Chemosensory protein regulates the behavioural response of Frankliniella intonsa and Frankliniella occidentalis to tomato zonate spot virus–Infected pepper (Capsicum annuum)

Many herbivorous insects rely on plant volatiles to locate their host plants. Vector-borne viral infections induce changes in plant volatiles, which render infected plants more attractive to insect vectors. However, the detailed mechanisms underlying the olfactory responses of insect vectors induced by the volatiles produced by virus-infected plants are poorly understood. Here, we show that volatiles emitted by pepper (Capsicum annuum) plants infected with tomato zonate spot virus (TZSV), particularly the volatile cis-3-hexenal, which is recognized by chemosensory protein 1 of the thrips Frankliniella intonsa (FintCSP1), are more attractive to F. intonsa than the volatiles emitted by non-infected pepper plants. FintCSP1 is highly abundant in the antenna of F. intonsa. Silencing of FintCSP1 significantly decreased electroantennogram responses of F. intonsa antennae to cis-3-hexenal and impaired thrips’ responses to TZSV-infected pepper plants and cis-3-hexenal, as assessed using a Y-tube olfactometer. Three-dimensional model predictions indicated that FintCSP1 consists of seven α-helixes and two disulfide bridges. Molecular docking analysis suggested that cis-3-hexenal is positioned deep inside the binding pocket of FintCSP1 and binds to residues of the protein. We combined site-directed mutagenesis and fluorescence binding assays and identified three hydrophilic residues, Lys26, Thr28, and Glu67, of FintCSP1 as being critical for cis-3-hexenal binding. Furthermore, CSP of F. occidentalis (FoccCSP) is also a key olfactory protein involved in modulating the behaviour of F. occidentalis to TZSV-infected pepper. This study revealed the specific binding characteristics of CSPs to cis-3-hexenal and confirmed the general hypothesis that virus infections induce changes in host volatiles, which can be recognized by the olfactory proteins of the insect vector to enhance vector attraction and this may facilitate viral spread and transmission.


Introduction
Most plant viruses that threaten agricultural crops rely on insect vectors for their transmission [1][2][3]. There are three types of insect transmission of plant viruses: nonpersistent transmission, semipersistent transmission, and persistent transmission [1,3]. In persistent transmission, insects retain the acquired virus for long periods (days to weeks), transmitting the virus often throughout the vector's lifespan [1,3]. A large body of literature has documented that viruses improve the nutrient composition of their host plants, which consequently increases the population growth parameters of virus vectors. Sometimes, viruses alter the profile of volatile compounds emitted by their hosts, making infected plants more attractive to insect vectors, thereby bringing more insect vectors to infected plants and thus promoting viral spread [4][5][6][7][8].
In certain host-virus-vector interaction, virus-induced volatile organic compounds (VOCs) alter emission of host which does not make the host plant more attractive to insect vectors [9]. Moreover, aphid vectors are initially more attracted to virus-infected plant, but they are then subsequently deterred from feeding on the host plant after a period of time [10,11]. Although virus-induced plant VOCs help insect vectors locate host plants [12][13][14], how VOCs modify insect behaviour is poorly understood. The insect olfactory system can identify minute amounts of volatile semiochemicals released from host plants or environments, which play an important role in mediating insect behavioural responses, such as seeking food, selecting oviposition sites, or avoiding enemies [15,16]. Olfaction in insects relies on olfactory proteins, including chemosensory proteins (CSPs), odorant binding proteins (OBPs), odorant receptors (ORs), ionotropic receptors, sensory neuron membrane proteins, and odorant degrading enzymes, which are all present in the sensillar lymph of the antennae, maxillary palps, legs, and other chemosensory structures [17][18][19]. CSPs are small, soluble polypeptides that are highly abundant in chemosensory organs of insects [16]. In insects, CSPs function in chemoreception, pheromone recognition, embryo development, regeneration of amputated limbs, solubilization of nutrients, insecticide resistance, and immune responses [16]. The best studied function of CSPs is in chemical communication. To be recognized, the odorous molecules must first pass through the pores or slits in the surface of the insect's antennae or maxillary palps to sensory lymphatic fluid, after which they bind to OBPs or CSPs to form a complex and then are transferred to the dendritic processes of the sensory neurons to finally activate membrane-bound ORs [17,20,21]. In addition, odorous molecules directly interact with ORs to initiate an olfactory-related signal transduction cascade. Thus, OBPs and CSPs commonly serve as ligand selectors, transporters, and solubilizers to trigger a signal or as deactivators once the olfactory signal has been perceived [17,22,23]. However, how these proteins function in insect vectors to locate virus-infected host plants have not been elucidated.
Some insect-borne plant virus-induced alterations in the emissions of host VOCs influence the behaviour of insect vectors and promote local virus transmission [24,25]. In this study, we selected tomato zonate spot virus (TZSV), a newly identified virus of the genus Orthotospovirus [26], and its thrips vector Frankliniella intonsa (Thysanoptera: Thripidae) to determine how virus infection induces changes in its plant host's VOCs to attract thrips vectors. TZSV, which causes substantial yield loss in Southwest China, was first isolated in the Yunnan Province in 2008 [27]. TZSV is transmitted by various thrips species: Frankliniella occidentalis, F. intonsa, F. schultzei, Thrips palmi, and T. tabaci [26]. Flower thrips (F. intonsa) is the dominant thrips species in several areas of China [28,29] (S1 Fig). Feeding on TZSV-infected plants significantly increased the population size of insect vectors [7]. Furthermore, F. occidentalis prefers infected plants over mock-inoculated plants, and the relative emissions of volatiles from TZSV-infected plants are significantly higher than those of mock-inoculated plants [30]. Thus, TZSV can induce changes in the emission of plant VOCs that attract thrips vectors, although the nature of the vector olfactory proteins that regulate their behavioural responses to virusinfected plants is unknown.
To elucidate the underlying regulatory mechanism, we performed a series of free-choice experiments to assess F. intonsa performance on virus-infected plants and the plant volatile cues induced by virus infection. Of four olfactory proteins (FintCSP1, FintCSP2, FintOBP, and FintOR), we demonstrate that FintCSP1 is a key olfactory protein for F. intonsa to perceive and locate TZSV-infected pepper (Capsicum annuum) plants. We also conducted fluorescence ligand binding assays to detect an interaction between specific volatiles and FintCSP1. We carried out homology modeling, molecular docking, site-directed mutagenesis, and ligand binding assays with FintCSP1 to identify the amino acids that contribute to the recognition between the VOC and its receptor. Moreover, CSP is also a key olfactory protein for F. occidentalis to perceive and locate TZSV-infected pepper plant odors. Our study reveals that a persistent plant virus manipulates the behaviour of its insect vector and this may promote or facilitate onward transmission of the virus. The results presented here provide a means to control the spread of plant viral diseases by modulating insect behaviour.

F. intonsa prefers volatiles emitted by TZSV-infected pepper plants
As the dominant thrips species in several areas of China [28,29], F. intonsa is also an important vector of TZSV (S1 Fig). To examine the response of F. intonsa to VOC cues from TZSVinfected pepper plants, we assessed the patterns of orientation preference of thrips exposed to plant odors (TZSV-infected or mock-inoculated pepper plants) in the absence of any other visual, taste, or contact cue. We determined that both females (75.0%) and males (71.7%) preferentially choose infected plants, with a significant difference from insects selecting mockinoculated plants (female, 25.0%, χ 2 = 14.000, P<0.05; male, 28.3%, χ 2 = 9.981, P<0.05; Fig  1A). This result illustrated the preference of F. intonsa for the odors of TZSV-infected pepper plants.
We previously showed that two volatile compounds, cis-3-hexenal and trans-2-hexenal, are uniquely emitted by TZSV-infected pepper plants compared to mock-inoculated plants [30]. Furthermore, four volatile compounds, i.e., m-cymene, (+)-2-carene, α-phellandrene, and αhumulene, which are released from TZSV-infected pepper plants, are produced in significantly higher amounts than in mock-inoculated plants [30]. To test whether any of these six volatile compounds might influence or contribute to F. intonsa preference for TZSV-infected pepper plants, we excised thrips antennae to perform electroantennogram (EAG) analyses. All six volatile compounds triggered an EAG response in our study, although thrips exhibited differences as a function of the odor and its concentration (Fig 1B). The EAG peak amplitudes of F. intonsa antennae to m-cymene (10 mg/L), cis-3-hexenal (10 mg/L), (+)-2-carene (1 mg/L), Choice of F. intonsa on pepper plants that were mock-inoculated (MP) or virus-infected (IP). Significant differences were determined by chisquared (χ 2 ) test (P < 0.05). (B) Electroantennogram (EAG) responses of F. intonsa antennae to volatile compounds. Different lowercase letters indicate significant differences between doses (P < 0.05). Ten individuals were assayed per dose and per compound. Data are means ± SE. (C) Preference tests of F. intonsa to volatile compounds. Data are means ± SE, expressed as a percentage of the number of thrips (n = 60) in a Y-tube olfactometer exposed to one of six volatile compounds induced by TZSV infection (right) or liquid paraffin (control, left). Significant differences were determined by χ 2 tests (P < 0.05). https://doi.org/10.1371/journal.ppat.1011380.g001

FintCSP1 is a key olfactory protein involved in modulating the attraction of F. intonsa to TZSV-infected pepper plant odors
To identify the key olfactory genes in F. intonsa that help the insects locate TZSV-infected pepper plants, we characterized their relative expression levels after exposure to TZSV-infected or mock-inoculated pepper plant odors. Based on GenBank data deposited at the National Center for Biotechnological Information (NCBI) database, we selected four olfactory genes, FintCSP1, FintCSP2, FintOBP, and FintOR. After exposing F. intonsa individuals to TZSV-infected plant odors for 12 h, FintCSP1 expression increased compared to insects exposed to mock-inoculated plant odors (P<0.05), while the other three genes did not reach statistical significance (Fig 2A). To further explore which F. intonsa olfactory genes participate in the host preference, we silenced each gene via RNA interference (RNAi) by injecting double-stranded RNAs (dsRNAs) (Fig 2B). In the Y-tube olfactometer assay, thrips with lower transcript levels for FintCSP2, FintOBP, or FintOR still exhibited a preference for TZSV-infected plants, as did insects injected with a dsRNA targeting the enhanced green fluorescent protein gene (EGFP) (P<0.05, Fig 2C). By contrast, dsFintCSP1-injected F. intonsa displayed no clear preference for either plant (χ 2 = 0.758, P = 0.384, Fig 2C). These results indicated that FintCSP1 functions in the olfactory response of F. intonsa to TZSV-infected pepper plants.
To clarify the function of FintCSP1 in F. intonsa in TZSV-infected pepper plant odor perception, we employed the olfactometer and EAG assays to evaluate the responses to the six volatile compounds mentioned earlier. Silencing of FintCSP1 only resulted in behavioural changes in F. intonsa in response to cis-3-hexenal in the olfactometer assay ( Fig 2D). In dsEGFP-treated thrips, 73.3% of individuals recognized cis-3-hexenal, with the remaining 26.7% moving toward the liquid paraffin control (χ 2 = 6.533, P = 0.011, Fig 2D). FintCSP1silenced thrips showed no preference between the liquid paraff in control and cis-3-hexenal (χ 2 = 0.000, P = 1.000, Fig 2D). To validate the altered behaviour observed for FintCSP1silenced F. intonsa in the olfactometer assay, we excised thrips antennae and exposed them to each volatile compound. Compared to injection with dsEGFP, FintCSP1-silenced antennae showed a significantly lower response to cis-3-hexenal (F = 2.777; df = 16, P = 0.028, Fig 2E). Moreover, cis-3-hexenal was the only compound to which the antennae of dsFintCSP1injected F. intonsa showed a different response relative to dsEGFP-injected thrips ( Fig 2E). These observations indicated that FintCSP1 is a key olfactory protein of F. intonsa to modulate the thrips' response to cis-3-hexenal.

FintCSP1 is abundant in antennae of F. intonsa
To investigate how FintCSP1 detects and transports volatile compounds emitted by TZSVinfected plants, we cloned and sequenced the full-length  [31][32][33]. Subsequently, we selected 31 CSPs from eight insect orders to construct a phylogenetic tree using the neighbor-joining method with Mega 7.0: Orthoptera (eleven sequences), Hemiptera (one sequence), Lepidoptera (one sequence), Thysanoptera (two sequences), Coleoptera (eight sequences), Hymenoptera (three sequences), Blattaria (two sequences), and Diptera (three sequences) (S4 Fig). This phylogenetic analysis showed that FintCSP1 clusters within a small branch close to CSP from F. occidentalis.
We determined FintCSP1 transcript levels in antennae at various developmental stages of F. intonsa by reverse transcription quantitative PCR (RT-qPCR), setting the levels in 1-day-old first-instar nymphs to 1 for normalization. FintCSP1 was expressed in all developmental stages examined, with the highest expression in the 2-day-old male adults, which was 56.05-fold higher than that in 1-day-old first-instar nymphs ( Fig 3A). FintCSP1 also accumulated in female adult antennae, as revealed by immunofluorescence microscopy and immunoelectron microscopy experiments (Fig 3B and 3C). The antennae of female adult F. intonsa contain scape, pedicel, and a long flagellum with five flagellomeres (I-V) ( Fig 3B). We detected strongly fluorescent foci at the three flagellomeres (I-III) of antenna, while no fluorescent signal was observed in the other parts ( Fig 3B). In addition, using immunogold labeling with FintCSP1-specific rabbit polyclonal antibody, we detected FintCSP1 proteins in sensilla basiconica at the flagellum I (Fig 3C), indicating that FintCSP1 is abundant in antennae.

FintCSP1 displays a moderate binding affinity to cis-3-hexenal
To characterize the biochemical binding activity of FintCSP1 to volatile substances, we produced a purified recombinant FintCSP1 (Fig 4). We detected a single band by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) with an apparent molecular weight between 10 and 20 kDa, which was consistent with its predicted molecular weight of 15.01 kDa (Fig 4).
N-phenyl-1-naphthylamine (1-NPN) is frequently used as a fluorescent probe to assess the affinity and specificity of olfactory proteins for potential ligands [19]. Binding plots for FintCSP1 indicated that the binding of 1-NPN is saturable, with a binding constant (K d ) of 6.07 ± 0.40 μM at pH 7.4 ( Fig 5A). Therefore, 1-NPN would provide an adequate fluorescent probe in competitive binding assays of purified FintCSP1 to different ligands. We thus used a competitive fluorescence binding assay to determine the binding affinity of FintCSP1 to cis-3-hexenal and its stereo-isomer trans-2-hexenal ( Fig 5B). We calculated the median inhibitory concentration (IC 50 ) and dissociation constant (K i ) values, which revealed that FintCSP1 displays a moderate binding affinity to cis-3-hexenal (K i = 28.68 ± 0.79 μM), while we observed
To understand how FintCSP1 interacts with host compounds, we conducted molecular docking predictions to identify the key sites in FintCSP1 that bind to cis-3-hexenal. The binding energy between FintCSP1 and cis-3-hexenal was -4.2759 kcal/mol (S3 Table). Cis-3-hexenal laid deep inside the binding pocket of FintCSP1, exhibiting bonded interactions (Fig 6C). Within the binding pocket, the oxygen atom of cis-3-hexenal, the possible hydrogen bond receptor, formed a hydrogen bond with the oxygen atom in the side chain of Ser84 in FintCSP1 (S3 Table). In addition, Van Der Waals interactions also formed between cis-3-hexenal and four residues, Lys26, Phe27, Thr28, and Glu67 (S3 Table). These interactions mainly contributed to the binding energy between cis-3-hexenal and FintCSP1, indicating that cis-3-hexenal tightly binds in the center of the FintCSP1 pocket and influences FintCSP1 activity.

Three amino acids in FintCSP1 affect its binding to cis-3-hexenal
To validate the molecular docking results and clarify the roles of the five predicted key amino acid sites of FintCSP1 in binding to cis-3-hexenal, we produced recombinant FintCSP1

CSP is also a key olfactory protein for F. occidentalis to perceive and locat TZSV-infected pepper plants
To validate the response of F. occidentalis to VOC cues from TZSV-infected pepper plants, we assessed the patterns of orientation preference of thrips exposed to plant odors (TZSV-infected or mock-inoculated pepper plants) in the absence of any other visual, taste, or contact cue. Both females (76.8%) and males (69.0%) preferentially chose infected plants, and the result was significantly different from that of insects selecting mock-inoculated plants (female, 23.2%, χ 2 = 16.071, P < 0.05; male, 31.0%, χ 2 = 8.345, P < 0.05; Fig 8A). This result illustrated the preference of F. occidentalis for the odors of TZSV-infected pepper plants. After exposing F. occidentalis individuals to TZSV-infected plant odors for 12 h, the expression levels of FoccCSP increased compared to insects exposed to mock-inoculated plant odors (P < 0.05, Fig 8B). Then, we silenced FoccCSP in F. occidentalis via RNAi as done in F. intonsa

PLOS PATHOGENS
TZSV manipulation of thrips behaviour in pepper plant ( Fig 8C). In the Y-tube olfactometer assay, FoccCSP-silenced F. occidentalis displayed no clear preference for either plant (Fig 8D). In addition, dsFoccCSP-silenced F. occidentalis displayed no clear preference for cis-3-hexenal or the liquid paraffin control (Fig 8E). Furthermore, compared to injection with dsEGFP, FoccCSP-silenced antennae showed a significantly lower response to cis-3-hexenal (F = 0.306; df = 5, P = 0.034, Fig 8F). These results indicated that FoccCSP functions in the olfactory response of F. occidentalis to TZSV-infected pepper plants.
To characterize the biochemical binding activity of FoccCSP to cis-3-hexenal, we produced a purified recombinant FoccCSP protein (S6 Fig). Binding plots for FoccCSP indicated that the binding of 1-NPN is saturable, with a K d of 6.36 ± 0.51 μM at pH 7.4 (S7 Fig). Recombinant FoccCSP specifically bound to cis-3-hexenal, with a K i of 23.42 ± 0.36 μM (Table 2). Furthermore, compared to other crystal structures of insect CSPs, S. gregaria CSPsg4 shared higher amino acid identity with FoccCSP (S8A Fig). The Ramachandran plot for FoccCSP indicated that nearly all the residues were located at the rational region (S5B Fig). Additionally, the simulated FoccCSP 3D structure consisted of seven α-helixes, and two disulfide bridges with the exact same residues as those of FintCSP1 with the exception of α5 residues (Glu85-Lys101) (S8A and S8B Fig).
The molecular docking results showed that the binding energy between FoccCSP and cis-3-hexenal was -4.0882kcal/mol (S4 Table). Cis-3-hexenal formed a suitable steric complementarity with the binding site of FoccCSP, exhibiting bonded interactions (S8C Fig). Within the binding pocket, the oxygen atom of cis-3-hexenal, the possible hydrogen bond receptor, formed a hydrogen bond with the nitrogen atom in the backbone of Phe27 in FoccCSP (S4 Table). In addition, Van Der Waals interactions also formed between cis-3-hexenal and seven residues, Lys26, Thr28, Thr29, Tyr31, Asp64, Glu67, and Gln87 (S4 Table). Moreover, to verify the results of molecular docking and clarify the key amino acids associated with binding activities of FoccCSP to cis-3-hexenal, we performed a site-directed mutation assay and prepared the purified FoccCSP mutant proteins (S6 Fig). (Fig 9A). Compared to the wild-type protein, three mutants, FoccCSP-Phe27Ala, FoccCSP-Thr28Ala, and FoccCSP-Thr29Ala, showed no affinity for cis-3-hexenal, while the FoccCSP-Val132Ala mutant had only a minor change in binding to cis-3-hexenal (Fig 9B and  Table 2). In addition, five mutants, FoccCSP-Lys26Ala, FoccCSP-Tyr31Ala,

TZSV manipulation of thrips behaviour in pepper plant
FoccCSP-Asp64Ala, FoccCSP-Glu67Ala, and FoccCSP-Gln87Ala, showed significantly decreased binding affinities to cis-3-hexenal ( Fig 9B and Table 2). These results indicated that eight amino acids in FoccCSP (Lys26, Phe27, Thr28, Thr29, Tyr31, Asp64, Glu67, and Gln87) affect its binding to cis-3-hexenal. Our results revealed a common mechanism by which TZSV induces the production of high amounts of specific plant volatiles that are recognized by thrips CSPs, increasing the attractiveness of infected plants to thrips vectors.

Discussion
Arboviruses rely on arthropod vectors for transmission among plant hosts [1,3]. To maximize their transmission rates, some plant viruses induce changes in the composition and volume of plant VOCs to make infected plants more attractive to insect vectors, which may promote disease transmission [2,5,12,14]. Orthotospoviruses, such as TZSV and tomato spotted wilt virus (TSWV), manipulate the volatiles produced by their host such that infected plants are more Behavioural responses of thrips to cis-3-hexenal (right) or liquid paraffin (control, left) after injection with dsFoccCSP or dsEGFP. Significant differences between the treatment (dsFoccCSP) and the control (dsEGFP) were determined by χ 2 test (P < 0.05). (F) EAG responses of dsFoccCSP-injected F. occidentalis to cis-3-hexenal. *, P <0.05, as determined by t-test between the treatment (dsFoccCSP) and the control (dsEGFP). https://doi.org/10.1371/journal.ppat.1011380.g008

PLOS PATHOGENS
TZSV manipulation of thrips behaviour in pepper plant attractive to vectors; however, infection by these viruses also negatively affects plant size and fruit quality [5,30]. However, the underlying mechanism of how these viruses manipulate host volatiles has not been elucidated. Here, we demonstrated that, in the absence of any visual, taste, or contact cues, infection with TZSV made pepper plants more attractive the thrips species F. intonsa and F. occidentalis, two important TZSV vectors, indicating that these insects possess an innate preference for the blend of volatile compounds emitted by TZSV-infected pepper plants (Figs 1A and 8A). We further showed that of the six volatile compounds produced by TZSV-infected pepper plants in greater amounts compared to mock-inoculated plants, only cis-3-hexenal attracted F. intonsa (Fig 1C). Cis-3-hexenal is usually produced in response to mechanical wounding or herbivory and attracts Anagrus nilaparvatae, an egg parasitoid of rice planthoppers (Nilaparvata lugens), at the very low concentration of 1:20,000 dilution [34,35]. In addition to raising the sensitivity off all webworm moth (Hyphantria cunea) antennae to pheromones [36], we showed here that cis-3-hexenal triggers an EAG response from F. intonsa antennae (Figs 1B and 2E), underscoring the involvement of the insect olfactory system in identifying this unique compound. In agreement, we detected abundant FintCSP1 in F. intonsa antennae, and exposure to TZSV-infected plant odors induced FintCSP1 and FoccCSP expression ( Chemosensory proteins play critical roles in the insect olfactory system, and are responsible for capturing and transporting outside odorants through hydrophilic lymph to olfactory receptors [37]. In our study, FintCSP1 was highly expressed in female adults and FintCSP1 mRNA levels increased rapidly up to day 5 after emergence and then began to decrease (Fig 3). Female thrips begin to lay eggs on day 2 after emergence, and peak egg laying occurs on day 5 after emergence [38,39]. These results, together with the fact that adult female insects first detect

PLOS PATHOGENS
TZSV manipulation of thrips behaviour in pepper plant their spawning sites through sensilla [40], suggest that FintCSP1 is involved in finding oviposition sites. Additionally, male thrips begin mating on day 2 after emergence [39]. Therefore, the high level of FintCSP1 expression in male thrips on day 2 might reflect the role of FintCSP1 in

PLOS PATHOGENS
TZSV manipulation of thrips behaviour in pepper plant mate seeking behaviour. Therefore, FintCSP1 plays an important role in the chemoreception of F. intonsa. Competitive binding methods have been employed to measure the binding affinity of CSPs to different odorants in vitro [41,42]. Here, we characterized the binding activity of FintCSP1 to cis-3-hexenal and its isomer trans-2-hexenal. The fluorescence binding assays showed that FintCSP1 specifically binds to cis-3-hexenal with high affinity (K i of below 30 μM), whereas FintCSP1 could not bind to trans-2-hexenal, indicating that thrips can distinguish between different isomers (S2 Table). These results, together with the observation that OBP5 from Apolygus lucorum strongly binds to cis-3-hexenal [43], suggest that cis-3-hexenal is an essential odor molecule in the chemoreception of insects. Site-directed mutagenesis of specific amino acid residues can also elucidate the detailed interactions between CSPs and odor molecules [37,41]. Hydrophilic and hydrophobic residues contribute to the interaction between proteins and odor molecules [44]. Our homology modeling and molecular docking analyses predicted that four hydrophilic residues in FintCSP1 (Ser84, Lys26, Thr28, and Glu67) and one hydrophobic residue (Phe27) have important roles in binding and recognition of cis-3-hexenal (Fig 6C and  S3 Table). These results, together with the individual mutations of Lys26, Thr28, and Glu67 to Ala that abolished the ability to bind cis-3-hexenal compared to the wild-type FintCSP1 ( Fig  7B and Table 1), suggest that hydrophilic residues and Van der Walls interactions are involved in the binding of FintCSP1 to this ligand. Thus, we conclude that TZSV-infected pepper plants induce the emission of a specific plant volatile (cis-3-hexenal) that is subsequently recognized by FintCSP1, which increases the attractiveness of infected plants to thrips vectors.
Generally, insects use the total amount of VOCs in specific ratios and/or specific compounds to locate their host plants [11,45]. Cis-3-hexenal belongs to the C6-compound hexanal group of green leaf volatiles, which is biosynthesized from linolenic acid by various enzymatic reactions in response to abiotic or biotic stresses. Additionally, cis-3-hexenal is converted to trans-2-hexenal (leaf aldehyde) and cis-3-hexenol (leaf alcohol) via enzymatic catalysis [46,47]. Cis-3-hexenal is also a chemical cue for tobacco hornworm (Manduca sexta) and is a feeding stimulant for M. sexta larvae [48]. Our results showed that cis-3-hexenal emitted from TZSVinfected plants enhanced thrips attraction to the plants. Based on this finding, and together with the observation that TZSV-infected plants are more preferred by thrips vectors [7], we infer that cis-3-hexenal is a key chemical for the attraction of adult thrips. Furthermore, these results also strongly suggest a potential application of cis-3-hexenal in plant protection. Color traps and scents are commonly used to control thrips [49,50]. However, color traps also catch numerous natural enemies of thrips [51], making it necessary to add scents to enhance the trapping of thrips but not their natural predators. Although cis-3-hexenal has been shown to attract one of the planthoppers' natural enemies, A. nilaparvatae [34], no studies have reported the effects of or the interaction between cis-3-hexenal and thrips natural enemies, such as Orius spp. (Heteroptera: Anthocoridae) [52]. Further investigations are needed to determine if cis-3-hexenal will pose a risk to Orius spp. populations, which would pave the way for the application of this attractant compound to fight thrips and their borne viruses.

PLOS PATHOGENS
TZSV manipulation of thrips behaviour in pepper plant faba) in an artificial climate incubator (PRX-250B, Saifu experimental instrument, Ningbo, China) (25˚C; 65 ± 5% relative humidity; and a photoperiod of 14 h light/10 h darkness, light intensity: 4800 lx). Tomato (Solanum lycopersicum) fruits infected with tomato zonate spot virus (TZSV) were provided by the Institute of Biotechnology and Germplasm Resources, Yunnan Academy of Agricultural Sciences, China. The virus from tomato fruit was mechanically inoculated onto pepper (Capsicum annuum) leaves (30-day-old). Healthy pepper plants were obtained locally and were grown in a greenhouse maintained at 25˚C under natural light conditions until they were used for virus transmission.

Selection preference of thrips for host plants
F. intonsa and F. occidentalis preference tests of plant volatile cues were performed using a glass Y-tube olfactometer according to the method described by Wang et al. (2014) with minor modifications [8]. A Y-shaped tube (1-cm inner diameter) consisted of a central tube (10 cm in length) and two arms (10 cm in length, offset by 60˚) connected to two different glass odor bottles. One glass odor bottle was 50 cm in height, with a 10-cm inner diameter that contained pepper plants (14-day post-viral infection); the other glass odor bottle was 5 cm in height, with a 3-cm inner diameter for plant VOCs. The airflow in the olfactometer (200 mL/min) was filtered through activated carbon and calibrated using a flow meter at the end of each arm. To remove all visual cues, the Y-tube was positioned horizontally in a black airtight cubic box (90 cm × 60 cm × 40 cm). The tested odor sources (pepper plant VOCs) were enclosed in the odor bottles 5 min prior to each test to allow the odors to fill the arms. Thrips (2-day-old, starved for 1 h) were individually placed in the central tube and allowed to choose between the two arms. When the thrips had climbed over 2/3 of the side arm and stayed there for more than 30 s, it was considered to have made a selection. If the thrips had not made a choice after being placed in the straight arm for 5 min, the insect behaviour was scored as "no selection," and it was not counted in the total data. The direction of the Y-tube arms and the silicone tubes were adjusted after five tested thrips to eliminate the influence of position. The Y-tube was replaced after 10 tested thrips to eliminate the possible influence of existing odors on thrips behaviour. Each treatment consisted of 20 thrips, with three replicates per treatment.

Expression analysis of olfactory genes in F. intonsa and F. occidentalis after exposure to TZSV-infected pepper plant odors
Fifty female F. intonsa or F. occidentalis individuals were placed in a clean and well-sealed glass bottle (45 mL, Cleman) containing leaves of TZSV-infected pepper plants (14-day post-viral infection) for 12 h. Glass bottles with mock-inoculated pepper plant leaves were used as a control. After 12 h of exposure, thrips were removed from the bottles and immediately frozen in liquid nitrogen for RNA extraction. Total RNA was isolated from 50 thrips with Trizol reagent (Ambion, Life Technologies, Carlsbad, CA, USA). First-strand cDNA was reverse transcribed from 1 μg of total RNA using the EasyScript Reverse Transcriptase Kit (Transgen Biotech, Beijing, China). Fintβ-actin (accession number: MT211604) and Foccβ-actin (accession number: XM_026432071.1) were selected as the reference genes; specific primers for Fintβ-actin, Foccβactin, FoccCSP, FintCSP1, FintCSP2, FintOBP, and FintOR were designed using Primer Premier 5.0 (S1 Table). Reverse transcription quantitative PCR (RT-qPCR) was performed using the GoTaq qPCR Master Mix (2X) kit (Promega, USA); each reaction was run as technical triplicates, while each sample was collected as biological triplicates. The expression changes of FintCSP1, FintCSP2, FintOBP, FintOR, or FoccCSP were calculated using the 2 −ΔΔCt method with normalization to Fintβ-actin or Foccβ-actin.

Expression pattern of FintCSP1 in antennae
The expression pattern of FintCSP1 in antennae was investigated by RT-qPCR. Total RNA was isolated from 200 antennae of 1-day-old first-instar nymphs, 1-day-old second-instar nymphs, 1-day-old pupae, 2-day-old female adults, 5-day-old female adults, 10-day-old female adults, 2-day-old male adults, 5-day-old male adults, and 10-day-old male adults. Total RNA extraction, first-strand cDNA synthesis, and qPCR methods were as described above. Fintβ-actin was the reference gene; qPCR was conducted with specific primers for Fintβ-actin and FintCSP1 (S1 Table). The relative transcript levels of FintCSP1 in antennae at various differental stages were determined according to the 2 −ΔΔCt method. Relative transcript levels were normalized to those in 1-day-old first-instar nymphs, which were set to 1 [53].

Immunofluorescence assays
To visualize FintCSP1 in F. intonsa antennae, female thrips antennae (2-day-old) were dissected under an optical microscope and fixed in 4% (w/v) paraformaldehyde (PFA) for 2 h and subsequently permeabilized with 2% (v/v) Triton X-100 (Sigma, USA) for 24 h at room temperature. The fixed samples were washed once with 100% ethanol and transferred to bleaching solution (100% ethanol: 30% H 2 O 2 ; 2:1 [v/v]) for 2 h. After washing, the thrips were incubated with anti-mouse antibodies against FintCSP1 conjugated to fluorescein isothiocyanate (FITC) in phosphate buffered saline (PBS) containing 3% (w/v) bovine serum albumin (BSA) for 2 h at 37˚C. The samples were washed three times with PBS, placed on a clean slide, and processed for immunofluorescence microscopy (Leica SP8, Germany).

Transmission electron microscopy analysis
Female thrips (2-day-old) antennae were dissected, fixed, dehydrated, and embedded as described previously [54]. The samples were sectioned using an ultramicrotome (Leica UC5, Germany) and incubated with FintCSP1-specific IgG and immunogold-labeled anti-goat antibodies against rabbit IgG that had been conjugated with 12-nm-diameter gold particles (Jackson). The samples were observed under an electron microscope (Hitachi HT-7000, Japan).

Bacterial expression and purification of FintCSP1 and FoccCSP
FintCSP1 and FoccCSP DNA were released from pEASY-T1/FintCSP1 and pEASY-T1/ FoccCSP, respectively, by restriction digestion with BamHI and XhoI and cloned into the pET-30a vector (Invitrogen). The resulting construct was transformed into Transetta (DE3) Escherichia coli competent cells (Transgen). The production of recombinant FintCSP1 and FoccCSP was induced by addition of isopropyl-β-D-1-thiogalactopyranoside (IPTG, 0.2 mM) at 15˚C and purified by Ni ion affinity chromatography (Sangon Biotech, Shanghai, China). The target protein was verified by 15% SDS-PAGE, and its concentration was determined with a BCA protein assay kit (Thermo Scientific, Rockford, IL, USA). The recombinant purified protein was stored at −20˚C until use.

Fluorescence competitive binding assays
The binding affinity of FintCSP1 and FoccCSP to the fluorescent probe N-phenyl-1-naphthylamine (1-NPN) was tested using a Spectramax M2 fluorescence spectrometer (Molecular Devices, USA) as reported previously [55]. Volatile compounds and 1-NPN were first dissolved to a concentration of 1 mM in HPLC-grade methanol and dissolved to their indicated final concentration (2-20 μM) to titrate 2 μM protein solutions resuspended in 50 mM Tris-HCl (pH 7.4). The 1-NPN/FintCSP1 or 1-NPN/FoccCSP mixture was excited at 280 nm, and the emission spectra were recorded at 415 nm. The binding constants (K d ) of FintCSP1 or FoccCSP to 1-NPN were calculated by Scatchard analysis using Prism 5 software (GraphPad, La Jolla, CA, USA). Assuming a protein activity of 100% and that the volatile compound binds to the protein in a 1:1 ratio at saturation, the dissociation constants (K i ) of FintCSP1 or FoccCSP for each ligand were calculated according to the IC 50  If the K i value was less than 20 μM, the tested ligands were considered to exhibit a strong binding affinity toward FintCSP1 or FoccCSP. A K i value between 20 and 50 μM indicated that ligands have a medium binding affinity toward FintCSP1 or FoccCSP. A K i value between 50 and 100 μM indicated that ligands have a weak binding affinity toward FintCSP1 or FoccCSP. K i values that exceeded 100 μM indicated that ligands have no binding affinity to FintCSP1 or FoccCSP.

RNA interference (RNAi) tests
The FintCSP1, FintCSP2, FintOBP, FintOR, FoccCSP, and enhanced green fluorescent protein (EGFP) coding sequences were obtained from GenBank under accession numbers MT199111, MT211602, OK067245, MT211603, AEP27186.1, and U55761.1, respectively. Primers were designed using Primer Premier 5.0, and the T7 promoter sequence (5 0 -TAATACGACTCAC TATAGGG-3 0 ) was added to the 5 0 end of each primer (S1 Table). DNA fragments containing T7 promoter sequences were amplified by PCR using pEASY-T1/target gene or pUC-EGFP (preserved in the laboratory) plasmid DNA as template. Then, a double-stranded RNA (dsRNA) was produced from each target gene using the HiScribe T7 Quick High Yield RNA Synthesis Kit (New England Biolabs) according to the manufacturer's instructions. Six hundred 2-day-old female F. intonsa or two hundred 2-day-old female F. occidentalis adults were anesthetized using CO 2 and microinjected with 40 nL (3 μg/μL) dsRNA (dsFintCSP1 dsFintCSP2, dsFintOBP, dsFintOR or dsFoccCSP or dsEGFP) using a CellTram Oil microinjector (Eppendorf, Hamburg, Germany). The injected thrips were kept on broad beans in an artificial climate incubator as described above and used in subsequent RNAi experiments. Three biological replicates were conducted for each treatment with each dsRNA.

Electroantennogram (EAG) assays
At 24 h after dsRNA injection, the electrophysiological responses of F. intonsa antennae to six volatile components (i.e., m-cymene, cis-3-hexenal, (+)-2-carene, trans-2-hexenal, α-phellandrene, and α-humulene) or F. occidentalis antennae to cis-3-hexenal were recorded using EAG. Antennae of female adult thrips were cut off at the base, and the distal terminus was carefully removed. The tested compounds were dissolved in liquid paraffin, with pure liquid paraffin used as a negative control. An Ag-AgCl 2 glass electrode and Kaissling electrode solution were used in the experiment according to Tian et al. [56]. The base of the antenna was inserted into the glass capillary filled with electrode solution at the end of the reference electrode; the top of the antenna was connected to the glass capillary at the end of the recording electrode to ensure no bubbles in the glass capillary. After successful connection, the antennae were stimulated with each chemical in the following order: liquid paraffin, volatile compounds, and liquid paraffin. Stimulus time was 0.2 s with each stimulus at 40 s intervals. At least 10 antennae were tested for each volatile compound. The EAG values to volatile compounds were calculated using the following equation: absolute value of EAG reaction = reaction value of the tested sample-average value of two control measurements.

Homology modeling and molecular docking
The template structure of CSP was identified through NCBI BLAST and downloaded from the RCSB Protein Data Bank (http://www.rcsb.org/) as the PDB identifier 2GVS [57]. Homology modeling of FintCSP1 or FoccCSP was conducted with MOE v2018.01 [58]. The protonation state of the protein and the orientation of the hydrogens were optimized by LigX, at a pH of 7.0 and temperature of 300 K. First, the target sequence was aligned to the template sequence, and 10 independent intermediate models were built. These different homology models were the result of the permutational selection of different loop candidates and side-chain rotamers. Then, the intermediate model that scored best according to the GB/VI scoring function was chosen as the final model. The final model was subjected to further energy minimization using the AMBER10: EHT force field.
MOE Dock was used for molecular docking analysis of cis-3-hexenal on to FintCSP1 or FoccCSP. The 2D structure of cis-3-hexenal was downloaded from PubChem and converted to a 3D structure in MOE through energy minimization as ligand. The predicted structure of FintCSP1 or FoccCSP was used as the receptor. Prior to docking, the force field of AMBER10: EHT and the implicit solvation model of Reaction Field (R-field) were selected. The "induced fit" protocol was selected, in which the side chains of the binding site in the receptor are allowed to move according to ligand conformations, and a constraint was applied on their positions. The weight used for tethering side-chain atoms to their original positions was 10. First, all docked poses were ranked by the London dG scoring function, then force field refinement was applied on the top 30 poses, followed by a rescoring with the GBVI/WSA dG scoring function. The conformation with the lowest binding free energy was identified as the best probable binding mode, which was visualized by PyMOL (www.pymol.org).

Site-directed mutagenesis
To verify the key amino acids of FintCSP1 and FoccCSP that bind to cis-3-hexenal, six amino acid residues of FintCSP1 were mutated (Lys26Ala, Phe27Ala, Thr28Ala, Glu67Ala, Ser84Ala, and the negative control residue Val132Ala) and nine amino acid residues of FoccCSP were mutated (Lys26Ala, Phe27Ala, Thr28Ala, Thr29Ala, Tyr31Ala, Asp64Ala, Glu67Ala, Gln87Ala, and the negative control residue Val132Ala) with the Fast Mutagenesis System Kit (Transgen Biotech, Beijing, China) using the pET-30a/FintCSP1 and pET-30a/FoccCSP plasmid DNA as template, respectively. Specific primers were designed using Primer Premier 5.0 according to the instructions of the Fast Mutagenesis System Kit (S1 Table). After validation by sequencing, each recombinant variant protein was produced and purified as described above, and their binding affinity to cis-3-hexenal was measured as described above. SPSS 21.0 software was used for data analysis (SPSS Inc., Chicago, IL, USA). The comparative analysis of FintCSP1 expression levels in different stages, and EAG recordings of F. intonsa antennae in response to different doses of volatile compounds used one-way analysis of variance (ANOVA) followed by least significant difference (LSD) multiple comparison analysis. The comparative analysis of FintCSP1, FintCSP2, FintOBP, FintOR, and FoccCSP expression levels and EAG values among different treatment groups used t-tests. In the behavioural assays, the chi-squared (χ 2 ) test was used for the olfactometry experiments. Competitive binding of wild-type FintCSP1, wild-type FoccCSP, and mutant variants with ligands were analyzed using Tukey's honestly significant difference (HSD) test.

PLOS PATHOGENS
TZSV manipulation of thrips behaviour in pepper plant