Virus-Induced Plant Volatiles Promote Virus Acquisition and Transmission by Insect Vectors

Rice dwarf virus (RDV) is transmitted by insect vectors Nephotettix virescens and Nephotettix cincticeps (Hemiptera: Cicadellidae) that threatens rice yield and results in substantial economic losses. RDV induces two volatiles ((E)-β-caryophyllene (EBC) and 2-heptanol) to emit from RDV-infected rice plants. However, the effects of the two volatiles on the olfactory behavior of both non-viruliferous and viruliferous N. virescens are unknown, and whether the two volatiles could facilitate the spread and dispersal of RDV remains elusive. Combining the methods of insect behavior, chemical ecology, and molecular biology, we found that EBC and 2-heptanol influenced the olfactory behavior of non-viruliferous and viruliferous N. virescens, independently. EBC attracted non-viruliferous N. virescens towards RDV-infected rice plants, promoting virus acquisition by non-viruliferous vectors. The effect was confirmed by using oscas1 mutant rice plants (repressed EBC synthesis), but EBC had no effects on viruliferous N. virescens. 2-heptanol did not attract or repel non-viruliferous N. virescens. However, spraying experiments showed that 2-heptanol repelled viruliferous N. virescens to prefer RDV-free rice plants, which would be conducive to the transmission of the virus. These novel results reveal that rice plant volatiles modify the behavior of N. virescens vectors to promote RDV acquisition and transmission. They will provide new insights into virus–vector–plant interactions, and promote the development of new prevention and control strategies for disease management.


Introduction
Plant viruses are most often transmitted by insect vectors including thrips, whiteflies, planthoppers, leafhoppers, and aphids [1], and they seriously threaten crop yield and lead to major economic losses [2]. Insect vectors, plant viruses, and host plants form complex multilevel interactions that affect virus dispersal [3]. For persistent plant viruses transmitted by insect vectors, studies have found that non-viruliferous insect vectors initially preferred virus-infected plants, but once the virus was acquired by the vectors, they showed a predilection for virus-free plants, and this phenomenon may accelerate the outbreak of virus [4]. Changes in host preference by insect vectors between virus-free and virus-infected plants after virus acquisition were first reported in the wheat-Rhopalosiphum padi-barley yellow dwarf virus (BYDV) pathosystem [5] and subsequently in several other pathosystems [6][7][8][9]. These studies indicated that changes in the host preference of insect vectors after virus acquisition might be associated with plant volatiles induced by virus infection [4,10].
Host plants release several different kinds of volatile organic compounds (VOCs) during infection by plant viruses including terpenes, sesquiterpenes, green leaf volatiles (GLVs), fatty acid derivatives, aromatics, and nitrogen-containing compounds, as well as the volatile plant hormones, methyl salicylate, and methyl jasmonate [11]. VOCs induced
To verify the involvement of rice volatiles mediating the olfactory behavior of N. virescens, WT (RDV-free) and WT-RDV (RDV-infected) rice plant odors were applied via two opposite arms of a four-chamber olfactometer. Results showed that the time nonviruliferous N. virescens spent in the four arenas was significantly different (Figure 2A, χ 2 = 60.816, p < 0.001, n = 41). The time non-viruliferous N. virescens invested in the arena containing WT-RDV plant odors was almost 569.3 s, whilst in the arena containing WT plant odors it was almost 239.7 s. The tracks of non-viruliferous N. virescens in the four arenas confirmed this discovery ( Figure S1A). These findings suggested that non-viruliferous N. virescens preferred WT-RDV plant odors over WT plant odors. The residence time of viruliferous N. virescens in the four arenas also showed significant differences ( Figure 2B, χ 2 = 67.936, p < 0.001, n = 44). Viruliferous N. virescens remained for almost 484.4 s in the WT plant odor area, and almost 270.8 s in the WT-RDV plant odor area. The walking path of viruliferous N. virescens in the four arenas is shown in Figure S1B. These results demonstrated that viruliferous N. virescens showed a significant preference for WT plant odors over WT-RDV plant odors. All these findings confirmed that rice volatiles influence the selection preferences between WT and WT-RDV rice plants of non-viruliferous and viruliferous N. virescens. All feeding choice assays were performed with fifte logical replicates, each of the fifteen new emergent adults. The statistical differences betw two treatments in the same column were indicated by asterisks (* p < 0.05; ** p < 0.01, Wil signed-ranks tests). Data are mean ± standard error.
To verify the involvement of rice volatiles mediating the olfactory behavior o rescens, WT (RDV-free) and WT-RDV (RDV-infected) rice plant odors were appl two opposite arms of a four-chamber olfactometer. Results showed that the time n uliferous N. virescens spent in the four arenas was significantly different ( Figure 2 60.816, p < 0.001, n = 41). The time non-viruliferous N. virescens invested in the aren taining WT-RDV plant odors was almost 569.3 sec, whilst in the arena containi plant odors it was almost 239.7 sec. The tracks of non-viruliferous N. virescens in th arenas confirmed this discovery ( Figure S1A). These findings suggested that nonerous N. virescens preferred WT-RDV plant odors over WT plant odors. The residen of viruliferous N. virescens in the four arenas also showed significant differences ( 2B, χ 2 = 67.936, p < 0.001, n = 44). Viruliferous N. virescens remained for almost 484 the WT plant odor area, and almost 270.8 s in the WT-RDV plant odor area. The w path of viruliferous N. virescens in the four arenas is shown in Figure S1B. These demonstrated that viruliferous N. virescens showed a significant preference for WT odors over WT-RDV plant odors. All these findings confirmed that rice volatiles inf the selection preferences between WT and WT-RDV rice plants of non-virulifero viruliferous N. virescens. ; WT-RDV, MH63-RDV (RDV-infected). All feeding choice assays were performed with fifteen biological replicates, each of the fifteen new emergent adults. The statistical differences between the two treatments in the same column were indicated by asterisks (* p < 0.05; ** p < 0.01, Wilcoxon's signed-ranks tests). Data are mean ± standard error. WT-RDV, MH63-RDV (RDV-infected). All feeding choice assays were performed with fifteen bio logical replicates, each of the fifteen new emergent adults. The statistical differences between th two treatments in the same column were indicated by asterisks (* p < 0.05; ** p < 0.01, Wilcoxon signed-ranks tests). Data are mean ± standard error.
To verify the involvement of rice volatiles mediating the olfactory behavior of N. v rescens, WT (RDV-free) and WT-RDV (RDV-infected) rice plant odors were applied vi two opposite arms of a four-chamber olfactometer. Results showed that the time non-vir uliferous N. virescens spent in the four arenas was significantly different ( Figure S1B. These result demonstrated that viruliferous N. virescens showed a significant preference for WT plan odors over WT-RDV plant odors. All these findings confirmed that rice volatiles influenc the selection preferences between WT and WT-RDV rice plants of non-viruliferous an viruliferous N. virescens.

Confirming the Attraction of EBC to Non-Viruliferous N. Virescens
Previous research showed that oscas1 rice plants infected with RDV could not emit more EBC [26]. Therefore, we used oscas1 mutant rice plants to verify the attraction of EBC to non-viruliferous N. virescens with feeding and plant odor preference studies. There was no significant preference for feeding selection by non-viruliferous N. virescens between WT (RDV-free) and oscas1 (RDV-free) rice plants, or between oscas1 (RDV-free) and oscas1-RDV (RDV-infected) rice plants during the entire investigation ( Figure 4A,B, p > 0.05). In contrast, the number of non-viruliferous vectors was more abundant on WT-RDV (RDV-infected) rice plants than that on oscas1-RDV (RDV-infected) rice plants (except for 2 h PI, P = 0.475) ( Figure 4C).  To further verify the attraction of EBC to non-viruliferous N. virescens, three plant odor preference assays were conducted. When rice volatiles from WT (RDV-free) and os-cas1 (RDV-free) were added via two opposite arenas of the olfactometer, the time nonviruliferous vectors spent in the olfactometer was significantly different ( Figure 5A, χ 2 = 27.000, p < 0.001, n = 38), but no significant difference between the two odors was found (p > 0.05). However, there was no significant preference by non-viruliferous vectors between oscas1 (RDV-free) and oscas1-RDV (RDV-infected) rice plant odors ( Figure 5B  oscas1, mutant rice plant (RDV-free); oscas1-RDV, mutant rice plant (RDV-infected). All feeding choice assays were performed with fifteen biological replicates, each of fifteen newly emerged adults. The statistical differences between the two treatments in the same column were indicated by asterisks (** p < 0.01, Wilcoxon's signed-ranks tests). Data are mean ± standard error.
To further verify the attraction of EBC to non-viruliferous N. virescens, three plant odor preference assays were conducted. When rice volatiles from WT (RDV-free) and oscas1 (RDV-free) were added via two opposite arenas of the olfactometer, the time non-viruliferous vectors spent in the olfactometer was significantly different ( Figure 5A, χ 2 = 27.000, p < 0.001, n = 38), but no significant difference between the two odors was found (p > 0.05). However, there was no significant preference by non-viruliferous vectors between oscas1 (RDV-free) and oscas1-RDV (RDV-infected) rice plant odors ( Figure 5B The walking tracks of non-viruliferous vectors in a four-field olfactometer with the three different assays can be found in Figure S3A-C. cas1 (RDV-free) were added via two opposite arenas of the olfactometer, the time nonviruliferous vectors spent in the olfactometer was significantly different ( Figure 5A, χ 2 = 27.000, p < 0.001, n = 38), but no significant difference between the two odors was found (p > 0.05). However, there was no significant preference by non-viruliferous vectors between oscas1 (RDV-free) and oscas1-RDV (RDV-infected) rice plant odors ( Figure 5B, χ 2 = 6.692, P = 0.082, n = 39). Non-viruliferous vectors walked more in the WT-RDV (RDV-infected) plants odor arena compared to the other three odor arenas (oscas1-RDV (RDV-infected) and two clean air odor fields) ( Figure 5C, χ 2 = 22.573, p < 0.001, n = 40). The walking tracks of non-viruliferous vectors in a four-field olfactometer with the three different assays can be found in Figure S3A-C.   The time viruliferous vectors spent in the control plant odor arena was significantly longer than that in the treated plant odor arena ( Figure 6B, χ 2 = 40.129, p < 0.001, n = 34), and the walking tracks of viruliferous vectors can be seen in Figure S4. during the first 4 h PI ( Figure 6A, 2 h, P = 0.858; 4 h, P = 0.064). However, viruliferous N. virescens preferred feeding on the control plants compared to treated plants at 8 h PI (Figure 6A, P = 0.007), and this performance was throughout the rest duration (24 h, P = 0.005; 48 h, P = 0.007; 72 h, P = 0.005).

EBC and 2-Heptanol Promote the RDV Acquisition and Transmission
The time viruliferous vectors spent in the control plant odor arena was significantly longer than that in the treated plant odor arena ( Figure 6B, χ 2 = 40.129, p < 0.001, n = 34), and the walking tracks of viruliferous vectors can be seen in Figure S4.

Discussion
The interactions among host plants, plant viruses, and insect vectors are multiple and complex [3,27,28]. Plant virus infection could alter chemical cues and behavioral changes in its host plants and insect vectors that enhance virus epidemic and dispersal [4,10,29,30]. The results from this present study support our hypothesis that EBC and 2-heptanol facilitate RDV acquisition and transmission by influencing the olfactory behavior of both nonviruliferous and viruliferous N. virescens independently. Firstly, non-viruliferous N. vi-

Discussion
The interactions among host plants, plant viruses, and insect vectors are multiple and complex [3,27,28]. Plant virus infection could alter chemical cues and behavioral changes in its host plants and insect vectors that enhance virus epidemic and dispersal [4,10,29,30]. The results from this present study support our hypothesis that EBC and 2-heptanol facilitate RDV acquisition and transmission by influencing the olfactory behavior of both non-viruliferous and viruliferous N. virescens independently. Firstly, non-viruliferous N. virescens preferred RDV-infected rice plants (odors) over RDV-free rice plants (odors), and conversely, viruliferous N. virescens preferred RDV-free rice plants (odors) to RDV-infected rice plants (odors). Secondly, EBC attracted non-viruliferous N. virescens and 2-heptanol repelled viruliferous N. virescens; revealed by a series of complementation assays. Thirdly, the two VOCs promote the rate of RDV acquisition and transmission by insect vectors.
Changes in host selection preference (conditional preference) by insect vectors may promote virion spreading and diffusion, especially for vectors transmitted plant viruses in persistent and circulative manners, such as BYDV and potato leaf roll virus (PLRV) transmitted by M. persicae, tomato spotted wilt virus (TSWV) transmitted by Frankliniella occidentalis, maize Iranian mosaic virus (MIMV) transmitted by Laodelphax striatellus, and tomato yellow leaf curl virus (TYLCV) and tomato severe rugose virus (ToSRV) transmitted by Bemisia tabaci [5,6,[31][32][33]. In the present study, we found that RDV modified the host selection preference of the vector N. virescens, which was consistent with our previous studies [8,26]. These findings suggested that the two vectors of RDV showed conditional preference. According to the general epidemiological model established by Gandon [32], and Shaw et al. [34], we infer that the conditional preference of N. virescens would aid the RDV dispersion.
Virus-induced plant volatiles may be attractive or repulsive to insect vectors, and thus play a key role in host plant-plant virus-insect vector interactions [11]. Most research has focused on the behavioral responses of insect vectors to blended VOCs from virusinfected plants, or artificially blended VOCs, which mimic the natural blends, or individual VOCs prominent in those blends [6,12,17,35,36]. However, the mechanism by which virusinduced plant volatiles, especially individual VOCs, affect the conditional preference of both non-viruliferous and viruliferous vectors has not been much explored. Here, we show that two rice VOCs induced by RDV infection mediated the olfactory behavior of vector N. virescens insects in different manners: EBC was attractive to non-viruliferous N. virescens causing them to prefer RDV-infected plants, while 2-heptanol was repulsive to viruliferous N. virescens causing them to feed on RDV-free plants. This finding is consistent with the role of the two VOCs in non-viruliferous and viruliferous N. cincticeps [26]. However, whether the two VOCs could exacerbate the diffusion of RDV is unclear. In this study, for the first time, we revealed that EBC could facilitate the acquisition of RDV by non-viruliferous N. virescens, and 2-heptanol would aid the spread of RDV by viruliferous N. virescens.
In this study, we have also shown that the impacts of EBC and 2-heptanol on the olfactory behavior of N. virescens are conversely dependent upon whether the insect is carrying the virus or not. Several recent studies found that the odorant binding protein (OBP) genes or olfactory receptor co-receptor (Orco) gene of insect vectors are the target genes of plant pathogens to mediate the different olfactory behavior of non-infected and infected vectors [46][47][48][49][50]. Therefore, we hypothesize that the OBP genes or Orco gene of N. virescens may be the target genes of RDV that influence the olfactory perception of the two VOCs and the behavioral discrimination by N. virescens, and these complex interactions need further investigation.
In summary, we show that EBC attracts non-viruliferous N. virescens resulting in their preference for RDV-infected rice plants over RDV-free rice plants, and 2-heptanol repels viruliferous N. virescens, resulting in their preference for RDV-free rice plants rather RDV-infected rice plants. Importantly, we show, for the first time, that virus-induced VOCs contribute to virus acquisition and transmission by insect vectors. However, these investigations are all conducted in laboratory conditions, and whether the two volatiles would be used in controlling this devastating disease of rice plants need serious and rigorous field experiments to test.

Insects and Rice Plants
The colony of N. virescens was originally collected from rice fields in Xundian, China, and was maintained on susceptible rice seedlings (Taichung Native1, TN1) in nylon cages (80-mesh, 45 cm 3 ) for several generations. A population of non-viruliferous or viruliferous N. virescens was established and maintained as previously described [8]. The population of non-viruliferous and viruliferous N. virescens was reared on TN1 (RDV-free) rice plants and RDV-TN1 (RDV-infected) rice plants respectively in a climate chamber at 26 ± 1 • C, 70 ± 5% relative humidity, under a regime of 14 h: 10 h (light: dark).
The wild-type (WT) (Minghui63) and oscas1 mutant (suppression of EBC synthase OsCAS via CRISPR-Cas9 system) [26] rice lines were used. RDV-infected rice plants were obtained and grown hydroponically in the greenhouse under natural lighting at a temperature of 26 ± 1 • C as previously described [8]

RDV Detection by RT-PCR
Total RNA was individually extracted from rice leaves or insects using Trizol reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer's procedure. cDNA was synthesized with 1 µg RNA using TransScript One-Step gDNA Removal and cDNA Synthesis SuperMix (Transgen, Beijing, China). The primers were designed based on the S8 fragment of RDV [51]. Forward primer, 5 -ATAGCTGGCGTTACGGCTAC-3 ; reverse primer, 5 -AAACCGTCCACCTGACTACG-3 . RT-PCR was performed in a 20 µL reaction containing 10 µL 2 × TransTaq HiFi PCR Super Mix (Transgen, Beijing, China), 2 µL cDNA template, 1 µL forward primer, 1 µL reverse primer, and 6 µL sterile H 2 O. The PCR was conducted with the following procedure: 94 • C for 3 min, 35 cycles of 94 • C for 30 s, 55 • C for 30 s, 72 • C for 2 min, and 72 • C for 10 min. The RDV infection status of rice plants or insects was confirmed by the RT-PCR results.

Studies on N. Virescens Feeding and Plant Odor Preferences
To understand the role of two VOCs ((E)-β-caryophyllene, EBC and 2-heptanol) in non-viruliferous and viruliferous N. virescens, for the first, the feeding and plant odor preference of N. virescens between WT (RDV-free) and WT-RDV (RDV-infected) rice plants needs to be definite. One WT and one WT-RDV rice plant were respectively planted into the device (containing Kimura B nutrient solution [52]) as previously described [8]. The device was a transparent polyethylene cylindrical cage (D = 18 cm, H = 50 cm) containing a plastic pot. Each cage contained two nylon mesh plugs (D = 5 cm) through which insects were transferred and to promote plant respiration ( Figure S5). Fifteen female adults (<24 h old) were starved for 4 h and then placed into each device, respectively. The number of non-viruliferous or viruliferous insects on each plant was recorded at 2, 4, 8, 24, 48, and 72 h post-inoculation (PI). Fifteen biological replicates were run for each test group.
A four-quadrant olfactometer with four arenas (Camsonar SIM-4, Camsonar Group Limited, UK) was used to determine the effects of host plant odors on the olfactory behavior of both non-viruliferous and viruliferous N. virescens, as previously described [26]. Odor materials (WT, RDV-free rice plants; WT-RDV, RDV-infected rice plants; clean air) were individually transferred into a glass bottle (D = 10 cm, H = 60 cm). All devices were connected using odorless PVC tubes (TYGON, LMT−55, Tokyo, Japan). One non-viruliferous or viruliferous female adult (starved for 4 h) was placed into the olfactometer and after 5 min, the location and activity of the insect were recorded for 20 min ( Figure S6). Data were analyzed using image-processing software (EthoVision XT 14, Noldus Information Technology, Wageningen, The Netherlands). A heat map was used to visualize the tracks made by the N. virescens in the four-odor area over the 20 min period (the same as below). If the residence time of the adult in the olfactometer was <95% of the total test time, the data were excluded. Forty-five biological replicates were carried out for each experiment.

The Effects of Individual VOCs on the Olfactory Behavior of N. Virescens
The effects of individual VOCs on the olfactory behavior of vectors were determined using a four-quadrant olfactometer (as described above). EBC (CAS: 87−44−5) and 2heptanol (2-hep) (CAS: 543−49−7) were purchased from Sigma-Aldrich. The concentrations of the two VOCs used for testing (0.01 µg µL −1 , 0.1 µg µL −1 , 1 µg µL −1 ) were as previously described [26]. The two VOCs were dissolved in paraffin oil. Then, 10 mL solution (VOCs in paraffin oil) was placed into an odor source bottle, and 10 mL paraffin oil was placed into the other three glass bottles individually. Forty-five biological replicates for each concentration of each VOC were carried out.

Validating the Role of VOCs in N. Virescens with Feeding and Plant Odor Preference Studies
RDV infection could not induce more EBC to be released from oscas1-RDV rice plants [26]. Thus, the oscas1 rice plants were used to further validate the attraction of EBC to non-viruliferous N. virescens. The feeding and plant odor preference studies were carried out as described above. The tested combinations of rice plants were (1) WT vs. oscas1; (2) oscas1 vs. oscas1-RDV; (3) WT-RDV vs. oscas1-RDV.
To verify the repellent effects of 2-heptanol on viruliferous N. virescens with feeding and plant odor preference studies, the tested combination was WT rice plant (added synthetic 2-heptanol (8.17 µg) in 10 µL of lanolin paste) vs. WT rice plant (added 10 µL of pure lanolin paste on rice leaves) [26].

The RDV Acquisition Rate by Non-Viruliferous N. Virescens and RDV Transmission Rate by Viruliferous N. Virescens
The testing device for RDV acquisition experiments was the same as for the feeding preference study above. The RDV acquisition rate by non-viruliferous N. virescens was compared using two combinations of rice plants: (1) WT-RDV vs. WT and (2) oscas1 vs. oscas1-RDV. Five biological replicates using three plants of each type and forty-five insects were carried out. Whether these insects carried RDV or not was confirmed as per the previous method [8]. The RDV acquisition rate was calculated from the RT-PCR results as above.
The RDV transmission rate by viruliferous N. virescens was determined using a combination of rice seedlings of WT + 2-heptanol vs. WT rice plant. For each group, 48 rice plants were used ( Figure S7). There were five biological replicates and each replicate contained forty-five insects. After 72 h, these insects were removed, and after a further 25 days, the RDV infection status of the rice plants was confirmed as above and the RDV transmission rate was calculated.

Statistical Analysis
All data were performed using SPSS 20.0 software. Feeding preference data were analyzed by Wilcoxon's signed-ranks tests [53]. The residence time of N. virescens in the four quadrants of the olfactometer, containing plant odors and individual VOCs, were analyzed using a non-parametric test (Friedman-ANOVA, p < 0.05), and the differences attributed to four fields by Wilcoxon-Wilcox as a post hoc test [54]. The RDV acquisition rate by non-viruliferous N. virescens and the RDV transmission rate by viruliferous N. virescens were analyzed using the Student's t-test.