Manipulation of Plant Defense Responses by the Tomato Psyllid (Bactericerca cockerelli) and Its Associated Endosymbiont Candidatus Liberibacter Psyllaurous

Some plant pathogens form obligate relationships with their insect vector and are vertically transmitted via eggs analogous to insect endosymbionts. Whether insect endosymbionts manipulate plant defenses to benefit their insect host remains unclear. The tomato psyllid, Bactericerca cockerelli (Sulc), vectors the endosymbiont “Candidatus Liberibacter psyllaurous” (Lps) during feeding on tomato (Solanum lycopersicum L.). Lps titer in psyllids varied relative to the psyllid developmental stage with younger psyllids harboring smaller Lps populations compared to older psyllids. In the present study, feeding by different life stages of B. cockerelli infected with Lps, resulted in distinct tomato transcript profiles. Feeding by young psyllid nymphs, with lower Lps levels, induced tomato genes regulated by jasmonic acid (JA) and salicylic acid (SA) (Allene oxide synthase, Proteinase inhibitor 2, Phenylalanine ammonia-lyase 5, Pathogenesis-related protein 1) compared to feeding by older nymphs and adults, where higher Lps titers were found. In addition, inoculation of Lps without insect hosts suppressed accumulation of these defense transcripts. Collectively, these data suggest that the endosymbiont-like pathogen Lps manipulates plant signaling and defensive responses to benefit themselves and the success of their obligate insect vector on their host plant.


Introduction
Most bacterial plant pathogens are not vectored by insects [1,2]. However, the few that are, require insect partners as intermediate hosts [1]. The vast majority of gram-negative bacterial plant pathogens appear to be dependent primarily on one insect vector, are vertically transmitted from parent to offspring, and are confined only to plant tissue in which their insect host feeds [1,[3][4][5][6][7][8]. Given that these pathogens have an obligate relationship with a specific insect vector and are vertically transmitted, these bacteria could be perceived as both an insect endosymbiont and a plant pathogen. To date, the impact of these gram-negative endosymbiont-like plant pathogens on the insect vector-plant interaction has not been studied.
Plants protect themselves against the diversity of herbivores and microbial pathogens by expressing an array of constitutive and induced defenses rendering the plant an inaccessible or unsuitable food source. The perception of attack and deployment of the induced defenses is primarily mediated by three well-studied defense-signaling pathways that are regulated by jasmonic acid (JA), salicylic acid (SA) and ethylene (ET) [9][10][11]. Herbivores and pathogens introduce a distinct set of elicitors and effectors that are perceived by host plants and these signals allow the plant to tailor its defense response to individual challengers [9][10][11][12][13][14].
Therefore, the nature of defenses elicited by endosymbiont-like pathogens in their host plants have the potential to profoundly impact the plant's interaction with the insect and/or ability to resist attacks by other pathogens or pests. If an herbivore can circumvent induced plant defenses or plant recognition by vectoring its endosymbiont associate into its host plant during feeding, it may have a selective advantage relative to insects feeding on uninfected plants. Alternatively, effectors from the endosymbiont may circumvent the plant recognition system, compromising plant immune responses and related insect and bacterial resistance in both the JA/ET-and SA-regulated defense pathways. Thus, the endosymbiont's modification of plant defenses could result in a more susceptible host plant for both symbiotic partners.
This study investigates whether the psyllid influences plant defense responses by vectoring Lps into tomato plants during feeding. This was first investigated by determining the genomewide changes in transcript abundance of tomato in response to Lps-infected B. cockerelli adults and instars. Second, to identify the effect of Lps on induced plant defenses, changes in levels of tomato defense transcripts regulated by the JA/ET and SA were determined after feeding by Lps-infected B. cockerelli and Lps infection in the absence of its vector. During plant defense trials we also measured Lps infection frequency and titer among different psyllid lifestages and frequency of vertical transmission via eggs to more fully characterize the microbe-insect-plant interaction.

Results
Genome-wide transcript profiles of tomato after continuous feeding by developing Lps-infected psyllids To understand the changes in tomato mRNA profiles after Lpsinfected B. cockerelli infestation, potato cDNA arrays were used in a reference design strategy [37]. Potato arrays were used because gene content, genome organization and nucleotide sequence conservation are similar among Solanaceous plant genomes [34].
Arrays were hybridized to cDNAs from tomato leaves after continuous feeding by 1 st and 2 nd instars (10 days), 3 rd , 4 th and 5 th instars (15 days), adult psyllids/egg deposition (2 days), and a nopsyllid control. Consistent with microarray data generated from aphid-infested plants [35,36], large changes in transcript abundance after psyllid infestation were not found and variation was high, resulting in no statistical differences in transcript levels among treatments (criteria: P,0.05 and a fold change of greater than 2). Despite the lack of significant differences, the top 149 expressed transcripts in each treatment are presented in Tables S1, S2, and S3. Non-synchronous development of psyllid infestations may have contributed to variation. For example, although each time point was dominated by insects at a particular stage in development, each infestation actually included eggs, developing instars as well as feeding adults. These mixed, infestations may have induced high variability due to cumulative life-stage effects on tomato mRNAs making life-stage specific effects difficult to resolve.
In order to examine qualitative transcript changes from our microarray data our statistical criteria was relaxed (P,0.1 and no fold change limit). Significant differences in 24 transcripts were evident at a P,0.1. The greatest numbers of significant transcripts were regulated by 3 rd -5 th instar feeding. Only 5 transcripts were up-regulated significantly (P,0.1), with putative functions in signaling, cell wall synthesis and protein synthesis (Table 1), while  9 transcripts were down-regulated, most with unknown function  (Table 1). Feeding by 1 st -2 nd instars down-regulated 2 transcripts significantly, a hypothetical and an oligosaccharide processing protein ( Table 1). The largest changes in transcript abundance were after adult feeding/egg deposition. Adults down-regulated transcripts related to signaling, cell wall synthesis/cell division and up-regulated mRNAs related to post-translational regulation (Table 1). Since microarray data could only be interpreted qualitatively the general trend of differential plant expression due to psyllid lifestage helped guide our sample design in the experiments below; the effect of discrete psyllid lifestages on tomato expression.

Tomato responses to graft-transmitted Lps
To monitor defense-response gene mRNAs among different psyllid lifestages qRT-PCR was performed. By using primers specific to tomato defense genes, more subtle changes in mRNA levels that were not detected using the heterologous potato arrays and the continuous feeding-design (above) were revealed. To this end, we monitored the defense gene transcripts encoding allene oxide synthase (AOS), proteinase inhibitor 2 (Pin2), phenylalanine lyase 5 (PAL5), and pathogenesis-related 1 (PR1)]. AOS is important for the synthesis of the defense hormone JA and Pin2, a defense transcript in the JA pathway, is known to inhibit Ser proteases critical for digestion of foliar proteins [10,37]. PAL is a key regulatory enzyme for the synthesis of phenolic compounds including SA. PR1 mRNAs accumulate in response to increased SA levels and was utilized as a SA-responsive marker transcript. Changes in the levels of defense gene mRNAs were determined in fully expanded tomato leaves after Lps graft infection and in response to Lps-infected psyllids. Since Lps-free psyllids were not available and could not be cured of infection in this study (discussed below), it is not possible to directly assess gene expression changes to Lps-free psyllids. However, inoculation of tomato plants with Lps by grafting allowed the impact of Lps infection without its vector to be determined.
Lps-infected or Lps-free scions were grafted to Lps-free root stocks with three mature leaves. Graft transmission of Lps was rapid ( Table 2). Within 24 hr after Lps inoculation via grafts, Lps was detected in the leaves of the grafted root stock (Table 2), whereas no Lps was detected in the plants using non-infected scions. Defense gene transcripts were measured in the root stock leaves from plants with Lps-infected and Lps-free (graft control) grafts. The graft control plants were used to account for any wound responses that occurred within the first 24 hr after grafting. Relative to the graft control plants, AOS and PR1 mRNA levels declined ( Fig. 1; P,0.05). While a decrease in PAL and Pin2 transcripts were noted, it was not significantly different from controls ( Fig. 1). Overall, Lps decreases transcript accumulation regulated by both JA-and SA-signaling.

Tomato responses to Lps in association with its psyllid vector
To assess the changes in tomato defense gene expression after vector feeding, individual leaves of tomato plants were infested for 24 hrs with 25 infected psyllid adults or nymphs in their 1 st -2 nd , 3 rd or 5 th instars. First, in regard to JA signaling, AOS mRNA abundance was influenced by the developmental stage of B. cockerelli that was feeding on the plant ( Fig. 2A; P = 0.0299); 1 st -2 nd instar feeding caused a significant increase in AOS mRNA levels relative to uninfested control plants ( Fig. 2A; P,0.05, Pairwise comparison LSD). In contrast, AOS transcripts were not significantly up-regulated after feeding by adults, 3 rd or 5 th instars relative to control plants ( Fig. 2A; P.0.05, Pairwise comparison LSD).
Similar to AOS, the JA defense transcript Pin2 was strongly influenced by the psyllid's developmental stage ( Fig. 2B; P,0.0001). Feeding by 1 st instars caused a significant ,2506 increase in Pin2 transcripts relative to uninfested leaves ( Fig. 2B; P,0.05, Pairwise comparison LSD). In contrast, relative to 1 st instar feeding, Pin2 mRNAs levels were more than ,5 fold lower in the 3 rd and 5 th instars and adult treatments; Pin2 transcript levels were not statistically different between 3 rd , 5 th instars, and psyllid adults ( Fig. 2B; P.0.05, Pairwise comparison LSD).
PAL5, a key biosynthesis enzyme in the SA pathway, is complexly regulated and its mRNAs accumulate in response to  feeding by some hemipterans and the methyl form of JA (MeJA), and is negatively regulated by SA [16]. Psyllid developmental stage had a significant impact on PAL5 transcript levels ( Fig. 3A; P = 0.0015). Compared to uninfested leaves, there was an increase in PAL5 mRNA levels after 1 st -2 nd and 3 rd instar feeding ( Fig. 3A; P,0.05, Pairwise comparison LSD). In contrast, 5 th instars and adult psyllids caused a significant reduction in PAL5 transcript levels relative to undamaged controls and 1 st instar feeding ( Fig. 3A; P,0.05, Pairwise comparison LSD). PR1 mRNAs are known to accumulate after biotic stress, SA and MeJA induction [16,38,39]. Abundance of PR1 depended on the psyllid life stage feeding ( Fig. 3B; P = 0.017). Within 24 hr of feeding, tomato plants infested with 1 st -2 n instars accumulated the highest levels of PR1 mRNA relative to other life stages of insect and undamaged controls ( Fig. 3B; P,0.05, Pairwise comparison LSD).

Efficiency of Lps transmission to tomato plants
Plants in 24 hour defense-response gene experiments were tested for Lps infection and disease symptoms. Lps transmission and disease symptoms were also observed after 72 hrs and 6 days of feeding to better characterize infection and disease symptom dynamics after 24 hrs of feeding for each lifestage. Lps was not detected in the uninfested control plants. Lps was successfully transmitted to tomato plants by all Lps-infected psyllid life stages   (Chi-square = 3.889, df = 3, sig. = 0.274, Table 2). Lps was detected in most psyllid-infested plants after 24 hr, 72 hr, or 6 days of feeding ( Table 2). The only insect life stage that did not transmit Lps at 100% efficiency was the adult (Table 2); adults transmitted Lps with an efficiency of 78%. In addition, the amount of time psyllids were allowed to feed on tomato plants and incubation time after graft inoculation (24 hr, 72 hr, and 6 days) had no significant effect on Lps infection success (Chi-square = 0.00, df = 2, sig. = 1.0; Table 2). Collectively these results indicate that Lps infection is rapid, occurring within 24 hr of infestation (Table 2). Infected and uninfected plants remained alive for over 72 days (the last day of observation) and Lps-infected plants developed qualitative psyllid yellows symptoms [8] within 3 weeks after infection relative to control plants.
Infection and titer of Lps in B. cockerelli throughout development Vertical transmission of Lps from infected gravid females to psyllid eggs was moderate to high and occurred in all six isofemale lines, ranging from 46.7-87.5% infection (Table 3). Since Lps is vertically transmitted from psyllid mother to egg at moderate to high frequencies (Table 3), we further assessed psyllid infection frequencies and titer throughout psyllid development for all psyllids used in 24 hr defense-response gene trials. For 72 hr and 6 day plant infection and symptom trials we also detected psyllid Lps infection frequency but not titer for different psyllid lifestages. For 24 hr defense-response gene trials Lps was detected in 96% of 1 st -2 nd and 3 rd instar nymphs, 97.4% of adult psyllids, and 100% of 5 th instar psyllids (N = 75 individuals per instar). Only eight psyllids out of the 300 used in all trials did not have detectable levels of Lps. These insects included: one 1 st -2 nd instar nymph from a 24-hr treatment, two 1 st -2 nd instar nymphs in a 6-day treatment, one 3 rd instar nymph from a 24-hr, 72-hr, and 6-day treatment, and two adults from a 24-hr treatment. Consequently the majority of psyllid individuals used in transmission trials and the 24-hr plant transcript treatments were infected with Lps, and infection frequency among all psyllid individuals was very high regardless of psyllid life stage.
Titer of Lps within psyllid individuals from the 24 hr defenseresponse trials depended on psyllid life stage; 5 th instar and adult psyllids had higher bacterium titers per insect cell relative to 1 st , 2 nd , and 3 rd instars (Fig. 4). There was a significant difference in Lps CT values among psyllid instars (F = 40.875, sig.,0.001, R 2 = 0.627, corrected total d.f. = 95, Fig. 4). Scheffe post-hoc tests showed that Lps CT values of fifth instar nymphs were not significantly different from adult stage individuals (P = 0.226). Both fifth instar and adult stage treatments were significantly lower in CT values (i.e. higher in titer) relative to 1 st and 2 nd instar nymphs and 3 rd instar nymphs (P,0.01; Fig. 4). Lps CT ratios of 1 st and 2 nd instar nymphs were not significantly different from 3 rd instar nymphs (P = 1.00). Overall Lps titer increased as the psyllid aged.

Anti-biotic treatment experiment of psyllids
In order to develop infected Liberibacter psyllaurous (Lps) and uninfected psyllid strains, techniques similar to those used to cure endosymbionts from aphids were used. The methods tested were ampicillin injection into host plants [40] and into gravid females [41]. Although these treatments were more tedious to conduct on sexual psyllids relative to clonal aphids, numerous attempts to cure Lps from psyllids using both techniques were unsuccessful.
When psyllids were reared on ampicillin-treated plants, Lps infection frequencies of 5 th instar psyllids initially declined to 50% but after three generations returned to the previous infection levels (100%). Microinjection of ampicillin into gravid psyllids caused high mortality of females relative to the buffer controls. Furthermore, the small numbers of females that survived the treatment and produced progeny were not cured of the Lps infection. The inability to recover Lps-free psyllids may be due to the fact that the psyllid line used in these studies may be codependent on Lps or that the antibiotic treatments, which are  successful in curing endosymbionts from pea aphids, were less effective in psyllids.

Discussion
Plant responses to the tomato psyllid and its endosymbiont-like pathogen Lps are complex and the data presented here suggest that Lps may influence the suite of plant defenses deployed upon psyllid infestation. When Lps is transmitted to tomatoes in the absence of its vector, there was a suppression of PR1 and AOS mRNAs relative to graph control plants. These data suggest that Lps suppresses some plant defenses regulated by JA and SA. In contrast to graft-transmitted Lps, all four defense mRNAs (PAL5, PR1, Pin2, AOS) significantly increased during the early developmental stages (1 st and 2 nd instar) of infected psyllid feeding. (Figs. 1,  2, 3). When older infected developmental lifestages were examined (3 rd and 5 th instars and adult psyllids) striking differences in tomato expression were revealed relative to 1 st and 2 nd instar treatments. Compared to young psyllid feeding s SA and JA associated transcripts were suppressed after feeding by older psyllids, similar to infected graphing treatments (Figs. 1, 2, 3). The activation of plant defense-response genes was inversely correlated with Lps relative titers in the insect vector. Psyllids in their 1 st , 2 nd and 3 rd instar supported smaller Lps populations relative to 5 th instars and adults (Fig. 4). The pattern of defense-response gene expression when Lps titer in psyllids was high emulated the response to Lps in isolation.
Collectively, these data indicate that tomatoes respond differentially to psyllids in different life stages. However, while well correlated, the titer of Lps in psyllids may only partially explain the life-stage specific variation in transcript abundance. This is suggested by the fact that Lps titers are low in 3 rd instar psyllids (Fig. 4), yet defense gene mRNAs accumulate to lower levels than after 1 st and 2 nd instar feeding ( Fig. 2A,B; Fig. 3B). This lifestage variation may be attributed to how efficient a particular psyllid instar can vector Lps into its host plant, or the elicitors or effectors that are delivered in the insect saliva or released due to plant cell damage [11,42]. Specific elicitors derived from developmental stage-specific feeding behaviors or insect secretions cannot be discounted based on our experimental design [43,44]. For example, psyllid elicitors unique to 1 st and 2 nd instar nymphs could be produced in oral secretions and/or through differences in damage compared to more mature psyllids. Future purification and characterization of elicitors/effectors involved should shed light on this relationship and may potentially broaden the list of exogenous cues recognized by the plant surveillance mechanism that trigger plant defense.
This study showed that Lps caused a substantial reduction of JA/ET-and SA-regulated plant defense transcripts relative to uninfected controls. Lps may suppress plant defenses by introducing small molecules or protein effectors to promote virulence. Pathogen-derived effectors can antagonize plant immunity through a diversity of processes, including alterations in mRNA levels [45][46][47][48]. Many of these effectors are transmitted from microbes by a Type III secretion system [49]. But similar to other Alphaproteobacteria pathogens, Liberibacter species do not possess a Type III secretion system and their associated effectors [50,51]. Nevertheless, Lps does possess a Type I secretion system and an exotoxin (a putative hemolysin protein; YP_004062617), which may be secreted by this system [51]. Lps also encodes for a FTR-1like high affinity iron transporter that is associated with plant virulence in other plant pathogens [51]. It is clear that the mechanisms of Lps evasion of plant defenses warrant further investigation.
This study has demonstrated the complexity of the tomatopsyllid-Lps association and its probable impacts on psyllid success. Like other beneficial or deleterious heritable insect endosymbionts that are facultative and specific to a certain insect phyla [52], Lps occurs sporadically among individuals in field populations [53,54] and is persistently associated with the tomato psyllid species. Nachappa et al. 2012 found that infected Lps tomato psyllids from the field displayed 1.66 lower fecundity and 1.56 lower nymphal survival on tomato relative to un-infected tomato psyllids collected from the field. Nevertheless two different psyllid nuclear and cytoplasmic genotypes were used during trials for infected versus un-infected comparisons and thus strain effects cannot be excluded as the major factors driving relative fitness effects.
Lps has an intimate and potentially long-term association with its insect host given the efficient vertical transmission of Lps to psyllid eggs (Table 3), increasing frequency of Lps infection throughout psyllid development [8,53], increasing titer of Lps during the psyllid life cycle (Fig. 4), and the phylogenetic conservation of other Liberibacter species to the psyllid phylum. Furthermore, Lps in the absence of its vector and psyllids with higher Lps titers (3 rd -5 th instars and adults) were able to suppress sentinel defense-response genes regulated by the JA-and SAsignaling pathways. This markedly contrasts to 1 st instars that have significantly lower Lps titers and strongly activate these sentinel defense response genes. Collectively, these data suggest that bacterial pathogens, like Lps, that are dependent on a specific insect vector may manipulate plant signaling and defensive responses to increase both their success and the success of their obligate insect vector on their host plant. In nature, mixed life stages of the tomato psyllid feed on a particular host plant, and consequently younger, less mobile instars may also benefit from this symbiotic interaction. Further studies using psyllids with and without symbionts on different host plants, and plants compromised in the different signaling pathways will determine whether or not Lps confers a selective advantage for psyllids.

Plant growth and insect rearing
Solanum lycopersicum L. cv. Moneymaker (Botanical Interests TM) seeds were planted in 473-mL pots with sterilized University of California (UC) soil mix three and fertilized with 1.25 mL of Osmocote Smart-Release flower and vegetable plant food fertilizerH (Scotts-Sierra Horticultural Products Company, Marysville, OH, USA) per plant. Plants were grown under greenhouse conditions (24-25uC) and metal halide grow lights (16 hr light:8 hr dark) were used to supplement natural lighting. The plants used in all experiments detailed below were one month old and at the four-leaf stage at the start of trials. One day before psyllid infestations, plants were sorted into treatment and control groups. Plants were paired by leaf developmental stage among treatment and control groups.
B. cockerelli (Texas line, TX-Lipsy) colonies were maintained as described in Hansen et al. [8]. Psyllid adults in the TX-Lipsy line maintained .99% infection with Lps and were reared on tomato plants [8]. Instars were determined as described in Hansen et al. [8]. To further understand the association between Lps and the tomato psyllid, vertical transmission of the bacterium via psyllid eggs was examined in six isofemale lines (Table 3). For trials examining Lps infection in isofemale line eggs, six 10 day old mated females reared on infected potato were each isolated in a clipcage on their own tomato host plant. After 24 hrs, gravid females were removed from plants and eggs were allowed to develop for three days on the host plant before they were sampled.
Before eggs were screened for Lps infection with PCR, the leaf and attached eggs were surface sterilized with 10% bleach (Clorox H, Oakland, CA) for five minutes and then triple rinsed with sterilized double distilled water. Eggs were then removed from the leaf using heat sterilized insect pins under a dissecting microscope. Care was taken not too severe or puncture the leaf surface during this process.

Whole plant infestations, RNA isolation, and microarray analysis
Because of limitations in technology we had at the time, we were not able to use more sensitive methods to examine global transcriptional profiles (for example RNAseq). However, trends were evident in results and we used these trends as an important justification for our sample design of the next round of trials using qPCR (next section).
Psyllid infestations of tomatoes were performed in a greenhouse (25-28uC, 16 hr light:8 hr dark). All plants were encased individually in mesh bags. For half of the plants (N = 16), twenty-five adult psyllid females were released within the mesh bags encasing the plants (infested). The remaining group of plants in mesh bags (N = 16) received no psyllids (control plants, uninfested). After two days, the adults were removed from the infested plants and eggs were allowed to hatch. Eggs did not hatch synchronously, however time points were chosen when most psyllids in an infestation were at a particular life stage. At each time point, leaflets were collected from four different infested or uninfested plants and pooled. Plants were not used for more than one collection time point. Infested leaflets were harvested at 0 days (0-d control), 2 days (adult feeding/egg deposition), 10 days (1 st -2 nd instar feeding), and 15 days (3 rd -5 th instar feeding) after infestation. Developmentally matched leaves were harvested from uninfested control plants at 0, 2, 10, and 15 days and are referred to as the uninfested control reference (day 0) and the day 2, 10 and 15 uninfested controls. Harvested leaves were flash frozen in liquid nitrogen and stored at 280uC until use. The time-course experiment was executed 3 times.
Total RNA was extracted from each tissue sample according to the protocol recommended by TIGR (The Institute for Genomic Research; Rockville, MD). Briefly, leaves were ground in liquid nitrogen and total RNA was isolated using a hot phenol protocol (http://jcvi.org/potato/sol_ma_protocols.shtml). Total RNA was treated with DNase and further purified using the SV total RNA isolation kit (Promega, Madison, WI, USA). RNA concentrations were determined spectrophotometrically and integrity was verified on a 1.2% formaldehyde agarose gel [55].
For global transcriptional profiles, the TIGR potato microarray (10 K version 3) was used containing ,10,000 annotated cDNA clones spotted as duplicates on the array. At the time the experiment was conducted (2004), cDNA microarrays were the best technology available for our budget. Detailed information about this microarray can be found at the TIGR web site (http:// jcvi.org/potato/sol_ma_microarrays.shtml). A potato array was used because Solanaceous plant genomes are similar with respect to gene content, genome organization and nucleotide sequence conservation. For example, see the transcriptional analysis of several Solanaceous plants performed by Robin Buell [56].
All steps of microarray processing to obtain raw data (cDNA production, cDNA labeling, microarray hybridization, data quantification) were carried out by the TIGR Expression Profiling Service according to published methods (http://jcvi.org/potato/ sol_ma_protocols.shtml). Briefly, hybridizations were conducted in a reference pool design. All uninfested (2, 10, 15 d) and infested (2, 10, 15 d) treatments (cDNA labeled with Cy3) were hybridized to slides with the reference (0 d) uninfested control pool (cDNA labeled with Cy5). Spot data was extracted using GENEPIX at TIGR (ver. 5.0 Pro: Axon Instruments, Union City, CA, USA). In compliance with MIAME standards our data output obtained from GENEPIX are publicly available and can be downloaded through a database maintained at the TIGR Web site (http:// www.tigr.org/tigr-scripts/sgedb/studies_SGED.pl).
Using the LIMMA ( [57]: http://bioinf.wehi.edu.au/limma/) package for the statistical software R, data sets were background corrected and normalized. Mean expression ratio from the duplicate spots per clone on each array were calculated prior to statistical analysis. To dissect true psyllid responsive transcripts from developmental changes in the plants, linear models were used with contrast matrixes, facilitating indirect comparisons between psyllid damaged tissue and developmentally matched controls. If the infested/reference ratio is greater than the control/reference ratio (a positive log2 (fold change)) value, a gene is designated as ''up regulated''. Similarly if the infested/reference ratio is less than the control/reference ratio (a negative log2 (fold change)) value, a gene will be considered down-regulated.
Linear models were implemented and FDR-adjusted P values were calculated in LIMMA to identify genes that may be differentially expressed. Because expression levels were low and variation was high, there were no statistical differences in transcript expression profiles among treatments (Criteria: P,0.05 and a fold change of greater than 2) after psyllid infestation compared to developmentally matched controls. For further analysis the top 149 transcripts expressed in each treatment are available in Table S1, S2, and S3. The details of the experiment and raw microarray data are available and can be downloaded from the TIGR Solanaceae Gene Expression Database (ftp://ftp.tigr.org/pub/data/s_tuberosum/SGED/ 078_Paine/).

Infestation of tomato with different life stages of Lpsinfected psyllids
One-month-old tomato plants were challenged with Lpsinfected psyllids of different developmental stages. For all infestations, 25 individual insects belonging to a particular instar were confined on a single leaflet per plant using a clip-cage [55] and allowed to feed for 24 hours. Four psyllid life stage inoculation treatments were evaluated: 1 st and 2 nd instar; 3 rd instar; 5 th instar; and adult. Three replicates were conducted for each insect developmental stage. Five healthy plants (psyllid and Lps-free) with clip cages were also included within trials as controls for psyllid infested plants. Each experiment replicate was isolated in an individual mesh insect cage (Bugdorm, MegaView Science Education Services Co., Ltd., Taichung, TAIWAN) to prevent other insects from contaminating plants during trials.
At the end of the infestation period, all psyllids and eggs were quickly removed from tomato leaves under a microscope before petiole and leaf midribs were cut from infested leaves (this is a crucial step since Lps is present in psyllid eggs [8]; Table 3). Leaf midribs and petioles were cut from leaf tissue using new sterilized razor blades, disposable Petri dishes, and disposable gloves for each sample to prevent sample contamination. Insects, leaf midribs/petioles and leaf blade were collected separately and flash frozen in liquid nitrogen and kept at 280uC until DNA and RNA extractions were performed. All psyllids used in trials were individually PCR screened for the presence of Lps after leaf samples were collected (see below).
Vector-free transmission of Lps to tomato leaves Numerous attempts to develop an uninfected psyllid line (Lpsfree line) from a naturally infected TX-Lipsy line were unsuccessful. Insects were treated with ampicillin through host plants and by direct injection into gravid females according to Douglas and Prosser [40] and Koga et al. [41], respectively. Infection came back after the first treatment and in the second there was high mortality of females and progeny from surviving females were not cured of the infection. Therefore, the impact of Lps on tomato gene expression had to be assessed in the absence of its vector.
Tomato plants infected with Lps were generated by grafting a Lps-infected scion to a non-infected tomato root stock. Control plants were constructed by grafting a non-infected scion to a noninfected root stock. All plants used in this experiment were one month old. Lps-infected scions were isolated from plants infested with psyllids for 2 weeks. Infected tomato plants were stripped of all leaves with a clean razor blade, triple rinsed with water to remove all psyllids, and sprayed with 2% Sunspray ultra-fine oil (Sunoco Inc., Aartselaar, Belgium) to kill any small nymphs and eggs potentially remaining on infected scions. Infected shoots were immediately tongue grafted to uninfected tomato plants (the root stock) 24 hours after insect removal; three whorls of leaves remained on the root stock. To control for the wounding that occurs during grafting, uninfected tomato shoots were grafted to the root stock (grafting control) as described above. Grafting treatments and controls were performed at the same time as psyllid life-stage treatments in an insect rearing room at 25+/21uC with a photoperiod of 14:10 (L:D) using metal halide grow lights. Grafting experiments were isolated in a separate mesh cage to protect plants from insect infestations.
Leaves from the root stocks of the Lps-infected scion plants (N = 5) or uninfected scion plants (N = 5) were collected at 24 hr, 72 hr, and 6 days after grafting to insure successful Lps infection. Lps presence in root stock leaves was determined with PCR (see below). Because Lps was detected in leaves of root stocks 24 hours after grafting to infected scions, RNAs were isolated and defense gene mRNA levels were determined for the 24-hr samples. The experiments were repeated three times.
DNA and RNA isolation DNA was extracted as described in Hansen et al. [8] from insects, eggs and plant midribs/petioles collected in the insect life stage and Lps graft experiment above. Bleach sterilized and autoclaved mortar and pestles were used to grind plant tissue in liquid nitrogen before plant extractions were performed, and 100 mg of ground leaf midrib/petioles were used for each extraction.
Total RNA was extracted from the 24-hour psyllid-infested and control leaves using a guanidinium thiocyanate-acid phenol based method [56]. To clean up the RNA, a modified cetyl trimethylammonium bromide (CTAB) method was used [58]. Briefly, total RNA was extracted in 500 mL of hot CTAB extraction buffer. Samples were then incubated at 65uC for 10 minutes, centrifuged and the supernatant was removed. The supernatant was re-extracted with 500 mL of hot 5% CTAB solution for an additional 20 minutes. Following centrifugation, the supernatant was removed and extracted with chloroformisoamyl alcohol (24:1 [v/v]) three times and precipitated overnight with 10 M LiCl. RNA integrity was verified on a 1.2% formaldehyde agarose gel [55] and with a microfluidic visualization tool (Bioanalyzer, Agilent Technologies, Santa Clara CA,USA, http://www.algilent.com) at UCR's Genomics Core.

PCR detection of Lps in the plant, insect and eggs
For Lps detection using PCR, two primer sets were used for two different Lps gene regions. The ribosomal intergenic space (IGS) was detected using 1611F and 480R primers [8] and 50 S rRNA region was detected using Bop-F and Bop-R primers (Table 4). To prevent pre-PCR carry-over contamination, a high fidelity polymerase (Fast Start High Fidelity PCR system, Roche, Indianapolis, IN, USA) with the ability to incorporate dUTP (Fermentas) was used for 50 S rRNA primers. This gene region has never been amplified before with dTTPs in contrast with the IGS primer set in our lab [59].
For both primer sets, PCR was performed in 25-mL reactions containing 1 mL of DNA template (concentration not determined), 0.2 mM each of dATP, dCTP, and dGTP, 0.4 mM of dTTP mix, 0.2 mM each primer, 16 PCR MgCl 2 -free buffer (Roche, Indianapolis, IN, USA), 5 mM MgCl 2 , 1 U polymerase (Roche, Indianapolis, IN, USA), and 1 U UDG when 50 s rRNA primers were used. For 1611F and 480R primers, amplifications were performed in a Mastercycler 5331 (Eppendorf, Hamburg, Germany) programmed as: an initial denaturing step of 95uC for 5 min; followed by 38 cycles of 95uC for 30 sec, 60uC for 50 sec, and 72uC for 1.5 min; and a final extension step of 72uC for 10 min. Thermocycler conditions for 50 s rRNA Bop-F and Bop-R primers (using UDG) were: an initial incubation of 37uC for 10 min, an initial denaturing/UDG deactivation step of 95uC for 10 min; followed by 38 cycles of 94uC for 30 sec, 60uC for 45 sec, and 72uC for 45 sec; and a final extension step of 72uC for 5 min. After PCR with 50 S rRNA primers, 0.1 U of uracil glycosylase inhibitor (New England Biolabs, Ipswich, MA, USA) was added to each PCR reaction. The PCR product was then incubated for 30 min at 37uC. Extraction and PCR negative controls and a PCR positive control were included in each assay.
Amplified DNA was visualized after electrophoresis on a 1% agarose gel stained with ethidium bromide (run at ,4.9 V/cm for at least 1 hr). Amplified Lps DNA from each primer set (one positive from each plant treatment, one positive from an egg from each isofemale line, and one positive from psyllids used during Insect DNA extracts positive for Lps presence were used to further quantify Lps titer within psyllids using quantitative PCR (qPCR) using a Rotor-Gene 3000 (Corbett Research, Rotor-gene 6.1) with a 72-well rotor was used for psyllid qPCR reactions. For qPCR, primers used for Lps quantification, which were developed in this study are: Y DRAG Q-PCR-IGS-7F and Y DRAG Q-PCR-IGS-7R (Table 4). DyNAmo HS, Sybr Green (Finnzymes, Lafayette, CO, USA) master mix, 0.9 mM of primer, 1 U of UDG, and 1 U of FastStart High Fidelity PCR System polymerase were all incorporated into a final volume of 18 mL master mix and 2 mL of psyllid DNA template extracted from individual psyllids. A single-copy psyllid reference gene (FORKHEAD) was also amplified with quantitative PCR to control for body mass variations between psyllid lifestages (QPCR-FKhead_90F and QPCR-Fkhead_188R; Table 4). Reaction conditions for qPCR were an initial incubation of 50uC for 2 min, an initial denaturing/UDG deactivation step of 95uC for 15 min; followed by 40 cycles of 94uC for 10 sec, 60uC for 20 sec, and 72uC for 30 sec; and a final melting step of 72-95uC with a 1uC increase every 5 sec after an initial 45 sec 1uC increase.
Relative quantification of bacterial titer was modified from Bustin [59] using LinRegPCR 11.1 for data [60,61]. After CT values were calculated for each qPCR technical replicate (four per sample) values were averaged per sample. To determine if there is a difference in Lps titer (CT, see above) between psyllid life stages an ANCOVA was used with FORKHEAD CT included as a covariate (to control for differences in psyllid biomass across life stage treatments). Post-hoc multiple comparison analyses of CT values were conducted between psyllid life stages using Scheffe's multiple comparisons. Parametric data were tested for normality using the Kolmogorov-Smirnov test. Type I error for all analyses is a = 0.01. All statistical analyses were conducted using SPSS Inc. 14.0 software.

Detection of tomato transcripts using quantitative reverse transcriptase-PCR (qRT-PCR)
Expression of transcripts associated with signaling and defensive tomato genes were quantified for both psyllid infested and grafting treatments, to tease apart the influence of Lps on the plant with and without its insect host. AOS (Allene oxide synthase), Pin2 (Proteinase inhibitor 2), PAL5 (Phenylalanine ammonia-lyase 5), PR1 (Pathogenesis-related 1; P6) and HSP90 (to confirm microarray data) mRNA levels were determined using quantitative reverse transcriptase-PCR (qRT-PCR). Three internal standard primer sets were evaluated for highest efficiency and lowest variation across samples; 18 S rRNA was used in experiments based on this criteria. Primers used for qRT-PCR were designed using Primer-Blast (http://www.ncbi.nlm.nih.gov/tools/ primer-blast/index.cgi?LINK_LOC = NcbiHomeAd).
Total RNA (1 mg) was DNase (Rq1, Promega, Madison, WI, USA) treated to clean up RNA and used as the starting material for the qRT-PCR experiments. The first-strand cDNA synthesis was carried out with the 'SuperScript' kit (Invitrogen Technologies, Carlsbad, CA, USA) using oligo-dT as primer and recommended methods in kit. Reactions were carried out using 5 mL of the 'Sybr green PCR master mix' (Applied Biosystems, Carlsbad, CA, USA), with 800 nM of primer, in the 7500 instrument (Applied Biosystems, Carlsbad, CA, USA). The PAL5, AOS, Pin2, PR1, and 18 S primers used for qRT-PCR are listed in Table 4.
The PCR reaction was initiated with incubation to 95uC for 10 min to activate the enzyme. Then, the following cycle was repeated 40 times: 95uC for 15 sec, 60uC for 15 sec, and 72uC for 15 sec. Three technical replicates were performed for each individual plant RNA sample, and a digital pipette was used for all qRT-PCR reactions. The raw CT values were averaged, quantified and analyzed according to the standard curve method (Applied Biosystems, Carlsbad, CA, USA), resulting in relative fold change. Psyllid-infested plant mRNA levels were expressed relative to undamaged controls with empty cages and Lps-graft-inoculated plant mRNA levels were expressed relative to graft-control plants.
Analysis of variance (ANOVA) was used to determine if there was a difference in transcript abundance in tomatoes infested with Lps-infected psyllids and infected with vector-free Lps. Post-hoc multiple comparison analyses of relative fold change values were conducted between psyllid-infested plants and Lps-inoculated plants using Least Significant Difference (LSD) Tests. All statistical analyses were conducted using SAS, version 9. Table S1 The top 149 tomato transcripts expressed after tomato psyllid1 st -2 nd instar feeding. (XLS)