A virus-induced gene silencing screen identifies a role for Thylakoid Formation1 in Pseudomonas syringae pv. tomato symptom development in tomato and Arabidopsis

In this study, we utilized Nicotiana benthamiana and virus-induced gene silencing (VIGS) to individually reduce the expression of over 4,000 genes. The silenced lines of N . benthamiana were then screened for altered response to purified COR. Using this forward genetics approach several genes were identified with altered responses to COR. These were designated as altered COR response ( ALC ) genes. When silenced, one of the identified genes, ALC1 , produced a hypersensitive/necrosis-like phenotype upon COR application in a Coronatine insensitive 1 ( COI1 ) dependent manner. To understand the involvement of ALC1 during the Pst DC3000-host interaction, we used the nucleotide sequence of ALC1 and identified its ortholog in Arabidopsis ( Thylakoid Formation1, THF1 ) and tomato ( SlALC1 ). In pathogenicity assays performed on Arabidopsis thf1 mutant and SlALC1 -silenced tomato plants, Pst DC3000 induced accelerated coalescing necrotic lesions. Furthermore, we showed that COR affects ALC1 localization in chloroplast in a COI1 -depenendent manner. In conclusion, our results show that VIGS-based forward genetic screen has potential to identify new players in COR signaling and disease associated necrotic cell death.


INTRODUCTION
In nature, plants come in contact with numerous microbes that are potential pathogens.
Active plant defense mechanisms, in general, involve a complex network of three genetically distinct signaling pathways, known as the salicylic acid (SA), jasmonic acid (JA), and ethylene (ET) pathways (Kunkel and Brooks, 2002;Glazebrook, 2005).
During a compatible interaction with a host, Pst DC3000 infection results in the activation of the JA signaling pathway (Zhao et al., 2003;Laurie-Berry et al., 2006;Uppalapati et al., 2007). This causes the suppression of the salicylic acid (SA) pathway owing to its antagonistic relation with the JA pathway (Kloek et al., 2001;Kunkel and Brooks, 2002;Zhao et al., 2003;Uppalapati et al., 2007). The suppression of the SA pathway during the Pst DC3000-host interaction is thought to be caused by COR, which functions as a molecular mimic of JAs (Feys et al., 1994;Bender et al., 1999;Staswick and Tiryaki, 2004).
Pst DC3000 causes disease on several plant species including tomato and Arabidopsis. A typical symptom on tomato leaves is bacterial speck, which includes necrosis surrounded by a chlorotic halo (Mittal and Davis, 1995;Zhao et al., 2003). In Arabidopsis, the infected area exhibits water-soaked lesions accompanied by diffused chlorosis (Mittal and Davis, 1995;Brooks et al., 2004). Pst DC3000 infection also causes chlorosis in other plants belonging to Brassicaceae family such as collard and turnip (Elizabeth and Bender, 2007). In addition to chlorosis, Pst DC3000-infected collard plants exhibit water-soaked lesions and anthocyanin, suggesting that Pst DC3000 elicits unique responses in different plants. Studies have shown that tomato plants inoculated with a COR-defective mutant of Pst DC3000 did not develop typical chlorotic symptoms; furthermore COR contributed to pathogen fitness and disease development in a SAindependent manner (Uppalapati et al., 2007). Tomato leaf tissues treated with purified 6 COR show chlorosis (Gnanamanickam et al., 1982;Uppalapati et al., 2005Uppalapati et al., , 2007. Unlike tomato, purified COR does not elicit chlorosis on Arabidopsis leaves (Bent et al., 1992;Mach et al., 2001). However, in Arabidopsis, COR is required for full disease symptom development and pathogen fitness in a SA-dependent manner (Kloek et al., 2001;Brooks et al., 2004). These results suggest that COR functions as an important virulence factor in tomato and Arabidopsis, although it functions differently in these hosts.
More recently we have demonstrated a role for COR-induced effects on photosynthetic machinery and ROS in modulating necrotic cell death during bacterial speck disease of tomato . Despite our present understanding of COR function, it is not clear how chlorosis impacts or benefits pathogen virulence.
Furthermore, the identity of host molecular targets for COR and the downstream signaling cascades that ensue are not well understood. Based on similarities between COR and JAs in terms of structure and function (Feys et al., 1994;Uppalapati et al., 2005), it seems likely that COR and JA interact with at least one common host receptor (Kastir et al., 2008). Thus, in addition to furthering our understanding of disease development, studies aimed at understanding the molecular mechanism of COR may provide information on JA-mediated plant defense.
In an effort to identify plant proteins that are the molecular targets of COR, we used a Tobacco rattle virus (TRV)-based virus-induced gene silencing (VIGS) as a fast-forward genetics tool (Liu et al., 2001a, b;Anand et al., 2007a) to screen a Nicotiana benthamiana cDNA library for genes that are involved in response to COR. We identified a N. benthamiana gene, ALC1, that when silenced displayed an unexpected hypersensitive/necrosis-like phenotype rather than a typical chlorotic phenotype in response to COR application. ALC1 has homology to an Arabidopsis gene, Thylakoid Formation1 (THF1; Wang et al., 2004). The pathogenicity assays performed in this study indicate that loss of ALC1/THF1 leads to accelerated cell death in response to Pst DC3000 infection in both tomato and Arabidopsis.

Application of purified COR on N. benthamiana leaves results in chlorosis
Unlike tomato, the efficiency of VIGS is quite uniform in N. benthamiana and therefore this host is suitable for large-scale fast-forward screening studies (Lu et al., 2003;del Pozo et al., 2004;Anand et al., 2007a). Purified COR, when spotted onto N.
benthamiana leaves at different concentrations (0.002-2 nmol in 2 μ l aliquots), produced a visible, confined chlorosis in a dose-dependent manner (Fig. 1A). 0.2 nmol concentration which produces a confined chlorosis phenotype was used for screening ( Fig. 1B). Based on these results we concluded that a VIGS-based approach in N.
benthamiana was suitable for screening silenced plants that show an altered chlorosis phenotype upon COR application; therefore, this approach was used to identify genes involved in COR-mediated signaling.

VIGS-based screening identifies several N. benthamiana genes with altered CORinduced response
To identify plant genes that are involved in COR signaling, a normalized N. benthamiana cDNA library cloned in pTRV2 and transformed into A. tumefaciens GV2260 was used 8 (Anand et. al, 2007a). N. benthamiana plants were individually inoculated with Agrobacterium containing TRV2 cDNA clones, along with an Agrobacterium strain containing TRV1, in duplicates, to silence their corresponding genes in N. benthamiana (Anand et al., 2007a). Two weeks after TRV inoculation, COR (0.2 nmol) was spotted on the leaves of silenced plants, and the phenotypes were recorded 5-7 days after COR application.
After screening ~4,000 cDNA clones, we identified five non-redundant cDNA clones that when silenced resulted in altered COR response (ALC) phenotype upon exogenous application of COR ( Fig. 1D S2A).
We termed Nb28C12 as altered COR response 1 (ALC1). The sequence information was then analyzed to predict gene function. A BLASTn search against the TIGR database using the NbALC1 sequence revealed 77% identity to an Arabidopsis gene named THF1 (Genbank ID AY899908); 92% identity to a potato gene that encodes a light-regulated chloroplast localized protein (Solanum tuberosum THF1, GenBank ID AY342161); 81% identity to a rice (Oryza sativa) gene encoding inositol phosphataselike protein (GenBank ID AY224446); and 79% identity to a wheat (Triticum aestivum) gene encoding Ptr ToxA binding protein (GenBank ID AY377991). To facilitate a more comprehensive comparative analysis of ALC1, we designed a primer pair to clone the full length ALC1 gene based on the sequence of tobacco (N. tabacum) ortholog (TC10126).
The cloned gene was then sequenced and the translated amino acid sequence was then aligned with orthologous plant protein sequences using ClustalW (http://www.ebi.ac.uk/clustalw/). As shown in supplemental Fig. S3A, N. benthamiana ALC1 shows strong sequence identity with orthologs from other species. N. benthamiana ALC1 also displays a higher degree of evolutionary relatedness with the tobacco ortholog when compared to other plant orthologs that were analyzed (Supplemental Fig. S3B).
The silencing of Nb28C12 resulted in a necrotic phenotype upon COR-treatment without a visible chlorosis (Fig. 1D).  Fig. S2B). However, majority of the leaves in ALC1 silenced tomatoes did not exhibit any obvious phenotype (Fig. S1B, middle panel), some of the older leaves (six weeks post TRV-inoculation) showed variegated coloration on the leaf surface (Fig. S1B, right panel). When purified COR (2 nmol) was exogenously applied, the silenced line showed a necrosis-like phenotype, whereas the control plants showed a typical confined chlorosis as expected (Fig. 3A).
Although we were fairly successful in transiently silencing the tomato ALC1 gene, a uniform and pronounced silencing, such as that observed in N. benthamiana, is often difficult to achieve in tomato (Ekengren et al., 2003;Ryu et al., 2004). Therefore, to achieve stable and uniform silencing and to confirm the necrosis phenotype induced by COR and Pst DC3000 on SlALC1-silenced tomato lines, we generated SlALC1 RNAi lines. We assayed three independent transgenic RNAi lines and all responded similarly to COR application and Pst DC3000 infection. Here, we discuss the data for one of the transgenic lines, 3-2. Results obtained from qRT-PCR indicated the transcript levels of Coronatine insensitive 1/jasmonic acid insensitive 1 (COI1/JAI1), a F-box protein is shown to be required for COR signaling in tomato and Arabidopsis (Feys et al., 1995;Zhao et al., 2003;Katsir et al., 2008). By using VIGS, we transiently silenced SlALC1 in To determine if the severe necrosis could be explained by a higher amount of bacterial growth in the silenced lines, the population of Pst DC3000 was monitored at 1, 3 and 5 dpi. Interestingly, the bacterial population on the transiently or stably silenced SlALC1 plants was not significantly different from that on the inoculated control plants

Arabidopsis thf1 mutant displays severe necrosis upon Pst DC3000 inoculation
As mentioned above, ALC1 is closely related to an Arabidopsis gene called THF1 (Fig.   S3A). An Arabidopsis thf1 mutant was previously identified and shown to have variegated leaves ( To determine whether THF1 has an effect on Pst DC3000-induced disease symptoms on Arabidopsis, we dip-inoculated (10 8 CFU/ml) or syringe-infiltrated (10 6 CFU/ml) the wild-type Col-0 and thf1 mutant with Pst DC3000. As expected, Col-0 showed water-soaked necrotic lesions accompanied by chlorosis (Fig. 6B). However, the thf1 mutant plants exhibited accelerated necrotic lesions without visible chlorosis( Fig.   6B). Complemented lines of the thf1 mutant and THF1-overexpressing plants displayed disease symptoms similar to the wild-type Col-0 after inoculation with Pst DC3000 (Fig.   6B). Interestingly, when the growth of Pst DC3000 was monitored at 0, 1, 2 and 4 dpi, no significant fold differences in the bacterial growth were observed between the wildtype Col-0, the thf1 mutant, complemented line of thf1 and the THF1 overexpression line ( Fig. 7). However, unlike in tomato, thf1 mutants supported slightly increased (1.5 fold) bacterial growth than the wild-type (Fig. 7). These results suggest that THF1 does not significantly contribute to the pathogen growth in Arabidopsis, at least for the duration of time the bacterial growth was monitored.
Necrosis occurred much earlier on Pst DC3000-infected thf1 leaves than on leaves of the wild-type Col-0 (data not shown). We therefore investigated whether the thf1 mutant had a weaker defense response and was more susceptible to biotic and abiotic stress because of defects in thylakoid formation (Wang et al., 2004). To investigate this, leaves of Col-0 and the thf1 mutant were infiltrated with two nonhost pathogens that do not infect Arabidopsis, P. syringae pv. tabaci and P. syringae pv. glycinea, and growth and symptoms were compared with a coronatine producing, P. syringae pv. maculicola, which is pathogenic to Arabidopsis (Dong et al., 1991;Cuppels and Ainsworth, 1995;Mishina and Zeier, 2006). As expected, the population of P. syringae pv. maculicola increased approximately 100-fold on both Col-0 and thf1 leaves by 3 dpi; however, neither P. syringae pv. glycinea nor tabaci multiplied to a significant level on Col-0 or thf1 plants (Supplemental Fig. S5A).
Arabidopsis Col-0 and the thf1 mutant were also monitored for symptom development in response to inoculation with P. syringae pvs. maculicola, glycinea and tabaci and the soft rot pathogen Erwinia carotovora subsp. carotovora. P. syringae pv.
maculicola induced chlorosis on Col-0 but not on thf1 mutant line (Supplemental Fig.   S5B). Neither Col-0 nor thf1 plants developed visible symptoms in response to P. syringae pvs. tabaci or glycinea (Fig. S5B). E. carotovora subsp. carotovora induced soft rot on both Col-0 and thf1 with no apparent difference in phenotypic response between the wild-type and the mutant line (Fig. S5B). Infiltration of leaves with cell death inducing agents such as NaCl (500 mM The JA pathway appears intact in Arabidopsis thf1 mutant plants after Pst DC3000 inoculation COR functions as mimic of JAs and mediates signaling via JA perception machinery in tomato and Arabidopsis (Feys et al., 1995;Zhao et al., 2003). Thus, it remained possible that the absence of chlorosis was due to disruption of the JA-dependent signaling pathway. Therefore, we used qRT-PCR to analyze transcript levels of Lipoxygenase2 (LOX2) and Plant defensin1.2 (PDF1.2). Transcripts of LOX2 and PDF1.2 were induced in both Col-0 and thf1 in response to Pst DC3000. Although expression of both genes was lower in thf1 mutant line, especially at 4 dpi, the JA pathway appears to be functional ( Fig. 8A, B) at the time points analyzed.
Chlorosis occurs due to the degradation of proteins in the chloroplast (Quirino et al., 2000) and the Arabidopsis CORI1 gene (encoding chlorophyllase) is induced upon COR or MeJA application (Benedetti et al., 1998), resulting in chlorophyll degradation (Benedetti and Arruda, 2002). The lack of chlorosis in thf1 could be due to repression of CORI1 as a result of loss of THF1 function. Thus, we analyzed CORI1 transcript levels in Pst DC3000 inoculated Col-0 and thf1 plants. CORI1 expression in Col-0 and thf1 was upregulated ~175 fold and ~75 respectively, 1dpi (Fig. 8D). These results further suggest that the chlorophyllase activity and JA dependent pathway are not severely affected in the thf1 mutant. Although we did not notice any visible chlorosis in COR or Pst DC3000 inoculated tissues (Fig. 6B), the COR-induced chlrophyllase activity suggests some degree of chlorophyll degradation in COR or Pst DC3000 inoculated thf1 plants.
Similarly, significant levels COR-induced chlorophyll degradation was observed in NbALC1-silenced plants (Fig. 2). Taken together, these results suggest that THF1 may be operating down-stream of COR-induce JA signaling and chlorosis.

Coronatine affects the localization of ALC1 in a COI1-dependent manner in N. benthamiana plants
To determine if coronatine directly affected ALC1, we monitored the effect of coronatine treatment on the localization/stability of ALC1. time intervals post-COR inoculation (Fig. 9). As previously reported for THF1, GFP-ALC1 localized to the chloroplast (Fig. 9A). Interestingly, within four hours upon COR application, ALC1 was destabilized/degraded as shown by the loss of GFP fluorescence ( Fig. 9A). ALC1 florescence was not detected even after 24 (Fig. 9A), 48 hours and 72 hours (time at which chlorosis is visible on the leaf; data not shown). It also noteworthy that the destabilization/degradation of GFP-ALC1 is seen only at the site of application of COR and nearby region but the leaf areas away from the region of coronatine application remains unaffected even after 24 and 48 hours (Supplemental Fig. S6). To rule out the possibility that COR application is leading to alterations in chloroplast structure, therefore resulting in non-specific effects on ALC1, we tested COR effects on GFP-RecA (Kohler et al., 1997), another chloroplast localizing protein (Fig. S7). Interestingly, COR application did not result in destabilization/degradation of GFP-RecA or GFP alone ( Fig.   S7; Fig. 9C).
The COI1-dependent nature of COR induced alterations in ALC1 localization were further confirmed in COI1-silenced N. benthamiana plants (Fig. 9B).Interestingly, in COI1-silenced plants an increased signal intensity of 35S::ALC1 GFP was observed following COR application (Fig. 9B). It is not clear if this is due to lack of COR activity upon COI1-silencing or a COI1-independnet activity of COR on ALC1. Silencing of COI1 did not affect the expression levels of 35S::ALC1 GFP florescence (Fig. S8).
However, silencing of COI1 abolished COR-induced destabilization/degradation of 35S::ALC1 GFP florescence (Fig. 9B) induced effects on chloroplast/ALC1 directly alters ALC1 localization or stability in a COI1-depenndet manner and therefore may affect its function.  , 1995;Zhao et al., 2003). Furthermore, necrosis spread beyond the region where COR was applied as early as 10 dpi, which is similar to the runaway cell death phenotype reported earlier in the Arabidopsis lsd1 mutant ( Fig.   2B; Jabs et al., 1996).

DISCUSSION
To determine the role of ALC1 in the development of symptoms in response to COR or Pst DC3000, we chose Arabidopsis since it is genetically tractable and a host of   , 2006). THF1 is also identified as an interactor of G-protein, GPA1 in Arabidopsis and shown to play a role in far-red irradiation preconditioned cell death (Huang et al., 2006;Wei et al., 2008). Our results using GFP-tagged ALC1 suggest that COR has direct effects on ALC1 and might target ALC1 to degradation in a COI1dependent manner (Fig. 8). Based on these results it is tempting to speculate if ALC1/THF1, localized in the chloroplast membrane may directly interact with COR.
ToxA is a determinant of virulence in P. tritici-repentis, a pathogen that causes the tan spot of wheat. Therefore, it is possible that COR may interact directly with ALC1/THF1 during the P. syringae pv. tomato-host interactions. Thus we speculate that ALC1 and THF1 (Wang et al., 2004), which is localized on the chloroplast is somehow involved in the maintenance of ROS homeostasis and therefore Arabidopsis thf1 mutants and ALC1 silenced tomato leaves are more sensitive to COR/pathogen induced ROS leading to accelerated cell death (necrosis) in tomato and Arabidopsis.
In conclusion, we have developed a VIGS-based forward genetic screen for identification of new targets involved in COR signaling. Although we set out to identify genes involved in COR induced chlorosis, we identified a gene, THF1,that when silenced causes necrosis upon COR application. We are presently screening a COR-responsive N.
benthamiana cDNA library to identify components involved in COR-induced chlorosis .
Although the precise role of THF1 in COR signaling pathway could be argued and need further confirmation, our results present a new role for chloroplast localized THF1 in bacterial speck disease development. expanded cotyledons were removed from the pots and were completely submerged in Agrobacterium mixture and vacuum infiltrated for 2 min as described earlier by Uppalapati et al. (2007). The seedlings were then transplanted into Professional Blend potting mixture (Sun Gro, Bellevue, WA). To improve the silencing efficiency, the remaining Agrobacterium culture was dispensed around the seedlings using the Agrodrench method (Ryu et al., 2004). Inoculated potted seedlings were then maintained in growth chambers for 2 days with a 12 h photoperiod (22°C, day; 18°C, night). Then the plants were moved to greenhouse and maintained at 14 h daylight at 25°C and 22°C

Plant materials, bacterial cultures and plant infections
at night for the next 10-14 days.

Generation of tomato ALC1 transgenic RNAi lines
For the generation of a tomato ALC1 RNAi line, the SlALC1 fragment (described above) was introduced into a GATEWAY-ready binary RNAi vector pK7GWIWG2(I) (Karimi et al., 2002)

Measurement of chlorophyll content
The chlorophyll content of leaf discs was measured as described in Arnon (1949) and . Two leaf discs (0.78 cm 2 each) were isolated six days postinoculation from leaves treated with water (mock control), 2µl of COR (0.2 nmol) and then macerated in liquid nitrogen, placed in 6 ml of acetone and incubated at 4°C in the dark for 12 h. Aliquots of total chlorophyll dissolved in acetone were mixed with hexane and 10 mM KOH at a ratio of 4:6:1 (v/v). Chl a and b were quantified spectrophotometrically using the formula described by Arnon (1949).

RNA isolation and reverse transcription-PCR (RT-PCR) analysis
Total RNA was isolated from leaves of N. benthamiana, tomato and Arabidopsis plants using TRIZOL reagent (Invitrogen, Carlsbad, CA, USA). The first strand cDNA was synthesized using oligo (dT) 15 primer and Omniscript RT kit (Qiagen, Valencia, CA, USA). For quantitative analysis of transcripts, primer pairs were designed using the Primer Express software (Applied Biosystems Inc., Foster City, CA) to amplify the target sequences. qRT-PCR was performed with ABI HT7900 machine using the SYBR Green method (Applied Biosystems Inc., Foster City, CA). PCR efficiency was determined using linear regression software LinRegPCR (Ramakers et al., 2003). In order to normalize the data, parallel reactions were run using the elongation factor-alpha (EF1α) primers as the endogenous control for Arabidopsis and actin or tubulin primers as the endogenous control for N. benthamiana and tomato (Supplemental Table S1). The relative transcript levels were quantified as described previously (Pfaffl, 2001).

Subcellular localization of ALC1
To transiently express ALC1 in N. benthamiana, the GATEWAY-ready pMDCC83 was used as a vector to generate a GFP fusion (Curtis and Grossniklaus, 2003). Full-length ALC1 sequence was amplified from N. benthamiana cDNA using the following gene specific primers:           Photographs were taken seven days after COR application.       Response of control and transiently silenced tomato lines to Pst DC3000. Pst DC3000 (5 x 10 6 CFU/ml) was spray-inoculated on control (TRV::GFP; left panels) and SlALC1 silenced (by VIGS; right panels) tomato plants. Photographs were taken after 5 and 10 days post inoculation (dpi). B. SlALC1-silenced transgenic RNAi line 3-2 and wild-type tomato plants were also spray-inoculated with Pst DC3000. Photographs were taken 7 dpi. Wangdi et al., Figure 5 A B  . The mutation in Arabidopsis THF1 has no effect on the growth of Pst DC3000. Arabidopsis leaves of the lines described in Fig. 5 were syringe-infiltrated with Pst DC3000 (10 6 CFU/ml), collected at intervals after inoculation, homogenized in water, and plated on KB medium to determine colony forming units (CFU). The error bars represent the standard deviation. The experiments were conducted at least three times with several replicates, and the data shown are representative of each experiment. Growth measurements with same letters showed no significant differences based on Fisher LSD values (p <0.005)  Four-week old Col-0 and thf1 mutant line were syringe-infiltrated with either Pst DC3000 (10 6 CFU/ml) or buffer (mock control). Total RNA was extracted from the leaves of the infected plants collected 1, 2 and 4 dpi and cDNA was synthesized for qRT-PCR analyses. The transcript levels were normalized against the elongation factor EF1α that was used as endogenous control as described by Pfaffl (2001). The transcript levels were quantified relative to the transcript levels on mock control which was assigned as 1. A,B. JA pathway genes (represented by LOX2 and PDF1.2); and (C) the chlorophyllase encoding gene CORI1 were activated in Pst DC3000 infected Col-0 and the thf1 mutant. All experiments were repeated at least three times. The data shown here represent the average of three biological replicates and three technical replicates with the standard deviation values shown as the error bars.