Persistence and metabolism of the diamide insecticide cyantraniliprole in tomato plants

Plant uptake and metabolism of pesticides are complex and dynamic processes, which contribute to the overall toxicity of the pesticides. We investigated the metabolic fate of cyantraniliprole, a new diamide class of insecticide, during various growth stages of tomato. Cyantraniliprole was the major residue in leaves, flowers, and fruits, with the relative metabolite-to-parent ratios maintained at < 10% up to 28 days after treatment (DAT). Mature leaves contained consistently higher residues of cyantraniliprole than young leaves throughout the study. Flowers contained the highest cyantraniliprole residues up to 21 DAT, then gradually decreased. Immature green fruits had the highest cyantraniliprole residues (5.3 ± 0.7 ng/g; 42 DAT), and decreased toward red ripening stages (1.4 ± 0.2 ng/g; 84 DAT). Metabolism of cyantraniliprole primarily occurred in the foliage, where 21 metabolites were tentatively identified. Flowers and fruits contained 14 and four of these metabolites, respectively. Major transformation pathways were characterized by ring closure, followed by N-demethylation, and glycosylation. Additionally, plant metabolism of cyantraniliprole was also associated with several minor phase-I, phase-II, and breakdown metabolites. The occurrence of these metabolites in plants varied as a function of tissue types and their developmental stages. Our study highlights a tissue-specific biotransformation and accumulation of metabolites of cyantraniliprole in tomato.

Pesticides are the mainstay of modern agriculture, enabling growers to reduce pest infestation in crops. However, widespread pesticide use has also prompted persistent public concerns over the adverse effects of residues on non-target organisms and human health. Partly driven by this concern, over the past decades, the amount of pesticides used in the United States has been reduced by approximately 40%, shifting toward pyrethroids and neonicotinoids, which are less toxic to birds and mammals 1 . Unfortunately, this shifting trend has exerted significant threats to some plant species and invertebrates (e.g., pollinators) 1 . Besides, incidences of resistance to pyrethroids and neonicotinoids have also been reported worldwide [2][3][4][5][6][7] , necessitating the introduction of new chemistries for pest management.
In this context, recently introduced diamide insecticides have been considered the most promising alternative, in large part due to their lower toxicity to beneficial arthropods, mammals, and pollinators [8][9][10] . Diamide insecticides target insects' ryanodine receptors and trigger uncontrolled release of internal calcium stores; consequently, the exposed insects suffer feeding cessation, lethargy, muscle paralysis, and eventually death 11 . Anthranilic diamides, such as cyantraniliprole and chlorantraniliprole, can be delivered via seed treatment, soil, or foliar applications to manage a broad-spectrum of chewing and sap sucking pests on several plant species [12][13][14] . Recent studies have suggested that soil application or seed treatment of cyantraniliprole and chlorantraniliprole could provide greater efficacy than foliar spray 15 or equivalent level of pest protection compared to neonicotinoid seed treatment 8 , thereby reducing negative impacts on non-target species and environmental health concerns, and a potentially long residual effect post-application 8,15 . Nevertheless, their persistence in soils may result in a prolonged exposure of the treated crops and pests to the insecticides, potentially leading to increased bioaccumulation of the residues inside plant tissues and insecticide resistance 16 . Although the long residual efficacy of anthranilic diamides against pest infestation has been well-studied 8,10,15 , the metabolic pathways and distribution patterns of the residues and metabolites in crops during this prolonged exposure is less known. Furthermore, root

Results and discussion
Residues of cyantraniliprole in xylem saps and plant tissues. No phytotoxicity was observed for the plants treated with cyantraniliprole throughout the study, and the treated plants showed equivalent growth and development compared to the control plants. Following soil application, cyantraniliprole was quickly absorbed by roots and transported acropetally via the evapotranspiration stream in xylem. The highest concentrations of cyantraniliprole in the xylem sap were observed at 7 and 14 DAT (38.3 ± 1.9 and 35.1 ± 3.7 ng/mL, respectively) and decreased gradually in subsequent samples, reaching 4.0 ± 0.7 ng/mL at 84 DAT (Fig. 1A). Cyantraniliprole was also detected in all plant tissues at all sampling intervals up to fruit maturity, with residue concentrations declining in the order of foliage > flowers > fruits (Fig. 1). In the foliage, there was significant interaction between leaf age and sampling interval in determining residue concentrations (p < 0.001). The highest concentrations were found in mature leaves, increased from 1120.3 ± 332.3 ng/g at 7 DAT to 1437.8 ± 525.8 ng/g at 28 DAT. During the same period (7-28 DAT), the concentrations of cyantraniliprole accumulated in young (apical) leaves were significantly lower than those in mature leaves (p < 0.001), but similar among young leaves sampled at different times (ranged from 303.0 ± 47.3 at 28 DAT to 392.4 ± 108.1 ng/g at 21 DAT). After 28 DAT, residue concentrations in mature leaves decreased considerably, but those in young leaves decreased only slightly as sampling progressed, thus reducing difference between young and mature leaves. At 84 DAT, residues in mature and young leaves were 327.1 ± 32.1 ng/g and 169.3 ± 49.8 ng/g, respectively. Flowers exhibited peak residue con-  www.nature.com/scientificreports/ centration at 21 DAT (526.3 ± 79.4 ng/g) but decreased rapidly, reaching 101.3 ± 8.4 ng/g at the last flower harvest (49 DAT). Concentrations of cyantraniliprole in flowers were similar to those in young leaves (p = 0.67). In fruits, cyantraniliprole residue decreased with sampling intervals from its highest concentration at the immature green stage (42 DAT; 5.3 ± 0.7 ng/g) to its lowest at the red ripening stage (84 DAT; 1.4 ± 0.2 ng/g). Since phloem mobility is absent, xylem flux was the predominant driving force that distributed cyantraniliprole throughout the plants following root uptake 15,25,27 . Accordingly, the higher residue concentrations in mature leaves compared with other tissues may be explained by the relative position and flow velocity for each respective plant tissues within the evapotranspiration stream 25 . Young leaves and flowers located in the positions of the farthest upward-transport along the xylem flux from soil pore water to plant apex. On the other hand, appreciable discrepancies in relative surface area of tomato leaves (6 m 2 /kg) and fruits (0.16 m 2 /kg) 25 suggested that the xylem flux for leaves to be significantly larger than for fruits. Tomato leaves transpire water through stomata on the surfaces, creating a driving force for water and soluble compounds flow into leaf tissues, as opposed to fruits, which lack stomata 28,29 . Consequently, significantly lower concentrations of cyantraniliprole were found in the fruits compared to young leaves and mature leaves at all sampling intervals (p < 0.001). The reduction in cyantraniliprole concentrations in all plant tissues with the progression of sampling time can be explained by growth dilution effect and plant metabolism in the respective tissues, as well as a lower cyantraniliprole uptake from the growing medium as supported by the change in concentrations of the insecticide in xylem sap.
Metabolic fate of cyantraniliprole in tomato plants. Due to the lack of reference standards, the metabolites of cyantraniliprole identified in the present study were confirmed at various confidence levels based on the framework proposed by Schymanski et al. 30 . The identification was based on the following criteria: (1) mass error < 5 ppm for the calculated and observed accurate m/z; (2) Cl/Br isotope patterns; (3) the proposed formulas obey the nitrogen rule 31 ; and (4) the proposed chemical structures can be fully explained by MS 2 fragment ions. Those metabolites with inadequate MS 2 characteristic fragments were assigned to the lowest confidence level (level 5).
In this study, a total of 21 metabolites were tentatively identified in different tissues of the tomato plants treated with cyantraniliprole, including those that were reported in the registration documents 32 as well as several new metabolites (Table 1). For example, the metabolite at m/z 455.0022 was detected in all treatments at a retention time of 11.04 min. The proposed elemental composition for this metabolite is C 19 H 12 BrClN 6 O, with a mass error of 1.1 ppm. Its isotope pattern was consistent with a molecule containing one bromine and one chlorine atom. Based on its accurate m/z and proposed formula, this metabolite was proposed as IN-J9Z38, a major metabolite of cyantraniliprole reported in previous studies [32][33][34] . MS 2 fragment ions were subsequently used to confirm the chemical structure (Fig. S2). Similarly, the accurate masses of metabolite TP651 (m/z 651.0608 at RT = 6.53 min and m/z 651.0616 at RT = 7.21 min) allowed the prediction of the chemical formula C 25 H 24 BrClN 6 O 8 with a mass error of 1.2 and 2.5 ppm, respectively. MS 2 fragmentation revealed identical product ions at m/z 489.00, which is the characteristic molecular ion of hydroxylated cyantraniliprole. Addition of an anhydroglucose moiety (162.05 Da) suggested the formation of glycosylated metabolites (Figs. S17 and S18). The same approach was used for identification and confirmation of the remaining metabolites ( Table 1). Out of 21 metabolites tentatively identified, 11 compounds were phase-I transformation products, six compounds were phase-II conjugates, and four compounds were breakdown products. In general, cyantraniliprole lacks the functional groups suitable for direct phase-II conjugation reactions (e.g., OH-, NH 2 -), which requires activation via phase-I reactions. This may explain the predominant presence of phase-I transformation products in the metabolite profile. Details of the proposed chemical structures and high-resolution mass spectra of all metabolites can be found in the Supporting Information (Figs. S2-S22).
Based on our results and previous reports 32,35,36 , the metabolic pathways involved in the transformation of cyantraniliprole in tomato plants are presented in Fig. 2. We observed that ring closure, followed by N-dem- On the other hand, cyantraniliprole also underwent hydroxylation at the N-methyl group to form IN-MYX98 which was subsequently glycosylated to TP651. There were two signals with similar accurate m/z 651.061 ± 0.001 Da, in which the MS 2 fragmentation gave rise to identical product ions at m/z 489.008 ± 0.001 Da, which is the characteristic molecular ion of hydroxylated cyantraniliprole ( Table 1). The isomeric signals of TP651 were denoted as TP651a www.nature.com/scientificreports/ were subsequently taken up. However, these metabolites were undetectable at all sampling intervals in xylem saps, suggesting that root-to-shoot translocation of cyantraniliprole metabolites was negligible, if any, and that transformation of cyantraniliprole in tomato plants primarily occurred in the above-ground plant tissues. Compared to young leaves, mature leaves were the major reservoir of cyantraniliprole metabolites. Cyantraniliprole metabolites were also detected in flower and fruit samples (Fig. 2); however, their occurrence in these tissues was quantitively lower than that in the foliage. In order to visualize the variations of the metabolites in different tissues and sampling intervals, the intensity of each metabolite was normalized to that of cyantraniliprole at the corresponding sampling interval and presented as the metabolite-to-parent (M/P) ratio. Relative abundances of the tentatively identified cyantraniliprole metabolites in young and mature leaves, flowers, and fruits are shown in Fig. 3. The corresponding peak area data prior to normalization can be found in the Supporting Information (Fig. S25). In the foliage, the M/P ratios of all detected metabolites were < 10% up to 28 DAT, indicating that cyantraniliprole was relatively recalcitrant to metabolism. IN-J9Z38, IN-MLA84, and TP619 were found to be the major metabolites in mature leaves, and their relative abundances increased throughout the sampling period. The highest M/P ratios of IN-J9Z38, IN-MLA84 and TP619 were approximately 24.9, 56.1, and 17.7% at 84 DAT, respectively. The increase of these major metabolites in mature leaves could be partly explained by the enhanced biotransformation of cyantraniliprole, as well as depleted store of cyantraniliprole in the growth medium toward the conclusion of the experiment, leading to significant reduction of the parent insecticide residues in mature leaves (Fig. 1B). In young leaves, the M/P ratios of both IN-J9Z38 and TP619 were < 10% across all sampling intervals, while those of IN-MLA84 was approximately 15.2% at 84 DAT (Fig. 3), likely attributed to lower enzyme activity compared to mature leaves 37 . Both IN-J9Z38 and IN-MLA84 were also present as major residues in studies with rotational crops (e.g., wheat forage, hay, straw, and soybean foliage), while primary crop studies reported the formation of several low abundance metabolites 32 . We observed the occurrence of TP619, tentatively identified as a glycosylated conjugate of IN-MLA84, as a major metabolite of cyantraniliprole in tomato plants for the first time in this study. Apart from the foliage, IN-J9Z38, IN-MLA84, and TP619 were also detected in the flowers sampled at different times, with the highest M/P ratios at approximately 5.5, 6.2, and 0.9%, respectively (Fig. 3). In the fruits, IN-J9Z38 was   (Fig. 3), potentially as a result of the significant decline of cyantraniliprole residues in tomato fruits during the same period (Fig. 1D). In general, our data suggested that ring closure (IN-J9Z38), followed by N-dealkylation (IN-MLA84) and glycosylation (TP619) were the predominant transformation pathways of cyantraniliprole across different growth stages of the tomato plants treated through soil drench. Additionally, there existed many minor metabolites which were detected mostly in the foliage and flowers of the treated tomato plants (Fig. 2). IN-J9Z38 and IN-MLA84 were proposed as the precursors of three dehalogenated metabolites, including IN-RNU71, TP363, and TP423. Their occurrence was observed in both leaves (M/P ratios < 5%) and flowers (trace levels). Reductive dehalogenation is a common transformation pathway of halogenated xenobiotics following plant uptake, as have previously been reported for polychlorinated biphenyls, polybrominated diphenyl ethers, 1,2,5,5,6,9,10-heptachlorodecane, and triclocarban in maize 38 , pumpkin 39,40 , and jalapeno pepper 41 plants, respectively.
Although hydroxylation of the parent xenobiotics, forming the metabolites that are more suitable for phase-II conjugation reactions, has been frequently observed, this transformation pathway might occur to a negligible extent in tomato plants upon soil application of cyantraniliprole. In this study, IN-MYX98 was the only hydroxylated cyantraniliprole detected in the tomato leaves and flowers, with the M/P ratios of < 0.5%. Similarly, hydroxylation at the cyano group to amide (IN-JCZ38) and carboxylic acid (IN-JSE76) metabolites also exhibited the highest M/P ratios of approximately 2.1% and 0.6% in mature leaves, respectively. Only trace levels of IN-JCZ38 were present in the flowers, while none of these hydroxylated metabolites were detectable in fruits at any developmental stages. In addition to TP619, other minor glycosylated conjugates were also observed, including TP633, TP651a, TP651b, and TP654. As shown in Fig. 3, these metabolites were largely present in mature leaves, with M/P ratios ranging 0.1-3.4%. Only TP651a was detected in flowers with M/P ratios of 0.3-0.8%.
In addition to the above-mentioned phase-I and phase-II metabolites, we also observed the presence of some breakdown products, such as IN-DBC80, IN-M2G98, and TP315 (Fig. 2). IN-DBC80 was detected at higher abundances in young leaves compared to mature leaves in all sampling intervals (Fig. 3). Its highest M/P ratios in young leaves were approximately 2.2% at 49 DAT, while the abundances in mature leaves were negligible. Interestingly, IN-DBC80 also exhibited similar M/P ratios in flowers, reaching approximately 2.5% at 49 DAT, and subsequently was found as the most abundant metabolite in fruits, with the M/P ratios ranging from 6.6 to 31.7% during 42-77 DAT; however, the M/P ratio of IN-DBC80 was abruptly decreased to approximately 9.5% www.nature.com/scientificreports/ at the red ripening stage (84 DAT). Another frequently detected breakdown metabolite was TP315, which was present in leaves and flowers (M/P ratios of < 2%), but undetectable in fruits at all sampling intervals. Collectively, the results obtained from this study indicated that cyantraniliprole is relatively recalcitrant to plant metabolism. Over the course of the experiment, cyantraniliprole was the major residue in different tissues up to fruit maturity. The biotransformation processes primarily occurred in the above-ground biomass, especially in the mature leaves. Cyantraniliprole → IN-J9Z38 → IN-MLA84 → TP619 were the predominant transformation pathways of cyantraniliprole in tomato plants, as evident by their ubiquitous presence in leaves, flowers, and fruits at different times after treatment. Additionally, biotransformation of cyantraniliprole in tomato plants was also associated with several metabolites, including phase-I, phase-II, and other breakdown products, in low abundance. To our best knowledge, this study captures for the first time the metabolic fate of a diamide insecticide over the lifespan of a crop plant.
In this study, the highest cyantraniliprole concentrations were approximately 5.3 ng/g in green fruits and 1.4 ng/g in ripening fruits, which were substantially lower than the Maximum Residue Limit set by the U.S. EPA for fruiting vegetables (2000 ng/g) 32 . Recently, there have been growing concerns about human health risks associated with the presence of xenobiotic metabolites in plants 42 . IN-J9Z38 has frequently been observed as a major metabolite via abiotic degradation and plant metabolism of cyantraniliprole, and is required to be reported as the residue of concern for risk assessment in processed commodities 32,36,43 . Our data indicated that metabolism of cyantraniliprole in plants via soil drench occurred slowly, in which the M/P ratios of IN-J9Z38 in mature leaves were approximately 5.3% at 28 DAT. On the other hand, IN-J9Z38 was undetectable in fruits. It has been known that glycosylated metabolites are likely to be hydrolyzed, or further transformed by the gut microbes during digestion 44 . However, the metabolite TP619 was only present in the fruits at trace levels, suggesting that it is of a minor concern. Accordingly, in tomatoes, cyantraniliprole residue should be the focus for risk assessment, since none of the metabolites seems to accumulate in fruits.

Materials and methods
Standard of cyantraniliprole (purity 98.2%) was purchased from Chemservice (West Chester, PA). The isotopelabeled d 3 -cyantraniliprole (purity > 95%) was purchased from Clearsynth (Mississauga, Ontario, Canada). MSgrade water, acetonitrile, and formic acid were purchased from Fisher Scientific (Waltham, MA). Analytical grade anhydrous sodium acetate (CH 3 COONa) and anhydrous magnesium sulfate (MgSO 4 ) were purchased Plant tissue harvest. Xylem sap, leaves, flowers, and fruits were collected at 7-to 14-day intervals from 7 to 84 DAT (Fig. 4, Table S1). At each interval (except 63 and 70 DAT), 5 plants from each treatment were randomly selected and destructively sampled for both mature and young leaves. Mature, fully expanded leaves at different sides of the middle section of the plants were sampled, whereas young leaves (apical leaves) were collected from top of the plants. Three to four leaves were harvested from each plant and combined to form two composite samples per plant-one representing young leaves and the other representing mature leaves. Flowers were only sampled at full-open (anthesis) stage from 14 to 49 DAT. Fruits were harvested from immature green (42 DAT) to red ripening (84 DAT), in which 2-3 fruits of similar size were collected from each plant at each sampling interval. The samples were kept on dry ice before transport to the laboratory and stored at − 80 °C until analysis.
Collection of xylem sap was performed as previously described by Lowe-Power et al. 45 . Plants were wellwatered each evening before morning sampling to build up root pressure. One to three hours after sunrise, the plants were detopped at approximately 7 cm above the soil surface using a sharp razor to yield a flat stump surface, on top of which xylem sap accumulated as a droplet. To minimize contamination, the sap droplet accumulated in the first 2 min of detopping was discarded, and the stump was washed with deionized water and gently blotted dry with Kimwipes tissues. For the next 20 min, the sap was frequently pipetted into a 2.0-mL centrifuge tube kept on ice. The tubes were then stored at − 80 °C until analysis. www.nature.com/scientificreports/ Sample preparation. Xylem sap was analyzed without further preparation. Leaves and flowers (approximately 0.5 g per sample) were weighed into 7-mL tubes. Prior to extraction, internal standard d 3 -cyantraniliprole was spiked into each sample at 120 ng/g fresh weight (fw). Sample homogenization was facilitated with the addition of 5 ceramic beads, 1.5 mL water, and 3.0 mL acetonitrile. The samples were then thoroughly homogenized at 6000 rpm for 4 repetitions of 30-s cycles using a homogenizer (Precellys ® Evolution, Bertin Instruments). The samples were centrifuged at 3000 rpm for 5 min, and the supernatant was transferred into new tubes containing 800 ± 5 mg MgSO 4 anhydrous and 200 ± 5 mg CH 3 COONa anhydrous. The samples were vortexed at 2500 rpm for 10 min and centrifuged at 3000 rpm for 5 min to obtain transparent acetonitrile phase. For matrix removal, aliquots of 1.0 mL of the acetonitrile supernatant were transferred into clean 2.0-mL tubes containing 200 ± 5 mg C18 sorbent and 150 ± 5 mg MgSO 4 anhydrous. The samples were shaken on a rotary shaker at 50 rpm for 5 min and centrifuged at 14,000 rpm for 10 min. Finally, the supernatant was transferred into 2.0-mL glass vials. For flowers, an additional concentration step (5 times) was performed by nitrogen blow-down and reconstitution to accurately verify the presence of low-abundant metabolites in the samples. Fruits were homogenized in a blender, and aliquots of 5.0 g pulp were weighed into 50-mL tubes. Prior to extraction, d 3 -cyantraniliprole was spiked into each sample at a concentration of 4 ng/g fw. Acetonitrile (10 mL) was then added into each tube and the samples were vortexed at 2500 rpm for 10 min. Next, 4.0 ± 0.1 g of MgSO 4 anhydrous and 1.0 ± 0.1 g of CH 3 COONa anhydrous were added, and the mixture was vortexed again for 10 min and centrifuged at 3000 rpm for 5 min. Aliquots of 1.5 mL of the supernatant were cleaned up with C18 sorbent and MgSO 4 anhydrous as described above. Finally, 1.0 mL clean supernatant were transferred into 2.0-mL glass vials, dried under nitrogen gas, and reconstituted in 0.1 mL acetonitrile.
Data were initially acquired using a full MS 1 scan on the Orbitrap within a mass range of 150-1200 m/z, with scan resolution set to 120,000 (FWHM), AGC of 400,000, and a maximum injection time of 50 ms. The acquired data were used for primary screening of the metabolite candidates. Fragmentation of the metabolite candidates was subsequently performed using data-dependent MS 2 collision-induced dissociation (CID) at 40% collision energy. The MS 1 master scan was operated with a scan range of 120-1000 m/z, with an Orbitrap resolution of 60,000 (FWHM), AGC of 200,000 and maximum injection time of 50 ms. The data-dependent MS 2 scan was performed for the m/z values listed in the targeted mass inclusion filter containing all metabolite candidates with a mass tolerance of 25 ppm. The scan resolution for MS 2 was set to 15,000 (FWHM), AGC of 100,000, and maximum injection time of 54 ms.
All samples were immediately analyzed after extraction in one single batch, with QC sample containing cyantraniliprole at 12.5 ng/mL injected after every 15 samples. The relative standard deviations of all QCs were from 1.3 to 5.9%. Quantification of cyantraniliprole residue in plant tissues was performed using data acquired from Orbitrap MS 1 scan in selected ion monitoring mode. Cyantraniliprole was identified by retention time (RT ± 0.1 min), accurate molecular ion at m/z 473.0123, and reference isotope ions at m/z 475.0101 and 477.0074 (mass errors < 5 ppm). Stock solutions of cyantraniliprole and d 3 -cyantraniliprole (1 µg/mL) were prepared in acetonitrile and stored at − 20 °C. Working standards (0.8-800 ng/mL) were freshly prepared prior to instrument analysis.
Recoveries of cyantraniliprole and potential plant matrix interference on quantification were determined by the ratio of internal standard d 3 -cyantraniliprole concentration in the final extract to the nominal spiked concentration in each sample (e.i., 120 ng/g for leaves and flowers, and 4 ng/g for fruits, respectively). Our preliminary trials indicated that the conventional QuEChERS matrix removal procedure resulted in significant loss of some cyantraniliprole metabolites (e.g., glycosylated conjugates). Therefore, in this study, matrix was partly removed using only C18 sorbent to essentially preserve the metabolite pool. As a result, plant matrix effects on quantitative analysis were still observed to different extent depending on the type and age of the tissues. Consequently, the concentration of cyantraniliprole was corrected to the respective recovery rate of d 3 -cyantraniliprole in each sample by multiplying the initial result obtained after analysis with a factor [100%/recovery %] 46 . In this study, the mean recoveries of d 3 -cyantraniliprole for young leaves, mature leaves, flowers, green fruits, and red fruits were 92.6 ± 8.3, 81.0 ± 9.9, 105.7 ± 8.5, 74.1 ± 17.1, and 63.2 ± 14.5%, respectively.
Limit of detection (LOD) of cyantraniliprole in each plant matrix was determined using spiked samples approach 47 . The untreated sample matrices were spiked with cyantraniliprole at a nominal concentration of 2 ng/ mL which was approximately five times the initial instrument LOD. The method LOD was then determined as follow: LOD ng/mL = t α × S s www.nature.com/scientificreports/ where t α is the Student's t-value appropriate for a single-tailed 99th percentile t statistic and a standard deviation estimate with n − 1 degrees of freedom; for seven injections, t α = 3.143. S s is the sample standard deviation of the replicate spiked sample analyses. Limit of quantification (LOQ) = 3.3 × LOD. The LOD and LOQ values of cyantraniliprole in different plant matrices and fresh biomass are provided in Table S2. On a fresh weight basis, the LOQ in young leaves, mature leaves, flowers, green fruits, and red fruits were 5.6, 6.2, 7.7, 0.3, and 0.4 ng/g, respectively.
Concentrations of cyantraniliprole in xylem sap were quantified on a Shimadzu Prominence UFLC coupled with a Shimadzu 8040-triple quadrupole mass spectrometer, applying ESI in positive mode. Chromatographic separation was performed at room temperature on a Kinetex XB-C18 column (150 × 3 mm, 2.6 μm, Phenomenex). Water and acetonitrile, both containing 0.1% formic acid, were used as the mobile phase A and B, respectively. The gradient program was: 0-2 min: 15% B; 8 min, 95% B; 8-11 min: 95% B; 11.01 min: 15% B; and re-equilibration at 15% B for 5 min. The flow rate was maintained at 0.4 mL/min and the injection volume was 2 µL. The interface voltage was set to 4.5 kV and nitrogen was used as nebulizing gas (3 L/min). The desolvation line and heat block temperatures were set to 250 and 400 °C, respectively. The MRM transition m/z 474.9 → 285.9 (CE = 15 V) was used for quantification, while the transitions m/z 474.9 → 176.9 (CE = 49 V) and m/z 474.9 → 111.9 (CE = 55 V) were used for confirmation. The LOD and LOQ values of cyantraniliprole in xylem sap were 0.3 and 1.1 ng/mL, respectively.
Data processing for metabolite identification. Screening of cyantraniliprole metabolites was performed as previously described 18 . Briefly, MS 1 scan data were processed by Compound Discoverer 3.1 software (Thermo Scientific), using a custom designed workflow as shown in Fig. S1. The m/z signals present in the nontreated samples were eliminated from being screened as cyantraniliprole metabolites. The metabolite candidates were further processed via a Cl/Br isotope filter, assuming the metabolites contained at least one Cl and/or one Br atom from the parent cyantraniliprole. Subsequently, Mass Frontier 8.0 software (Thermo Scientific) was used to generate the possible chemical formulas for the metabolite candidates identified by Compound Discoverer, with mass-accuracy errors of < 5 ppm for the molecular ion [M + H] + . Following the primary identification of the metabolite candidates, the proposed structures were further confirmed by the characteristic fragment ions (mass-accuracy errors < 5 ppm) acquired by data-dependent MS 2 fragmentation. Annotation of the fragment ions was performed using Mass Frontier 8.0 and CFM-ID 3.0 web server 48 . The identification confidence was ultimately classified following the framework proposed by Schymanski 30 .
A list of cyantraniliprole metabolites was compiled from the literature (Table S3). The metabolites identified by Compound Discoverer were subsequently searched against the compiled list of metabolites using their accurate m/z and the corresponding MS 2 fragment ions. The remaining metabolites were annotated as newly identified transformation products (TPs). Due to the lack of reference standards, relative quantification of the tentatively identified metabolites (intensity cutoff at approximately 10 4 , with signal-to-noise ratio > 10) compared to the parent cyantraniliprole were reported. The chromatograms of cyantraniliprole and metabolites at the levels of quantification are provided in Figs. S23 and S24.
Statistical analysis. Concentrations of cyantraniliprole in plant tissues were calculated based on fresh weight (fw). Analysis of variance (ANOVA) with Sidak's or Tukey's HSD post hoc tests were performed using GraphPad Prism 8.4.3 (GraphPad Software, San Diego, CA) to detect significant differences and separate means among plant tissues and sampling intervals (α = 0.05).
Ethics statement. The tomato seeds used in this study is commercially available in the United States; therefore, this study does not contain any research requiring ethical consent or approval. Insecticide treatments and collection of plant samples complied with relevant institutional, national, and international guidelines and legislation.

Data availability
Data supporting the findings of this manuscript are available from the corresponding author upon reasonable request.