To conduct a comparative analysis of the roles of MAPKKK genes in defense responses, we first sought to clone N. benthamiana MAPKKK genes belonging to group A. Group A includes four subgroups, A1–A4, and contains many genes involved in plant defense responses. To obtain one N. benthamiana MAPKKK gene homolog from each group A subgroup, we selected the following: the Arabidopsis MAPKKK genes AtMEKK1 20 , AtMAPKKKγ, and AtMAPKKKε1 21 22 as representatives of the A1, A2, and A4 subgroups, respectively. For the A3 subgroup, which includes NtNPK1 23 and AtANP1, we were able to amplify N. benthamiana cDNA fragments using NtNPK1-specific primers; the amplified gene turned out to be 98.0% identical in nucleotide sequence to that encoding the NtNPK1 kinase domain, which has been well characterized for its role in defense responses 7 . Therefore, the A3 subgroup was excluded from further study.

Using the nucleotide sequences encoding the highly conserved kinase domains of AtMEKK1 19 , AtMAPKKKγ, and AtMAPKKKε1 as queries for BLAST searches against the N. benthamiana, N. tabacum, and Solanum lycopersicum EST databases, we obtained three EST sequences: N. benthamiana TC15397, N. tabacum BP133312, and S. lycopersicum BI931567, respectively. Based on these sequences, we designed specific primers to isolate full-length N. benthamiana MAPKKK cDNA clones, as described in the Materials and Methods. The cloned N. benthamiana MAPKKK genes are predicted to encode proteins with highly conserved kinase domains and more divergence in other regions, a general feature of plant MAPKKK genes 4 .

Next, we used the amino acid sequences of the kinase domains of our newly cloned MAPKKKs and those of other previously identified MAPKKKs to construct a phylogenetic tree (Figure 1A). Our newly cloned MAPKKK genes were tentatively designated after the name of the most closely related genes in the phylogenetic tree. Thus, our A2 MAPKKK homolog was designated NbMAPKKKγ after A. thaliana AtMAPKKKγ. At a late phase of this study, an A4 MAPKKK homolog, N. benthamiana NbMAPKKKε, was reported 9 . Although our A4 homolog shares high sequence similarity with NbMAPKKKε, the genes were not identical (96.6% or 95.0% identity at the nucleotide or amino acid level, respectively) and differed in length. Because these results indicated that A4 MAPKKK homologs in N. benthamiana consisted of at least two genes, our A4 homolog was designated NbMAPKKKε2 after NbMAPKKKε. In fact, from the recently released N. benthamiana draft genome sequence (http://solgenomics.net/), only two contigs that show high sequence homology (about ~90%) with NbMAPKKKε2 were retrieved and each of these sequences corresponded to NbMAPKKKε2 and NbMAPKKKε, respectively. Although our A1 MAPKKK homolog was most closely related to MsOMTK1, which was designated after the functional features of its gene products, designating this gene “NbOMTK1” was apparently incorrect, as we do not know if our A1 homologous gene product and MsOMTK1 have a similar function. The next most closely related genes were AtMEKK1 and Brassica napus BnMAPKKKβ1. Therefore, our A1 homolog was designated NbMAPKKKβ after BnMAPKKKβ1, a gene of the A1 subgroup, in accordance with other newly cloned genes NbMAPKKKγ and NbMAPKKKε2. The NbMAPKKKβ, NbMAPKKKγ, and NbMAPKKKε2 cDNA sequences determined in this study were deposited in the DNA Data Bank of Japan (DDBJ) under the accession numbers AB649283, AB649284, and AB649285, respectively.

The domain structures of the three novel MAPKKK genes are shown in Figure 1B. The NbMAPKKKβ, NbMAPKKKγ, and NbMAPKKKε2 cDNA are 2477, 2794, and 4708 bp in length, respectively, and are respectively predicted to encode proteins of 564, 651, and 1395 amino acids. Kinase domains are found in the C-terminal regions of NbMAPKKKβ and NbMAPKKKγ and the N-terminal region of NbMAPKKKε2. The C-terminal region of NbMAPKKKε2 contains two ARM (armadillo/β-catenin-like repeat) domains. The domain structures of these proteins are similar to those of their respective A. thaliana homologs.

Overexpression of the kinase domain of the tomato homolog of NbMAPKKKα, SlMAPKKKα, in N. benthamiana leaves by agroinfiltration has been reported to induce pathogen-independent cell death 8 . To investigate whether our three newly identified MAPKKKs also possessed cell death-inducing activity, we performed transient overexpression analysis of each kinase domain using agroinfiltration. The transient overexpression of the kinase domains of NbMAPKKKβ and NbMAPKKKγ induced pathogen-independent cell death in the infiltrated area (Figure 2A). In the infiltrated areas, cell death was associated with significant hydrogen peroxide production, which is detected by a characteristic brown color that emerges upon 3,3'-diaminobenzidine (DAB) staining. These results were confirmed by measuring the extent of cell death using an ion leakage assay (Figure 2B), which showed significantly increased ion leakage in the NbMAPKKKβ- and NbMAPKKKγ-overexpressing areas. No significant difference in the level of ion leakage was observed for NbMAPKKKβ vs. NbMAPKKKγ. In contrast, overexpression of the kinase domain of NbMAPKKKε2 failed to induce cell death or hydrogen peroxide production and failed to increase the level of ion leakage. Western blot analysis using anti-Myc monoclonal antibody (Millipore, Billerica, MA, USA) was performed to assess the expression levels of these kinase domains (Figure 2C). Specific signals of NbMAPKKKβ and NbMAPKKKε2 kinase domains were detected, but no accumulation of NbMAPKKKγ kinase domain was found. However, we suppose that the NbMAPKKKγ kinase domain was accurately expressed, even at very low levels, because its overexpression rapidly induced cell death. Western blot analysis also showed that the accumulated expression level of NbMAPKKKε2 kinase domain, whose overexpression did not induce cell death, was much greater than the expression levels of the NbMAPKKKβ and NbMAPKKKγ kinase domains. Because these results indicated that NbMAPKKKε2 is not involved in the induction of pathogen-independent cell death, we excluded NbMAPKKKε2 from further analysis.

To further investigate the involvement of NbMAPKKKβ and NbMAPKKKγ in pathogen-independent cell death, we conducted transient overexpression experiments using their full-length open reading frames (ORFs). As shown in Figure 2D, overexpression of NbMAPKKKβ or NbMAPKKKγ induced cell death. The observed level of ion leakage was similar for overexpression of NbMAPKKKβ, NbMAPKKKγ, and NbMAPKKKα, which was used as a positive control (Figure 2B).

Several previous studies have demonstrated that kinase activity is required for the triggering of pathogen-independent cell death by transient overexpression of a component of a MAPK cascade 8 24 25 . To examine whether NbMAPKKKβ- and NbMAPKKKγ-induced cell death also required kinase activity, we constructed full-length and kinase domain MAPKKK mutants deficient in ATP binding. In these mutants, the essential conserved lysine residue (K) in the ATP-binding site of the kinase domain was replaced with a methionine (M) 8 . As shown in Figure 2A and 2D, none of these K → M mutants induced cell death when overexpressed. This result was confirmed by DAB staining (Figure 2A and 2D) and ion leakage assays (Figure 2B). Also, the expression of these kinase domain K → M mutants was confirmed by western blot analysis with anti-Myc monoclonal antibody (Figure 2C). These results indicate that cell death induced by the transient overexpression of N. benthamiana NbMAPKKKβ or NbMAPKKKγ is dependent on their kinase activities.

In light of the above finding that overexpression of full-length NbMAPKKKβ or NbMAPKKKγ or their kinase domains can induce cell death, we expected that silencing of the genes encoding these MAPKKKs would suppress virally induced PCD. Therefore, we used a TRV-based VIGS system 17 to silence NbMAPKKKβ and NbMAPKKKγ. Successful silencing of each gene was confirmed by analyzing the expression of NbMAPKKKβ, NbMAPKKKγ, and NbMAPKKKα (control) using real-time reverse transcription (RT)-PCR analysis (Figure 3A). Also, Southern blot analyses using a kinase domain-specific probe of each gene revealed that NbMAPKKKβ and NbMAPKKKγ exist as single-copy and multiple-copy genes, respectively, in the N. benthamiana genome (Additional file 1: Figure S1). In the N. benthamiana draft genome, two contigs showing high sequence homology (about >90%) with NbMAPKKKγ were obtained, suggesting that there are at least two copies of NbMAPKKKγ homologs. Thus, we assume that a single gene is specifically targeted in NbMAPKKKβ-silenced plants, but one or more highly similar genes might be targeted in NbMAPKKKγ-silenced plants. Hereafter, we use the term “NbMAPKKKγ” to indicate NbMAPKKKγ and/or its homologous gene(s) except for transient overexpression experiments. The gene-silenced plants did not display any obvious phenotypic differences compared with non-silenced (VIGS vector) control plants (Figure 3B), suggesting that neither NbMAPKKKβ nor NbMAPKKKγ is involved in normal plant growth and development. In contrast, silencing of NtNPK1, a MAPKKK involved in cell plate formation in plant cytokinesis and in N gene-mediated HR cell death, causes severe stunting of plants 7 .

NbMAPKKKβ- and NbMAPKKKγ-silenced plants and control non-silenced plants were subsequently inoculated with PlAMV-Li1, which induces PCD-associated necrosis in N. benthamiana. In the PlAMV-Li1-infiltrated areas of non-silenced plants, we observed characteristic necrotic symptoms and brown color staining, indicating the accumulation of hydrogen peroxide. However, in the Li1-infiltrated NbMAPKKKβ-silenced plants, cell death and hydrogen peroxide production were completely compromised. Also, in NbMAPKKKγ-silenced plants, cell death and hydrogen peroxide production were suppressed completely and partially, respectively, despite only about 20% decrease in the abundance of NbMAPKKKγ transcripts (Figure 3C). Suppression of cell death in the NbMAPKKKβ- and NbMAPKKKγ-silenced plants was confirmed using ion leakage assays (Figure 3D). The suppression of cell death in NbMAPKKKγ-silenced plants despite the slight decrease in its mRNA can be explained by a strict requirement of a high level of NbMAPKKKγ protein in cell death. Although it is possible that residual NbMAPKKKγ protein contributes to the decreased level of hydrogen peroxide production, its level might be too low to induce cell death. These results indicate that NbMAPKKKβ and NbMAPKKKγ are involved in hydrogen peroxide production and PCD induced by PlAMV-Li1.

Together with the findings of our previous study showing that NbMAPKKKα is involved in PlAMV-Li1-induced PCD 19 , the above findings demonstrate that NbMAPKKKβ and NbMAPKKKγ are also essential for this virally induced PCD. To determine the relationships among these three genes in the PCD signaling pathway, we designed epistasis experiments combining transient overexpression and silencing of combinations of these genes. In these experiments, plants with a specific VIGS-silenced MAPKKK gene were agroinfiltrated with an Agrobacterium culture expressing a different MAPKKK gene, and the level of ion leakage was measured. Prior to the epistasis experiments, a preliminary experiment was performed to determine the minimum turbidity of Agrobacterium inoculum for each gene sufficient to induce complete cell death. Complete cell death was observed at the following cell densities at OD₆₀₀: 0.5, NbMAPKKKα; 1.0, NbMAPKKKβ; and 0.05, NbMAPKKKγ. Although each MAPKKK was overexpressed at different concentrations of Agrobacterium inocula, no significant difference in the basal ion leakage level was observed among the different turbidities (ranging from 0.05 to 1.0) of Agrobacterium inocula that expressed the GUS gene in wild-type plants (Additional file 2: Figure S2). For NbMAPKKKβ, cell death induced by both the full-length ORF and kinase domain was slow and weak, but cell death induced by the kinase domain was more often observed than that induced by the full-length ORF. In the ion leakage assay shown in Figure 2B, this tendency was detected, although the difference in cell death induced by the full-length ORF and kinase domain of NbMAPKKKβ was not statistically significant. Previous reports have shown that the substrate specificity of MAPKKK is not affected when only the kinase domain is transiently activated 5 8 26 . Hence, the kinase domain of NbMAPKKKβ was used for further analysis, instead of the full-length ORF.

As shown in Figure 4A, ion leakage levels induced by transient expression of the NbMAPKKKβ kinase domain or full-length NbMAPKKKγ were lower in NbMAPKKKα-silenced plants than in VIGS vector-only control plants. Cell death induced by both NbMAPKKKβ and NbMAPKKKγ was consistently significantly suppressed in NbMAPKKKα-silenced plants compared to control plants (data not shown). This result suggests that NbMAPKKKα either acts downstream of both NbMAPKKKβ and NbMAPKKKγ in cell death induction or plays an essential role in codependent activation of both NbMAPKKKβ and NbMAPKKKγ. In contrast, the ion leakage levels induced by transient expression of full-length NbMAPKKKα or NbMAPKKKγ were similar in NbMAPKKKβ-silenced plants and control plants (Figure 4B), suggesting that NbMAPKKKα and NbMAPKKKβ are not codependent in their activation; rather, NbMAPKKKβ functions upstream of NbMAPKKKα in the cell death signaling pathway. This result led us to postulate that NbMAPKKKβ functions upstream of NbMAPKKKγ. As expected, in NbMAPKKKγ-silenced plants, cell death and the ion leakage level induced by the transient overexpression of the NbMAPKKKβ kinase domain were significantly compromised compared to control plants (data not shown and Figure 4C, respectively). Cell death and the ion leakage induced by transient overexpression of NbMAPKKKα, however, were similar in NbMAPKKKγ-silenced and control plants (data not shown and Figure 4C, respectively), confirming that NbMAPKKKβ functions upstream of NbMAPKKKγ and that NbMAPKKKγ functions upstream of NbMAPKKKα. In this epistasis analysis, ion leakage levels were closely associated with the intensity of cell death at all combinations of these genes. Thus, these three MAPKKK genes form a linear signaling pathway leading to PCD in which NbMAPKKKβ and NbMAPKKKα function as the furthest upstream and downstream components, respectively, of the three MAPKKK components.

N. benthamiana plants were grown in a growth chamber at 25 °C. To virally induce PCD, plants were inoculated with the binary vector pLi1, which contains the full-length cDNA of the Li1 isolate of plantago asiatica mosaic virus (PlAMV) downstream of the cauliflower mosaic virus 35 S promoter 33 .

To retrieve partial MAPKKK cDNA sequences, a BLAST search was performed using the amino acid sequences of the kinase domains of the Arabidopsis MAPKKK genes AtMEKK1, AtMAPKKKγ, and AtMAPKKKε1 as queries against the N. benthamiana, N. tabacum, and S. lycopersicum databases of the Gene Index Project at the Computational Biology and Functional Genomics Laboratory Web site (http://compbio.dfci.harvard.edu/tgi/cgi-bin/tgi/Blast/index.cgi) and the tobacco BY-2 EST clone database of the RIKEN BioResource Center (http://www.brc.riken.go.jp/lab/epd/blast/index.shtml). The consensus nucleotide sequences for the retrieved partial cDNA sequences and query sequences were used to design a pair of specific primers for each of the three MAPKKK genes. Each pair of specific primers was used for RT-PCR amplification of a cDNA fragment from total RNA extracted from N. benthamiana leaves. The RT-PCR products were gel-purified and cloned into the pGEM-T easy vector (Promega, Madison, WI, USA) for sequencing. New primers were then designed based on the regions of the obtained cDNA sequences that were nonhomologous to the query sequences (to prevent nonspecific amplification) and used for specific 5′- and 3′-rapid amplification of cDNA ends (RACE) using a GeneRacer Kit (Invitrogen, Carlsbad, CA, USA) according to the standard protocol provided by the manufacturer. The sequences of primers used in this study are listed in Table 1. cDNA sequences of these MAPKKK genes were determined from at least three independent clones.

ORF and motif analyses were performed using the ORF Finder (http://www.ncbi.nlm.nih.gov/gorf/gorf.html) and CD-Search (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) programs, respectively. Phylogenetic analysis was performed using MEGA3.1 based on a multiple alignment created using ClustalW. The MAPKKK gene sequences used in phylogenetic analysis are listed in the GenBank database under the following accession numbers: AtMEKK1 (NM_116919), AtMEKK2 (NM_116917), AtMEKK3 (NM_116916), AtMEKK4 (NM_117272), BnMAPKKKβ1 (AJ010093), AtANP1 (NM_100771), AtANP2 (NM_104370), AtANP3 (NM_111477), NtNPK1 (D26601), AtMAPKKKγ (NM_126084), AtYODA (AY357949), NbMAPKKKα (AY500155), AtMAPKKKα (NM_179472), AtMAPKKKε1 (NM_112199), AtMAPKKKε2 (NM_111677), MsOMTK1 (AJ575100), BnMAPKKKα1 (AJ010091), BnMAPKKKε1 (AJ238845), SlMAPKKKε (GU192457), and NbMAPKKKε (GU205153). Homo sapiens HsMEKK1 (AF042838) was used as an outgroup.

The coding regions of NbMAPKKKα, NbMAPKKKβ, and NbMAPKKKγ were amplified by RT-PCR using primer pairs Kp-NbMAPKKKa-1F/Xh-NbMAPKKKa-1836R, Kp-NbMAPKKKb-1F/EcV-NbMAPKKKb-1695R, and Kp-NbMAPKKKg-1F/Xh-NbMAPKKKg-1956R, respectively. The kinase domains of NbMAPKKKβ, NbMAPKKKγ, and NbMAPKKKε2 were amplified using primer pairs Kp-NbMAPKKKbkinase-F/EcV-NbMAPKKKbkinase-R, Kp-NbMAPKKKgkinase-F/Xh-NbMAPKKKgkinase-R, and Kp-NbMAPKKKekinase-F/Xh-NbMAPKKKekinase-R, respectively.

Loss-of-function mutants of NbMAPKKKβ and NbMAPKKKγ were obtained by substituting methionine for the essential lysine in the ATP-binding site in the kinase domain. These mutants were obtained by site-directed mutagenesis using primers NbMAPKKKb-K381M-F and NbMAPKKKb-K381M-R for NbMAPKKKβ and NbMAPKKKg-K374M-F and NbMAPKKKg-K374M-R for NbMAPKKKγ. Each PCR-amplified MAPKKK gene fragment was subcloned into the pEarleyGate 203 vector under 35 S promoter 34 via the pENTA entry vector 35 with LR Clonase II Enzyme Mix (Invitrogen). GUS gene was also subcloned into the pEarleyGate 203 vector under the 35 S promoter. These plasmid vectors were transformed into Agrobacterium tumefaciens strain EHA105. Agroinfiltration was performed as previously described 27 .

VIGS was performed as described previously using pTV:00 derivatives and pBintra6 17 . NbMAPKKKα silencing was induced using pTV:NbMAPKKKα 19 . For NbMAPKKKβ and NbMAPKKKγ silencing, the 390-bp KpnI–PvuII fragment of the 5′-terminal region of the PCR-amplified full-length NbMAPKKKβ ORF and the 576-bp KpnI–EcoRV fragment of the PCR-amplified full-length NbMAPKKKγ ORF were introduced into pTV:00 in the antisense direction to generate pTV:NbMAPKKKβ and pTV:NbMAPKKKγ, respectively. The NbMAPKKKβ ORF was amplified using primers Kp-NbMAPKKKb-1 F and EcV-NbMAPKKKb-1695R, and the NbMAPKKKγ ORF was amplified using primers Kp-NbMAPKKKg-1 F and Xh-NbMAPKKKg-1956R.

The methods used for total RNA isolation, cDNA synthesis, and quantitative analysis of gene expression were the same as those previously described 18 . Expression of 18 S rRNA was used as the normalizer. The primers used for real-time RT-PCR were NbMAPKKKa-255F and NbMAPKKKa-393R for NbMAPKKKα, NbMAPKKKb-1179F and NbMAPKKKb-1286R for NbMAPKKKβ, NbMAPKKKg-903F and NbMAPKKKg-984R for NbMAPKKKγ, and Nb18S-193F and Nb280R for Nb18S rRNA.