We developed a noninvasive experimental setup to visualize VOC-mediated plant communication. Petri dishes were separated in half using a porous barrier which enables air to move freely within the petri dish. Insects were only able to eat plants on one side of the dish. On half of the dish, we grew five Emitter plants for two weeks. Insects were added at the time of the experiment. On the other half of the dish, we grew two Receiver plants (Figure 1a). Fluorescence was analyzed continuously using time-lapse imaging.

To visualize VOC-mediated communication between plants, we used insect infested Wild-type (WT) Arabidopsis as Emitter plants. For Receiver plants, we used transgenic Arabidopsis containing Yellow Fluorescent Protein (YFP) fused to the promoter of VSP1 (VSP1-Receiver) (Betsuyaku et al., 2018; Kinoshita and Betsuyaku, 2018; Kinoshita et al., 2019). Arabidopsis Columbia accession (Col-0) for both Emitter and Receiver plants were used. Col-0 accessions do not produce detectable amounts of GLVs because of their lack of functional HYDROPEROXIDE LYASE (HPL), a key enzyme in GLV synthesis (Duan et al., 2005).

When Emitter plants were infested with diamondback moth (DB moth) (Plutella xylostella), a specialist caterpillar for the Brassicaceae family which includes Arabidopsis, VSP1-Receiver displayed an intense fluorescent signal (Figures 1b-d; Supplementary Figure 1; Supplementary Movies S1, S2). Fluorescence was detectable at the onset of an insect attack indicated by initial decline of surface area exhibiting chlorophyll autofluorescence (Figures 1b, c; Supplementary Figure 2). For example, one hour after DB moth infestation, less than five percent of Arabidopsis-Emitter leaves were damaged (Figure 1c). Surprisingly, VSP1-Receiver displayed approximately 30% of peak fluorescence at this early timing (Figure 1b-c). There was also greater VSP1-Receiver signal intensity when the number of intact plants increased suggesting signaling amplification within groups of plants (Figure 1e).

Fluorescent intensities were measured and plotted along the horizontal section of the leaf base to analyze signal distribution within VSP1-Receiver. Fluorescence declined around the midvein but was highly expressed in the tissue surrounding the midvein (Figures 1f, g). This suggested that insect defense in the lamina tissue surrounding midvein was activated within Receiver plants via the detection of VOC (Caldwell et al., 2015). In a parallel experiment, WT Arabidopsis was subjected to polyphagous oriental armyworm (armyworm) (Mythimna separata). This was done to demonstrate whether the VSP-1 expression pattern we observed using DB moth was specific to specialist herbivore insects.

We found that while VSP1-Receiver plants displayed an increase of fluorescence when exposed to armyworm-infested Emitter plants, its signal was weaker when compared to the DB moth experiments (Figures 1d, h, i; Supplementary Figure 1; Supplementary Movie S3). It is likely that relatively large armyworms consumed Emitter plants faster than they could release sufficient quantities of VOC. In any case, the signal followed a similar pattern to that of DB Moth. To ensure that these signals were not caused by the respiration of insects, we verified that neither carbon dioxide nor extra humidity induced a clear fluorescent signal (Figure 1h). There was a slight decrease of fluorescence in response to increased carbon dioxide. Insect respiration likely dampens rather than induces signal. We also confirmed that fluorescent signals are detected even when holes were made in the Petri dish to accelerate ventilation (Supplementary Figure 3).

GLUTAMATE RECEPTOR-LIKE proteins (GLRs) are necessary to express JA-inducible anti-herbivory genes in plants under insect attack (Mousavi et al., 2013; Toyota et al., 2018). So, we sought to test whether GLR genes in Emitter plants are required to release VOC.

When double mutant glr3.3glr3.6 was used as an Emitter plant, VSP1-Receiver plants exhibited reduced fluorescence when compared to the WT Emitter experiments (Figures 1d, i; Supplementary Figure 1). Fluorescent signal from VSP1-Receiver did not elevate when exposed to the scent of herbivore insects by itself (Figure 1i). Emitter genes GLR3.3 and GLR3.6 were required to communicate with Receiver plants.

Prior studies demonstrate that plants increase their [Ca²⁺]_cyt levels after exposure to terpenoids and Green Leaf Volatiles, GLVs (Hu et al., 2019; Wang and Erb, 2022; Aratani et al., 2023). We examined whether natural VOC without GLV from plants under herbivore insect stress increase calcium waves in Receiver plants. Using time-lapse imaging, we visualized [Ca²⁺]_cyt increase in Receiver using Arabidopsis expressing GCaMP3 (Gcamp3-Receiver), a protein-based fluorescent calcium sensor (Toyota et al., 2018). Fluorescent signal was detectable in the midvein and extended toward the tip of the leaf in Receiver plants (Figures 2a, b). For both DB moth and armyworm, this corresponded with previous findings that [Ca²⁺]_cyt travels through the Emitter plant’s vascular system (Toyota et al., 2018). Receiver plants appear to mirror the Emitter response to insect damage without being physically damaged.

We quantified the speed of [Ca²⁺]_cyt increase by measuring length of the signal increase along the midvein from the base of the leaf. [Ca²⁺]_cyt increase moved markedly (more than two orders of magnitude) more slowly than what was observed in Emitter plants. The peak speed of [Ca²⁺]_cyt increase was 118 ± 15 μm/min. within a Receiver plant’s leaf in response to DB Moth damaged Emitter plants (Figure 2b). In Emitter plants, [Ca²⁺]_cyt increase moved approximately 6 mm/min when Arabidopsis was damaged by another lepidopteran insect, cabbage butterfly (Pieris rapae) (Toyota et al., 2018). Emitter plants appear to respond faster than Receiver plants to warn its own leaves of imminent insect threats.

Emitter-GLRs are required for the upregulation of fluorescent signal in VSP1-Receiver (Figure 1i) as well as the production of calcium signals in Emitter plants (Nguyen et al., 2018; Toyota et al., 2018). VOC from infested Wild-type (WT) Emitter plants did not increase fluorescent signal around the midvein of Gcamp3-Receiver/glr3.3glr3.6 (Figure 2c; Supplementary Figure 4). This confirms that Receiver GLR3.3 and GLR3.6 are necessary to activate calcium signaling in the Receiver plants. GLR is localized within vascular bundles (Toyota et al., 2018). It is likely that VOCs enter plants through the stomata (Aratani et al., 2023), and diffuse across mesophyll cells which encase vascular bundles where GLRs are expressed (Nguyen et al., 2018). Additionally, VOC likely interacts with proteins expressed in guard and/or mesophyll cells. These proteins activate GLR expressed in contact cells in xylem parenchyma and phloem where calcium signals originate (Toyota et al., 2018). For instance, an analog of the terpene caryophyllene interacts with the transcriptional repressor TOPLESS (TPL) (Nagashima et al. 2019). Accordingly, TPL and TPR4 (TPL RELATED 4) are highly expressed in stomata guard and mesophyll cells (Winter et al., 2007).

Emitter GLRs are necessary to upregulate VSP1 expression in Receiver plants (Figures 1b, h, i). When Emitter plants contained glr3.3 and glr3.6 mutations, the region of [Ca²⁺]_cyt increase in Receiver plants did not extend toward the tip of the leaf when compared to WT Emitter (Figure 2d, Supplementary Figure 5). Emitter GLRs are therefore necessary to produce calcium waves in Receiver plants.

We analyzed how plant communication occurs without GLVs by isolating chemical compounds found in VOC using gas chromatography mass spectrometry (GC-MS). The median number of detected compounds was 22 (range 14–24) in damaged Emitter plants, compared to 10 (range 6–13) in undamaged controls. Principal component analysis clarified that biological replicates clustered tightly (Figures 3a, b). Volcano plot analysis identified chemical compounds whose levels changed significantly in close proximity to damaged Emitter plants (Figure 3c). Among these, 23-chemical components increased significantly when plants were infested with DB moth (Supplementary Table 1). GLVs like (E)-2-hexenal, (Z)-3-hexenal, and (Z)-3-hexenyl acetate were absent due to their lack of functional Hydroperoxide Lyase (HPL), the enzyme responsible for GLV synthesis (Duan et al., 2005).

Octanal, an aldehyde known to attract the natural enemies of herbivore insects, was detected in our analysis (Figure 3c). For example, octanal is released by cabbage when infested with pierid butterfly larvae (Birkett et al., 2003). Hydrocarbons like tridecane, pentadecane, dodecane, cyclohexadecane were also found (Figure 3c). These hydrocarbons were detected as a part of VOC emissions in various plant species, inclusive of rice and tomato (Errard et al., 2015; Yi et al., 2023; Gokila et al., 2024; Yasa et al., 2024). While these VOC were detected in other plant species, their significance was likely missed in terms of their function as a facilitator of plant communication under duress from herbivore insects.

Notably, acetophenone, a major compound found in Alyssum (Lobularia maritima) (Brassicaceae) was identified (Figure 3c). Volatiles from Alyssum attract Cotesia vestalis, a parasitoid wasp and natural enemy of DB moth (Zubkov and Kouznetsov, 2023). Acetophenone is known to extend C. vestalis’ lifespan, and increase C. vestalis’ parasitism rate (Chen et al., 2020). Acetophonone enables Brassicaceae plants to provide an important benefit to Cotesia vestalis without extending these same benefits to DB moths. Further investigation is needed to clarify whether these compounds play a direct role in triggering the GCaMP3 calcium waves and VSP1-YFP expression.

Emission of octanal, four aforementioned hydrocarbons, and acetophenone required GLRs. Volcano plot analysis using GLR mutants did not detect significant differences between Emitter plants with and without DB moth samples (p < 0.05, fold change ≥ 2, Supplementary Table 2). Although relationships between alkane biosynthesis and GLR remain unclear, acetophenone is produced via β-oxidative pathway (Zhai et al., 2025). Octanal is derived from lipid oxidation (Liang et al., 2022). At the same time, GLR triggers jasmonic acid biosynthesis and this requires both LOX-mediated lipid oxidation and subsequent β-oxidation steps (Li et al., 2005; Toyota et al., 2018). It is plausible that GLR-dependent signaling promotes the accumulation of both compounds by modulating these metabolic activities although direct regulatory links are not yet established.

Arabidopsis thaliana ecotype Columbia-0, which is publicly available from Arabidopsis Biological Resource Center (https://abrc.osu.edu/) was used as Wild type (WT). Transgenic lines GCamp3 and GCamp3/glr3.3glr3.6 were described by Toyota et al (Toyota et al., 2018). Transgenic line VSP1-YFP was previously documented (Betsuyaku et al., 2018; Kinoshita and Betsuyaku, 2018; Kinoshita et al., 2019). Columbia-0 ecotype was graciously provided by Dr. Tohru Ariizumi, University of Tsukuba.

Surface sterilized Arabidopsis seeds were grown for three weeks in Murashige and Skoog (MS) media consisting of half-strength MS, 0.5% sucrose, and 0.6% Phytagel (Kinoshita et al., 2012). Plants were grown at 22°C under 16-h/8-h light/dark cycle. After one week, seven seedlings were transferred to media inside a Petri dish divided in half by a plastic divider and grown for two additional weeks. For Figure 1e, 22 seedlings were transferred. Three-week-old seedlings were used in the experiments. This divider had 1 mm holes. The plastic divider in the Petri dish was manufactured at the University of Tsukuba Engineering Workshop Division.

Armyworm (Mythimna separata) was reared according to Kuramitsu et al (Kuramitsu et al., 2019). Silkmate (Ehime Sanshu) was used as an artificial diet. Diamondback moths (Plutella xylostella) were reared using komatsuna leaves. Second and third instar armyworm larvae and fourth instar diamondback moth larvae were used.

M205FA automated stereomicroscope (Leica Microsystems) with a motorized stage and DFC7000T color CCD camera (Leica Microsystems) was used to acquire images. Apparatus was controlled by LasX software (Leica Microsystems). A metal halide bulb (Leica EL6000) was used as an excitation light source. Chlorophyll auto fluorescence and YFP signals were detected using Texas Red and YFP filters respectively (both from Leica Microsystems) (Kinoshita and Betsuyaku, 2018; Kinoshita et al., 2019). Excitation emission wavelengths of YFP and Texas Red filters were 510/20–560/40 nm and 560/40–610 LP nm respectively.

In all analyzes, except Figures 1b, c, the first cycle was recorded without insects. Six DB moth larvae or two armyworm larvae were used in each experiment (with the exception of Figure 1e). Equal numbers of insects were used for control without Emitter (Figures 1i, 2d). For the carbon dioxide control experiments, a Petri dish with Receiver was opened and placed in a plastic box containing approximately 800 ppm of carbon dioxide (Figure 1h). This environment was created by placing dry ice in the plastic box. The Petri dish was left open for one minute before being sealed and analyzed using time-lapse microscopy. To measure the effect of moisture on control experiments, we placed a cotton ball with 40 μl of ultra-pure water, which is the approximate weight of six DB moths, into a Petri dish with Receiver and analyzed them using time-lapse microscopy (Figure 1h). For experiments using our automatic fluorescent signal quantification method (Kinoshita et al., 2019), diamondback moth was added at the beginning of recordings (Figure 1b, c). In the experiment regarding ventilation, we used a 0.6 mm drill to drill three holes on the Emitter side and five holes on the Receiver side of the lid of the petri dish (Supplementary Figure 3).

Experimental research using plants and insects in our study complies with relevant international, national, and institutional rules and guidelines.

Red channel images from Texas Red filter and white light images were exported from LasX (Leica Microsystems). Green channel images from the YFP filter were exported and analyzed using FIJI (NIH) by superposing binarized Texas Red images to define plant shapes. These superposed images were used to extract plant-specific signals. Shadows in merged images at the intersection of the adjacent frames were adjusted using the BaSic tool (Peng et al., 2017).

For Figure 1c, the “Analysis” function of LasX was used to calculate the Texas Red area from red channel images. Mean YFP signal intensity from the whole plant was analyzed using green channel images (Kinoshita et al., 2019). To prevent the leaves from growing upward in the dark, 300 milliseconds of white light was applied for every YFP image taken on a single tile. As for the overlap between leaves or insects, 20 millisecond exposure for Texas Red imaging immediately followed five seconds of YFP exposure.

The remaining YFP signal intensity was quantified using images from the green channel with FIJI (NIH) software. For Figures 1h, i, fluorescent signal intensities of fully developed, non-overlapped whole leaves were measured. For Figures 1f, g, signal intensity at the time of Fmax was analyzed at the leaf blade closest to the petiole. Signal intensity was plotted along the horizontal orientation of a leaf using FIJI. Relative positions were used: the center of the leaf was 0.5. The edges on both ends were 0 and 1. Timing of maximum mean fluorescent signal was used for Figures 1e, i. For Figure 2c, mean signal intensity for the increased fluorescent region within midvein was compared against the rest of the leaf. For Figures 2b, d, length of the region of signal increase along the midvein from base of the leaf was measured using threshold of 6-94. For Figure 2b, data for 1.5 hours was used, for Figure 2d, data at 0.5 hours was used. For Figure 1e, acrylic box of 193 mm (W) x 103 mm (H) x 25 mm (D) was used. Box was separated into three sections using a porous divider. The first section contained five WT Emitters, the second section contained 15 WT Arabidopsis plants, and the third section contained two VSP1-Receivers. Twenty DM moths were added to the Emitter section after the first frame was recorded. Recording was conducted for 20-hours in one-hour intervals. Fluorescent signal intensities of fully developed, non-overlapped whole leaves were measured. The brightest two leaves were analyzed for each plant. Statistics and graphs were done using Spitzer et al. (2014) and R package (Team, R.C., 2019).

Volatile components were collected via passive absorption onto a Monotrap (RGC18 TD, GL-Science, Tokyo, Japan) for three hours at room temperature. After the volatiles were absorbed, samples were transferred to a 1.5-ml vial and stored at − 20 °C until analysis. Headspace volatiles collected using the Monotrap were analyzed by gas chromatography-mass spectrometry (GC–MS, GC: Agilent 7890A/MS5977B MSD, Agilent Technologies, CA, USA) with an HP-5MS UI capillary column (30 m, 0.25-mm ID, 0.25-μm film thickness; Agilent Technologies) equipped with a thermal-desorption system, cooled injection, and cold trap (Gerstel). The GC was maintained at 40 °C for 3 min., increased to 150 °C at a rate of 10 °C/min., then to 280 °C at 20 °C/min., and held at this temperature for five minutes. Helium was the carrier gas at a constant flow of 1.1 ml/min. The compounds were tentatively identified using data contained in the NIST Mass Spectral Library, 2017 release. GC-MS data were deconvoluted using Unknowns Analysis software (ver. B.09.00, Agilent Technologies) and aligned using Mass Profinder Professional (ver. 14.9, Agilent Technologies). Only entities present in at least 60% of replicates from one condition were included in subsequent analyzes. To determine changes in components in each condition, the data was subjected to principal component analysis using software R (version 4.1.0). To determine which components varied between conditions, we performed volcano plot analysis to compare with and without infested Emitter (P value 0.05, fold change 2) in the Mass Profinder Professional software. Compound identification was performed by comparing mass spectra to the NIST 2017 Library.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.