Functional Transient Expression of Epitope-Tagged N Protein
The N-mediated response can be observed in tobacco (Nicotiana tabacum) cv Samsun (NN) as a hypersensitive response (HR) after transient expression of the TMV P50 elicitor (Figure 1A). In N-transgenic Nicotiana benthamiana plants (line 310A, carrying N under the control of its native promoter), there is no HR but the N response is manifested as resistance against a green fluorescent protein–tagged version of TMV (TMV:GFP). TMV:GFP induces green fluorescent foci on the inoculated leaves of nontransgenic plants that are not produced on 310A (Figure 1B).
Transiently expressed hemagglutinin (HA)- and myc-tagged versions of the N protein under the control of the cauliflower mosaic virus (CaMV) 35S promoter and terminator were also functional in the transient expression assay. Thus, transient expression of HA-tagged N genomic sequence in tobacco cv Petite Havana (nn) generated a P50-dependent HR (Figure 1A), and in nontransformed N. benthamiana it suppressed TMV:GFP (Figure 1B). The HR response was not visible when HA-tagged N was expressed in tobacco cv Petite Havana (nn) in the absence of P50 (Figure 1A). The same results were obtained with a myc-tagged version of the protein and with similar constructs coupled to the promoter from N. However, except where stated, the experiments described below were with the 35S promoter constructs. In all instances, the constructs had the 35S rather than the N transcriptional terminator; nevertheless, the encoded wild-type N proteins were functional mediators of TMV resistance. This finding is in contrast with the previous report that the N genomic 3' sequence is required for proper N function (Dinesh-Kumar and Baker, 2000). A possible explanation for the difference is our use of a transient assay rather than stable transformation for the expression of N.
N Protein Oligomerizes in Response to Elicitor
We transiently coexpressed HA- and myc-tagged versions of N to determine whether N can self-associate. These constructs were expressed with either the TMV P50 elicitor or, as a control, the CP from Potato virus Y, which does not elicit any type of N-mediated response. In the presence of the control CP or with N-HA expressed alone, N-HA did not coimmunoprecipitate with N-myc and vice versa (Figures 2A and 2C). However, in the presence of the P50 elicitor (Figure 2A), the two forms of N coimmunoprecipitated. This N protein coprecipitation is not an artifact of 35S overexpression, because HA- and myc-tagged constructs under the control of the N native promoter interacted similarly in the presence of P50 (Figure 2B). There was no interaction of N with a functional HA-tagged version of NRG1 (Figure 2C) in these assays or with a functional GFP-tagged version of the P50 elicitor (Figure 2D).
Next, we addressed the elicitor-mediated oligomerization of N in different genetic backgrounds in which the N-mediated response was compromised. To do so, we performed our experiments in plants silenced for EDS1
, and SGT1
, genes required for N-mediated resistance (Peart et al., 2002a
). Tobacco rattle virus
(TRV)-induced silencing of EDS1
had no effect on the P50-dependent interactions of N protein. The N-HA and N-MYC proteins coimmunoprecipitated in extracts of TRV:EDS1- or TRV:NRG1-infected plants, as in plants infected with an empty TRV (TRV:00) (Figure 3A
). These results were consistent in three independent experiments, and additional experiments confirmed the predicted loss of TMV resistance in plants infected with the EDS1- and NRG1-silencing constructs (Peart et al., 2002a
). Therefore, from these results, we can rule out the possibility that oligomerization is a consequence of the resistance response, and we conclude that elicitor-mediated oligomerization of N protein is upstream of EDS1 and NRG1 in the N-mediated response to TMV.
The level of N protein in the P50-elicited samples was consistently higher than in nonelicited samples (CP) (see Figures 2, 3, 4C, and 5C), suggesting that N is stabilized or solubilized in the presence of the P50 elicitor. However, after silencing of SGT1 with TRV:SGT1, in contrast with the results with TRV:EDS1 and TRV:NRG1, there were lower levels of soluble N and the P50-induced N oligomerization could not be detected (Figure 3A), even after overexposure of the protein gel blot (Figure 3B). The levels of N were so low that we cannot draw any conclusions about the role of SGT1 in N oligomerization. However, these results indicate that SGT1, either directly or indirectly, plays a role in the stabilization of N.
Mutations in Conserved Motifs Affect P50 Elicitor-Mediated Oligomerization and N Protein Stabilization
The P-loop motif (Figure 4A) in the NBS is likely involved in nucleotide binding in NBS-LRR proteins (Tameling et al., 2002) and is necessary for their function in disease resistance (Dinesh-Kumar et al., 2000; Bendahmane et al., 2002). To test the P-loop role in N self-association, we created a 35S promoter N construct with the mutation GK221,222AA. The mutant protein was stable and, as expected, failed to trigger a resistance response against TMV (Figure 4B) in either the HR or TMV:GFP resistance assay method. This mutant also lost the ability to oligomerize or coprecipitate with the wild-type protein in the presence of the elicitor (Figure 4C) and did not increase in abundance after P50 elicitation. From these results, we conclude that elicitor-mediated oligomerization and stabilization of N requires the presence of an intact P-loop motif. Furthermore, our data show that the CP does not reduce the abundance of the N protein (Figure 4C).
By contrast, a second conserved motif, RNBS-A, does not affect the coprecipitation of N, although it is required for N protein function in resistance assays. RNBS-A is located between the P-loop and kinase-2 motifs, and its consensus sequence differs between TIR and non-TIR R proteins (Meyers et al., 1999, 2003). Close inspection of an alignment of known functional TIR R proteins identified the presence of a putative LXXLL motif (Leo and Chen, 2000) inside the RNBS-A motif (Figure 5A). Because LXXLL motifs are known to participate in protein–protein interactions, we reasoned that this motif might play a role in the N-mediated oligomerization.
We created a 35S-driven N construct with the RNBS-A mutation LL270,271AA (NAA; Figure 5). The same mutation in the equivalent LXXLL motif of the mammalian NOD protein CTIIA caused loss of function and affected its oligomerization ability (Sisk et al., 2001). NAA produced a stable protein but failed to mediate resistance against TMV (Figure 5B). Although this construct occasionally caused a very weak HR in response to P50 in tobacco, it consistently allowed virus multiplication in the TMV:GFP resistance assay. Unlike wild-type N, this mutant protein in the elicited sample (P50) was consistently less abundant than in the nonelicited sample (CP) (Figure 5C), indicating that the RNBS-A motif influences the elicitor-mediated protein stabilization. However, the RNBS-A mutant retained the ability to oligomerize in response to the P50 elicitor (Figure 5C). Thus, neither elicitor-mediated protein stabilization nor resistance is an automatic consequence of N protein oligomerization.
TIR Domain Interactions
To further investigate the N interactions, we coexpressed epitope-tagged versions of the TIR, NBS, LRR, TIR-NBS, and NBS-LRR domains of N in the presence and absence of P50. In some instances (e.g., with NBS), the domains were unstable and protein could barely be detected (see Supplemental Table 1 and Supplemental Figure 1 online); however, of the heterotypic combinations of stable domains, none formed a functional complex, as described previously for domains of Rx (Moffett et al., 2002) (P. Mestre, unpublished data). However, there was a homotypic interaction of TIR domains (Figure 6B). Coimmunoprecipitation of TIR domains did not require and was not affected by the presence of P50 (see Supplemental Figure 2 online). The TIR coprecipitation was specific because the TIR of N interacted only weakly or did not interact with the TIR domains of the RPS4 or Bs4 NBS-LRR proteins (see Supplemental Figure 3 online). A summary of all of the homotypic and heterotypic interactions tested in the resistance and pull-down assays is presented in Supplemental Tables 2 and 3 online.
We tested the significance of the TIR interactions by mutation of TIR domain amino residues at predicted solvent-exposed sites that may play a role in signaling and protein–protein interactions (Figure 6A
). All nine mutants produced stable proteins, and three of them (R24A, S80A, and P110Y) were compromised in the HR assay in tobacco leaves and in the resistance assay against TMV:GFP in N. benthamiana
; data not shown). In the HR assay, the R24A mutant was completely inactive and the S80A and P110Y mutants induced a very weak HR. All three mutants completely failed to induce resistance in the TMV:GFP assay. The corresponding mutants of full-length N retained their ability to oligomerize in response to the P50 elicitor (Figure 6D
), although the interaction was weakest for mutant R24A. Similarly, the isolated TIR domain mutants exhibited homotypic interactions, although, as with the full-length proteins, the TIR domain of R24A was the weakest interactor (Figure 6E
). These results are as predicted if the coprecipitation of full-length N is mediated by homotypic interactions of the TIR domain. They also are the predicted outcomes if oligomerization of N is an early event in the elicitor-mediated activation of the disease resistance pathway.