Plant ERD2-like proteins function as endoplasmic reticulum luminal protein receptors and participate in programmed cell death during innate immunity

Authors


(e-mail yuleliu@mail.tsinghua.edu.cn).

Summary

The hypersensitive response (HR), a form of programmed cell death (PCD), is a tightly regulated innate immune response in plants that is hypothesized to restrict pathogen growth and disease development. Although considerable efforts have been made to understand HR PCD, it remains unknown whether the retrograde pathway from the Golgi to the endoplasmic reticulum (ER) is involved. Here we provide direct genetic evidence that two Nicotiana benthamiana homologs, ERD2a and ERD2b, function as ER luminal protein receptors and participate in HR PCD. Virus-induced gene silencing (VIGS) of ERD2a and/or ERD2b caused escape of ER-resident proteins from the ER, and resulted in plants that were more sensitive to ER stress. Silencing of ERD2b delayed HR PCD induced by the non-host pathogens Xanthomonas oryzae pv. oryzae and Pseudomonas syringae pv. tomato DC3000. However, both silencing of ERD2a and co-silencing of ERD2a and ERD2b exacerbated HR PCD. Individual and combined suppression of ERD2a and ERD2b exaggerated R gene-mediated cell death. Nevertheless, silencing of ERD2a and/or ERD2b had no detectable effects on bacterial growth. Furthermore, VIGS of several putative ligands of ERD2a/2b, including the ER quality control (ERQC) component genes BiP, CRT3 and UGGT, had different effects on HR PCD induced by different pathogens. This indicates that immunity-related cell death pathways are separate with respect to the genetic requirements for these ERQC components. These results suggest that ERD2a and ERD2b function as ER luminal protein receptors to ensure ERQC and alleviate ER stress, thus affecting HR PCD during the plant innate immune response.

Introduction

In response to pathogen attack, plants have evolved various defense mechanisms to protect themselves. One is the hypersensitive response (HR), a form of programmed cell death (PCD), which is characterized by rapid death of plant cells at and immediately surrounding infection sites (Lam, 2004; Jones and Dangl, 2006). HR PCD may be triggered by a wide variety of pathogens and is often associated with plant resistance to pathogen infection (Martin et al., 2003). Although considerable efforts have been made to understand HR PCD, the molecular and cellular mechanisms involved are still unclear (Mur et al., 2008).

Non-host resistance is the most common and durable form of disease resistance exhibited by an entire plant species to a specific parasite or pathogen (Fan and Doerner, 2012). Xanthomonas oryzae pv. oryzae (Xoo) and Pseudomonas syringae pv. tomato (Pst) DC3000 are two well-known non-host pathogens that induce non-host HR PCD in tobacco plants (Li et al., 2004; Wei et al., 2007). HopQ1-1 has been reported to be the sole avirulence determinant for Pst DC3000-triggered HR in N. benthamiana (Wei et al., 2007). Harpin and HrpD6 have been reported to determine Xoo-triggered HR in tobacco plants (Li et al., 2004; Guo et al., 2010). Recent studies in Arabidopsis suggest that non-host resistance consists of multiple layers of innate immunity and protects plants from the vast majority of potentially pathogenic microbes (Fan and Doerner, 2012). By screening a virus-induced gene silencing (VIGS) library (Liu et al., 2005), we found that suppression of N. benthamiana ERD2b, a homolog of the yeast ER luminal protein receptor ERD2p, alters Xoo-induced non-host HR PCD.

In yeast and mammals, ERD2 proteins are responsible for ER protein retention by retrieval of ER luminal proteins. Many ER luminal proteins carry a conserved C-terminal tetrapeptide sequence, typically, HDEL in yeast and KDEL in mammals. However, plant ER luminal proteins may contain either HDEL or KDEL as the retention signal (Hadlington and Denecke, 2000). The HDEL/KDEL tetrapeptide motifs are necessary and sufficient for retention of ER luminal proteins (Denecke et al., 1992; Napier et al., 1992). In yeast and humans, ERD2 recognizes C-terminal HDEL/KDEL-like motifs of proteins that have escaped from the ER to the Golgi and retrieves these proteins back into the ER (Munro and Pelham, 1987; Lewis et al., 1990; b; Semenza et al., 1990; Lewis and Pelham, 1992; Wilson et al., 1993). However, whether plant ERD2-like proteins function as ER luminal protein receptors in plants has not been proven.

Although the role of ERD2-like proteins in yeast and mammalian cells has been widely investigated, little is known about their role in the whole-organism context. Impairment of the KDEL receptor in transgenic mice perturbed ER quality control (ERQC) and led to development of dilated cardiomyopathy due to ER stress, and such mice generally die earlier later (Hamada et al., 2004). The ERD2b mutation in Arabidopsis was reported to impair CRT3 protein accumulation and compromise Elongation factor Tu receptor EFR)-mediated pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI) (Li et al., 2009). Furthermore, ERQC was reported to be involved in EFR-mediated PTI in Arabidopsis, Xa21-mediated PTI in rice and N-mediated resistance in tobacco (Caplan et al., 2009; Li et al., 2009; Lu et al., 2009; Nekrasov et al., 2009; Saijo et al., 2009; Chen et al., 2010; Park et al., 2010). However, it is not clear whether ERD2-like proteins and ERQC are involved in HR PCD.

Here, we report that N. benthamiana ERD2a and ERD2b function as ER luminal protein receptors and participate in HR PCD during plant immunity.

Results

Identification of N. benthamiana ER luminal protein receptors

To identify novel genes involved in Xoo-induced HR PCD, we performed a high-throughput VIGS screen in N. benthamiana plants as previously described (Liu et al., 2005). Of 500 cDNAs that were silenced, we found that VIGS using the cDNA clone C27 delayed Xoo-induced HR PCD (see below). Sequence analysis of clone C27 revealed that it encodes a 215 amino acid protein that shares highest identity (83%) with the putative Arabidopsis ER luminal protein receptor ERD2b (Li et al., 2009). This gene was called NbERD2b. We further cloned its closest homolog, NbERD2a, and found that the putative NbERD2a protein shares 73% identity with NbERD2b. NbERD2a and NbERD2b share 44–50% amino acid identity with yeast ER lumen protein receptor (ERD2p) and the human KEDL receptors HsKDELR1 and HsKDELR2 (Semenza et al., 1990; Pfluger et al., 2005) (Figure S1a). Both NbERD2a and NbERD2b are predicted to have a typical G-protein-coupled receptor (GPCR) structure with seven transmembrane domains (Figure S1b).

ERD2a and ERD2b function as orthologs of yeast ERD2

To test whether ERD2a and ERD2b function in retrieval of ER luminal proteins, we cloned the genes encoding them into the LEU2-containing yeast expression vector GS315, and transformed them into yeast strain ΔLE26A, which lacks a chromosomal copy of ERD2 (ScERD2) but contains ERD2 from Kluvveromyces lactis on a URA3 plasmid (Semenza et al., 1990; Lee et al., 1993). ERD2 is an essential gene in yeast, and loss of URA3 is a prerequisite for yeast growth on plates containing 5-fluoroorotic acid (FOA), so only strains transformed with an active gene complementing ERD2 and without URA3 are able to grow on these plates. Yeast ΔLE26A strains transformed with NbERD2a or NbERD2b were able to grow on FOA plates, as did yeast transformed with ScERD2 (Figure 1, left). By contrast, strain ΔLE26A failed to grow on the plates. To confirm that growth on FOA plates was indeed due to introduction of functional ERD2 and loss of URA3, we checked the above yeast strains growing on FOA plates and the ΔLE26A strain growing on plates lacking uracil or leucine. Strains transformed with the LEU2 plasmid containing NbERD2a, NbERD2b or ScERD2 were able to grow on plates lacking leucine (Figure 1, middle) but not on plates lacking uracil (Figure 1, right). By contrast, the ΔLE26A strain failed to grow on plates lacking leucine but grew on plates lacking uracil. These results suggest that both Nb ERD2a and Nb ERD2b are orthologs of yeast ERD2 and can function as ER luminal protein receptors in yeast.

Figure 1.

 NbERD2a and NbERD2b complement the lethal phenotype of the yeast erd2 deletion mutant.
The parental yeast strain, ΔLE26A (P), and strains transformed with S. cerevisiae ERD2 (ScERD2), N. benthamiana ERD2a (NbERD2a) or N. benthamiana ERD2b (NbERD2b) in the LEU2-containing plasmid GS315 were tested on plates containing FOA. Colonies that grew on the FOA plate and the parental yeast strain (P) were tested for growth on plates lacking leucine or uracil.

ERD2a and ERD2b localize to the Golgi and ER

To determine the subcellular localization of ERD2a and ERD2b, we generated the fusion constructs ERD2a–YFP and ERD2b–YFP driven by the CaMV 35S promoter by fusing YFP to the C-terminus of ERD2a and ERD2b, respectively. These fusion proteins were transiently co-expressed with the Golgi marker construct G-cb in N. benthamiana (Nelson et al., 2007). ERD2a–YFP and ERD2b–YFP (yellow) are evident as punctate structures that overlap with punctae indicated by the Golgi marker G-cb (cyan) (Figure S2, arrowheads). It has been reported that the Golgi marker G-cb shows weak ER labeling and strong Golgi labeling, as observed in our assay (Nelson et al., 2007). We observed a network distribution of both ERD2a and ERD2b on the ER (Figure S2, yellow). These results suggest that ERD2a and ERD2b localize to the Golgi and ER, consistent with their predicted role as ER luminal protein receptors for traffic between the Golgi and the ER.

ERD2a and ERD2b function as ER luminal protein receptors in plants

To test whether both ERD2a and ERD2b function as ER luminal protein receptors in plants, we examined their role in retrieving ER-localized HDEL-tagged GFP (GFP–HDEL). We used the tobacco rattle virus (TRV) VIGS vector (Liu et al., 2002) to silence ERD2a and/or ERD2b in 16c plants expressing GFP–HDEL (Ruiz et al., 1998). Silencing of ERD2a (VIGS-ERD2a) or ERD2b (VIGS-ERD2b) alone had a weak effect on GFP fluorescence. By contrast, co-silencing of ERD2a and ERD2b (VIGS-ERD2a+2b) eliminated most of the GFP fluorescence and the upper leaves were red under ultraviolet light (Figure 2a). Confocal microscopy was performed to observe the ER network as revealed by GFP–HDEL green fluorescence. Slightly weaker GFP fluorescence was observed in the ER network of ERD2b-silenced plants compared to non-silenced control plants, but no obvious change was observed for ERD2a-silenced plants (Figure 2a, lower panel). However, ERD2a/2b co-silenced plants showed a dramatic decrease in fluorescence intensity. Real-time RT-PCR analysis showed that neither combined nor individual silencing of ERD2a and/or ERD2b had an obvious effect on GFP mRNA levels (Figure S3). We assessed GFP protein amounts by Western blotting and detected two forms of GFP (approximately 28 and 27 kDa) in all samples. Total GFP protein content was similar in ERD2a-silenced and control plants and slightly lower in ERD2b-silenced plants. However, total GFP protein content in ERD2a/2b co-silenced plants was dramatically lower (Figure 2b).

Figure 2.

 ERD2a and ERD2b function cooperatively as ER luminal protein receptors to mediate ER retrieval of HDEL-tagged GFP.
(a) Effect of silencing of ERD2a and/or ERD2b on GFP–HDEL fluorescence in 16c plants (GFP–HDEL transgenic plants). Leaves were photographed at 14 days post-inoculation (upper panel). The fluorescence intensity of ER networks labeled with GFP–HDEL in epidermal cells was observed by confocal microscopy (lower panel, scale bar = 2 μm). GFP fluorescence is green, and chlorophyll shows red autofluorescence under UV light.
(b) Effect of silencing of ERD2a and/or ERD2b on GFP–HDEL levels in total (left), soluble (S100, middle) and microsomal (P100, right) protein fractions in 16c plants. Lanes 1–4 represent samples from control, ERD2b-silenced, ERD2a-silenced and ERD2a/2b co-silenced 16c plants, respectively. Coomassie brilliant blue (CBB) was used as a loading control. The relative band intensity of GFP was normalized to the Coomassie brilliant blue control; results are means ± SD for three independent triplicate experiments, with mean control values set to 1.00. Values in lanes 1–4 are 1.00, 0.72 ± 0.14, 0.96 ± 0.09 and 0.39 ± 0.12 (< 0.05) for total protein; 1.00, 0.67 ± 0.07, 0.97 ± 0.11 and 0.41 ± 0.13 (< 0.05) for S100; 1.00, 0.48 ± 0.10, 0.90 ± 0.12 and 0.22 ± 0.11 (< 0.05) for P100. The gels were probed using anti-GFP, anti-V-H-ATPase (vacuolar H-ATPase subunit, a vacuolar membrane marker) and anti-PEPC (phosphoenolpyruvate carboxylase, a cytosolic marker) antibodies as indicated.

To test whether individual or combined silencing of ERD2a and/or ERD2b caused GFP–HDEL escape from the ER, we isolated the soluble fraction (S100) and the microsomal fraction (P100) as previously described (Tamura et al., 2003). We found that 28 kDa GFP existed in both microsomal and soluble fractions, but 27 kDa GFP was detected exclusively in the soluble fractions in all samples. The 28 kDa form is believed to be GFP–HDEL, whereas it has been reported that the 27 kDa form is truncated GFP–HDEL in which the C-terminal peptide including the HDEL sequence has been removed (Brandizzi et al., 2003; Tamura et al., 2004; Zheng et al., 2004). On delivery to a vacuole or the cytoplasm, 28 kDa GFP may first be processed into 27 kDa GFP prior to complete degradation. Lower GFP protein levels were observed in microsomal and soluble fractions from ERD2b-silenced and ERD2a/2b co-silenced plants compared to non-silenced control plants (Figure 2b, middle and right). In addition, ERD2b-silenced plants accumulated less 28 kDa GFP in the microsomal fraction compared to control plants (Figure 2b, right). Furthermore, 28 kDa GFP levels in the microsomal fraction were much lower in ERD2a/2b co-silenced plants. Because GFP fluorescence was observed in the ER network, we hypothesize that only 28 kDa GFP in the microsomal fraction and not that in the soluble fraction can lead to green fluorescence. These results suggest that ERD2b silencing and ERD2a/2b co-silencing result in escape of 28 kDa GFP from the ER into vacuoles or the cytoplasm, and both ERD2a and ERD2b are required for retrieval of HDEL-tagged GFP in plants.

We also investigated the effect of ERD2a and ERD2b on the retrieval of ER-localized KDEL-tagged GFP (GFP–KDEL). We generated transgenic N. benthamiana plants expressing GFP–KDEL (termed GK plants). GFP–KDEL localizes to the ER (Boevink et al., 1996). We performed similar silencing experiments in GK plants as described above for 16c plants. GK plants subjected to ERD2a silencing or infected with the TRV vector alone showed similar levels of GFP fluorescence. By contrast, the upper leaves of ERD2b-silenced plants appeared red, like those of ERD2a/2b co-silenced plants, under ultraviolet light 2 weeks after inoculation (Figure 3a). Confocal microscopy showed that ERD2b silencing and ERD2a/2b co-silencing dramatically reduced GFP fluorescence in the ER network (Figure 3a, lower panel). Furthermore, silencing of ERD2b caused a dramatic decrease in GFP–KDEL in both the total and subcellular protein fractions, and ERD2a/2b co-silencing caused an even greater decrease (Figure 3b). Real-time RT-PCR showed that neither combined nor individual silencing of ERD2a and/or ERD2b had an obvious effect on GFP mRNA (Figure S3). These results suggest that both ERD2a and ERD2b are required for retrieval of KDEL-tagged GFP in plants.

Figure 3.

 ERD2a and ERD2b function cooperatively as ER luminal protein receptors to mediate ER retrieval of KDEL-tagged GFP.
(a) Effect of ERD2a and/or ERD2b silencing on GFP–KDEL fluorescence in GK plants (GFP–KDEL transgenic plants). Leaves were photographed at 14 days post-inoculation (upper panel). The fluorescence intensity of ER networks labeled with GFP–KDEL in epidermal cells was observed by confocal microscopy (lower panel, scale bar = 2 μm).
(b) Effect of ERD2a and/or ERD2b silencing on GFP–KDEL levels in total (left), S100 (middle) and P100 (right) protein fractions in GK plants. Lanes 1–4 represent samples from control, ERD2b-silenced, ERD2a-silenced and ERD2a/2b co-silenced GK plants, respectively. Coomassie brilliant blue (CBB) was used as a loading control. The relative band intensity of GFP is normalized to the Coomassie brilliant blue control; values are mean ± SD for three independent triplicate experiments, with mean control values set to 1.00. Values in lanes 1–4 are 1.00, 0.32 ± 0.12 (< 0.05), 0.82 ± 0.17 and 0.11 ± 0.15 (< 0.05) for total protein; 1.00, 0.30 ± 0.15 (< 0.05), 0.36 ± 0.20 and 0.08 ± 0.05 (< 0.05) for S100; 1.00, 0.13 ± 0.06 (< 0.05), 0.88 ± 0.15 and 0.05 ± 0.07 (< 0.05) for P100.

Taken together, our data clearly demonstrate that ERD2a and ERD2b are required for HDEL/KDEL-mediated retrieval of ER luminal proteins as ER luminal protein receptors in plants. In addition, VIGS of ERD2b eliminated most of the GFP fluorescence for GK plants but not for 16c plants, suggesting that ERD2b may have preferential affinity for KDEL-tagged proteins in plants.

ERD2a and ERD2b synergistically affect ER stress

Yeast ERD2 and its mammalian homologs are involved in ER stress and the unfolded protein response (Yamamoto et al., 2001). To test whether ERD2a and ERD2b have a role in mediating ER retrieval of endogenous ER luminal proteins and ER stress, we analyzed the effect of their silencing on gene expression of the ER stress marker BiP, an ER luminal protein. Both individual and combined silencing of ERD2a and/or ERD2b caused increases in BiP mRNA levels of approximately five- and eightfold, respectively, but only ERD2a/2b co-silencing caused an increase in the total protein level (Figure 4a,b). Furthermore, we found an increase in BiP protein in the soluble fraction and a reduction in the microsomal fraction in plants subjected to individual and combined ERD2a/2b silencing compared to control plants. These results suggest that more BiP escapes from the ER in plants subjected to silencing (Figure 4b). Together, these results suggest that both ERD2a and ERD2b have an important role in mediating ER retrieval of endogenous ER luminal proteins.

Figure 4.

 ERD2a and ERD2b synergistically regulate ER stress.
(a) Real-time RT-PCR results showing the effect of ERD2a and/or ERD2b silencing on BiP mRNA levels. Data are shown as means ± SD for three independent triplicate experiments, with mean control values set to 1.0 (*< 0.05, **< 0.01 versus control). Actin was used as the internal control.
(b) Effect of silencing of ERD2a and/or ERD2b on BiP levels in total (left), soluble (S100, middle) and microsomal (P100, right) protein fractions. Coomassie brilliant blue (CBB) was used as a loading control. Lanes 1–4 represent samples from control, ERD2b-silenced, ERD2a-silenced and ERD2a/2b co-silenced N. benthamiana plants, respectively. Coomassie brilliant blue was used as a loading control. The relative band intensity of BiP is normalized to the Coomassie brilliant blue control; values are means ± SD for three independent triplicate experiments, with mean control values set to 1.00. Values in lanes 1–4 are 1.00, 0.94 ± 0.14, 1.02 ± 0.07 and 1.80 ± 0.11 (< 0.05) for total protein; 1.00, 3.12 ± 0.11 (< 0.05), 3.27 ± 0.10 (< 0.05) and 3.74 ± 0.15 (< 0.05) for S100; 1.00, 0.62 ± 0.09 (< 0.05), 0.79 ± 0.13 (< 0.05) and 0.25 ± 0.08 (< 0.05) for P100.
(c) Plant response to the ER stressor DTT (30 mm). Photographs were obtained 24 h after injection.
(d) Cell death analysis. Data represent the percentage the death area versus the inoculated area as means ± SD (= 5). Significance was assessed using Student’s t-test (*< 0.05, ***< 0.001 versus control; †††< 0.001 versus ERD2a/2b-co-silenced plants). Cell death experiments were performed three times.

To further confirm the role of ERD2a and ERD2b in ER stress, we treated silenced plants with dithiothrietol (DTT), an ER stress inducer that inhibits disulfide bond formation (Martinez and Chrispeels, 2003). We found that plants subjected to ERD2a and/or ERD2b silencing were more sensitive to DTT and exhibited enhanced cell death; co-silencing caused even more severe cell death compared to control plants 24 h after treatment (Figure 4c,d). Together, these results suggest that both ERD2a and ERD2b function synergistically to alleviate ER stress by mediating ER luminal protein retrieval.

ERD2a and ERD2b have different roles in non-host resistance-associated HR PCD

We determined mRNA levels of ERD2b and ERD2a in the resistance response induced by Xoo. In time-course experiments, we found that mRNA levels of both ERD2b and ERD2a increased rapidly in the early stages of Xoo infection (Figure 5a).

Figure 5.

 Different roles of ERD2a and ERD2b in HR PCD induced by non-host pathogens.
(a) Real-time RT-PCR results showing that ERD2b and ERD2a mRNA levels increased rapidly in the early stages of the Xoo-induced resistance response. Relative levels are shown as means ± SD for three independent triplicate experiments, with mean values at time zero set to 1.0 (*< 0.05, **< 0.01 versus control). Actin was used as the internal control.
(b) Effect of silencing of ERD2a and/or ERD2b on HR PCD induced by Xoo and Pst DC3000. For better HR visualization, photographs were taken at 24 hpi for Xoo and 36 hpi for Pst DC3000 after ethanol clearing.
(c) Quantitative representation of HR PCD as percentage of cell death area versus inoculated area in plants subjected to ERD2a and/or ERD2b silencing. Data shown are means ± SD for three independent experiments (*< 0.05, ***< 0.001 versus control).
(d) Effect of silencing of ERD2a and/or ERD2b on growth of Xoo and Pst DC3000. Bacteria grown in planta were extracted from leaves at 3 days post-inoculation. Data shown are means ± SD for three samples from one of two individual experiments with similar results.

To further elucidate the role of both ERD2a and ERD2b in Xoo-induced HR PCD, we first performed VIGS to silence these two genes in N. benthamiana plants. These silenced plants did not exhibit any obvious developmental defects. Real-time RT-PCR was used to evaluate silencing levels, with Actin as an internal reference. Silencing of ERD2a or ERD2b alone reduced the corresponding mRNA levels by approximately 68 and 90%, respectively, but had no significant effect on mRNA levels of the other homolog. ERD2a/2b co-silencing reduced mRNA levels of ERD2a and ERD2b by 98% and 85%, respectively (Figure S4). These results suggest that silencing of ERD2a and ERD2b is effective and gene-specific. We then inoculated ERD2a- and/or ERD2b-silenced N. benthamiana plants with Xoo strain Pxo99. In non-silenced control plants, Xoo induced HR PCD at 12 h post-inoculation (hpi), as expected. However, no HR PCD was observed in ERD2b-silenced plants at 24 hpi, suggesting that ERD2b silencing delayed Xoo-induced HR PCD. Surprisingly, increased cell death and an increased percentage of cell death area versus inoculation area were observed in ERD2a-silenced and ERD2a/2b co-silenced plants (Figure 5b, upper panel, Figure 5c and Table 1). However, we found no effect of silencing of ERD2a and/or ERD2b on Xoo growth (Figure 5d, upper panel). These results suggest that ERD2a and ERD2b may have different roles.

Table 1.   Summary of effect of silencing of ERD2a and/or ERD2b on HR
HR trigger ERD2b-silenced ERD2a-silenced ERD2a/2b co-silenced
XooDelayIncreasedIncreased
Pst DC3000DelayIncreasedIncreased
N–TMV-p50IncreasedIncreasedMore increased
Cf9–Avr9IncreasedIncreasedIncreased
Pto–AvrPtoIncreasedIncreasedMore increased

Pst DC3000 is a well-recognized non-host pathogen of N. benthamiana that can cause rapid HR PCD (Keith et al., 2003). To further test the role of ERD2a and ERD2b in non-host resistance responses, we inoculated ERD2a- and/or ERD2b-silenced plants with Pst DC3000. HR was induced at 24 hpi on non-silenced control plants, as expected. However, even at 36 hpi, no significant cell death was observed in ERD2b-silenced plants (Figure 5b, lower panel, Figure 5c and Table 1). By contrast, we observed more severe cell death and a higher percentage of cell death area versus inoculation area in ERD2a-silenced and ERD2a/2b co-silenced plants than in control plants (Figure 5b, lower panel, Figure 5c and Table 1). However, we found no effect of silencing of ERD2a and/or ERD2b on Pst DC3000 growth (Figure 5d, lower panel). These results indicate that ERD2a and ERD2b play different roles in non-host pathogen Xoo and Pst DC3000-induced PCD, and there may be uncoupling of HR PCD from restriction of pathogen growth.

Silencing of ERD2a and ERD2b exacerbates R gene-mediated cell death

As ERD2a and ERD2b have different roles in non-host resistance-associated HR PCD, we hypothesized that ERD2a and ERD2b may play a similar role in HR PCD induced by R gene-mediated resistance. We tested the roles of ERD2a and ERD2b in HR PCD by transiently expressing the N gene elicitor TMV-p50 (Erickson et al., 1999) in N-containing N. benthamiana (NN) plants. Surprisingly, we did not observe a delay in TMV-p50-induced HR in ERD2b-silenced NN plants. By contrast, both individual and combined silencing exacerbated TMV-p50-induced cell death, and co-silencing caused more severe HR PCD (Figure 6a,b and Table 1). Furthermore, co-silencing of ERD2a/2b induced a higher percentage of cell death area versus inoculation area than individual silencing did (Figure 6b and Table 1).

Figure 6.

 Silencing of ERD2a and ERD2b exacerbates R gene-mediated HR cell death.
(a) Transient expression of TMV-p50 in NN plants, and of Cf9 with Avr9, and Pto with AvrPto in Nbenthamiana plants. For better HR visualization, photographs were taken 3 days post-inoculation with TMV-p50 and 40 hpi with Cf9/Avr9 and Pto/AvrPto after ethanol clearing.
(b) Quantitative representation of HR PCD as a percentage of cell death area versus inoculated area in plants subjected to ERD2a and/or ERD2b silencing. Data shown are means ± SD (= 5 for p50/N and = 4 for Cf9/Avr9 and Pto/AvrPto). *< 0.05, **< 0.01, ***< 0.001 versus control; ††< 0.01, †††< 0.001 versus ERD2a/2b-co-silenced plants. In all cases, experiments were performed at least three times.

To further test the roles of ERD2a and ERD2b in HR PCD associated with other resistance pathways, we transiently co-expressed the fungal resistance gene Cf9 and its corresponding avirulence gene Avr9 (Hammond-Kosack and Jones, 1997) and the bacterial resistance gene Pto and its corresponding avirulence gene AvrPto (Pedley and Martin, 2003) to induce HR PCD. These treatments induced HR PCD in all N. benthamiana plants. Both individual and combined silencing of ERD2a/2b exacerbated HR PCD (Figure 6a,b and Table 1). HR PCD induced by co-expression of Pto and AvrPto was more severe in co-silenced than in individually silenced plants (Figure 6a,b and Table 1). These results suggest that ERD2a and ERD2b act additively to affect R gene-mediated HR PCD.

Roles of ERQC machineries in HR PCD

As both ERD2a and ERD2b are required for HDEL/KDEL-mediated retrieval of ER luminal proteins, most of which are ERQC components, it is plausible that at least some of the ERQC components play a role in pathogen-induced HR PCD. To test this hypothesis, we silenced three genes of the ERQC machinery (BiP, CRT3 and UGGT) in N. benthamiana plants. In silenced plants, BiP, CRT3 and UGGT mRNA levels were reduced by approximately 91, 70 and 75%, respectively, compared to the levels in non-silenced control plants (Figure S5). In contrast to ERD2a/2b, silencing of BiP caused some developmental defects, including leaf curling to a fist shape and the appearance of many pits on leaves. Cessation of uppermost leaf formation and cell death were even observed on individual leaves (Figure S6), demonstrating that the BiP gene is essential for plant growth and development. As these defects were only observed for newly developing plant parts, the lower leaves could still be used for pathogen inoculation. The CNX–CRT–UGGT (calnexin–calreticulin–UDP-glucose:glycoprotein glucosyltransferase) complex has been implicated in protein folding in plants (Denecke et al., 1995; Crofts and Denecke, 1998; Pattison and Amtmann, 2009). Silencing of CRT3 or UGGT did not cause any obvious developmental defects before pathogen inoculation. Silencing of BiP but not CRT3 or UGGT resulted in delays in HR PCD induced by Xoo (Figure 7a,c and Table 2). By contrast, silencing of CRT3 or UGGT but not BiP resulted in a delay in HR PCD induced by Pst DC3000 (Figure 7b,c and Table 2). However, we found no difference in Xoo and Pst DC3000 growth between silenced and control plants (Figure 7d). These results suggest that ERQC plays an important role in pathogen-induced PCD.

Figure 7.

 Roles of ERQC components in HR PCD.
(a) Effect of silencing of BiP, CRT3 or UGGT on HR PCD induced by Xoo.
(b) Effect of silencing of BiP, CRT3 or UGGT on HR PCD induced by Pst DC3000. For better HR visualization, photographs were taken at 24 hpi for Xoo and 36 hpi for Pst DC3000 after ethanol clearing.
(c) Quantitative representation of HR PCD as a percentage of cell death area versus inoculated area in plants subjected to ERQC component silencing. Data shown are means ± SD for three independent experiments (*< 0.05, **< 0.01 versus control).
(d) Effect of silencing of BiP, CRT3 or UGGT on growth of Xoo and Pst DC3000. Bacteria grown in planta were extracted from leaves at 3 days post-inoculation. Data shown are means ± SD for three samples from one of two individual experiments with similar results.

Table 2.   Summary of effect of silencing ERQC components on HR
HR trigger BiP-silenced CRT3-silenced UGGT-silenced
XooDelayRetainedRetained
Pst DC3000RetainedDelayDelay

Discussion

Plant ERD2-like proteins function as ER luminal protein receptors

We identified N. benthamiana ERD2-like proteins ERD2a and ERD2b as orthologs of yeast ERD2 and the human KDEL receptor. In yeast and mammals, ERD2-like genes encode integral membrane proteins localized in the Golgi and ER that are required for retrieval of ER luminal proteins from the secretory pathway (Lewis et al., 1990; Semenza et al., 1990). These receptor types have also been found in other organisms (Lewis and Pelham, 1990; Lee et al., 1993; Pfluger et al., 2005; Li et al., 2009). Arabidopsis contains seven ERD2-like genes, and it has been shown that ERD2a (At1g29330) complements the yeast erd2 lethal phenotype (Lee et al., 1993). However, experimental proof that plant ERD2-like proteins can also function as ER luminal protein receptors in plants was lacking. Our study provides direct evidence that the N. benthamiana ERD2-like proteins ERD2a and ERD2b act as ER luminal protein receptors, and their ER retrieval roles depend on HDEL and KDEL signals. First, ERD2a and ERD2b share 44–50% amino acid identity with the yeast and human ER luminal protein retention receptors ScERD2p, HsKDELR1 and HsKDELR2 (Semenza et al., 1990; Pfluger et al., 2005) (Figure S1a). Second, the tobacco ERD2-like genes ERD2a and ERD2b complement the lethal phenotype of the yeast erd2 deletion mutant. Third, subcellular localization results for C-terminal YFP fusion proteins suggest that both ERD2a and ERD2b are localized to the Golgi and ER, in line with their potential role as ER luminal protein receptors for trafficking between these two organelles. Fourth, ERD2a/2b silencing affected protein levels and localization of ER-targeted HDEL/KDEL-tagged GFP in transgenic plants (microsomal versus soluble fractions). Fifth, a proportion of BiP, a HDEL-terminated endogenous ER resident, escaped from the ER in ERD2a- and/or ERD2b-silenced plants. In this study, we did not assess the effect of ERD2a/2b silencing on ER retention of plant endogenous KDEL-terminated ER resident proteins because no antibodies are available and there is no evidence indicating that such ER residents are involved in HR PCD. Further studies are required to confirm whether plant ERD2-like proteins actually contribute to ER retrieval of endogenous KDEL-containing proteins.

Mice with KDEL receptor deficiency develop dilated cardiomyopathy associated with ER stress, and generally die at approximately 14 months (Hamada et al., 2004), which indicates that KDEL receptors are essential for animal growth and development. By contrast, ERD2a/2b co-silenced plants exhibited no obvious developmental phenotypes under normal growth conditions, similar to Arabidopsis ERD2b mutants (Li et al., 2009). However, we cannot exclude the possibility that ERD2a and ERD2b play roles in plant development because VIGS was performed using adult plants, and this technique only knocked down rather than completely knocking out endogenous gene expression. Furthermore, there may be functional redundancy among N. benthamiana ERD2-related paralogs (ERPs) because ERPs other than ERD2a/2b exist in the Nicotiana EST database.

Given the existence of multiple ERPs in plants, it has been suggested that different retrieval signals (e.g. HDEL versus KDEL) could be recognized by different ERPs (Hadlington and Denecke, 2000). In Arabidopsis, mutation of ERD2b specifically affected the accumulation of CRT3, but not CRT1 and CRT2, although they all carry a C-terminal HDEL signal (Li et al., 2009). Here, we showed that NbERD2a and NbERD2b are functionally redundant in ER retrieval of HDEL-tagged GFP. Individual silencing of ERD2a had only a weak effect on ER escape by HDEL/KDEL–GFP. However, individual silencing of ERD2b resulted in ER escape by most KDEL-tagged GFPs but fewer HDEL-tagged GFPs in transgenic plants, suggesting that ERD2b may play a major role in retrieval of ER proteins that carry a C-terminal KDEL signal, even though ERD2a also contributes to this process. Our results suggest that different ERPs could have different abilities to recognize different retrieval signals.

We noted that AtERD2b and NbERD2b show some differences in subcellular localization and clients even though they share up to 83% amino acid sequence identity. NbERD2b was localized to both the Golgi and ER in our study, whereas AtERD2b is localized to the Golgi (Li et al., 2009). Perhaps this discrepancy arose because the YFP reporter was fused to the C-terminal end of NbERD2b in our study, whereas Li et al. (2009) placed the YFP reporter between the predicted signal peptide and the mature part of the coding sequence of AtERD2b. In addition, BiP is a client of NbERD2b in our study, but not of AtERD2b (Li et al., 2009). The latter difference may be species-specific. It would be interesting to investigate the biological significance of these differences.

In the present study, ERD2a/2b co-silencing eliminated almost all GFP fluorescence in transgenic plants expressing GFP–HDEL or GFP–KDEL. Furthermore, silencing of ERD2b alone eliminated most of the GFP fluorescence in transgenic GK plants expressing GFP–KDEL. ER escape of ER-targeted GFP may be evaluated by observing GFP fluorescence in transgenic plants under UV light. Thus, transgenic plants expressing ER-targeted GFP may be a powerful tool for dissecting the ER retention pathway.

Plant ERD2-like proteins contribute to ER stress

In yeast and mammals, ERD2/KDEL receptors participate in ER stress or the unfolded protein response, which is an evolutionarily conserved mechanism whereby cells respond to stress conditions (Beh and Rose, 1995). Arabidopsis ERD2a transcription is induced under various stress conditions (Bar-Peled et al., 1995), indicating that plant ERD2-like genes may be involved in ER stress. BiP has been used widely as an indicator of the onset of ER stress. BiP over-expression in plants can alleviate ER stress (Leborgne-Castel et al., 1999). We found that both individual and combined silencing of ERD2a and ERD2b resulted in a significant increase in BiP mRNA levels, and combined silencing also caused a slight increase in total protein levels. However, ERD2a/2b co-silenced plants were more sensitive to treatment with the ER stressor DTT. We found more BiP in the soluble fraction but less BiP in the microsomal fraction in plants subjected to individual or combined ERD2a/2b silencing compared to the levels in control plants. These results suggest that some BiP escaped from the ER, and only BiP in the ER is responsible for alleviating ER stress. Therefore, ERD2a and ERD2b may cooperate to regulate ER stress through BiP and/or other ER chaperones.

Roles of ERD2-like proteins in cell death

Pathogen infection induces ER stress and BiP expression (Jelitto-Van Dooren et al., 1999). We assumed that ERD2-like proteins may be initially activated to raises ER chaperone levels to relieve ER stress when plants are challenged with pathogens or other stresses (Vitale and Boston, 2008). One of the strategies to supplement the requirement for ER chaperones is to minimize chaperone loss by accelerating the retrieval process. We found that expression of ERD2a and ERD2b was transiently up-regulated during Xoo-induced HR. Co-silencing of ERD2a/2b resulted in more robust HR PCD induced by all the treatments tested in this study, suggesting that ERD2a and ERD2b are involved in alleviating pathogen-induced ER stress. Activation of ER stress may be either cytoprotective or pro-apoptotic (Capitani and Sallese, 2009). Moreover, we found that ERD2b silencing delayed HR cell death induced by Xoo or Pst DC3000 even though both ERD2a silencing and ERD2a/2b co-silencing exacerbated PCD induced by Xoo or Pst DC3000 and gene-mediated PCD tested. This different phenotype may be due to different strengths of ER stress induced by silencing of ERD2a and ERD2b. Prolonged ER stress always results in ER-associated cell death, consistent with our observation that Xoo and Pst DC3000 finally induce cell death, even in ERD2b-silenced plants.

It has been reported that the KDEL receptor mediates a retrieval mechanism that contributes to ERQC in mammals (Yamamoto et al., 2001). According to studies in mammalian and yeast systems, three main machineries mediate ERQC (Kleizen and Braakman, 2004; Buck et al., 2007): (i) the luminal-binding protein (BiP) complex, consisting of Hsp70 family chaperone BiP and Hsp40 family co-chaperone ERdj3 (Jin et al., 2008, 2009), (ii) the calnexin–calreticulin–UDP-glucose:glycoprotein glucosyltransferase (CNX–CRT–UGGT) complex (Denecke et al., 1995; Crofts and Denecke, 1998; Ruddock and Molinari, 2006; Pattison and Amtmann, 2009), and (iii) protein disulfide isomerases (PDIs) and thiol oxidoreductases, involving formation of disulfide bonds between free thiol groups in the same or different peptides (LaMantia et al., 1991; Wilkinson and Gilbert, 2004; Sakoh-Nakatogawa et al., 2009). We tested the role of these three different ERQC machineries in HR PCD induced by Xoo or Pst DC3000. Interestingly, silencing of BiP but not of CRT3 or UGGT delayed HR PCD induced by Xoo, while silencing of CRT3 and UGGT but not of BiP delayed HR PCD induced by Pst DC3000. However, we did not observe any effect of silencing of any of three PDI enzymes (1, 2 or 11) on HR PCD induced by either Pst DC3000 or Xoo (data not shown). These results imply that the factors involved in Xoo-induced HR PCD may depend on BiPs but not on CRT3/UGGT N-glycan decoration for function, while the factors involved in Pst DC3000-induced HR PCD may depend on CRT3/UGGT N-glycan decoration but not on BiPs for function. As silencing of PDIs had no effect on HR PCD induced by either of the bacterial strains, the factors involved in HR PCD induced by Xoo and Pst DC3000 may not contain disulfide bonds that are crucial for function. However, we cannot exclude the possibility that the factors involved in HR PCD induced by Xoo and Pst DC3000 may contain disulfide bonds that are crucial for function, because lack of a phenotype in plants with individually silenced PDIs may be due to genetic redundancy or incomplete silencing. These results reflect the possibility that HR induced by different pathogens requires distinct sets of ERQC components. It has been reported that several membrane receptor kinases regulate cell death (He et al., 2007; Kemmerling et al., 2007; Caplan et al., 2009; Gao et al., 2009), and protein secretion is involved in plant immunity (Bhat et al., 2005; Wang et al., 2005; Kalde et al., 2007; Kwon et al., 2008a,b). The separate cell death pathways may engage different membrane and/or secretory proteins that differentially require ERD2-like proteins and other ERQC components. We observed no effect of silencing of two ERD2-like genes and their putative ligands on bacterial growth in non-host plants. The results suggest that these two ERD2-like genes and their putative ligands may regulate HR PCD, but not non-host resistance.

ERD2a and ERD2b may regulate HR PCD through ERQC mechanisms. It is possible that ERD2b silencing lowers the function of the ER-resident proteins BiP, CRT3 and UGGT, which in turn reduces both BiP-dependent HR induced by Xoo and CRT3/UGGT-dependent HR induced by Pst DC3000. It has been reported that ERQC is required for PTI via regulation of biogenesis of the pattern recognition receptor EFR in Arabidopsis (Li et al., 2009; Lu et al., 2009; Nekrasov et al., 2009; Saijo et al., 2009). Arabidopsis ERD2b mutation eliminates accumulation of CRT3 and compromises EFR-mediated PTI (Li et al., 2009). However, no EFR homolog was found in N. benthamiana, suggesting the existence of an EFR-independent but salicylic acid-dependent immunity that requires ERQC components as reported previously (Saijo et al., 2009). In addition, BiP3 over-expression in rice reduces XA21-mediated resistance but does not affect the general rice defense response or cell death (Park et al., 2010). It was also reported that ERQC is involved in gene-mediated resistance (Caplan et al., 2009). Silencing of CRT3 reduces N-mediated resistance through a plasma membrane-localized receptor-like kinase (IRK) (Caplan et al., 2009). ERD2a/2b may also regulate HR PCD through ER chaperones. Silencing of ERD2a and/or ERD2b slightly reduced BiP accumulation in the ER, and this may have an effect on biogenesis of some proteins involved in HR PCD. In addition, some BiP was transported out of the ER for degradation in ERD2a/2b-silenced plants. Some ligands of BiP or other ER chaperones may be regulators of HR PCD and may be transported together into vacuoles or the cytoplasm for degradation (Pimpl et al., 2006). However, we cannot rule out the possibility that ERD2a/2b regulate HR PCD through other signaling pathways, because the KDEL receptor is involved in activation of Src family kinases in mammalian systems (Pulvirenti et al., 2008; Sallese et al., 2009).

Experimental procedures

Plant materials and plasmids

Details of the plant materials and plasmids used are provided in Data S1.

Yeast complementation assay

cDNA fragments of NbERD2a, NbERD2b and ScERD2 were cloned into pGS315, a LEU2-containing yeast expression vector (Pfluger et al. 2005). Yeast complementation assays were performed as described by Pfluger et al. (2005).

VIGS and pathogen-induced cell death assays

VIGS was performed as described by Liu et al. (2002). NN plants were used for TMV-p50 assays and N. benthamiana plants were used for other cell death assays. Xoo strain Pxo99 was resuspended in sterile water to an OD600 of 0.01, and inoculated into leaves. Leaves were subjected to 30 mm DTT inoculation for ER stress assays. All other cell death assays were performed as previously described (Liu et al., 2005). For a better view, leaves infected with Xoo, Pst DC3000, TMV-p50, Cf9/Avr9 and Pto/AvrPto were treated with ethanol (Gurlebeck et al., 2005). Inoculated areas are determined based on the edge of water-soaked areas generated by the infiltrations. Cell death areas and inoculated areas were calculated using a standard Photoshop CS4 method (Adobe, http://www.adobe.com). For bacterial growth tests, Xoo and Pst DC3000 were diluted to 106 and 105 cfu ml−1 using sterile water and 10 mm MgCl2, respectively, and inoculated into leaves. Four discs of 6 mm diameter were collected per sample at the time points indicated, and were used for pathogen counting as described by Peart et al. (2002).

Protein extraction and immunoblot analysis

Total, soluble and microsomal protein fractions were extracted from leaves as reported by Tamura et al. (2004). Rabbit polyclonal antibody against GFP (sc-8334; Santa Cruz Biotechnology, http://www.scbt.com) was used at a dilution of 1:5000 for GFP detection. Rabbit polyclonal antibody against BiP (sc-33757; Santa Cruz Biotechnology) was used at a dilution of 1:500. Rabbit polyclonal antibodies against Arabidopsis vacuolar H-ATPase subunit E (ab50208; Abmart, http://www.abmart.cn) and horseradish peroxidase-conjugated Phosphoenolpyruvate Carboxylase (PEPC, 200-4363; Rockland, http://www.rockland-inc.com) were used as subcellular markers for P100 and S100 fractions, respectively, at a dilution of 1:2000. Goat anti-rabbit horseradish peroxidase-conjugated IgG (BS13278; Bioword, http://www.bioworld.com) was used at a dilution of 1:20 000. The signal was detected by the enhanced chemiluminescence method. Protein concentrations were determined using a BCA kit (Pierce, http://www.piercenet.com). The protein marker SM0871 (Fermentas, http://www.fermentas.com) was used to estimate the relative molecular weight of protein bands. Coomassie brilliant blue was used as a loading control, and the relative band intensity was estimated using TotalLab, http://www.totallab.com/ version 2009 software (http://www.totallab.com).

RT-PCR and mRNA quantification

mRNA levels of N. benthamiana ERD2a, ERD2b and BiP were monitored by real-time RT-PCR. Total RNAs were extracted from leaves using TRNzol solution (Tiangen, http://www.tiangen.com) and then treated with RNase-free DNase I (Fermentas, http://www.fermentas.com) to remove potential DNA contamination. Real-time RT-PCR was performed using a SYBR Green kit (Applied Biosystems, http://www.appliedbiosystems.com) according to the manufacturer’s recommendations on a Bio-Rad CFX system (http://www.bio-rad.com).

Subcellular localization

Agrobacterium cultures containing ERD2a–YFP or ERD2b–YFP transient expression vectors were inoculated into leaves of N. benthamiana together with the Golgi marker construct G-cb (Nelson et al., 2007). Confocal imaging was performed 48 hpi using an inverted Zeiss LSM 710 laser scanning microscope (http://www.zeiss.com).

Statistical analysis

One-way anova followed by a Tukey test was used to assess real-time RT-PCR results, and a paired Student’s t-test was used for analysis of other data. Differences were considered statistically significant at < 0.05.

Accession numbers:

Sequence data from this article can be found in the GenBank/TGI EST data libraries under the accession numbers: NbERD2a (GU388433), NbERD2b (GU388432), AtERD2a (At1g29330), At ERD2b (AT3G25040), HsKDELR1 (NP_006792), HsKDELR2 (NP_006845), ScERD2p (NP_009513), NtBiP (TC5409), NbCRT3 (TC7352), NbUGGT (BP133373) and NbActin (TC13230).

Acknowledgements

We thank S.P. Dinesh-Kumar for thoughtful comments and critical reading of the manuscript, and Michael Lewis, S. Kamoun, J. Haseloff and D. Baulcombe for providing materials. This work was supported by the National Natural Science Foundation of China (grant number 30930060), the National Basic Research Program of China (grant number 2011CB910100) and the National Transgenic Program of China (grant numbers 2011ZX08009-003 and 2011ZX08010-002). We have no conflict of interest to declare.

Ancillary