A Comprehensive Mutational Analysis of the Arabidopsis Resistance Protein RPW8.2 Reveals Key Amino Acids for Defense Activation and Protein Targeting

Natural Mutation-Guided Mutagenesis of RPW8.2 Reveals Three Amino Acids Critical for Induction of Cell Death and Disease Resistance

A previous study on intraspecific polymorphism at RPW8.2 showed that compared with the functional RPW8.2 allele from accession Ms-0 (designated R82^Ms-0), Thr-64 to Ser (T64S), Asp-116 to Gly (D116G), and Thr-161 to Lys (T161K) substitutions are present in alleles from most accessions that are susceptible to powdery mildew (Figure 5 in Orgil et al., 2007). This implies that Thr-64, Asp-116, and/or Thr-161 in R82^Ms-0 may contribute to the defense function of RPW8.2. By contrast, the R82^Bg-1 allele contains the highest number (13) of nonsynonymous substitutions, including those resulting in T64S, D116G, and T161K amino acid replacements; however, plants of accession Bg-1 showed an intermediate infection phenotype with more extensive powdery mildew–induced cell death (Orgil et al., 2007; Wang at al., 2009). This suggests that some amino acid substitutions in R82^Bg-1 may offset the phenotypic effect produced by T64S, D116G, and T161K. To assess the functional impact of the three common substitutions and others found in R82^Bg-1, we conducted site-directed mutagenesis using R82^Ms-0 as template and generated RPW8.2 mutants whose protein products contain one of the T64S, D116G, T161K, E59K, and RRR90-92KKK substitutions. The rationale for the RRR90-92KKK mutation is that the LRRR motif in the mammalian potassium channel is a 14-3-3 binding site for promoting its cell surface expression (Yuan et al., 2003); therefore, it is possible that 89-92LRRR in RPW8.2 may serve a similar role because RPW8.2 has been shown to interact with 14-3-3 lambda (Yang et al., 2009).

Unless otherwise indicated, for making these five and many other RPW8.2 mutant constructs described in this study (see Figure 1 for a schematic illustration), we made in-frame translational fusion of yellow fluorescent protein (YFP) to the C terminus of the protein encoded by each RPW8.2 mutant. The DNA constructs were placed under control of the promoter of R82^Ms-0 and introduced into accession Col-gl (Columbia-0 [Col-0] carrying the glabrous mutation). Col-gl lacks RPW8.1 and RPW8.2 and is susceptible to the adapted powdery mildew isolate Golovinomyces cichoracearum UCSC1 (Xiao et al., 2001) and thus is ideal for functional evaluation of the RPW8.2 mutant constructs. We generated at least 30 independent T1 transgenic lines for each DNA construct and examined (1) the frequency of lines displaying spontaneous HR-like cell death (SHL) and powdery mildew–induced cell death (HR) and resistance and (2) the subcellular localization of the mutant proteins. As shown in Figure 2 and Table 1, very few (2.4%) T1 transgenic plants expressing YFP-tagged R82^Ms-0 developed SHL before inoculation, while about a quarter (25.4%) displayed apparent resistance (disease reaction score < 1 to 2; Xiao et al., 2005) and HR as small lesions visible to the naked eye. In comparison, none of the T1 lines expressing YFP-tagged RPW8.2 containing E59K (designated R82^E59K), R82^RRR90-92KKK, or R82^T161K exhibited SHL, while the percentage of lines showing resistance and HR was only slightly reduced to 22.4, 21.5, and 21.3%, respectively. In addition, confocal laser scanning microscopy showed that the three mutant proteins exhibited typical EHM-specific localization (see Supplemental Figures 1A to 1C online). These results suggest that these substitutions do not significantly alter the function of RPW8.2. By contrast, we observed a high frequency of SHL (37.1%) and HR (51.8%) in T1 lines expressing R82^T64S (Figures 2F and 2G, Table 1). This result suggests that either Thr-64 or Ser-64 may be phosphorylated and there is a functional consequence of this phosphorylation. Sequence analysis with NetPhos 2.0 (http://www.cbs.dtu.dk/services/NetPhos/) suggested that Thr-64 (score = 0.950) but not Ser-64 (score = 0.185) is likely to be phosphorylated. To test if this is the case, we made two additional mutants, R82^T64E, in which Thr is changed to a phosphomimetic Glu, and R82^T64A, in which Thr is changed to Ala. T1 lines transgenic for R82^T64E showed a slightly higher rate of SHL (10.9%), but the percentage of plants that showed HR and resistance (21.2%) was comparable to that of the R82^Ms-0 T1 population (Figure 2H, Table 1). By contrast, the rates and severity of SHL (46.7%) and HR (53.3%) in the R82^T64A T1 lines were comparable to those of the R82^T64S T1 lines (Figures 2I and 2J, Table 1). Taken together, these results suggest that phosphorylation of RPW8.2 at Thr-64 may indeed play a critical role in preventing inappropriate cell death in the absence of powdery mildew pathogens.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Schematic Illustration of RPW8.2 Mutants Constructed and Tested.

RPW8.2 mutant genes were in-frame fused with YFP and tested in Arabidopsis by stable expression. For site-specific mutagenesis, only the functionally important mutants are highlighted above the schematic RPW8.2 sequence; the remaining mutants are indicated by gray bars underneath. For NAAIRS replacement, five of the 29 constructs show significant reduction in EHM targeting (gray bar). For deletion analysis, constructs in black show EHM localization; constructs in gray show no or little EHM localization. Internal deletions are shown by dashed line flanked by two dark circles. The two shaded regions delimited by the mutational analyses are critical for EHM localization of RPW8.2. Stars indicate the functionally informative breaking points. The black arrow indicates the minimal sequence construct in the bottom containing 60 amino acids. TMS-R82 andTMA-R82, RPW8.2-YFP constructs in which the TMD is replaced by that of SYP122 or ACBP2. WT, wild type.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Cell Death and Disease Reaction Phenotypes of Site-Directed RPW8.2 Mutants.

(A) Representative leaves of Col-gl lines transgenic for 35S:YFP or P_RPW8.2:RPW8.2-YFP (line R2Y4) as control. Plants were inoculated with Gc UCSC1 and pictures taken at 7 d postinoculation (dpi).

(B) to (O) Representative mildew-infected leaves at 7 or 10 d postinoculation or uninfected plants expressing the indicated YFP-tagged RPW8.2 mutants from the RPW8.2 promoter in comparison with Col-gl or Col-gl transgenic for 35S:YFP. Arrowheads indicate spontaneous HR-like cell death lesions; arrows indicate big necrotic lesions resulting from merge of small lesions.

Next, we examined the phenotypic effect of the D116G mutation. Out of 121 T1 lines transgenic for R82^D116G, none had SHL and only three lines (2.5%) showed weak HR and moderate resistance to powdery mildew, and the remaining lines lacked SHL or HR and were susceptible (Figure 2E, Table 1). Confocal microscopy showed that R82^D116G was expressed in the majority of these T1 lines in which the mutant protein was also specifically targeted to the EHM (see Supplemental Figure 1D online), excluding the possibility that this phenotype is due to the lack of expression or mislocalization of R82^D116G. These results indicate that Asp-116 is a critical residue for the defense function of RPW8.2 and reinforce the notion that EHM targeting of RPW8.2 is controlled by mechanisms most likely independent of those governing RPW8.2’s defense function (Wang et al., 2009).

To examine if there is interaction between the T64S, D116G, T161K, and E59K mutations, we constructed five double-site RPW8.2 mutants. T1 lines transgenic for R82^E59K/T161K were phenotypically similar to the respective single-site mutant lines and to the R82^Ms-0 wild-type control lines (Table 1). Not surprisingly, T1 lines transgenic for R82^T64S/T161K showed enhanced SHL and HR similar to those seen in R82^T64S lines, and T1 lines transgenic for R82^E59K/D116G or R82^D116G/T161K were phenotypically similar to those transgenic for R82^D116G. Interestingly, none of the T1 lines transgenic for R82^T64S/D116G developed SHL and only 16.2% of them exhibited delayed (∼1 d) HR and moderate resistance (Figure 2K; disease reaction score 2), suggesting that D116G can largely suppress T64S-mediated SHL and constitutive defense.

To identify additional amino acid substitutions likely responsible for the pronounced SHL and fungus-induced massive HR cell death phenotype of the Bg-1 accession (Orgil et al., 2007; Wang et al., 2009), we constructed RPW8.2 mutants carrying multiple-site substitutions (Table 1) by overlapping PCR using a fragment from R82^Bg-1 that contains the desired mutations and an overlapping fragment(s) from R82^Ms-0 for the rest. While T1 lines transgenic for R82^{H19Q/S45T/V50I/Q52K/E59K} showed phenotypes similar to those of the R82^Ms-0 wild-type control lines, adding T64S and F56L to this mutant construct resulted in enhanced SHL and HR in the T1 lines comparable to that seen in T1 plants transgenic for R82^T64S (Table 1). This suggests that these six mutations (i.e., H19Q, S45T, V50I, Q52K, F56L, and E59K) do not have significant contribution to the pronounced cell death phenotype of Bg-1 plants. We then made two RPW8.2 mutant constructs that contain T64S and D116G plus K70E/E77V/L89Q or K70E/E77V/T161K. T1 lines transgenic for either of these two constructs showed more frequent HR than R82^T64S/D116G T1 lines but less frequent HR than R82^Ms-0 T1 plants (Table 1), implying that K70E, E77V, and L89Q or T161K mutations may partially contribute to the more extensive HR found in plants expressing R82^Bg-1 (Wang et al., 2009). Furthermore, we made two RPW8.2 mutants carrying seven substitutions, E59K/T64S/V68F/K70E/E77V/D116G/T161K or E59K/T64S/V68F/K70E/E77V/L89Q/D116G. T1 plants transgenic for either construct (the latter in particular) showed much more extensive SHL (31.7% for the first and 60.7% for the second) and HR (38.2 and 77.5%) than T1 transgenic lines of R82^Ms-0 (Table 1). Based on these observations, we reasoned that (1) L89Q in combination with other six mutations may have a more pronounced cell death–inducing effect on R82^Bg-1 than does T161K in the same context; and (2) V68F must account for a major increase of the pro-cell death activity of the two RPW8.2 mutants. To test the latter speculation, we made the R82^V68F single-site mutant and found that T1 lines transgenic for this construct indeed developed more severe and more frequent SHL (47.7%) and HR (51.1%) (Table 1; Figures 2L and 2M). Interestingly, mildew-induced cell death in these plants was not apparently associated with resistance as indicated by moderate fungal mass on the infected leaves (Figure 2M). This phenotype is reminiscent of the extensive necrotic cell death of mildew-infected Bg-1 plants (Wang et al., 2009). To see if D116G can suppress V68F-mediated cell death, we made the R82^V68F/D116G double-site mutant. Interestingly, the cell death and disease phenotypes of this mutant were similar to those of the R82^V68F single-site mutant (Figures 2N and 2O), indicating that V68F is epistatic to D116G. All of the above-described RPW8.2 mutants with site-directed mutations, when detectable in invaded cells, exhibited normal EHM localization (see Supplemental Figure 1 online), indicating that none of these mutations significantly affects RPW8.2 trafficking properties.

The N Terminus of RPW8.2 Is Required for Protein Stability and Contributes to EHM Targeting Specificity

To search for domains or motifs in RPW8.2 responsible for the EHM-specific targeting, we conducted systematic and comprehensive mutational analyses. We first evaluated the role of the N-terminal domain of RPW8.2 in EHM targeting. In a previous study, we found that two RPW8.2 mutant proteins truncated for either the N-terminal 22 or 30 amino acids were expressed at extremely low levels and localized as a few puncta in an unknown compartment (Wang et al., 2010). Because the N-terminal 22 amino acids are predicted to be a transmembrane domain (TMD) (Xiao et al., 2001), it was no surprise that removal of the entire TMD resulted in protein instability and mislocalization. However, whether the TMD serves as a signal peptide and/or whether specific residues in the TMD contribute to RPW8.2’s EHM targeting is not known. To this end, we made a series of N-terminal deletion mutants (Figure 1) and evaluated their defense function and subcellular localization in stable transgenic lines. Surprisingly, we found that mutant R82^∆5-12, in which amino acids 5 to12 (VAAGGALG) were deleted, largely retained normal EHM-specific localization in ∼65% of the invaded cells. However, in the remaining invaded cells, most of the mutant protein was detected as variable-sized spots or patches around the haustorial neck and/or penetration site, although some signal was still detected as small puncta at the EHM or diffused into the EHM (Figures 3A and 3B). Strikingly, in some invaded cells, we detected YFP signal as numerous small puncta in a fireworks-like distribution of 10 to 30 µm in diameter radiating from the penetration site where the signal was more concentrated (Figure 3C). In some cases, such a fireworks-like distribution was confined by a weak propidium iodide (PI)–positive ring (see Supplemental Figure 2A online). Intriguingly, in the same infection site, the mutant protein was also found in the EHM (inset in Figure 3C; see Supplemental Movie 1 online). From a side view, most of the small puncta were arranged at the same horizontal level around the fungal penetration site, and intensive signal was also observed at the haustorial neck (indicated by an arrow in Figure 3C). This raised a question as to in what compartment the fireworks-like domain resides. The localization patterns for R82^∆5-12 were essentially replicated in the R82^∆5-14 mutant in which amino acids 5 to 14 were deleted (Figure 3D; see Supplemental Figure 2B online), with the latter having slightly less frequent (∼50%) normal EHM-specific localization (inset in Figure 3D). In some rare cases, R82^∆5-14 was only found in the haustorial neck region (see Supplemental Figure 2C online), which is the portion of the EHM that may be first synthesized. This observation implies that this subdomain of the EHM is the preferred destination of R82^∆5-14 and/or the overall level of R82^∆5-14 accumulation is too low to allow signal of R82^∆5-14 detectable in the EHM encasing the main body of the haustorium. To further characterize the fireworks-like domain, we introduced a plasma membrane (PM) marker PIP2A-mCherry (Nelson et al., 2007) into Col-gl plants expressing R82^Ms-0, R82^∆5-12, or R82^∆5-14 for colocalization analysis. As expected, R82^Ms-0 was located at the EHM encasing the haustorial complex. However, the membrane compartment marked by PIP2A-mCherry could not reach into the surrounding region of the haustorium such that a hollow was observed (see Supplemental Figure 2D online). Conversely, puncta containing R82^∆5-12 and R82^∆5-14 were able to localize and/or move through this domain to reach the EHM (Figures 3C and 3D; see Supplemental Figure 2 online). These observations support the formation of a special fungal penetration-perturbed membrane domain around the penetration site where R82^∆5-12 and R82^∆5-14 are localized.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

The N-Terminal TMD Is Required for EHM-Specific Localization of RPW8.2.

Col-gl transgenic lines expressing each of the four N-terminal deletion RPW8.2 mutants tagged with YFP were inoculated with Gc UCSC1. Infected leaves were collected at 2 d postinoculation, stained with PI, and subjected to confocal microscopy. The YFP signal is pseudo-colored in green and PI-stained structures in red. Images are representative Z-stack projections of 15 to 65 optical sections. h, haustorium; n, nucleus; p, penetration site. Bars = 10 μm.

(A) Mutant R82^∆5-12 in some cells showed reduced EHM localization as reflected by small puncta and aggregates at or peripheral to the EHM, particularly at the haustorial neck region (arrow).

(B) R82^∆5-12 as protein aggregate (arrow) was seen around the penetration site (top panel) and/or surrounding the haustorial neck (low panel).

(C) R82^∆5-12 was found as small puncta in a fireworks-like domain centering in the penetration and as small puncta or diffuse signal at the EHM in the same cell (inset). Note that when viewed horizontally, the R82^∆5-12–labeled fireworks-like domain seemed to be a thin layer aligning to the cell wall (bottom panel). The haustorial neck region (arrow) was also labeled by R82^∆5-12.

(D) Mutant R82^∆5-14 was also found in the fireworks-like domain in some cells, while it was also found at the EHM in other cells (inset).

(E) Mutant R82^∆5-15 was found in puncta not apparently associated with the EHM, indicating it is incapable of EHM targeting.

(F) Mutant R82^∆5-18 was exclusively found as varied sized puncta in the nucleus.

We then made one more amino acid deletion to make R82^∆5-15 in which amino acids 5 to 15 (VAAGGALGLAL) were removed. Imaging analysis of multiple T1 lines transgenic for R82^∆5-15 showed that this mutant protein completely lost its EHM targeting since it was only seen as randomly distributed puncta unrelated to the EHM (Figure 3E). This suggests that removal of amino acids 5 to 15 may lead to a complete loss of membrane anchorage of R82^∆5-15 or Leu at 15 is essential for EHM targeting. Finally, we made and functionally evaluated another mutant R82^∆5-18 in which amino acids 5 to 18 (VAAGGALGLALSVL) were deleted. Interestingly, R82^∆5-18 was found to be exclusively nuclear localized (Figure 3F). This observation suggests that (1) 15-LSVL-18 in the predicted TMD is required for EHM localization of RPW8.2, and (2) RPW8.2 may contain a nuclear localization signal (NLS) that could render nuclear localization to R82^∆5-18.

Our above observations seem to support a certain role for amino acids 5 to 14 (VAAGGALGLA) in RPW8.2’s EHM-specific localization. However, because these 12 amino acids are part of the predicted TMD, the compromised EHM-specific localization of R82^∆5-12 and R82^∆5-14 may be simply due to a defect of these mutant proteins in membrane insertion. To test this speculation, we replaced the TMD (amino acids 2 to 18; IAEVAAGGALGLALSVLH) of RPW8.2 by a TMD from validated PM-localized proteins. We selected the TMD (285-WTCFAILLLLIIVVLIVVFT-304) of SYP122, a syntaxin localized to the PM but not found in the EHM (Assaad et al., 2004), and the N-terminal TMD (7-LAQSVILGLIFSYLLAKLISIVV-29) of the PM-localized ACBP2, which has been demonstrated to enable PM localization of YFP (Li and Chye, 2003). We obtained multiple lines transgenic for either TMD^SYP122- R82^∆2-18 or TMD^ACBP2-R82^∆2-18 and found that these two fusion proteins were distributed as diffuse signal or small puncta aligning the cell wall and peripheral to or at the EHM (see Supplemental Figure 3 online). However, we rarely detected strong and diffuse YFP signal at the EHM from these two TMD-replaced fusion proteins, and none of the transgenic lines expressing these two constructs showed resistance to Gc UCSC1.

The above observations, together with our previous results (Wang et al., 2010), indicate that the N terminus (1 to 22 amino acids) of RPW8.2 comprises a special TMD or more likely a signal peptide that is not only important for RPW8.2's membrane anchorage but may also contribute to its EHM targeting efficiency and protein stability and, consequently, its defense function. Meanwhile, these results also suggest that the C-terminal portion (amino acids 23 to 174) of RPW8.2 may contain an EHM targeting signal(s) that probably determines RPW8.2's EHM-specific localization.

The C Terminus of RPW8.2 Ensures Efficient EHM Targeting by Suppressing Nuclear Localization

To search for a putative EHM targeting signal, we conducted a systematic deletion analysis from the C-terminal end of RPW8.2. Previously, we found that 12 RPW8.2 alleles contain a single base pair indel that results in frame shift and a truncation of 28 to 34 amino acids at the C termini of the deduced proteins (Orgil et al., 2007). However, the functional consequence of these truncations is not known. As an entry point for our C-terminal deletion analysis, we first made R82^∆138-174, which encodes a mutant RPW8.2 lacking the C-terminal 37 amino acids. T1 lines transgenic for R82^∆138-174 displayed cell death and resistance phenotypes similar to those of T1 lines transgenic for R82^Ms-0 (see Supplemental Figure 4A online). However, to our surprise, while this fusion protein exhibited normal EHM localization in some cells, it was also found in the nucleus in other cells (Figures 4A and 4B). The nuclear localization was evidenced by the presence of YFP puncta of varied sizes in nuclei stained by both PI (Figure 4C) and 4′,6-diamidino-2-phenylindole (Figure 4H). We thus speculated that the C-terminal 37 amino acids of RPW8.2 may suppress RPW8.2’s nuclear localization to ensure efficient EHM targeting. To obtain further evidence for this, we made nine additional C-terminal truncation mutants (i.e., R82 ^Δ120-174, R82 ^Δ116-174, R82 ^Δ115-174, R82 ^Δ113-174, R82 ^Δ112-174, R82 ^Δ111-174, R82 ^Δ110-174, R82 ^Δ109-174, and R82 ^Δ88-174) in which the C-terminal 55, 59, 60, 62, 63, 64, 65, 66, or 87 amino acids were respectively truncated (Figure 1; see Supplemental Table 1 online). For the first five mutants, we found that, as the size of the truncation increases (from 55 to 63 amino acids), the respective mutant proteins exhibited decreased EHM localization while gaining more nuclear localization (Figures 4E and 4F). Starting from R82^∆111-174, the remaining four mutant proteins completely lost EHM targeting and became exclusively nuclear localized (Figures 4G and 4H). All mutant proteins failed to activate cell death and disease resistance (see Supplemental Figure 4B online). These observations, when added to earlier results, indicate that (1) amino acids 1 to 111 in RPW8.2 must contain an EHM targeting signal; (2) Leu-111 is absolutely critical for EHM targeting, as removing it in the R82^∆111-174 mutant abolished EHM targeting, whereas keeping it in R82^∆112-174 achieved EHM targeting, albeit at a low frequency; (3) there must be at least one NLS in the N-terminal half (amino acids 1 to 110) of RPW8.2; and (4) the C-terminal tail (amino acids 138 to 174) of RPW8.2 suppresses nuclear localization of RPW8.2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

The C-Terminal Portion Is Required for Suppression of Nuclear Localization of RPW8.2.

YFP signal from tagged RPW8.2 mutant proteins is pseudo-colored green, PI-stained structures are red, and DAPI-stained nuclei are blue. Images are representative Z-stack projections of 15 to 45 optical sections. h, haustorium; n, nucleus. Bars = 10 μm.

(A) Nuclear localization of R82^∆138-174.

(B) EHM localization of R82^∆138-174.

(C) Epidermal cells expressing YFP alone as control to show that both nuclei and the haustorium were stained red by PI.

(D) EHM and nuclear localization of R82^Δ120-174.

(E) EHM and nuclear localization of R82^∆113-174.

(F) Nuclear localization and weak EHM targeting of R82^Δ112-174.

(G) Exclusive nuclear localization of R82^Δ111-174.

(H) Nuclear localization of R82^Δ110-174. Note both the haustorium and nucleus were stained blue by DAPI.

(I) Nuclear localization of R82^Δ88-174.

We noted that the YFP signal in the nucleus representing the above-mentioned C-terminally truncated RPW8.2 proteins was often detected in multiple patterns ranging from faint homogenous distribution to small puncta (<0.1 μm) and to large aggregates (up to 1 μm) in different cells (see Supplemental Figure 5 online), suggesting a dynamic nature for the localization.

Intriguingly, we noticed that R82^∆120-174 was also found in stromule-like membranes that wrap around and connect individual plastids (Figure 5A), in addition to EHM and nuclear localization. Stromules are defined as stroma-filled, dynamic tubular structures extending out from the envelope membrane of different plastid types (Köhler et al., 1997; Waters et al., 2004). This unexpected novel localization was further signified by our observations that additional RPW8.2 mutant proteins were also found in the stromule-like membrane structures (see later text).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

RPW8.2 Variants Are Targeted to the PSM.

YFP signal from tagged RPW8.2 mutant proteins is pseudo-colored green, PI-stained structures are red, and autofluorescent chloroplasts or plastids are blue. Images are representative Z-stack projections of 15 to 45 optical sections. h, haustorium. Bars = 10 μm.

(A) R82^∆120-174 was observed in the PSM tightly associated with and connecting individual plastids (asterisks). Note that there are bulges or lumps close to the encased plastids or in the middle of the membrane strand.

(B) and (C) R82^∆43-97 was found in the PSM as puncta at or peripheral to the EHM.

(C) R82^∆43-97 was targeted at the EHM as shown by diffuse YFP signal in the EHM.

(D) to (G) Localization of mutant proteins R82^∆43-97 (D), R82^∆43-115 (E), R82^∆43-135 (F), and R82^∆43-141 (G) in epidermal cells containing a haustorium. Note the big dots/patches (arrows) in some cells.

(H) R82^∆65-93 was found in ring structures (asterisks) associated with chloroplasts in mesophyll cells.

(I) R82^{∆65-93+∆138-174} was occasionally found both in a PSM (asterisks) and the EHM and the tubules that connect these two compartments.

(J) R82^{∆65-93+∆138-174} was observed in the PSM similarly as R82^∆120-174. The bulged portion is indicated by arrows.

(K) and (L) Colocalization analysis of YFP-tagged R82^{∆65-93+∆138-174} expressed from the RPW8.2 promoter and a stromule marker (a plastid-targeting [PT] signal in fusion with mCherry; Nelson et al., 2007) expressed from the 35S promoter in stable Arabidopsis Col-0 plants. While some R82^{∆65-93+∆138-174} signal seemed to colocalize with PT-mCherry in thin stromule strands (arrows in [K]), most of it was found in the periphery of PT-mCherry labeled plastids or stromule protrusions (arrows in [L]).

Two Internal Regions of RPW8.2 Are Required for EHM Targeting, and RPW8.2 May Be Targeted to the Peristromule Membrane

Based on our N- and C-terminal deletion analyses, it seemed that neither the TMD at the N terminus nor the C-terminal portion (amino acids 112 to 174) of RPW8.2 encodes the EHM targeting signal, although both are required for efficient EHM targeting. To identify the EHM targeting signal, we made 11 internal deletion constructs in which we kept the N-terminal 1 to 42 or 1 to 64 amino acids intact and combined it with different sized C-tails (Figure 1). We first removed amino acids 43 to 93 to make the R82^∆43-93 mutant. This mutant protein retained EHM-specific localization, although the expression level was generally low as reflected by weak YFP signal largely in punctate and occasionally diffuse signal at the EHM (Figures 5B and 5C). This indicates that amino acids 43 to 93 are not essential for EHM targeting, although they may contribute to protein stability. We then made and evaluated four additional mutants R82^∆43-97, R82^∆43-115, R82^∆43-135, and R82^∆43-141 (Figure 1). As shown by representative images in Figures 5D to 5G, all these four mutant proteins showed rare and poor EHM localization and were found in small puncta to big patches (2 to 3 µm) in the cytoplasm. These observations, together with those from earlier deletion analyses, suggested that (1) the region from amino acids 94 to 111 contributes to the EHM targeting specificity and (2) either amino acids 1 to 42 (the N terminus) or amino acids 135 to 174 (the C terminus) contribute to the residual EHM targeting observed with the above four mutant proteins in comparison with R82^∆43-93. We thus subsequently made R82^∆65-93 to see if a longer N-terminal portion could improve the EHM targeting specificity. We failed to notice any obvious improvement in EHM targeting specificity and efficiency for R82^∆65-93. Instead, we occasionally found this mutant protein in ring structures surrounding plastids in epidermal cells (Figure 5H), similar to the localization pattern of R82^∆120-174. Next, we further deleted the C-terminal 37 residues (amino acids 138 to 174) using R82^∆65-93 as template to make R82^{∆65-93+∆138-174} in order to examine if removal of the C terminus compromises EHM targeting in R82^∆65-93. As shown in Figure 5I, R82^{∆65-93+∆138-174} was still occasionally found in the EHM, suggesting that the EHM targeting specificity in R82^∆43-93 (and other three mutants) is not conferred by the C-terminal 37 amino acids. Combining the results from all the deletion analyses described thus far, we inferred that two regions (i.e., the N terminus [amino acids 1 to 42] and particularly an internal region [amino acids 94 to 111]) are mainly responsible for EHM-specific localization (Figure 1).

Interestingly, R82^{∆65-93+∆138-174} was more frequently found in the stromule-like membrane structures that tightly wrap around and connect individual plastids in uninfected epidermal cells in some T1 lines that had high levels of constitutive expression (Figure 5J). This localization pattern was observed for R82^∆120-174 (Figure 5A), and in both cases, there were big bright spots indicative of high-level protein accumulation at one spot of the rings or in the strands bridging two different plastids (Figures 5A and 5J). Most strikingly, we observed that in some haustorium-invaded cells R82^{∆65-93+∆138-174}-labeled stromule-like membrane strands apparently connected plastids and the haustoria (Figure 5I).

Stromules are tubular protrusions of the plastid envelope membrane, and to our knowledge, proteins found in stromules are all plastid-targeted proteins, although a myosin motor protein has been shown to be associated with stromules in Nicotiana benthamiana (Sattarzadeh et al., 2009). Our above findings raised a question as to whether some RPW8.2 mutant proteins are truly targeted to stromules. Thus, we examined if R82^{∆65-93+∆138-174} colocalizes with a canonical stromule marker protein. By coexpressing a stromule marker protein (i.e., a plastid targeting signal in fusion with mCherry) (Nelson et al., 2007) in the same background, we found that R82^{∆65-93+∆138-174} was localized at the periphery of, but not exactly colocalized with, the stromule marker (Figures 5K and 5L). Taken together, our observations that several RPW8.2 mutant proteins were targeted to a membrane wrapping around plastids and peripheral to stromules suggest the existence of a novel interfacial membrane between the cytoplasm and plastids or plastid-derived stromules. We tentatively named this novel membrane the peristromule membrane (PSM). Given that several RPW8.2 mutant proteins were found in the PSM and the EHM (Figure 5), we speculate that the PSM may share certain common membrane characteristics with the EHM.

Two Short Motifs Enriched in Basic Residues Are Essential for EHM Targeting

To further characterize the EHM targeting signals in the two regions delimited by the deletion analyses described above, we conducted a NAAIRS scan across the entire RPW8.2 protein sequence. NAAIRS is a six–amino acid (Asn-Ala-Ala-Ile-Arg-Ser) segment frequently found in both α-helices and β-sheets, and replacement of six consecutive residues with NAAIRS is thus believed to minimize gross disruptions in secondary structures (Wilson et al., 1985). NAAIRS scanning has been used successfully to identify critical elements in the Arabidopsis SNI1 protein (Mosher et al., 2006) and the potato (Solanum tuberosum) resistance protein Rx (Rairdan et al., 2008). To this end, we made 29 NAAIRS replacement mutants covering almost all amino acids of RPW8.2 (Figures 1 and 6A). Each of the 29 NAAIRS mutants was fused with YFP at the C terminus and expressed in Col-gl plants from the RPW8.2 promoter (see Supplemental Table 1 online). Among these NAAIRS mutants, three showed significant reduction in EHM targeting. The first mutant R82^N20-25, in which amino acids 20 to 25 (EAVKRA) were replaced by NAAIRS, was rarely (<5% infected epidermal cells) found in the EHM (Figure 6B). Instead, the mutant protein was more frequently found in ring structures peripheral to chloroplasts in mesophyll cells (Figure 6C), which is reminiscent of the localization pattern of RPW8.1 (Wang et al., 2007). The second mutant R82^N26-31, in which amino acids 26 to 31 (KDRSVT) were replaced by NAAIRS, showed even less obvious EHM targeting, as reflected by the formation of various sized puncta in or peripheral to the EHM (Figure 6E), or randomly distributed in the invaded cell (Figure 6D). The third mutant, R82^N95-100, in which amino acids 95 to 100 (RKKFRY) were replaced with NAAIRS, showed similar localization patterns as R82^N26-31, with puncta ranging from being barely visible to as big as 2 to 3 µm in diameter docked around the haustorium, likely incapable of fusing into the EHM (Figures 6F and 6G). T1 lines (>30) transgenic for each of the three mutants were all susceptible, indicating that these mutant proteins are also defective in defense activation. Interestingly, these three NAAIRS replacements were located in two small regions, amino acids 20 to 31 and amino acid 95 to 100, each of which respectively falls into the two bigger regions, amino acids 1 to 42 and amino acids 94 to 115, defined as largely responsible for EHM targeting by our deletion analyses (Figure 1). In addition to these three mutants, three other mutants (i.e., R82^N8-19, R82^N32-37, and R82^N89-94) also showed loss of (for R82^N8-19) or obvious reduction (the remaining two) in EHM targeting. Apart from R82^N8-19 in which 12 amino acids in the TMD were replaced by NAAIRS (which likely affected the TMD properties), the remaining two had the replacements close to the two regions amino acids 20 to 31 or amino acids 95 to 100 most critical for EHM targeting, suggesting the neighboring residues may also be important for efficient EHM targeting. All these six mutants were nonfunctional in defense activation, as none of the T1 lines transgenic for each of these constructs displayed SHL, HR, or resistance to powdery mildew (see Supplemental Table 1 online).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

NAAIRS Scanning Identified Short Amino Acid Motifs in RPW8.2 Important for EHM Targeting.

YFP signal from tagged RPW8.2 mutant proteins is pseudo-colored green, PI-stained structures are red, and autofluorescent chloroplasts or plastids are blue. Images are representative Z-stack projections of 15 to 45 optical sections. h, haustorium; p, penetration site. Bars = 10 μm.

(A) The RPW8.2 protein sequence and positions of the 29 NAAIRS replacement and potential functional motifs and residues. Short lines underneath the amino acid sequences indicate positions of NAAIRS replacements; green lines indicated mutants with more or less normal EHM localization; red lines indicate mutants largely defective in EHM targeting; black lines indicates mutants without detectable protein accumulation. Two R/K-R/K-x-R/K motifs critical for EHM targeting are shaded pink; two putative NLSs are shaded green; two putative NESs are shaded purple.

(B) The NAAIRS replacement mutant R82^N20-25 was only rarely observed in the EHM. In most cases, it was found as varied sized puncta peripheral to the haustorium.

(C) R82^N20-25 was occasionally observed in ring structures surrounding chloroplasts in the mesophyll cells.

(D) and (E) R82^N26-31 is largely defective in EHM targeting, forming protein aggregates (arrows) unrelated to the haustorium (D) and/or small puncta at or peripheral to the EHM (E).

(F) and (G) R82^N95-100 is largely defective in EHM targeting, forming big protein aggregates (arrows) and small puncta at or peripheral to the EHM.

(H) R82^{N20-25+N95-100} is defective in EHM targeting, forming small puncta in the cytoplasm and protein aggregates (arrows) at the penetration site where the papilla was stained red by PI.

(I) An optical section from (H) with bright field showing the presence of the haustorium. Arrows indicate protein aggregates.

(J) R82^{N20-25+N95-100} was occasionally observed in ring structures surrounding chloroplasts in the mesophyll cells.

(K) R82^{N26-31+N95-100} is defective in EHM targeting, forming small puncta in the cytoplasm and around the haustorium, and big aggregates (arrow) at the penetration site.

(L) An optical section from (K) with bright field showing the presence of the haustorium.

The remaining 23 NAAIRS mutants showed a range of phenotypes both in terms of cell death activation and EHM targeting (see Supplemental Table 1 online). For example, T1 seedlings transgenic for R82^N137-142 (ISTKLD replaced by NAAIRS) and R82^N143-148 (KIMPQP replaced by NAAIRS) started to develop massive SHL in rosette leaves when they were 2 to 3 weeks old and died before bolting (see Supplemental Figures 4C and 4D online). This observation suggested that the two mutant proteins have higher capacity to trigger inappropriate cell death at early stages of plant development. Unfortunately, we were unable to detect any YFP signal in seedlings transgenic for these two mutant constructs. Similarly, we also failed to detect YFP signal in mutant R82^N83-88 (VEENAE replaced by NAAIRS); however, we found that 10 out of 32 of the T1 plants transgenic for this construct developed SHL (see Supplemental Figure 4E online) and showed resistance to powdery mildew. R82^N38-43 also failed to accumulate to a detectable level and Col-gl lines transgenic for this mutant construct were also susceptible. The remaining 19 NAAIRS mutants did not activate SHL. However, some were able to induce normal HR and resistance to powdery mildew (see Supplemental Table 1 online). Not surprisingly, all of these NAAIRS mutants showed normal EHM targeting, even though there was variation in YFP signal intensity, most likely reflecting varied protein stability among these mutant proteins (see Supplemental Table 1 online).

Because none of three single NAAIRS replacements in the two defined regions amino acids 20 to 31 and amino acids 95 to 100 completely abolished the EHM targeting of RPW8.2, it is likely that the these two regions may additively contribute to EHM targeting. To test this, we made double NAAIRS replacement mutants, R82^{N20-25+N95-100} and R82^{N26-31+N95-100}, and examined their subcellular localization in stable transgenic T1 plants. As anticipated, these two double mutant proteins were mostly found in puncta unrelated to the EHM in the invaded cells, with bigger patches close to the haustorial neck region (Figures 6H to 6L), indicating a complete loss of EHM targeting. Occasionally, in some invaded cells, the mutant proteins were seen as punctate or diffuse signals in the proximity of haustoria (see Supplemental Figures 6A and 6B online). In addition, similar to R82^N20-25 (Figure 6C), R82^{N20-25+N95-100} was also found to form punctate rings surrounding chloroplasts in mesophyll cells (Figure 6J). Collectively, these data further suggest that the two regions, amino acids 20 to 31 and amino acids 95 to 100, indeed comprise two small regions critical for EHM targeting of RPW8.2.

R/K-R/K-X-R/K Represents the Core EHM Targeting Signal

A common feature of the two regions required for EHM targeting is that they both are enriched in basic residues, each containing four Arg/Lys residues. We thus decided to assess the importance of the basic residues and a few others in this motif using site-directed substitution with Ala. Among eight single-site mutations (i.e., R24A, K26A, D27A, R28A, R95A, K96A, K97A, and F98A), only R28A and F98A resulted in the formation of protein aggregates as shown by various-sized puncta at or peripheral to the EHM (Figures 7A and 7B), indicating that Arg-28 and Phe-98 are important for vesicle docking to or fusion with the EHM, the remaining six mutations did not grossly affect EHM localization of the respective mutant proteins (see Supplemental Figures 6C and 6D and Supplemental Table 1 online). In addition, we made 15 single- or multiple-site Ala substitutions at residues from 100 to 127. Except for Y100A, which did not have detectable expression, and L111A, which significantly reduced EHM targeting (thus agreeing with results from the C-terminal deletion analysis; Figure 4), the remaining 13 mutations did not significantly affect RPW8.2’s EHM targeting and defense function (see Supplemental Table 1 online). Next, we made three two-site mutations with a single site in each of these two regions. Mutant R82^R24A+R95A, in which Arg-24 and Arg-95 were both replaced by Ala, formed big protein aggregates at or around the EHM, implying a defect in fusion of R82^R24A+R95A vesicles with the EHM (Figure 7C). Mutant R82^R24A+K97A, in which Arg-24 and Lys-97 were replaced by Ala, was partially compromised in its ability to localize to the EHM. However, it seemed to be restricted to the portion of the EHM that wraps the haustorial neck with weak signal forming a ring around the penetration site (Figure 7D). Most significantly, mutant R82^K26A+R95A, in which Lys-26 and Arg-95 were replaced by Ala, largely failed in targeting to the EHM, as evidenced by randomly distributed puncta in the haustorium-invaded cells (Figure 7E), indicating that these two basic residues together are essential for EHM targeting. Close examination of these two regions identified a common R/K-R/K-x-R/K motif (shaded in pink in Figure 6A). Based on the above results, we propose that the R/K-R/K-x-R/K–containing motifs in amino acids 20 to 30 and amino acids 95 to 100 probably comprise the core EHM targeting signals in RPW8.2. We tentatively named amino acids 20 to 30 EHM-TARGETING SIGNAL1 (ETS1) and amino acid 95 to 100 ETS2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Site-Directed Mutagenesis at the Two R/K-R/K-x-R/K–Containing Motifs Provided Further Evidence for Their Role in EHM Targeting.

YFP signal from tagged RPW8.2 mutant proteins is pseudo-colored green, and PI-stained structures are red. Images are representative Z-stack projections of 15 to 45 optical sections. h, haustorium. Bars = 10 μm.

(A) R82^R28A is largely defective in EHM targeting, forming small puncta and big aggregates (arrows) at or peripheral to the EHM.

(B) R82^F98A is largely defective in EHM targeting, forming small puncta and patches at or peripheral to the EHM.

(C) R82^R24A+R95A is largely defective in EHM targeting, forming small puncta and big aggregates (arrows) at or peripheral to the EHM.

(D) R82^R24A+K97A is also largely defective in EHM targeting. YFP signal was mainly detected in the portion of the EHM around the haustorial neck (arrow). Weak YFP signal was also observed in a ring structure surrounding the penetration site and possibly in nearby PM (asterisks).

(E) R82^K26A+R95A is defective in EHM targeting as shown by formation of varied sized puncta in the cytoplasm unrelated to the EHM.

Defining the Minimum Amino Acid Sequence in RPW8.2 for EHM Targeting

Our data from deletion/truncation, NAAIRS replacement, and site-specific mutational analyses indicate that the putative TMD or signal peptide, ETS1, and ETS2 are necessary for efficient targeting of RPW8.2 to the EHM. To determine if these three domains/motifs are sufficient for EHM targeting, we made R82^{∆43-93+∆113-174} that contains amino acids 1 to 42 (the TMD and ETS1) and amino acids 95 to 112 (ETS2) (60 amino acids in total; Figure 1) using R82^{∆43-93+∆138-174} as template. All transgenic lines (>24) expressing this construct were as susceptible as Col-0 wild type, indicating this mutant protein is not functional in defense. However, this protein was clearly detected as diffuse and punctate signal at the EHM (Figures 8A to 8C). We thus conclude that the 60 amino acids containing the putative TMD or signal peptide and two ETSs are both necessary and sufficient for EHM localization. Interestingly, YFP signal from this fusion protein was also found around the nucleus (Figure 8B) and in the PSM connecting plastids with the nucleus (Figure 8D) in the same epidermal cells. In some epidermal cells, YFP signal was also found in punctate ring structures surrounding plastids in uninfected epidermal cells (Figure 8E). These features of subcellular localization of R82^{∆43-93+∆113-174} are similar to several RPW8.2 mutants, such as R82^{∆65-93+∆138-174} (Figures 5H to 5L), which collectively suggests that the EHM and the PSM may share certain common characteristics that enable localization of several RPW8.2 mutant proteins.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Minimal Sequence Requirement of RPW8.2 for EHM Targeting and a Model for Intracellular Trafficking Pathways in Epidermal Cells Invaded by Haustoria of Powdery Mildew.

YFP signal from the tagged RPW8.2 mutant protein is pseudo-colored green, PI-stained structures are red, and autofluorescent chloroplasts are blue. Images are representative Z-stack projections of 15 to 45 optical sections. h, haustorium; n, nucleus. Bars = 10 μm.

(A) to (C) R82^{∆43-93+∆113-174} is capable of reaching the EHM as mostly diffuse signal (A) or both diffuse and small punctate signal ([B] and [C]). Insets are single optical sections. Note weak YFP signal was also detected in ring structures surrounding the nucleus and plastids (arrows).

(D) R82^{∆43-93+∆113-174} was detected in the PSM 20 min after treatment with 10 µM abscisic acid. Note the YFP-positive stromule-like membrane filaments (arrows) extended from plastids into the nucleus.

(E) R82^{∆43-93+∆113-174} was detected in punctate rings surrounding plastids (asterisks) labeled by a stromule marker PT-mCherry (Nelson et al., 2007) (pseudo-colored red).

(F) A cartoon illustrating vesicle trafficking in a leaf epidermal cell invaded by a haustorium of powdery mildew. During the biogenesis of the EHM, major trafficking at the trans-Golgi network is oriented toward the EHM (1), while vesicle trafficking to PM (2) and other organelles such as the nucleus remain active. Vesicles loaded with protein cargos containing EHM targeting motifs such as RPW8.2 are specifically targeted to the EHM, while vesicles containing proteins without a strong sorting signal may passively enter the major EHM-oriented trafficking pathway (1). A common feature(s) may be shared between the EHM and the PSM, which may provide a trafficking cue for RPW8.2’s localization. The PSM may further serve as a trafficking highway for RPW8.2 to reach the EHM (3). In addition, RPW8.2 contains NLSs and export signals; consequently, some RPW8.2 mutant proteins are nuclear localized. This raises a possibility that a small portion of RPW8.2 may be nuclear localized for defense activation (4) and suggests that balanced trafficking forces may be required to ensure EHM targeting of RPW8.2. It has been proposed that some PM-localized proteins may be translocalized to the EHM via the endocytic vesicle trafficking pathway (Lu et al., 2012) (5); however, definitive evidence remains to be provided.