Plant cells induce an array of defense responses upon perception of pathogen-derived molecules. General elicitors that are characteristic of whole classes of microorganisms, such as bacterial flagellin, can trigger resistance (Gómez-Gómez and Boller, 2002). However, disease resistance is often governed by a gene-for-gene interaction (Dangl and Jones, 2001), in which plants carrying a resistance (R) gene specifically recognize a pathogen carrying a corresponding avirulence (Avr) gene. R gene–mediated resistance, or race-specific resistance, is typically accompanied by rapid localized cell death at the site of infection called the hypersensitive response (HR).

Cladosporium fulvum is a biotrophic fungus that causes leaf mold in tomato (Lycopersicon esculentum). The tomato–C. fulvum interaction is useful for studying the molecular basis of race-specific resistance (Rivas and Thomas, 2002). Tomato Cf genes confer resistance to specific races of C. fulvum through recognition of Avr peptides that are secreted into the leaf apoplast during infection. For example, the R genes Cf-4 and Cf-9 from tomato confer resistance to strains of C. fulvum that express Avr4 and Avr9, respectively. Avr4 and Avr9 encode precursor proteins that are processed upon secretion to mature proteins of 86 and 28 amino acids, respectively (Van den Ackerveken et al., 1993; Joosten et al., 1994). Cf genes encode type I transmembrane glycoproteins carrying extracytoplasmic Leu-rich repeats (LRRs), a single pass membrane-spanning region, and a short cytoplasmic domain that has no similarity to known signaling domains (Thomas et al., 1998). The Cf-4 and Cf-9 proteins have identical C-terminal regions with sequence variation confined to their N-terminal LRRs. Extensive binding studies have failed to detect a direct interaction between Cf-9 and Avr9 (Luderer et al., 2001), suggesting that perception of Avr9 by Cf-9 is indirect and possibly involves a high affinity binding site that has been detected in membranes of solanaceous plants (Kooman-Gersmann et al., 1996).

Cf-4 and Cf-9 confer an Avr-dependent HR in tobacco (Hammond-Kosack et al., 1998; Thomas et al., 2000). We used Cf-9–transgenic tobacco and derived cell suspension cultures to characterize a variety of early responses to the Avr9 elicitor (Rivas and Thomas, 2002). These responses include changes in ion fluxes (Piedras et al., 1998; Blatt et al., 1999), production of active oxygen species (Piedras et al., 1998), and the activation of mitogen-activated protein kinases and a calcium-dependent protein kinase (CDPK) (Romeis et al., 1999, 2000, 2001). These events are accompanied by massive changes in gene expression (Durrant et al., 2000), which likely lead to a reprogramming of the cell toward defense. However, the molecular mechanisms by which Cf-9 activates these responses are not well understood.

Many signaling components acting downstream of R genes have been identified by forward genetic screens in tomato, barley (Hordeum vulgare), and especially in the model plant Arabidopsis thaliana (Martin et al., 2003). In tomato, Rcr-1 and Rcr-2 were found to be required for the function of Cf-9 (Hammond-Kosack et al., 1994), but the phenotypes associated with mutations in these genes did not allow molecular identification of the genes. The Rcr3 gene product is required for Cf-2 function but not the function of other Cf genes (Dixon et al., 2000). Rcr3 encodes a secreted Cys protease and likely functions upstream of Cf-2 (Krüger et al., 2002). Reverse genetics has been more successful in identifying genes required for Cf-9 function. Virus-induced gene silencing (VIGS) in Nicotiana benthamiana was used to show that silencing of the NtCDPK2 gene family causes a reduced HR mediated by Cf-4 and Cf-9 (Romeis et al., 2001). The HRs elicited by Cf-4/Avr4 and Cf-9/Avr9 were also suppressed by VIGS of SGT1 (Peart et al., 2002b). SGT1 is required for the function of SCF (Skp1-Cul1-F-box) ubiquitin ligases and other multiprotein complexes, indicating that ubiquitylation is important for Cf-mediated defense responses. VIGS of CITRX, which encodes a thioredoxin that interacts with the C-terminal cytoplasmic domain of Cf-9, causes an accelerated Cf-9/Avr9-triggered HR and causes increased resistance of tomato to C. fulvum (Rivas et al., 2004). The negative regulatory role of CITRX is important for the defense response of Cf-9 but not Cf-2.

In this study, we sought to identify additional components required for Cf-mediated defense responses. We previously used cDNA-amplified fragment length polymorphism (AFLP) analysis to identify transcripts whose expression patterns are rapidly altered during the Cf-9/Avr9-mediated defense response in tobacco (N. tabacum) cell cultures (Durrant et al., 2000). Many Avr9/Cf-9 rapidly elicited (ACRE) genes are predicted to encode regulatory proteins, including protein kinases and transcription factors. We have reannotated the ACRE genes and used VIGS to silence 42 ACRE genes. We have identified three genes that are important for generating the HRs elicited by Cf-9/Avr9 and Cf-4/Avr4. One of these genes encodes a protein kinase called ACIK1. ACIK1 was found to be an important component of the Cf-9 and Cf-4 signaling pathways but not other R gene pathways.

We previously reported sequence similarity for 37 induced and five repressed cDNA-AFLP fragments (42 in total) (Durrant et al., 2000). We have refined the BLAST search criteria and recently searched updated genomic and EST databases for all 257 sequenced fragments. We used a two-step strategy: first, we searched nucleotide sequence databases with ACRE gene sequences, and when these revealed tomato or tobacco EST homologs, a longer sequence could then be used to search the Arabidopsis, rice (Oryza sativa), and other protein sequence databases. We have found similarities for 136 additional cDNA-AFLP fragments (178/257 in total), which represent ~110 distinct genes (see Supplemental Table 1 online). Based on homology, transcript tags were classified into functional groups (Table 1; see Supplemental Table 1 online). Prominent amongst these groups are putative regulators of protein degradation (F-box, U-box, RING-H2, and BTB/POZ containing proteins), transcription factors (WRKY, Myb, C2H2 zinc finger, and AP2-type), and protein kinases, including receptor-like kinases. Other interesting examples include genes predicted to encode calcium binding proteins, protein phosphatases, lipases, an NDR/HIN1-like protein, AAA-type ATPases, a ligand-gated ion channel, a Cf-type R protein, a synaptobrevin-like protein, and a subunit of the Exo70 exocytosis complex.

Table 1. Functional Classification of cDNA-AFLP Tag Sequences Altered in Cf-9/Avr9 Response

This updated bioinformatics study provides a much more complete picture of the types of genes elicited by Cf-9/Avr9 than was previously described (Durrant et al., 2000) and allowed a more informed choice of which ACRE genes to use for functional studies.

Table 2. Full-Length and Partial ACRE cDNA Sequences Used for VIGS

Figure 1. Identification of ACRE Genes Required for Cf-9– and Cf-4–Dependent HR.

The predicted NtACIK1 protein contains 442 amino acids with a molecular mass of 50.2 kD. A Blastp search revealed that NtACIK1 has all 11 subdomains required for Ser/Thr protein kinase activity (Figure 2B). The kinase domain is flanked by two short nonkinase domains that have no obvious similarities to proteins of known function (Figure 2A). Because the protein sequence does not contain any membrane-spanning region or signal peptides, it is likely that NtACIK1 is either a cytoplasmic or nuclear kinase. Alignments of NtACIK1 with putative orthologs in tomato (LeACIK1) and Arabidopsis (AtACIK1a [At2g05940] and AtACIK1b [At5g35580]) showed an overall amino acid sequence identity of 60 to 75% (Figure 2B). The strongest similarity between the sequences was found in the kinase domains, which were 80 to 87% identical. Both the flanking N-terminal and C-terminal regions are highly divergent, sharing only 31 to 40% identity.

Figure 2. Sequence of NtACIK1 Protein and Alignments of Homologs.

Database searches revealed that the kinase domain of NtACIK1 is similar to those of receptor-like kinase domains. In Arabidopsis, ~620 of the ~1100 predicted protein kinases are receptor-like (Shiu and Bleecker, 2001). Many kinase sequences classified as receptor-like kinases lack an apparent signal sequence or transmembrane domain and are termed receptor-like cytoplasmic kinases. NtACIK1 and its orthologs are most related to a subgroup of 46 Arabidopsis kinases called receptor-like cytoplasmic kinase-VII (Shiu and Bleecker, 2001; Shiu et al., 2004). Members of this subgroup include NAK, APK1a, APK1b, APK2a, and ARSK1, which are a set of partially characterized protein kinases of unknown function (Hirayama and Oka, 1992; Moran and Walker, 1993; Hwang and Goodman, 1995; Ito et al., 1997). This subgroup includes PBS1, which is an Arabidopsis protein kinase required for recognition of AvrPphB from Pseudomonas syringae pv phaseolicola (Swiderski and Innes, 2001). MLPK, a membrane-anchored protein kinase involved in Brassica self-incompatibility signaling (Murase et al., 2004), is also highly related in sequence to this subgroup. The relationship between ACIK1 and other plant protein kinases was analyzed by construction of a phylogenetic tree (Figure 2C). Although several programs were used to estimate phylogenies, the tree shown was built using Bayesian evolutionary tree estimation (Sjölander, 1998). Within the clade containing ACIK1 proteins (node N28), all trees were similar; a neighbor joining tree gave bootstrap values of 73%, joining ACIK1 sequences to ARSK1 (node N26), and 82% joining to the clade containing the NAK kinase (node N27). The Bayesian evolutionary tree estimation produced the tree topology most widely recognized as being correct for the more diverse kinases. In summary, this analysis indicated that NtACIK1 and its orthologs cluster with ARSK1 (Hwang and Goodman, 1995), but in a branch distinct from NAK, MLPK, and PBS1.

Figure 3. Expression Patterns of NtACIK1 and LeACIK1 after Elicitation by Avr9 and Wounding.

We also investigated the expression pattern of the putative NtACIK1 homolog from tomato, LeACIK1 (Figure 2B). The full-length cDNA sequence of a tomato homolog, LeACIK1, was identified in GenBank (AF332960). As observed for NtACIK1, LeACIK1 was strongly induced in tomato plants by both wounding [infiltration with control IF(−Avr9)] and Avr9-elicitation [infiltration with IF(+Avr9)] (Figure 3C). We therefore concluded that LeACIK1 might be the tomato homolog of NtACIK1.

Figure 4. LeACIK1 Is a Functional Protein Kinase.

Figure 5. ACIK1 Is Required for Both Cf-9– and Cf-4–Dependent HRs in N. benthamiana and Tobacco.

We also investigated the effect of silencing NtACIK1 in Cf-9 transgenic tobacco plants. Because VIGS cannot be used in N. tabacum, we generated NtACIK1-RNAi hairpin constructs under the control of the 35S promoter for gene silencing in tobacco leaf tissue. As a negative control, we used a nonhairpin construct that contained one arm of NtACIK1 cDNA and one arm of NtCDPK2 cDNA (Romeis et al., 2001), which cannot form a hairpin structure for silencing. These constructs were transiently delivered into tobacco leaves by agroinfiltration. Four days after agroinfiltration, tobacco leaves were infiltrated with IF(+Avr9) and scored for development of the HR. In four out of six independent experiments, we observed strongly reduced HR symptoms in the leaf half expressing the NtACIK1-hairpin construct as compared with the control leaf half (Figure 5C). This again correlated with weaker stress-induced NtACIK1 transcript accumulation in the silenced leaf half (Figure 5D). Thus, using a second independent and nonviral silencing method, we could confirm the requirement of NtACIK1 for the Cf-9/Avr9-induced HR.

Figure 6. ACIK1 Specifically Regulates the Cf-9– and Cf-4–Dependent HRs.

To further test the specificity of ACIK1 function, N- and Rx-mediated resistance was tested in N. benthamiana plants silenced for NbACIK1. The N gene encodes an NB-LRR R protein that mediates resistance to Tobacco mosaic virus (TMV). Transgenic N. benthamiana plants expressing either the N gene (Peart et al., 2002a) or the Rx gene (Bendahmane et al., 1999) were used in these experiments. Inoculation of GFP-tagged strains of TMV (TMV-GFP) or PVX (PVX-GFP) onto N. benthamiana plants results in widespread growth of either virus that can be observed under UV light as accumulation of GFP fluorescence (Peart et al., 2002a). No green fluorescence was detected after inoculation of N-transgenic N. benthamiana with TMV-GFP or after inoculation of Rx-transgenic N. benthamiana with PVX-GFP (Figure 6, top row, columns 5 and 6; Peart et al., 2002a). In three independent experiments, each using six leaves, no effect on TMV or PVX spread was observed in NbACIK1 silenced plants because in neither case was green fluorescence detected in the inoculated leaves 5 DAI (Figure 6, middle row, columns 5 and 6). By contrast, plants that were silenced with TRV:SGT1 showed pronounced GFP fluorescence (Figure 6, bottom row, columns 5 and 6). ACIK1 therefore does not appear to play a role in resistance to TMV or PVX.

Taken together, these results indicate that ACIK1 function is likely to be specific for Cf-9– and Cf-4–mediated defense responses.

Figure 7. LeACIK1 Is Required for Cf-9–Mediated Resistance to C. fulvum Expressing Avr9.

Table 3. Proportion of TRV-Silenced Tomato Leaves Showing HR or C. fulvum Growth

To test the effect of silencing on Cf-9–mediated resistance, tomato plants were dip-inoculated with C. fulvum spores 3 weeks after silencing induction (Figure 7B, Table 3). The C. fulvum strain used expresses the β-glucuronidase (GUS) gene to allow easy visualization of fungal growth. After an additional 3 weeks, leaves were stained with 5-bromo-4-chloro-3-indoyl-β-glucuronide (X-gluc) to score fungal growth macroscopically and microscopically. As expected, no X-gluc staining and no fungal growth could be observed in Cf-9 tomato plants that were either untreated or silenced with the empty vector control (Figure 7B; data not shown). By contrast, Cf0 tomato plants that were untreated or silenced with the control TRV:00 had extensive fungal growth as detected by X-gluc staining (Figure 7B; data not shown). The dark blue stain was a result of fungal hyphae, growing intracellularly and emerging again through the stomata. Leaves from Cf-9 plants silenced with either TRV:LeACIK1 or TRV:Cf-9 showed significantly increased staining as compared with the empty vector controls, indicating that silencing of LeACIK1 caused a breakdown of Cf-9–mediated resistance. Fungal hyphae could be detected in LeACIK1- and Cf-9–silenced leaves under the microscope, although in both cases considerably less hyphae were observed than in infected Cf0 plants (Figure 7B, bottom panels). Also, no reproductive structures emerging from the stomata were observed, indicating that LeACIK1 and Cf-9 silencing only cause partial loss of resistance. Silencing with the NtACIK1 sequence had no effect on Cf-9–mediated resistance. These results were obtained in three independent experiments using multiple leaves from two to three silenced tomato plants. We confirmed that LeACIK1 mRNA and Cf-9 mRNA were reduced in the respective silenced leaves using RT-PCR (Figure 7C). Notably, under our conditions, silencing in tomato plants seemed to be less effective relative to N. benthamiana, and the reductions of LeACIK1 and Cf-9 mRNA levels were always partial.

In this study, we focused on the role of the protein kinase ACIK1 in the plant defense response. Future studies will focus on the roles of ACRE-189 and ACRE-276, which are predicted to be involved in protein degradation via ubiquitylation.

So far, we cannot distinguish if the required ACIK1 function is a result of preexisting protein or protein derived from induced transcript. However, it is likely that important regulation of ACIK1 activity is achieved through posttranslational modifications that occur upon elicitation, involving changes in protein kinase activity, subcellular localization, and/or protein stability. Unfortunately, we were unsuccessful in detecting endogenous ACIK1 protein in planta using antibodies generated against NtACIK1 (data not shown). It is possible that ACIK1 protein levels are maintained at relatively low levels. We were also unable to detect elicitor-induced changes in ACIK1 protein kinase activity or protein localization using transiently expressed epitope-tagged versions of ACIK1 (data not shown).

VIGS can be effective with as little as 25 nucleotides of sequence identity between the insert in the virus vector and the target RNA (Thomas et al., 2001). Because we used 5′- or 3′-RACE ACIK1 cDNA fragments for VIGS in N. benthamiana, we cannot exclude silencing of other closely related genes. The same would apply for the gene-silencing experiments using transient expression of an NtACIK1-hairpin construct because 550 bp of coding sequence was used that included part of the region encoding the kinase domain. However, for VIGS in tomato we used only the last 150 bp of sequence corresponding to the nonconserved region at the C terminus of the protein beyond the kinase domain. Searches of the tomato EST database (The Institute for Genomic Research) revealed no significant sequence identities with this region. We therefore did not expect any cross-silencing of related kinases, and the effect of VIGS is likely to be specific for LeACIK1.

The kinase domains of ACIK1 orthologs cluster with receptor-like kinases (Shiu and Bleecker, 2001). Receptor-like kinases have been implicated in both race-specific and general defense responses. The Xa21 gene from rice specifies race-specific resistance against Xanthomonas oryzae (Song et al., 1995), whereas the Arabidopsis FLS2 gene is required for recognition of the general bacterial elicitor flagellin (Gómez-Gómez and Boller, 2000). The Xa21 and FLS2 proteins each consist of extracellular LRRs, a single-pass transmembrane domain, and a cytoplasmic Ser/Thr kinase domain. Because Cf-9 does not contain any apparent signaling domains, we expect that at least one additional protein provides the signaling function. In this scenario, Cf-9 and ACIK1 could be part of a receptor-like kinase complex, with Cf-9 providing the extracellular LRRs for recognition and ACIK1 providing the cytoplasmic kinase for signaling. Such an interaction between a transmembrane protein containing an extracellular LRR domain and a protein kinase has been well described. In Drosophila, the Toll protein interacts via an adaptor complex with the Pelle protein kinase (Hoffmann and Reichhart, 2002). Similarly, in mammals, Toll-like receptors such as TLR2, TLR4, or TLR5 interact via the adaptor molecule MyD88 with the protein kinase IRAK (Akira et al., 2001; Underhill and Ozinsky, 2002) to form functional receptor-like kinases.

Another possible role for LeACIK1 is the phosphorylation of negative regulators of the signaling pathway leading to defense responses. Our laboratory recently identified a thioredoxin, CITRX, as an interactor of the Cf-9 resistance protein and as a negative regulator of Cf-9–mediated signaling responses in tomato (Rivas et al., 2004). To activate the corresponding signaling pathways, negative regulators such as CITRX have to be inactivated or degraded. Proteolysis in the eukaryotic cytosol typically involves substrate ubiquitylation mediated by, for instance, the SCF complex or other E3 ligases. Polyubiquitylated proteins are then targeted for degradation via the 26S proteasome (Hochstrasser, 1995). It is known that many substrates of SCF complexes must be phosphorylated before they can bind to the SCF complex and become polyubiquitylated (Deshaies, 1999; Kumar et al., 2003). The requirement for substrate phosphorylation creates an opportunity for differential regulation of the stability of SCF substrates. Elicitation would result in an increase of ACIK1 protein kinase activity, leading to the phosphorylation and specific degradation of its substrates and induction of the corresponding signaling pathways.

A third possible role for ACIK1, which is not mutually exclusive with the first two options, is as a component of a protein kinase cascade. Two well-studied kinase families participating in phosphorylation cascades during plant defense responses are the mitogen-activated protein kinases (MAPKs) and the CDPKs. In the Cf-9/Avr9 system, Romeis et al. (1999) described the activation of the two MAPKs SIPK (salicylic acid–induced protein kinase) and WIPK (wound-induced protein kinase). It has been shown that suppression of certain MAPK activities resulted in reduced plant defense responses. In tomato, Pto-mediated resistance was compromised by silencing the genes encoding the two MAPK kinases MEK1 and MEK2 and the two MAPKs WIPK and NTF6 (Ekengren et al., 2003). In N. benthamiana, silencing of either SIPK or WIPK compromised N gene–mediated TMV resistance (Jin et al., 2003), and silencing of both SIPK and WIPK reduced the resistance to the bacterial pathogen Pseudomonas cichorii (Sharma et al., 2003). In Arabidopsis, silencing of MPK6 reduced basal and race-specific resistance to various pathogens (Menke et al., 2004). Similarly, CDPKs were shown to be required for plant defense responses. In Cf-9 transgenic tobacco plants, NtCDPK2 was activated after Avr9 elicitation, and VIGS of NtCDPK2 in N. benthamiana plants resulted in reduced Cf-9– and Cf-4–mediated HRs (Romeis et al., 2001). The tomato Pto kinase interacts with the Ser/Thr kinases Pti1 (Zhou et al., 1995), Adi2, and Adi3 (Bogdanove and Martin, 2000). This suggests that Pto is involved in phosphorylation cascades independent of the described MAPK or CDPK pathways. It is possible that ACIK1 may also take part in an as yet unidentified signaling cascade. Future work on the identification of the downstream targets and of interacting proteins will lead to a better understanding of the role of ACIK1 in the plant defense response.

For viral inoculation, second leaves of 2- to 3-week-old N. benthamiana seedlings were pierced with a toothpick that was previously touched to an Agrobacterium colony. After an additional 3 to 4 weeks, during which PVX spread systemically, the third or fourth leaves were analyzed for an HR phenotype. Individual leaves were infiltrated on one side with Agrobacterium containing binary vectors expressing Cf-9 and Avr9 (p9/456/Avr9) or expressing Cf-4 and Avr4 (p3/456/Avr4) (Thomas et al., 2000).

Infection of plants by agroinfiltration was done as previously described (Ratcliff et al., 2001), except that cultures containing pTV00-derived constructs were mixed with those containing pBINTRA6 in a 10:1 ratio before infiltration. The fourth and fifth leaves above the viral inoculated leaves were analyzed for ACIK1 transcript levels and for defense-related phenotypes. For elicitation of the HR, R genes and their cognate avirulence genes were coexpressed using Agrobacterium-mediated transient expression. The binary constructs containing the various R/Avr combinations have been described elsewhere: Cf-9/Avr9, Cf-4/Avr4 (Thomas et al., 2000), Pto/AvrPto (Peart et al., 2002b), and Rx/PVX-coat protein (Peart et al., 2002a). TMV-GFP and PVX-GFP inoculation of transgenic N. benthamiana plants carrying the N or the Rx gene, respectively, as well as visual detection and imaging of GFP fluorescence were performed as described by Peart et al. (2002a).

Three to four weeks after the onset of silencing, plants were treated with paclobutrazol and dip-inoculated with 5 × 10⁶ spores of Cladosporium fulvum race 4 GUS as previously described (Balint-Kurti et al., 1994; Thomas et al., 1997). Fungal growth was scored 3 weeks later by GUS staining and analyzed by light microscopy (Axiophot; Zeiss, Jena, Germany). In parallel, leaves of silenced tomato plants were infiltrated with IF with or without Avr9 and were analyzed for development of HR.

Sequence data from this article have been deposited with the EMBL/GenBank data libraries under accession numbers listed in Table 2. The accession numbers for the other sequences described in this article are as follows: NtACIK1 (AY220481), LeACIK1 (AF332960), AtACIK1a (AAL07094), and AtACIK1b (NP_198408).

We thank David Baulcombe's laboratory for the PVX and TRV vectors used for VIGS and for providing the GFP-PVX and GFP-TMV constructs as well as N- and Rx-transgenic N. benthamiana plants. We also thank Shelley Hepworth, Thorsten Nürnberger, and Frédéric Brunner for helpful comments on the manuscript. This research was supported by the Gatsby Charitable Foundation. O.R. was supported by a Human Frontiers Science Program fellowship. A.A.L. was supported by a European Community Marie Curie Fellowship (EC Grant HPMF/CT-2001-01288). L.F.-L. was supported in part by a National Science Foundation grant to K.S.