Mutations in an Atypical TIR-NB-LRR-LIM Resistance Protein Confer Autoimmunity

In order to defend against microbial infection, plants employ a complex immune system that relies partly on resistance (R) proteins that initiate intricate signaling cascades upon pathogen detection. The resistance signaling network utilized by plants is only partially characterized. A genetic screen conducted to identify novel defense regulators involved in this network resulted in the isolation of the snc6-1D mutant. Positional cloning revealed that this mutant contained a molecular lesion in the chilling sensitive 3 (CHS3) gene, thus the allele was renamed chs3-2D. CHS3 encodes a TIR-NB-LRR R protein that contains a C-terminal zinc-binding LIM (Lin-11, Isl-1, Mec-3) domain. Although this protein has been previously implicated in cold stress and defense response, the role of the LIM domain in modulating protein activity is unclear. The chs3-2D allele contains a G to A point mutation causing a C1340 to Y1340 substitution close to the LIM domain. It encodes a dominant gain-of-function mutation. The chs3-2D mutant is severely stunted and displays curled leaf morphology. Additionally, it constitutively expresses PATHOGENESIS-RELATED (PR) genes, accumulates salicylic acid, and shows enhanced resistance to the virulent oomycete isolate Hyaloperonospora arabidopsidis (H.a.) Noco2. Subcellular localization assays using GFP fusion constructs indicate that both CHS3 and chs3-2D localize to the nucleus. A third chs3 mutant allele, chs3-3D, was identified in an unrelated genetic screen in our lab. This allele contains a C to T point mutation resulting in an M1017 to V1017 substitution in the LRR–LIM linker region. Additionally, a chs3-2D suppressor screen identified two revertant alleles containing secondary mutations that abolish the mutant morphology. Analysis of the locations of these molecular lesions provides support for the hypothesis that the LIM domain represses CHS3 R-like protein activity. This repression may occur through either autoinhibition or binding of a negative defense regulator.


INTRODUCTION
In order to evade pathogen infection plants have developed complex resistance mechanisms. Conserved pathogen features necessary for the microbial lifestyle, collectively known as pathogenassociated molecular patterns (PAMPs), can be recognized by pattern recognition receptors on the plant cell surface (Zipfel and Felix, 2005). PAMP detection elicits a signaling cascade leading to the activation of PAMP-triggered immunity (PTI; Jones and Dangl, 2006). However, during the process of infection successful pathogens are able to release effector molecules into the plant host cell that can inhibit PTI, attenuating defense responses. To compensate, plants have evolved many RESISTANCE (R) genes encoding proteins that can recognize these effector molecules, thereby rapidly and effectively inducing defense responses to prevent the spread of infection. These responses may include the accumulation of the defense hormone salicylic acid (SA), induction of PATHOGENESIS-RELATED (PR) genes, production of reactive oxygen species (ROS), reinforcement of the cell wall, and a type of localized cell death referred to as the hypersensitive response (HR; Hammond-Kosack and Jones, 1996).
The majority of plant R proteins possess a central nucleotidebinding (NB) site, a number of leucine-rich repeats (LRRs) at the C terminus, and either a coiled-coil (CC) or Toll/interleukin1receptor-like (TIR) domain at the N terminus (Dangl and Jones, 2001). There are roughly 150 R genes encoding NB-LRR proteins distributed throughout the Arabidopsis genome (Meyers et al., 2003). Signaling through many CC-NB-LRR R proteins requires NON-SPECIFIC DISEASE RESISTANCE 1 (NDR1) as an intermediate (Aarts et al., 1998), while signaling through most TIR-NB-LRR R proteins typically involves ENHANCED DISEASE SUS-CEPTIBILITY 1 (EDS1), PHYTOALEXIN-DEFICIENT 4 (PAD4), and SENESCENCE-ASSOCIATED GENE 101 (SAG101; Feys et al., 2005). These pathways appear to later converge and result in the deployment of similar defense mechanisms.
One key step in resistance signaling is the accumulation of the defense hormone SA. Downstream of SA synthesis, NON-EXPRESSOR OF PATHOGENESIS-RELATED GENES 1 (NPR1) functions as an essential positive regulator of resistance (Cao et al., 1994;Rairdan and Delaney, 2002). Through its interactions with the TGA family of basic leucine zipper transcription www.frontiersin.org factors, NPR1 modulates the expression of PR genes (Dong, 2004). The npr1-1 mutation abolishes SA-induced PR gene expression and pathogen resistance, and does not induce the expression of the pPR2-GUS reporter gene (Cao et al., 1994). Thus, a suppressor screen to identify negative regulatory components involved in resistance signaling can be carried out in the npr1 genetic background quite simply by using pPR2-GUS reporter gene expression for mutant isolation. Early screens resulted in the identification of SUPPRESSOR OF NPR1, INDUCIBLE 1 (SNI1), SUPPRESSOR OF NPR1, CONSTITUTIVE 1 (SNC1), SUPPRESSOR OF SALICYLIC ACID INSENSITIVE (SSI ) genes, and SUPPRESSOR OF NIM1-1 (SON1; Li et al., 1999Li et al., , 2001Shah et al., 1999Shah et al., , 2001Kim and Delaney, 2002;Shirano et al., 2002). Recent independent screens have led to the identification of SNC2, SNC3, SNC4, and SNC5 (renamed SUPPRESSOR OF RPS4-RLD; SRFR1) as intermediates in plant pathogen resistance Li et al., 2010;Zhang et al., 2010).
In this study, we report the identification and characterization of a snc6-1D mutant isolated from an npr1-1 suppressor screen. SNC6 encodes CHS3, a TIR-NB-LRR protein with an additional LIM (Lin-11, Isl-1, Mec-3;Freyd et al., 1990) domain at the C terminus. Thus, snc6-1D was renamed chs3-2D. The chs3-2D mutant is dwarfed and displays curled leaf morphology. It constitutively induces the expression of PR genes, shows enhanced oomycete resistance, and accumulates SA. Two revertant mutants were isolated through a chs3-2D suppressor screen. Another gainof-function mutant allele, chs3-3D, has also been identified in our lab as part of an independent mos4 (modifier of snc1, 4) suppressor screen. Analysis of the locations of these mutations provides further insight into the structural and functional details of the different domains of this unique R-like protein.

IDENTIFICATION AND GENETIC CHARACTERIZATION OF THE snc6-1D npr1-1 MUTANT
In plants, NPR1 is an essential regulatory component that functions downstream of salicylic acid (SA) induction in the systemic acquired resistance (SAR) pathway (Dong, 2004). In order to identify novel negative defense regulators independent of NPR1, a suppressor screen (described previously by Gao et al., 2008) was carried out in the npr1-1 background. The snc6-1D npr1-1 mutant was identified by its constitutive expression of pPR2-GUS ( Figure 1A). When the snc6-1D npr1-1 mutant (with pPR2-GUS) was backcrossed with SNC6 npr1-1 (with pPR2-GUS), the F1 progeny displayed the mutant morphology and constitutively expressed the pPR2-GUS reporter gene (data not shown), indicating that snc6-1D is a dominant mutation. Additionally, the snc6-1D npr1-1 mutant displays dwarfed morphology and curled leaves ( Figure 1B). These morphological traits are similar to but much more severe than those observed in the autoimmune R gene mutant snc1 (Li et al., 2001).

PR GENE EXPRESSION AND PATHOGEN RESISTANCE IN snc6-1D npr1-1
In snc6-1D npr1-1, constitutive pPR2-GUS reporter gene expression was observed primarily in the leaves and petioles ( Figure 1A). The induction of pPR2-GUS likely occurs via an NPR1-independent pathway in this mutant. When the expression levels of endogenous PR1 and PR2 were examined using RT-PCR analysis, both PR genes showed constitutively enhanced expression in snc6-1D npr1-1 (Figures 1C,D).

SA ACCUMULATION IN snc6-1D npr1-1
The dwarfed size and curled leaf phenotypes observed in the snc6-1D npr1-1 mutant, similar to those displayed in snc1 plants, are typical of plants with elevated SA levels (Bowling et al., 1997). High-pressure liquid chromatography was used to measure the endogenous levels of both free SA and salicylic acid β-glucoside (SAG), the storage form of the defense hormone that enables the release of free SA as required by the plant. Levels of free SA were ∼5 times higher in snc6-1D npr1-1 than in wild type or npr1-1, and levels of total SA (SA + SAG) were ∼25 times higher than in wild type and ∼5 times higher than in npr1-1 ( Figure 1F). SA accumulation observed under uninduced conditions in snc6-1D npr1-1 is likely partly responsible for the constitutive PR gene expression and disease resistance observed in the mutant.

MAP-BASED CLONING OF THE snc6-1D LOCUS
To map the snc6-1D locus, snc6-1D npr1-1 (Col-0 ecotype) was crossed with wild type (SNC6 NPR1) Ler ecotype. Rough mapping showed the snc6-1D mutation is located on the top of chromosome 5 between markers T24H18 and F17K4. Further fine mapping narrowed the location of the mutation between markers MPI7 and MRG7, an interval of approximately 120 kb (Figure 2A).
Sequencing of PCR fragments covering this region revealed a single G to A point mutation in the fifth exon of At5g17890 ( Figure 2B). A mutation in this gene had previously been identified and characterized in a screen for temperature sensitive mutants, and the gene was named CHILLING SENSITIVE 3 (CHS3; Yang et al., 2010). Thus, the snc6-1D allele was renamed chs3-2D. CHS3 is predicted to encode a protein of 1613 amino acids containing TIR, NB, LRR, and LIM domains ( Figure 5B). The chs3-2D mutation results in a C1340 to Y1340 substitution close to the LIM domain. Since chs3-2D is dominant, we hypothesized this might be a gain-of-function mutation.

chs3-2D IS A GAIN-OF-FUNCTION MUTATION
To determine the genetic nature of the chs3-2D mutation, wild type plants were transformed with either CHS3-GFP or chs3-2D-GFP fusion constructs under the control of its endogenous promoter. When transformed with CHS3-GFP, the transformants displayed wild type morphology, PR gene expression, and H.a. Noco2 resistance (two independent representative lines are shown in Figures 3A-D). In contrast, lines transformed with chs3-2D-GFP displayed mutant morphology, constitutive PR gene expression and enhanced resistance to H.a. Noco2. This indicates that chs3-2D is a gain-of-function mutation resulting in constitutive disease resistance. Although chs3-2D-GFP is able to confer autoimmunity, we were not able to observe the GFP signal of the fusion Frontiers in Plant Science | Plant-Microbe Interaction

FIGURE 1 | Phenotypic characterization of snc6-1D npr1-1. (A)
Expression of the pPR2-GUS reporter gene, present in wild type (WT) and npr1-1 mutant background, in WT, npr1-1, and snc6-1D npr1-1 plants. Seedlings were grown on MS media for 14 days prior to staining for GUS activity. The three photographs shown were taken at different magnifications in order to effectively capture GUS staining, thus seedling size should not be compared between panels. (B) Morphology of wild type (WT), npr1-1, and snc6-1D npr1-1 plants. Representative 4-week-old soil-grown plants were photographed. (C,D) Endogenous expression of (C) PR1 and (D) PR2 relative to Actin 1 in WT, npr1-1, and snc6-1D npr1-1. Values presented are averages of three replicates ± SD. (E) Growth of H.a. Noco2 on WT, npr1-1, and snc6-1D npr1-1 mutants 7 days post-infection. Values presented are averages of three replicates ± SD. (F) Endogenous free and total SA levels in WT, npr1-1, and snc6-1D npr1-1. Plants were grown on soil for 4 weeks and leaf tissue was collected for SA extraction. SA levels were analyzed using high-pressure liquid chromatography. Values presented are averages of four replicates ± SD. protein in transgenic plants, probably due to the very low protein levels.

CHS3 AND chs3-2D LOCALIZE TO THE NUCLEUS
Nuclear localization of CHS3 is suggested by WoLFpSORT, although the PredictProtein program indicates that the protein does not contain a known nuclear localization signal (Rost et al., 2003;Horton et al., 2007). To experimentally examine the subcellular localization of CHS3, Arabidopsis mesophyll protoplasts were transformed with either CHS3-GFP or chs3-2D-GFP fusion constructs. GFP fluorescence was detected exclusively in the nucleus of protoplasts for both constructs (Figure 4). No obvious differences in GFP fluorescence subcellular localization were observed.

IDENTIFICATION OF ADDITIONAL chs3 ALLELES
A chs3-2D suppressor screen was undertaken to find mutant alleles causing a reversion to wild type morphology. Two revertant chs3 alleles were identified ( Figure 5A). Sequence analysis revealed that the first allele, chs3-2D-r1, contains a C to T point mutation in the LRR domain resulting in an L716 to F716 substitution while the second allele, chs3-2D-r1, has a G to A point mutation in the LRR-LIM linker region that causes an E1007 to G1007 substitution. An additional chs3 gain-of-function allele, chs3-3D, was identified from a separate mos4 suppressor screen undertaken in our lab (the mos4 suppressor screen will be described in future publications). Similar to chs3-2D, the chs3-3D mutant displays a dwarf, curled leaf phenotype and constitutively expresses the pPR2-GUS www.frontiersin.org reporter gene (Figure A1 in Appendix). This allele contains a C to T point mutation in the LRR-LIM linker region that causes an M1017 to V1017 substitution. The locations of these mutations are presented along with the locations of the chs3-1  and chs3-2D mutations ( Figure 5B).

DISCUSSION
The chs3-2D mutation characterized in this study was shown to suppress the enhanced disease susceptibility phenotypes of npr1-1 through the activation of constitutive defense responses. The chs3-2D npr1-1 double mutant displayed dwarfed, curled leaf morphology similar to that of snc1, a well characterized autoimmune mutant. Additionally, chs3-2D npr1-1 constitutively expressed the pPR2-GUS reporter construct as well as endogenous PR1 and PR2, and accumulated the defense hormone SA. The molecular lesion responsible for these mutant phenotypes is located close to the Cterminal LIM domain of the CHS3 R-like protein, and was shown to be a gain-of-function mutation. The mutation does not seem to alter the nuclear localization of the protein. Two revertant alleles were isolated and an additional gain-of-function mutant allele (chs3-3D) was identified in separate genetic screens. The mutation sites in these gain-of-function and loss-of-function alleles provide further understanding as to the potential function of the enigmatic LIM domain of CHS3.

chs3-2D MUTANT PHENOTYPES SUBTLY DIFFER FROM THOSE OF chs3-1, chs2 /rpp4, AND ssi4
The chs3-1 mutant exhibits enhanced freezing tolerance and constitutive resistance against Pseudomonas syringae pv. tomato (Pst ) DC3000 when grown at 16˚C, yet when grown at 22˚C these phenotypes are mitigated . Another chilling sensitive R protein mutant, chs2, contains a dominant point mutation in the NB-ARC1 domain of RPP4 (Huang et al., 2010). Like chs3-1, the gain-of-function resistance phenotypes of this mutant are only observable at temperatures below 16˚C. Similarly, the chs3-2D   not shown). However, constitutive disease resistance is maintained in chs3-2D npr1-1 mutant plants grown at 22˚C (Figure 1). As well, the chs3-1 mutation is recessive in the Col-0 background while the chs3-2D mutation is dominant, which is more typical of gain-offunction mutations. These data indicate that chs3-2D is a more severe gain-of-function allele of CHS3. The chs3-2D allele may be of use in future genetic studies of resistance signaling as its constitutive defense response phenotypes are more stable, exhibited at both 16 and 22˚C (data not shown).
The ssi4 mutant identified in an earlier npr1 suppressor screen displays an autoimmune phenotype similar to that of chs3-2D, caused by a semi-dominant point mutation in the NB domain of a TIR-NB-LRR R protein (Shirano et al., 2002). However, rather than displaying chilling sensitivity, this mutant is sensitive to humidity; at a relative humidity of 95%, the dwarfed size and constitutive resistance associated with this mutation are abolished (Zhou et al., 2004). Taken together with the chilling sensitive mutants, this indicates that disease resistance and environmental stress response pathways are intricately tied in plants, although the degree of overlap and the precise signaling events required for each pathway remain unclear.

CHS3 IS A NUCLEAR PROTEIN
Wild type CHS3 fused to GFP localizes to the nucleus, and this subcellular localization is unaltered by the chs3-2D mutation (Figure 4). This is not a surprising finding as a number of R proteins, including barley Mla10, tobacco N, and Arabidopsis RPS4 and SNC1, have been shown to localize to the nucleus, although in some cases nuclear localization requires the presence of effector proteins (Burch-Smith et al., 2007;Shen et al., 2007;Wirthmueller et al., 2007;Cheng et al., 2009). As well, a number of LIM proteins localize to the nucleus (reviewed in Kadrmas and Beckerle, 2004). The expression of the CHS3-GFP fusion construct in the Col-0 wild type background confers wild type resistance, whereas chs3-2D-GFP expressed in the wild type background confers constitutive defense response induction (Figure 3). The constitutive resistance phenotypes observed in the chs3-2D-GFP transgenic plants provide further evidence that this protein is correctly localized. The lack of observable GFP fluorescence in transgenic plants indicates that CHS3-2D is of very low abundance but likely has high defense activation activity. We did not observe differences in localization in the protoplast system between wild type and chs3-2D-GFP. However, that does not exclude the possibility that in plants quantitative localization differences contribute to differences in defense response activation.

INSIGHTS INTO LIM DOMAIN FUNCTION
CHS3 belongs to the TIR-NB-LRR subclass of R proteins. In terms of R protein functionality, the presence of the N-terminal TIR domain determines which downstream intermediates will be involved in resistance signaling (Feys et al., 2005). Studies of the flax L6 R protein show that the TIR domain self-associates, likely forming a homodimer, and is required for resistance signaling (Bernoux et al., 2011). The TIR domain has also been implicated in the induction of the cell death response (Frost et al., 2004;Weaver et al., 2006;Swiderski et al., 2009). Likewise, the NB domain appears to play a critical role in resistance signaling and the provocation of cell death (Rairdan et al., 2008). It acts in concert with the ARC1 and ARC2 subdomains to bind nucleotides, and by this means regulates R protein activity (Tameling et al., 2006). Effector www.frontiersin.org recognition specificity possibly derives from LRR domain diversity (Ellis et al., 1999;Dodds et al., 2001;Shen et al., 2003;Rairdan and Moffett, 2006). Intramolecular interactions between the different domains are believed to be crucial for precise R protein functionality (Rairdan and Moffett, 2006).
The N terminus and LRR domain are thought to be involved in the regulation of localized cell death, as suggested by domain swap experiments between tomato R protein homologs Mi-1.1 and Mi-1.2 (Hwang et al., 2000). Studies on Rx, a Solanum tuberosum R protein required for resistance to Potato Virus X, found that autoactivation was conferred upon deletion of the LRR domain or by specific point mutations in the LRR or NB domains (Bendahmane et al., 2002). Domain swap experiments between Rx and the homologous R protein Gpa2 indicate that interactions between the LRR and ARC2 domains may be required for autoinhibition of the R protein in the absence of a pathogen (Rairdan and Moffett, 2006). These experiments set a precedent for the repression of R protein activity via intramolecular interactions. While the detailed functional roles of the above domains are still being elucidated, comparatively little is known about the additional C-terminal LIM domain possessed by CHS3.
Genes encoding LIM domain-containing proteins have been identified in all well-studied eukaryotic genomes, yet most domain characterization studies have been done in animals. While early experiments demonstrated LIM domain-specific DNA-binding activity in vitro (Baltz et al., 1996;Nishiya et al., 1998), LIM domains are thought to serve primarily as protein-protein interaction interfaces and are involved in dimerization (Feuerstein et al., 1994;Schmeichel and Beckerle, 1994;Arber and Caroni, 1996). Animals possess a large number of diverse LIM proteins that have been implicated in a variety of biological processes, including the regulation of actin organization, axon guidance, cell-fate determination, and cell signaling (reviewed in Kadrmas and Beckerle, 2004). Plants contain fewer LIM proteins, and in Arabidopsis LIM proteins are encoded by genes from two distinct subfamilies. The first consists of six genes encoding proteins with two LIM domains that are homologous to animal CRP proteins (Eliasson et al., 2000). These proteins bind actin filaments and regulate actin cytoskeleton organization (Papuga et al., 2010). The genes in the second subfamily encode DA1 and seven DA1-related (DAR) proteins, including CHS3 (DAR4), which are plant-specific proteins each containing a single conserved LIM domain (Li et al., 2008). While DA1 is involved in seed and organ size regulation and CHS3 is associated with resistance signaling and cold response , the role of the LIM domain in modulating these activities remains ambiguous.
The locations of the constitutive resistance-inducing chs3 mutations provide some insight. The chs3-1 allele carries a point mutation in an intron-exon junction in a domain of unknown function close to the LIM domain, possibly resulting in a truncated protein due to splicing discrepancies . The chs3-2D mutation is caused by a C1340 to Y1340 substitution close to the LIM domain. Additionally, a chs3-3D mutation identified in an unrelated mos4 suppressor screen results from an M1017 to V1017 substitution in the LRR-LIM linker region (Figure 5). The distribution of these mutations throughout a number of domains highlights the importance of the C terminus in regulating R-like activity in CHS3. The entire C terminus may be required for autoinhibition, with the regions surrounding the LIM domain playing an essential role in maintaining correct folding of the LIM to confer proper repression. On the other hand, repression may be a result of binding to a negative regulator. The mutation sites may all be located in close proximity in the three-dimensional protein, perhaps as part of a binding site for an interacting repressor protein.
The mutation sites resulting in a reversion to wild type morphology in chs3-2D mutant plants give pause for thought. One revertant contains an L716 to F716 substitution in the LRR domain. Due to the complex intramolecular interactions between the TIR, NB, and LRR domains required for R protein activation, this mutation may result in loss of R protein activity. The other revertant allele contains a G to A point mutation that causes an E1007 to G1007 substitution in the LRR-LIM linker region, in close proximity to the chs3-3D molecular lesion. Plants heterozygous for either chs3-2D-r1 or chs3-2D-r2 in the chs3-2D background exhibit an intermediate morphological phenotype (data not shown), indicating that these mutations are semi-dominant. This is similar to what was observed for the CHS3 loss-of-function T-DNA insertion alleles chs3-2, chs3-3, chs3-4 ( Figure 5B; Yang et al., 2010), thus providing support for chs3-2D-r1 and chs3-2D-r2 being loss-of-function mutations as well. A similar R gene dosage effect has been observed in plants heterozygous for the snc1 mutant, as the stunted morphological trait is recessive while the constitutive PR gene expression trait is semi-dominant (Li et al., 2001).
It is possible that the gain-of-function mutations surrounding the LIM release the self-inhibition of the LIM motif while certain loss-of-function mutations enhance its repressive activity through their different impacts on protein structure and conformation. Alternatively, these mutations may have opposing effects on protein-binding sites. The gain-of-function mutations may weaken binding platforms for negative regulators, while certain loss-of-function mutations may stabilize these interaction sites. It is also possible that the revertant mutations result in a loss of constitutive defense response activation simply by destabilizing the CHS3 protein or disrupting the whole protein structure leading to complete loss-of-function of the protein. Yang et al. (2010) previously hypothesized that the LIM domain acts to repress R protein activity in CHS3 in the absence of pathogenic effector molecules through intramolecular interactions. Indeed, intramolecular autoinhibitory activity of the LIM domain has been observed with a number of animal LIM proteins (Nagata et al., 1999;Garvalov et al., 2003), and as stated above autoinhibition of R protein activity via intramolecular interactions has been observed in Rx (Rairdan and Moffett, 2006). The unique LIM domain of CHS3 may add another layer of complexity to this self-regulation. Thus, in one proposed model for LIM domain regulation of CHS3 R protein activity, the LIM domain binds directly to the N terminus of CHS3 in the absence of a pathogen, preventing R protein activation. Upon the detection of Avirulence (Avr) proteins by the LRR domain a conformational change may take place, alleviating inhibition imposed by the LIM domain. However, Frontiers in Plant Science | Plant-Microbe Interaction as the LIM domain commonly serves as a protein-binding interface in animals (Schmeichel and Beckerle, 1994), it may function similarly in plants. In an alternative model, the LIM domain serves as a binding site for a distinct negative regulator of CHS3 activity in the absence of a pathogen, with this interaction being disrupted by Avr protein detection.

A PUTATIVE MODEL FOR LIM FUNCTIONALITY
There is mounting evidence to suggest that the LIM domain functions in the repression of R protein activity in CHS3 in the absence of pathogens. However, the detailed biochemical mechanism of how this inhibition occurs and under what conditions is still unclear. The extent of the interactions between the LIM domain and the other domains in CHS3 is also not understood. Future structural analysis of CHS3 may lend precise insights into the functional role of its LIM domain.
The npr1-1 suppressor screen was previously described by Gao et al. (2008). Briefly, seeds from npr1-1 plants containing the pPR2-GUS reporter gene were treated with EMS. The seeds from approximately 1300 M1 plants were grown on MS media, yielding approximately 26 000 M2 plants. Half of the seedlings from a given line were tested for a gain of pPR2-GUS reporter gene expression by GUS staining. Seeds were collected from the remaining plants of those lines displaying constitutive pPR2-GUS expression. The M2 plants of these lines were sprayed with an H.a. Noco2 suspension (50 000 spores/mL) and scored 7 days post-treatment.
SA extraction and measurement were carried out using HPLC according to procedures described previously (Li et al., 1999).

MOLECULAR CLONING
The full-length CHS3 gene including the promoter region but lacking the stop codon was PCR amplified from Col-0 genomic DNA and cloned into a pCAMBIA1305-GFP vector in-frame to the N terminus of GFP using KpnI and BamHI restriction sites. The CHS3-GFP expression clone was sequenced to confirm in-frame fusion. Transformation was done using the floral dip method (Clough and Bent, 1998), and transformants were selected on MS medium supplemented with 30 μg/mL hygromycin. Oneweek-old seedlings were used to prepare protoplasts and confocal microscopy was used to analyze GFP fluorescence as previously described (Yoo et al., 2007). The same procedure was carried out for chs3-2D.

ACKNOWLEDGMENTS
We would like to thank Oliver Dong and Virginia Woloshen for careful reading of the manuscript. This work was supported by grants from the Chinese Ministry of Science and Technology to Yuelin Zhang and from the National Science and Engineering Research Council (NSERC) to Xin Li. Kaeli C. M. Johnson is supported by an NSERC Alexander Graham Bell Canada Graduate Scholarship (M). www.frontiersin.org