The mealybug Phenacoccus solenopsis suppresses plant defense responses by manipulating JA-SA crosstalk

Induced plant defenses against herbivores are modulated by jasmonic acid-, salicylic acid-, and ethylene-signaling pathways. Although there is evidence that some pathogens suppress plant defenses by interfering with the crosstalk between different signaling pathways, such evidence is scarce for herbivores. Here, we demonstrate that the mealybug Phenacoccus solenopsis suppresses the induced defenses in tomato. We found that exogenous JA, but not SA, significantly decreased mealybug feeding time and reduced nymphal performance. In addition, constitutive activation of JA signaling in 35s::prosys plants reduced mealybug survival. These data indicate that the JA signaling pathway plays a key role in mediating the defense responses against P. solenopsis. We also found that mealybug feeding decreased JA production and JA-dependent defense gene expression, but increased SA accumulation and SA-dependent gene expression. In SA-deficient plants, mealybug feeding did not suppress but activated JA accumulation, indicating that the suppression of JA-regulated defenses depends on the SA signaling pathway. Mealybugs benefit from suppression of JA-regulated defenses by exhibiting enhanced nymphal performance. These findings confirm that P. solenopsis manipulates plants for its own benefits by modulating the JA-SA crosstalk and thereby suppressing induced defenses.

nymphal performance on cotton 21 . However, the underlying mechanism involved in the suppression of JA defense responses by P. solenopsis remains unknown. Considering that P. solenopsis may also activate the induction of SA-dependent responses and show enhanced performance on SA-treated plants 21 , we hypothesized that suppression of JA defense by P. solenopsis might be mediated by crosstalk with the SA signaling pathway. To test this hypothesis, we used wild-type tomato that is a host plant for P. solenopsis and two transgenic tomato lines including NahG with defective SA biosynthesis and 35S::prosys with constitutive JA signaling. First, we examined the effects of exogenous JA, SA, and herbivory treatments on the performance of P. solenopsis larvae. Second, we used the electronic penetration graph (EPG) technique to record the P. solenopsis feeding behavior on undamaged plants and plants treated with JA, SA, and herbivory. Third, we quantified the accumulation of endogenous JA and SA as well as the transcript levels of JA-and SA-dependent genes in host plants in response to P. solenopsis feeding. Finally, we examined the survival rate of P. solenopsis on NahG, 35S::prosys, and their corresponding wild-type plants, and quantified JA and SA in the two transgenic lines in response to P. solenopsis feeding. Our results demonstrate that P. solenopsis feeding enhances SA accumulation, which suppresses the JA signaling pathway. As a consequence of this herbivore modification of JA-SA crosstalk, the performance of P. solenopsis nymphs is enhanced.

Methods
Plants and insects. Wild-type tomato (Lycopersicon esculentum) cv Moneymaker (MM) is the parental line for the SA-non-accumulating NahG mutant. Wild-type tomato (Solanum lycopersicum) cv Castlemart (CM) is the parental line for the transgenic tomato line 35s::prosys in which JA signaling is constitutive. 35S::prosys seeds were collected from a 35S::prosys homozygote that had been backcrossed five times to its wild-type line cv Castlemart 22,34 . Tomato seedlings were grown in 500-ml pots containing a commercial potting mix (Fafard Growing Mix 1, Agawam, MA), and were kept in an insect-free greenhouse compartment under natural light and 28/ 24uC. Plants with four to five fully expanded leaves were used for experiments.

Plant treatments
(1) Chemical treatment: Jasmonic acid or salicylic acid (Sigma-Aldrich) was dissolved in 0.5 mL of acetone and dispersed in water (containing 0.1% Tween 20) to produce a 1.0 mM JA or SA solution. Each plant was sprayed with 1.0 mL/leaf of the JA or SA solution with a hand-sprayer. Twenty-four hours later, JA-or SA-treated plants were used for experiments. (2) Mealybug treatment: A mixture of third-instar nymphs and newly emerged adults (40 in total) of P. solenopsis were carefully transferred onto each plant and allowed to feed freely for 6, 12, 24, 72, or 120 h. After that, leaf samples were collected for phytohormone analysis. Plants pre-infested with 40 P. solenopsis adults (2-to 4-days old) for 120 h were used for performance bioassays, the EPG experiment, and gene-expression analysis. (3) Control treatment: In a preliminary experiment, we found that treatment with a water solution (including 0.5 mL of acetone and 0.1% Tween 20) had no effect on the performance of P. solenopsis reared on tomato plants. A similar experiment was performed to determine the effects of control, JA, SA, and herbivory treatments on P. solenopsis developmental duration (from egg to last molt). The procedure was the same as that described in the previous paragraph except that developmental duration was recorded twice daily until adults emerged, and each treatment was represented by five replicate plants.
EPG recording of P. solenopsis feeding behavior on tomato plants. P. solenopsis stylet penetration activities on the leaves of control plants or plants treated with JA, SA, or herbivory (mealybugs) were recorded using a four-channel DC-EPG system (Giga-4; Wageningen, The Netherlands). The method used for recording was the same as that described by Huang et al. (2014) 23 . In brief, 7-to 8-day-old adult females were carefully collected from tomato plants. The dorsal wax was partially and gently removed from each specimen with a fine brush to enable the attachment of a gold wire electrode (18 mm diameter, 2 cm long). Experiments were conducted in a Faraday cage in the laboratory at 27 6 2uC. The recordings were started around 08:30 hours and continued for 12 h under florescent light. For each treatment, 20 recordings, each done with a separate mealybug on a separate plant, were used for characterization and analysis of the EPG signals using the EPG analysis worksheet.
Quantification of endogenous JA and SA. Endogenous JA and SA in tomato plants were quantified as described by Wei et al. (2013) 24 and Almeida-Trapp et al. (2014) 25 . In brief, plant material (250-300 mg) was frozen and ground in liquid nitrogen. The resulting powder was mixed with 4 mL of HPLC grade methanol (Sigma-Aldrich) and kept at 220uC for 12 h. For quantification, [9,10]-dihydro-JA (300 ng) and d6-SA (500 ng) were added as internal standards. JA, SA, and their internal standards were partitioned to an aqueous phase by centrifugation and vaporization. After three rounds of freezing and thawing, the aqueous phase was centrifuged, and the pH of the supernatant was adjusted to 3. Quantitative real-time PCR. Total RNA was extracted and purified as described by   19 . First stand cDNA was synthesized from 200 ng RNA using a First-Strand cDNA Synthesis Kit (Bio-Rad, Hercules, CA, USA) according to the manufacturer's instructions. To quantify LoxA, Pin2, PR1, and GluA transcript levels in different samples, real-time quantitative RT-PCR was performed. The real-time PCR was carried out on an ABI 7500 Real Time PCR System (Applied Biosystems, Foster City, CA) with a 96-well rotor. The amplification reactions were performed in a 20-mL final volume containing 10 ml of iQ TM SYBRH supermix (BioRad, Hercules, CA), 0.8 mL of forward primer (5 mM) and reverse primer (5 mM) pairs, and 2 mL of cDNA first-strand template. Thermal cycling conditions were 5 min at 95uC followed by 40 cycles of 15 s at 95uC and 15 s at 58uC. The gene-specific primers, which were designed and checked as described previously 29 , were as follows: for LoxA (U09026; F: 59-TGGTAGACCACCAACACGAA-39, R: 59-GACCAAAACGCTCGTCTCTC-39); for Pin-2 (AY129402; F: 59-TGATGCCAAGGCTTGTACTAGAGA-39, R: 59-AGCGGACTTCCTTCTGAACGT-39); for PR-1a (M69247; F: 59-GAGGGCA-GCCGTGCAA-39, R: 59-CACATTTTTCCACCAACACATTG-39); for GluA (M80604; F: 59-TCA GCAGGGTTGCAAAATCA-3, R: 59-CTCTAGGTGGGTAG-GTGTTGGTTAA-39); and for GAPDH (U93208; F: 59-CTCCATCACAGCCAC-TCAGA-39, R: 59-TTCCACCTCTCCAATCCTTG-39). All reactions were run in duplicate, and average values were used in the analyses. Normalized gene expression was calculated using the 2 2DCt method with GAPDH as the endogenous control gene, and values were subsequently log 2 transformed.
Statistical analysis. Fisher's protected least significant difference (PLSD) test of ANOVA was used to analyze phytohormone data and the data for development time and survival rate. The gene expression data were statistically analyzed by one-way ANOVA. Proportional data were arcsine square root transformed for analysis and back-transformed to percentages for presentation. EPG data were transformed where appropriate (square-root transformation for number of occurrences and natural log transformation for duration) and analyzed by one-way ANOVA.

Results
Performance of P. solenopsis on tomato plants. After 14 d of infestation, the survival rate of P. solenopsis nymphs on control plants was 78.3 6 3.3%, and survival on JA-, SA-treated, or P. solenopsis-infested plants did not significantly differ from survival on control plants (F 3, 12 5 1.14, P 5 0.37; Fig. 1A). In contrast, survival at 21 d significantly differed among the four treatments (  Feeding behavior of P. solenopsis as indicated by EPG. EPG detects the electrical fluctuations caused by probing behavior of mealybugs during foraging and feeding 26,27 . Three main phases during stylet penetration have been defined: the pathway phase, the xylem phase, and the phloem (or sieve element) phase 28 . Given that the probing behavior of P. solenopsis has seldom been recorded in the xylem phase 23,27 , we mainly analyzed the occurrence of cell puncture (as indicated by the rate at which potential dropped) in the pathway phase and the duration and insertion rate in the phloem (or sieve element) phase. Based on the EPG data, the mealybug treatment did not affect the feeding behavior of P. solenopsis adults ( Table 1). The JA treatment significantly reduced the total feeding time in the phloem phase but did not affect the feeding rate in the phloem phase, the number of probes, or the rate of cell puncture ( Table 1). The SA treatment significantly increased the rates of feeding in the phloem phase and cell puncture in the pathway phase, but did not affect the total feeding time in the phloem phase or the number of probes ( Table 1).
Quantification of endogenous JA and SA in host leaves. The amount of JA was significantly lower in P. solenopsis-infested leaves than in control leaves at 12 h after infestation (P 5 0.008) but not at 6, 24, 72, or 120 h after infestation (6 h: P 5 0.88; 24 h: P 5 0.08; 72 h: P 5 0.42; 120 h: P 5 0.08) ( Fig. 2A). This is consistent with previous findings that the effect of herbivory on endogenous JA level generally occurs within 12 h after infestation 8,29 .
To test whether SA signaling plays a key role in mediating the suppression of JA defensive responses by P. solenopsis, we further examined the JA and SA levels in NahG (SA-deficient) and 35s::prosys (JA-overexpression) plants infested with P. solenopsis. For NahG plants, the amount of JA was significantly higher in infested leaves than in control leaves at 12, 24, and 72 h after infestation (12 h: P 5 0.004; 24 h: P 5 0.002; 72 h: P 5 0.001) (Fig. 3A). In contrast, the amounts of SA in control NahG plants were consistently low (,9 ng g 21 fresh weight), and P. solenopsis infestation of NahG plants did not increase SA amounts at 12, 24, and 72 h after infestation (12 h: P 5 0.12; 24 h: P 5 0.20; 72 h: P 5 0.91) (Fig. 3B).
For 35s::prosys plants, the amounts of JA were significantly higher in infested leaves than in control leaves at 72 and 120 h after infestation (72 h: P 5 0.04; 120 h: P 5 0.015) (Fig. 4A). Similarly, the amounts of SA were significantly higher in infested leaves than in control leaves at 72 and 120 h after infestation (72 h: P 5 0.001; 120 h: P 5 0.001) (Fig. 4B).
Changes in gene-expression in response to P. solenopsis feeding. To further investigate the effects of P. solenopsis feeding on JA-or SA-dependent defense responses, we examined the transcript levels of two JA-dependent genes (LoxA and Pin-2) and SA-dependent  genes (PR-1a and GluA). Lipoxygenase (LOX) is a key enzyme in the biosynthesis of JA along the octadecanoid pathway 30 . Proteinase inhibitor II (Pin-2), a serine proteinase inhibitor with trypsin and chymotrypsin inhibitory activities, is known to confer insect resistance in many Solanaceae plants 31 . In tomato, LoxA and Pin-2 are marker genes of the JA signaling pathway 33 . PR-1a and GluA are known as two pathogenesis-related genes, which are mainly regulated by the SA signaling pathway 32,33 . Mealybug feeding significantly suppressed the expression of Pin-2 (F 1, 5 5 8.47, P 5 0.04) but not of LoxA (Fig. 5A). In contrast, mealybug feeding significantly induced the expression of PR-1a and GluA (PR-1a: F 1, 5 5 9.17, P 5 0.04; GluA: F 1, 5 5 12.32, P 5 0.02) (Fig. 5B).
Performance of P. solenopsis on tomato mutants. Survival of P. solenopsis were significantly lower on SA-deficient NahG plants than on wild-type MM plants at 14 (Fig. 6B).

Discussion
The results of the present study show that application of JA but not SA significantly decreases the feeding time of the mealybug P. solenopsis in the phloem tissue and consequently decreases nymphal performance. In addition, constitutive expression of JA signaling in 35s::prosys plants reduces P. solenopsis survival. These data demonstrate that the JA signaling pathway plays a key role in mediating the defense responses against P. solenopsis in tomato. This result is consistent with a previous finding that the JA-dependent defense pathway may be involved in basal defense against P. solenopsis in cotton 21     indicated by the accumulation of JA in infested SA-deficient NahG plants and significantly higher levels of SA in infested WT plants. Moreover, P. solenopsis survival is better on control WT plants than on SA-deficient NahG plants and on 35s::prosys plants that constitutively express JA, which suggests that manipulation of the pathways by nymphs allows them to perform better. Taken together, our results indicate that P. solenopsis can manipulate plants for its own benefit by modulating JA-SA crosstalk in order to suppress induced defenses. The JA signaling pathway, including the wound hormones JA and JA-Ile, is widely recognized as a key regulator of plant defense against insect herbivores 1,34,35 . Several recent studies have found that insects can suppress JA-mediated defense responses to enhance their performance on food plant 11,12,17,19 . These suppression mechanisms often involve some herbivore effectors, including specific enzymes from oral secretions 36 , egg-derived elicitor 12 , and vectored microorganisms 14,37 . For example, caterpillars of the beet armyworm Spodoptera exigua can suppress the accumulation of JA and the induction of JA-dependent genes in Arabidopsis by releasing effectors in oral secretions 38 . Glucose oxidase in the saliva of Helicoverpa zea caterpillars was the first effector identified that reduces JA-regulated nicotine production in Nicotiana tabacum 36 . Similarly, phloem-feeding insects like aphids and whiteflies can inhibit the JA defense responses in different plant species although the effectors involved have not been identified 11,39,40 . However, the mechanism by which plants respond to herbivore effectors and eventually mediate the suppression of JA signaling has not been extensively studied.
Considering that pathogens can reduce plant defenses by manipulating antagonistic crosstalk between JA and SA signaling pathways 41 , researchers have speculated that insects can circumvent plant defenses in a similar way 18 . This possibility has been supported by several studies. For instance, oviposition by the cabbage butterfly Pieris brassicae induces SA accumulation and reduces the induction of JA-regulated defense genes in Arabidopsis and leads to the enhanced performance of Spodoptera littoralis; that the suppression of JA-dependent genes and enhancement of S. littoralis performance were not observed in the SA-deficient mutant sid2-1 indicates that SA mediates this phenomenon 12 . In tomato, larvae of Colorado potato beetle, Leptinotarsa decemlineata, use bacteria in their oral secretions to decrease JA accumulation, decrease JA-responsive antiherbivore defenses, and increase SA production; these effects were not observed, however, in SA-deficient plants 13 . In the present study, we found that the expression level of the JA-dependent gene Pin-2 and JA production were significantly lower in P. solenopsis-infested plants than in non-infested plants, which indicates that, unlike herbivores in the two above cases, P. solenopsis does not prevent induction of JA-regulated defenses but reduces the constitutive level of JA defenses. This is consistent with the previous finding that P. solenopsis feeding strongly reduces the constitutive level of JA-regulated gossypol in cotton 21 . Our data further showed that if SA signaling is blocked, P. solenopsis does activate JA accumulation. Similar phenomena were also observed in the interaction between the whitefly Bemisia tabaci and Arabidopsis. The expression level of the JA-regulated defense genes PDF1.2 and VSP1 is significantly lower in B. tabaci-infested plants than in non-infested plants, indicating that B. tabaci reduces the constitutive levels of defense genes; however, once SA synthesis or its action is impaired, B. tabaci feeding does activate the induction of the two genes 11,19 . Considering that chewing insects cause extensive tissue damage whereas phloem-feeding insects cause minimal tissue damage 18 , the above cases indicated that the mode of mechanical damage caused by different herbivores possibly might be an additional factor that influences the strength of the reciprocal antagonism between JA and SA pathways. Collectively, these data indicate that herbivore modulation of the antagonistic crosstalk between the JA and SA signaling pathways might be more intricate than previously thought 10 .   Our results showed that P. solenopsis feeding did not affect the expression level of LoxA, which encodes a key enzyme involved in JA biosynthesis 30 , but decreased the expression level of JA-dependent Pin-2 and endogenous JA level. Based on these data we speculated that the suppression of JA defense responses by P. solenopsis is likely located downstream of JA biosynthesis pathway. This speculation is supported by the observation that B. tabaci nymphs induce the upstream JA-responsive genes (LOX2 and OPR3), but suppress the downstream JA-responsive gene (VPS1 and PDF1.2) in Arabidopsis 11,19 , indicating that nymph feeding can suppress the downstream components of the JA signaling pathway. Future experiments should attempt to explore the target genes manipulated by phloem-feeding insects in the JA signaling pathway.
In addition to the role of SA in the suppression of JA defense responses, the consistent accumulation of SA induced by P. solenopsis feeding might benefit the development of larvae. One reason is that exogenous SA treatment strongly increases the feeding rate of P. solenopsis in the phloem tissue (Table 1), which might directly contribute to rapid larval development (Fig 1B). Alternatively, given that SA pathway plays a central role in plant defense against microbial and fungal pathogens 6 , the accumulation of SA near the feeding site might protect the leaf tissue from infection and ensure that the tissue retains its full nutrient content (i.e., is not degraded) during larval feeding.
Although considerable SA accumulated in response to P. solenopsis infestation, JA accumulation was also strongly induced by P. solenopsis infestation of the 35s::prosys plant. Because JA signaling is constitutively enhanced in the 35s::prosys plant 22 , we speculate that if the JA signaling is activated before SA signaling, the tomato plant might become insensitive to SA-mediated suppression of JA signaling by mealybug infestation. Our speculation is supported by a previous study in which simultaneous activation of the JA and ET signaling before induction of the SA response rendered Arabidopsis plants insensitive to SA-mediated suppression of JA-responsive gene expression 42 . These findings support the view that the sequence of hormone activation affects realization of the antagonism between JA and SA crosstalk 7,43 . Alternatively, we cannot rule out the possibility that the enhanced SA level induced by mealybug feeding is insufficient to suppress JA accumulation in the 35s::prosys plant.
The mealybug P. solenopsis has been rapidly spreading throughout South China, where it is causing serious economic losses to cotton production 20 . Our data confirm that, at least in the laboratory, P. solenopsis can circumvent plant defenses by modulating the JA-SA crosstalk. This capability has possibly facilitated the rapid invasion of mealybugs in China and elsewhere 21 . Future studies should determine whether the modulation of the JA-SA antagonism by P. solenopsis occurs in the field, and if so, whether it enhances P. solenopsis population development.