coerced by an evil culprit - the brown citrus aphid (Toxoptera citricida). It causes damage to citrus mainly in three ways:

• The brown citrus aphids (Toxoptera citricida) suckcitrus sap through its piercing-sucking mouthparts and plundering the citrus nutrients;
• The honeydew excreted after ingesting high-sugar substances nourishes microorganisms on the leaf surface ;
• Transmitting the Citrus Tristeza Virus (CTV).

In response to this highly threatening pest, our team has designed a comprehensive citrus aphid management system, APHiGO, to achieve the goal of protecting citrus.

It is mainly divided into the following three independent and collaborative BioBrick modules:

• Early detection system: Converts honeydew information into methyl salicylate, which is then collected for real-time monitoring of aphid density;
• Trunk injection RNAi system: Efficient RNA molecules are protected and delivered by functionalized VLPs to establish long-term effective systemic protection within the plant;
• Engineered Metarhizium anisopliae control system: A targeted therapeutic agent that enables rapid emergency eradication of outbreak populations.

In March 2025, our team went to Shanwei, the main citrus-producing area in Guangdong, to conduct research. There, we found that citrus aphids had caused significant damage to citrus. After communicating with farmers and conducting literature research, we learned that aphids are tiny and often stay on the undersides of leaves, making it impossible for farmers to detect their presence in time and apply pesticides. In response, our team designed sucrose-responsive engineered Bacillus subtilis based on the honeydew excreted by brown citrus aphids, and measured the limit of methyl salicylate through hardware. After conversion, farmers can use an app to monitor real-time changes in aphid population density.

Population dynamics analysis of Toxoptera citricida

When communicating with citrus farmers, they reflected that aphid outbreaks always come suddenly. To gain a deeper understanding of this phenomenon, we developed an aphid-natural enemy interaction model that takes into account temperature-driven and spatial diffusion factors. We also conducted continuous observations of the continuous population dynamics of the brown citrus aphid in the laboratory, with the observation object being the aphids on a single leaf of laboratory citrus seedlings, and the continuous observation lasted for more than 200 hours.

Fig 1. Laboratory Observation on the Population Dynamics of Aphids.

The experimental data clearly show two stages of aphid population change: during the growth stage, the population continuously increases but consists entirely of wingless forms; during the outbreak stage (after 128 hours, around 5 days), winged aphids begin to appear and migrate away in large numbers (Figure 1). This is consistent with the model analysis indicating that the population surges and explodes within 5 days, suggesting a high degree of agreement between the experimental data and the model simulation.

Meanwhile, it can be observed that aphids are tiny in size and mostly stay on the undersides of leaves. By the time the human eye can clearly detect them, the optimal control window has already passed. Therefore, we propose an intelligent monitoring module that can track population dynamics in real time and predict the outbreak time window.

Production of Methyl Salicylate in Response to Sucrose Gradient

Aphids excrete honeydew immediately after feeding, and the amount of honeydew is positively correlated with population density, with sucrose being one of its main components. Based on this, we designed and constructed a biosensor capable of responding to sucrose signals. We induced engineered Bacillus subtilis with 0%, 4%, and 8% (w/v) sucrose for 18 hours, and quantified methyl salicylate using liquid chromatography-mass spectrometry. According to the standard curve, we confirmed the production of methyl salicylate and determined its yields to be 1 ng·mL⁻¹/OD, 1.379 ng·mL⁻¹/OD, and 1.595 ng·mL⁻¹/OD, respectively (Figure 2).

Fig 2. Expression level of methyl salicylate(a) Methyl salicylate standard curve(b) Methyl salicylate concentrations induced by different concentrations. Since sucrose induction inhibits the growth of Bacillus subtilis to varying degrees, the expression levels were normalized using OD values after measurement to exclude the influence of cell number. Among them, the OD600 values induced by 0%, 4%, and 8% were 4.102, 3.263, and 2.194, respectively.

This experiment preliminarily identified that the engineered bacteria can express methyl salicylate. Additionally, it preliminarily quantified that the engineered bacteria can respond to changes in sucrose concentration, thereby expressing different concentrations of methyl salicylate, providing guidance for subsequent rigorous quantitative experiments.

Monitoring Range of Methyl Salicylate by Hardware

We designed hardware that can measure the ppm value of methyl salicylate in the environment, convert the data into aphid population density, and feed it back to the APP, enabling farmers to monitor the occurrence of pests in real time. We used liquid paraffin as a control and measured the ppm values of methyl salicylate solutions from the stock solution to 10000× dilution using the hardware to evaluate the concentration range of methyl salicylate monitored by the hardware.

We diluted the methyl salicylate solution with a mass fraction of 99% using liquid paraffin to prepare 10×, 100×, 1000×, and 10000× dilutions, with each group having a volume of 30 mL. Cotton gauze was soaked in the corresponding solutions and placed in a closed space. We then observed the changes in the MeSA (methyl salicylate) parameter values of the hardware. Once these values stabilized within a fluctuating range, 10 sets of data were recorded, and a concentration gradient curve was plotted.

After measurement, the ppm value of liquid paraffin is 0.

Fig 3. Figure of Detection Results for Methyl Salicylate Concentration Gradient.

The experimental results showed that after the methyl salicylate at the current mass concentration was diluted by 10,000×, the monitoring value of the hardware fluctuated between 9–15 ppm, which was the minimum limit for the hardware measurement (as shown in Figure 3). Through conversion based on the mass-density relationship, the minimum detectable concentration of methyl salicylate by our hardware was calculated to be 7.47–7.74 μg·mL⁻¹.

This experiment has initially detected the concentration limit of methyl salicylate that the hardware can monitor, and verified the feasibility of connecting the hardware to the back-end application. To better support the detection application of methyl salicylate expressed by engineered bacteria, the hardware will be optimized in the future to achieve higher sensitivity and a lower limit of detection.

We utilized RNAi technology to design a series of RNA molecules with lethal effects against Toxoptera citricida (brown citrus aphids). To enhance the stability of RNA molecules during delivery in complex environments and improve their uptake efficiency within aphid tissues, we engineered MS2 virus-like particles (VLPs) incorporating aphid gut-targeting peptides and penetration peptides. The self-assembly of RNA molecules with the VLPs is facilitated by the specific interaction between the VLPs and pac-RNA.

Determination of Lethal Efficiency

To clarify the lethal effects of the dsRNAs targeting CHS, CP, and CYP450 that we designed and prepared, we used enzyme-free water as control. Each experimental group was treated with the corresponding RNA at a concentration of 800 ng/μL, with 3 biological replicates set for each group. Observations were made every 12 hours for a total of 5 days.

Fig 4. Cumulative mortality rates of brown citrus aphids at 6 h, 12 h, 24 h, 36 h, 48 h, 60 h, 72 h, 84 h, 96 h, 108 h, and 120 h after treatment with 800 ng/μl dsCHS, dsCP, dsCYP450, dsF2, dsF3, tri-shRNA, bi-amiRNA and in the control group.

The results showed that after 5 days, the corrected mortality rate of aphids in the dsCHS treatment group reached 20.48%, that in the dsCP treatment group was 30%, that in the dsCYP450 treatment group was 31.4%, that in the tri-shRNA treatment group further increased to 49.5%, and that in the bi-amiRNA treatment group was as high as 60.2% (Figure 4).The results demonstrated that the designed RNA molecules exhibited lethal effects on the brown citrus aphid (Toxoptera citricida).

Silence efficiency measurement

The control group and the brown citrus aphids treated with the corresponding RNA at 800 ng/μL were collected for determination of silencing efficiency.

Fig 5. (a)he relative expression levels of CHS, CYP450, and CP19(CP) in Aphis citricidus after treatment with 800 ng/μl dsCHS, dsCYP450, and dsCP for 120 h, respectively, compared to the control. (b)Relative expression levels of CHS and CYP450 in brown citrus aphids treated with 800 ng/μl dsF2 for 120 h, compared with the control. (c) Relative expression levels of CHS, CYP450, and CP19(CP) in brown citrus aphids treated with 800 ng/μl dsF3 for 120 h, compared with the control. Error bars represent the standard deviation (SD) derived from at least three biological replicates. All datasets were statistically analyzed using a Student's t-test (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

The results showed that the gene silencing rates were as follows: 27% for CHS after dsCHS treatment, 43% for CP after dsCP treatment, 30% for CP after dsCYP450 treatment, 31.5% for CYP450 after dsF2 treatment, and 35.3%, 20%, and 45% for CHS, CYP450, and CP respectively after dsF3 treatment (Figure 5).The silencing results validated the efficacy of the RNA molecules at the molecular level.

Characterization of Apparent Phenomena

CHS (chitin synthase) plays a crucial role in forming the insect exoskeleton and midgut peritrophic membrane. During the molting process, chitin-rich structures within these tissues must be reconstructed to accommodate the insect’s continuous growth. Silencing of the CHS gene disrupts this process, leading to molting defects that impair growth and development, ultimately resulting in death. Compared with the normal mortality observed in Aphis citricidus, CHS gene silencing specifically induces molting abnormalities (Figure 6).

Fig 6. the figure shows the images taken under a stereomicroscope at 40x magnification: (a) Normal death phenomenon in the Control group of brown citrus aphids; (b) Individuals of brown citrus aphids that died during the molting process after being treated with 800 ng/μl dsF3.

In the Results section, we have presented the basic characterization of MS2 virus-like particles (VLPs), their stability in an enzymatic environment, and the feasibility of delivery via trunk injection. However, strategies and methods for efficiently delivering them into adult aphids remain to be established. To achieve efficient delivery of amiRNA into aphids, we fused GBP3.1 (a 12-amino-acid aphid intestinal binding peptide: TCSKKYPRSPCM) and TAT (an 11-amino-acid cell-penetrating peptide: YGRKKRRQRRR) to the dimeric capsid protein (CP), thereby endowing VLPs with the ability to target and penetrate aphid intestinal tissues. For a simple and intuitive verification of the activity of these two functional peptides, we constructed a fusion protein system using enhanced Green Fluorescent Protein (EGFP) as the reporter molecule. The GBP3.1-EGFP-TAT fusion protein was successfully expressed and purified using a prokaryotic expression vector (pET28a_GBP3.1-EGFP-TAT-Histag) (Figure 7). As a control, we constructed the plasmid pET28a_EGFP-Histag, which successfully expressed the EGFP-Histag fusion protein (Figure 8).

Fig 7. SDS-PAGE Coomassie brilliant blue staining analysis of the purified target protein from plasmid pET28a_GBP3.1-EGFP-TAT-Histag. Lane MW is the protein molecular weight marker; Lane 1 is the bacterial culture supernatant, Lane 2 is the bacterial pellet; Lanes 3 and 4 are the washing solutions during the purification of the target protein, and Lanes 5–7 are the purified eluates. The theoretical molecular weight of the target protein is 36263.922 Da (approximately 36 kDa), and the actual size of the electrophoresis band is about 36 kDa, which is consistent with the theoretical molecular weight.

Fig 8. SDS-PAGE Coomassie Brilliant Blue staining analysis of the purified target protein from plasmid pET28a_EGFP-Histag. Lane MW is the protein molecular weight marker; Lane 1 is the bacterial culture supernatant, Lane 2 is the bacterial pellet; Lanes 3 and 4 are the washing solutions during the purification of the target protein, and Lanes 5–7 are the purified eluates. The theoretical molecular weight of the target protein is 32078.27 Da (approximately 32 kDa), and the actual size of the electrophoretic band is about 32 kDa, which is consistent with the theoretical molecular weight.

The purified fusion protein was administered to aphids via feeding, with a separate EGFP group set up as the negative control.In Vivo Imaging was performed using a fluorescence microscope (Figure 9), and tissue sections stained with an enhanced Green Fluorescent Protein (EGFP)-specific antibody for immunofluorescence were observed via a laser confocal microscope (Figure 10). The results showed that, compared with the control group, the GBP3.1-EGFP-TAT fusion protein treatment group exhibits a significantly enhanced retention of green fluorescent signals in aphids. This indicates that TAT and GBP3.1 can effectively carry EGFP to penetrate the intestinal tissue barrier of aphids and enter the interior of cells, with GBP3.1 demonstrating a certain degree of tissue targeting specificity. These results confirm that GBP3.1 and TAT can jointly mediate the transport of exogenous proteins across the intestinal barrier of aphids to achieve intracellular delivery. This experiment verified the targeting and penetration functions of GBP3.1 and TAT peptides in live aphids at the protein level, providing critical functional evidence for the subsequent construction of an MS2 VLP delivery system fused with the two peptides and loaded with amiRNA.

Figure 9: Fluorescent microscope live imaging of aphid bodies. The left panel shows the negative control group (EGFP group): aphids were fed artificial diet containing enhanced green fluorescent protein (EGFP). The right panel shows the experimental group (GBP3.1-EGFP-TAT group): aphids were fed artificial diet containing GBP3.1-EGFP-TAT fusion protein. The upper part of each group shows bright-field imaging, and the lower part shows fluorescence imaging, with corresponding views. The scale bar represents 0.5 mm. Chitin generates weak, nonspecific green fluorescence under 488 nm excitation (as indicated by the arrow pointing to the aphid's legs in the negative control group). However, the experimental group shows significant and widespread green fluorescence throughout the aphid body, while only weak nonspecific fluorescence due to chitin is observed in the negative control group. This indicates that TAT and GBP3.1 can synergistically deliver and retain EGFP effectively within the aphid body.

Fig 10. Laser confocal immunofluorescence images of aphid gut tissue. The left image shows the negative control group (EGFP group): aphids were fed artificial diet containing enhanced green fluorescent protein (EGFP), followed by immunofluorescence (IF) staining with an EGFP antibody. The right image shows the experimental group (GBP3.1-EGFP-TAT group): aphids were fed artificial diet containing GBP3.1-EGFP-TAT fusion protein, followed by immunofluorescence staining with an EGFP antibody. The scale bar represents 50 μm. Blue fluorescence indicates DAPI staining of cell nuclei, while green fluorescence represents the immunofluorescence signal after EGFP antibody incubation.

In the Results section, we have demonstrated the successful delivery of VLPs to tree leaves using a red ink-VLPs mixed formulation. The experimental design and implementation were based on a series of prior investigations into the mechanisms of exogenous molecular transport in trees. The detailed process and conclusions are as follows:

It is known that substance transport within trees primarily relies on endogenous mechanisms such as free diffusion, transpiration, pressure-flow transport between source and sink tissues, and molecular exchange. However, whether exogenous molecules can be effectively transported via these mechanisms requires experimental validation. To investigate this, we selected trees with consistent growth (to control experimental variables), removed the upper half of the trunk, and immersed the cut surface in a series of red ink concentrations with different time gradients. We then collected transverse sections from the tree trunks of each treatment group and calculated the proportion of the red-stained area relative to the total cross-sectional area. The results showed (Figure11-12):

• When the red ink concentration is low, the concentration gradient is the main driving force for solution diffusion.
• As concentration increases, the diffusion rate into the phloem significantly accelerates.
• Once the concentration reaches a certain threshold, the diffusion rate increases at a slower pace due to the limited capacity of the phloem, eventually showing a logarithmic growth trend.

These results indicate that exogenous molecules can be transported within trees, even in the absence of pressure-flow transport between source and sink tissues (due to the removal of the upper half of the tree, which disrupts the source-sink structure). This experimental investigation provides key evidence for the feasibility of delivering exogenous proteins (such as target VLPs) into trees via injection.

Figure 11. Immersion of tree trunk cross-sections in red ink with gradient concentrations. The results show that higher red ink concentrations lead to a higher degree of staining.

Fig 12. Immersion of tree trunk cross-sections in red ink with gradient concentrations. The results show that the longer the immersion time, the higher the degree of staining.