Here, we describe elicitor-mediated oligomerization and stabilization of the N protein as novel processes associated with resistance against TMV (Figures 2 to 4). The oligomerization and stabilization appear to be separate processes because they are differentially affected by mutations in the RNBS-A motif of N (Figure 5). We infer that oligomerization is functionally significant because, like the N resistance response, it is dependent on an intact P-loop motif (Figure 4). Moreover, there was a correlation between N oligomerization and the elicitation of resistance in that activation of resistance with either mutant or wild-type N was always associated with oligomerization. However, oligomerization is not sufficient to trigger the resistance response (Figure 5), and we deduce that additional interactions of oligomerized N are required for N function. In addition, because silencing of a presumed downstream signaling component in the N signaling pathway (EDS1) has no effect on either the oligomerization or the stabilization of N (Figure 4), it is likely that these are early events associated with the elicitor activation of N. A model of N oligomerization-induced activation of disease resistance is described in more detail below.
The results from the mutational analysis of the TIR domain are in agreement with the proposed signaling function of the TIR. The R24A, S80A, and P110Y mutations have evidently resulted in loss of function in the signaling domain but have caused only partial loss (R24A) or no loss of the oligomerization function. This proposed separation of oligomerization and signaling functions is reinforced by our analysis of an RNBS-A mutant (Figure 5). It is also consistent with the conclusion from EDS1 and NRG1 silencing (Figure 3) that N oligomerization is an early event in the sequence of events leading to P50-elicited disease resistance.
Our finding that the silencing of SGT1 resulted in low levels of N soluble protein is consistent with, although does not prove, the possibility that SGT1 is involved in the elicitor-induced stabilization of N. SGT1 interacts with the RAR1 and HSP90 cofactors of disease resistance, and it has been proposed that RAR1 and SGT1 are cochaperones of HSP90 in the folding of R proteins (Shirasu and Schulze-Lefert, 2000; Schulze-Lefert, 2004). This idea is supported by several lines of evidence: RPM1 levels are reduced in Arabidopsis rar1 or hsp90.2 mutants (Tornero et al., 2002; Hubert et al., 2003); the amount of RPS2 is reduced in Arabidopsis rar1 (Belkhadir et al., 2004); silencing of HSP90 in N. benthamiana causes reduced levels of Rx (Lu et al., 2003); RAR1 controls the steady state level of Mla proteins in barley (Hordeum vulgare) (Bieri et al., 2004).
However, R protein stabilization, SGT1 function, and disease resistance are not always associated in the same way. The RPM1 NBS-LRR protein, for example, which confers resistance against Pseudomonas syringae strains, is unlike N in that it is destabilized when resistance is elicited (Boyes et al., 1998). In this instance, the role of SGT1 may be as a cofactor in a RAR1-dependent degradation mechanism (Holt et al., 2005) rather than as a R protein stabilizer, as implied by our analysis of N. To reconcile these apparently conflicting results, we propose that SGT1 and associated proteins are part of a system for controlling the level of R proteins through either positive or negative regulation.
A model of N activation is shown in Figure 7. The initial events are changes to the conformation of N. We considered the possibility that these changes could be analogous to the disruption of CC and the LRR domain interactions that are associated with the elicitation of Rx-mediated resistance (Figure 7, left branch). However, in an extensive analysis (see Supplemental Tables 2 and 3 online), there was no evidence for intramolecular interactions by coexpression of N domains. Perhaps the intramolecular interactions in N are weaker than in Rx and are not effective for proteins expressed in trans. Alternatively, it is possible that elicitor-mediated activation of N involves interactions with other as yet unidentified proteins or, perhaps, a change in subcellular localization of the protein (Figure 7, right branch).
Our model proposes that these initial conformational changes would cause N to oligomerize in complexes that are required for the activation of the EDS1-dependent and perhaps other response pathways, leading to virus resistance and HR (Figure 7
). In principle, the TIR interactions (Figure 6
) could mediate the oligomerization process if they are exposed by the elicitor-induced changes to N. Presumably, the isolated TIR domains, being free of the rest of the N protein, would not be masked by a subcellular location or other domains in N and would be available to interact even in the absence of elicitor. We show the homotypic N interactions in Figure 7
as being direct, but we emphasize that they could be indirect and dependent on host factors that have not yet been identified. It is also possible that domains of N, in addition to the TIR, may be involved in the oligomerization process. It is possible that, as with Toll and TLRs, the TIR domain interactions are secondary to oligomerization at other more C-terminal domains (Xu et al., 2000
; Hu et al., 2004
; Sun et al., 2004
). Unfortunately, the isolated NBS domains of N were not stable in the transient assay, and we could not assay their potential for homomeric interactions.
How could oligomerization of N activate the virus resistance and HR pathways? It is unlikely to be simple induced proximity of TIR domains, because the isolated TIR domains interact but do not induce an HR. Accordingly, as with other members of the NOD family, expression of the N-terminal domain alone does not activate response pathways (Inohara and Nunez, 2003). Perhaps, as described for the NOD protein CTIIA (Sisk et al., 2001, and references therein), the NBS domain in the N oligomer is a scaffold for components of the signaling pathway (Figure 7). This possibility is consistent with our finding that an RNBS-A mutant oligomerizes in response to elicitor but does not trigger a resistance response (Figure 5): the role of RNBS-A would be to interact with signaling components and not in oligomerization.
The elicitor-induced oligomerization and stabilization of N, being upstream of EDS1 (Figure 3), are the earliest identified molecular features of the N resistance pathway; therefore, they are useful for positioning other processes in the sequence of recognition and response mechanisms. Here, for example, we have shown that the CC-NBS-LRR protein NRG1 is likely to act downstream of the elicitor recognition process because elicitor-induced oligomerization/stabilization of N occurs in the NRG1-silenced plants. As overexpression of NRG1 induces responses that are not dependent on EDS1 (Peart et al., 2005), it is likely that NRG1 is downstream or, perhaps more likely, independent of EDS1 in the N pathway.
We were able to detect the N oligomerization in our transient assay system because the elicitor-induced cell death is absent. In other experimental systems, including the Rx CC-NBS-LRR protein, the elicitor-induced cell death is rapid and prevented us from detecting the transiently expressed proteins (Moffett et al., 2002). The Rx cell death response can be suppressed, for example, in an SGT1-silencing background or with P-loop mutants (Moffett et al., 2002), but these conditions are not suitable for the detection of N oligomerization (Figures 3 and 4). Therefore, it is an open question whether other NBS-LRR proteins oligomerize in response to elicitor. Clearly, to explore the similarity of plant NBS-LRR R proteins with NOD proteins of animals, it will be necessary to extend the analyses of protein interactions to a range of other R proteins. It would also help to have more detailed analysis of N and more information about structure and structure–function relationships.