A chloroplastic protein inhibits defense-induced cell death and is destabilized by activation of a disease resistance protein.
AbstractOne branch of plant immunity is mediated through nucleotide-binding/Leu-rich repeat (NB-LRR) family proteins that recognize specific effectors encoded by pathogens. Members of the I2-like family constitute a well-conserved subgroup of NB-LRRs from Solanaceae possessing a coiled-coil (CC) domain at their N termini. We show here that the CC domains of several I2-like proteins are able to induce a hypersensitive response (HR), a form of programmed cell death associated with disease resistance. Using yeast two-hybrid screens, we identified the chloroplastic protein Thylakoid Formation1 (THF1) as an interacting partner for several I2-like CC domains. Co-immunoprecipitations and bimolecular fluorescence complementation assays confirmed that THF1 and I2-like CC domains interact in planta and that these interactions take place in the cytosol. Several HR-inducing I2-like CC domains have a negative effect on the accumulation of THF1, suggesting that the latter is destabilized by active CC domains. To confirm this model, we investigated N′, which recognizes the coat protein of most Tobamoviruses, as a prototypical member of the I2-like family. Transient expression and gene silencing data indicated that THF1 functions as a negative regulator of cell death and that activation of full-length N′ results in the destabilization of THF1. Consistent with the known function of THF1 in maintaining chloroplast homeostasis, we show that the HR induced by N′ is light-dependent. Together, our results define, to our knowledge, novel molecular mechanisms linking light and chloroplasts to the induction of cell death by a subgroup of NB-LRR proteins.
In response to pathogens and pests, plants have evolved multiple sensor systems to activate defense mechanisms (Dangl et al., 2013; Dodds and Rathjen, 2010). One such mechanism is based on disease resistance (R) proteins belonging to the nucleotide-binding/Leu-rich repeat (NB-LRR) family. These proteins display a conserved protein structure, with a central NB domain and LRRs at the C terminus (Collier and Moffett, 2009; Takken and Goverse, 2012). NB-LRR proteins fall into three major clades, which correlate with the domains present at their N termini (Shao et al., 2014). TIR-NB-LRRs harbor a domain with similarity to Drosophila Toll and human IL-1 receptors (TIR), while most non-TIR-NB-LRRs harbor a coiled-coil (CC) domain and are therefore commonly referred to as CC-NB-LRRs. In addition, NB-LRR proteins with N-terminal CC domains related to the RPW8 protein (CCR-NB-LRRs) appear to function downstream of pathogen-sensing NB-LRRs (Peart et al., 2005; Bonardi et al., 2011; Collier et al., 2011).
As immune receptors, NB-LRRs recognize specific pathogen-encoded proteins known as effectors that are either delivered to, or synthesized in (in the case of viruses), the host cytosol (Chisholm et al., 2006). Following recognition of an effector, NB-LRRs induce effector-triggered immunity, which often culminates with the onset of a form of programmed cell death known as the “hypersensitive response” (HR; Dangl et al., 2013). The HR involves plant-specific characteristics, including vacuolization and chloroplast disruption (Coll et al., 2011). Several reports suggest that activation of different effector-triggered immunity features are mediated by NB-LRR proteins depending on their cytoplasmic or nuclear localization, and that the activation of the HR requires NB-LRR activity in the cytoplasm (Slootweg et al., 2010; Tameling et al., 2010; Heidrich et al., 2011; Bai et al., 2012). Development of the HR is also dependent on reactive oxygen species (ROS) produced by plasma membrane NADPH oxidases, as well as within organelles such as chloroplasts, mitochondria, and peroxisomes (Coll et al., 2011). However, the molecular mechanisms coupling the NB-LRR activation and the production of ROS remain unclear.
Structural analyses of NB-LRR proteins have indicated that the C-terminal LRR domain confers recognition specificity, while the central NB domain plays a role in regulation and possibly downstream signaling (Rairdan et al., 2008; Elmore et al., 2011; Takken and Goverse, 2012). The variable N termini have been implicated in at least two key NB-LRR functions. In many cases, recognition of pathogens is mediated through plant recognition co-factors that interact both with the N-terminal TIR or CC domain of an NB-LRR and the cognate effector protein (Collier and Moffett, 2009; Saunders et al., 2012; Lewis et al., 2013). Expression of the N-terminal domains of some NB-LRRs has also been shown to induce cell death, including TIR (Frost et al., 2004; Weaver et al., 2006; Swiderski et al., 2009; Krasileva et al., 2010; Bernoux et al., 2011), CC (Maekawa et al., 2011; Wang et al., 2015), and CCR (Collier et al., 2011) domains. These observations have led to the question of whether the N termini of NB-LRR proteins are involved in pathogen recognition, signaling, or both.
The I2 disease resistance gene of tomato (Solanum lycopersicum) encodes a CC-NB-LRR protein that confers resistance to race 2 of the vascular wilt fungus Fusarium oxysporum f sp lycopersici (Ori et al., 1997; Simons et al., 1998) through recognition of the xylem-secreted effector Avr2 (Houterman et al., 2007, 2009). I2 homologs (or I2-like) constitute a monophyletic clade found only in Solanaceous species (Pan et al., 2000; Couch et al., 2006). In potato (Solanum tuberosum), 26 I2-like genes have been identified (Jupe et al., 2012), including the R3a gene that confers resistance to isolates of the oomycete Phytophthora infestans that carry the RXLR effector Avr3a (Huang et al., 2004; 2005). Several members of the I2-like family have also been characterized for their antiviral properties. In pepper (Capsicum annuum), L gene alleles L1 to L4 recognize overlapping subsets of Tobamovirus coat protein (CP; Tomita et al., 2011) and Tobamoviruses are correspondingly classified into pathotypes (P0, P1, P1,2, P1,2,3, and P1,2,3,4), based on their ability to infect Capsicum plants carrying different L alleles (Genda et al., 2007). Tobacco Mosaic Virus (TMV) and Tomato Mosaic Virus (ToMV) belong to pathotype P0 and are therefore recognized by all L proteins. In Nicotiana sylvestris, the N′ gene also confers resistance to Tobamoviruses (Saito et al., 1987, 1989; Culver and Dawson, 1989); however, this I2-like protein has a unique recognition spectrum that protects against ToMV, but not TMV (Sekine et al., 2012).
In a systematic analysis of CC domains derived from Solanaceous NB-LRRs, we found that several I2-like CC domains induce cell death. To further characterize the signaling pathway associated with this response, we conducted yeast two-hybrid (Y2H) screens and identified the protein Thylakoid Formation1 (THF1) as an interacting partner for the CC domains of N′, R3a, and L3. THF1 is a chloroplastic protein, but it is encoded by a nuclear gene that is conserved in all oxygenic photoautotrophs, from cyanobacteria to flowering plants (Wang et al., 2004; Keren et al., 2005). The THF1 protein was first identified in photosystem II (PSII) preparations of the cyanobacterium Synechocystis sp PCC 6803 (Kashino et al., 2002) and appears to interact with and regulate PSII components (Huang et al., 2013; Yamatani et al., 2013). THF1 has also been shown to be involved in plant-pathogen interactions, as it is a direct target of the phytotoxin ToxA, a cell death-inducing protein of the necrotroph Pyrenophora tritici-repentis (Manning et al., 2007, 2010). THF1 has also been implicated in the response to coronatine (COR) during the interaction between tomato and Pseudomonas syringae (Wangdi et al., 2010). THF1 has been reported to have a complex organellar distribution, being detected in the envelope, thylakoid membrane, and stroma of Arabidopsis chloroplasts (Peltier et al., 2004, 2006). However, co-immunoprecipitation (Co-IP) and bimolecular fluorescence complementation (BIFC) experiments indicated that THF1 and I2-like CC domains interact in the cytosol. Interestingly, several I2-like CC domains negatively affected the accumulation of THF1, an effect that correlated with their propensity to induce cell death. Transient expression and gene silencing data further indicated that THF1 functions as a negative regulator of cell death that modulates the HR induced by a full-length version of N′. Consistent with the established function of THF1 in maintaining chloroplast homeostasis, we show that the HR induced by N′ is light-dependent. Overall, our data suggests that upon activation, I2-like proteins interact with THF1 in the cytosol, leading to its destabilization, and that diminished levels of THF1 then affects homeostasis of chloroplasts, which in turn contributes to the induction of a light-dependent HR.
RESULTS The CC Domains of I2-Like Proteins Induce Cell DeathWe previously reported that the CCR domains of NB-LRR proteins NRG1 and ADR1 induce rapid cell death upon transient expression in Nicotiana tabacum (tobacco) and N. benthamiana (Collier et al., 2011). This activity appeared to be specific to CCR domains, as the CC domains from several other NB-LRRs, including I2 and the I2-like protein R3a, did not induce cell death in tobacco (Collier et al., 2011). Recently, the resistance gene N′ was cloned (Sekine et al., 2012) and in studying its CC domain (N′CC), we observed that it also induces cell death upon transient expression in tobacco (Fig. 1A). Although the cell death induced by N′CC is slightly delayed compared to NRG1CCR, with tissue collapse starting around 30 h versus 24 h, respectively (Fig. 1A), the differences in responses are no longer apparent after 48 h. As reported previously (Collier et al., 2011), transient expression of the CC domains from closely related I2-like proteins (I2CC and R3aCC) or more divergent NB-LRRs (RxCC, Tm-2aCC, and Bs2CC) did not induce cell death in tobacco, despite being well expressed (Fig. 1B).
Figure 1.The CC domain of N′ induces cell death in N. tabacum (tobacco). Agrobacterium-mediated transformation was used to express HA-tagged CC domains of I2-like or non-I2-like NB-LRRs, as well as the CCR domain of NRG1. A, Tobacco leaf sectors were photographed at 24-h intervals following infiltration. B, Tissues from infiltrated sectors were sampled after 24-h expression (20 h for NRG1CCR) and total protein extracts were subjected to IB with anti-HA antibodies. Ponceau staining demonstrates equal loading between the various protein samples.
To further investigate the cell-death response induced by N′CC, we tested the CC domains of additional I2-like proteins, including homologs (I2-5CC, I2-7CC, I2-28CC, and I2-31CC) identified from tobacco (Couch et al., 2006), as well as the L proteins (L1CC, L2CC, L3CC, and L4CC) from pepper (Tomita et al., 2011). Upon transient expression in tobacco, we found that only N′CC induced a clear cell-death response (Supplemental Fig. S1A), even though all CCs were readily detectable by immunoblotting (IB; Supplemental Fig. S1B). In contrast, transient expression in N. benthamiana revealed that the CC domains of several I2-like proteins induce a strong cell-death response (Fig. 2A). Notably, expression of R3aCC, I2CC, I2-7CC, and I2-31CC induced cell death with strength and timing similar to N′CC. Cell-death symptoms were also visible in tissues infiltrated with I2-5CC, albeit with less intensity. A very weak cell-death response was sometimes observed for I2-28CC, but never with the CC domains of the L proteins, even after ethanol-clearing of infiltrated leaves (Fig. 2A). IB again confirmed expression of all CCs (Fig. 2B). Although it remains unclear why N′CC is the only CC domain to induce cell death in tobacco, these findings indicate that the induction of cell death is a common property of I2-like CCs.
Figure 2.Multiple I2-like CCs induce HR in N. benthamiana. HA-tagged CC domains, CCR domains, or pBIN61 EV were expressed in N. benthamiana leaves by agroinfiltration. A, After 72-h expression, infiltrated leaves were photographed before (left panels) and after ethanol clearing (middle panels). Thumbnail leaf diagrams indicate the positioning of each CC domain and I2-like proteins are denoted in red (right panels). B, Tissues from infiltrated sectors were sampled after 24-h expression and total protein extracts were subjected to IB with anti-HA antibodies. Ponceau staining demonstrates equal loading between the various protein samples.
The ability of several I2-like CCs to induce cell death suggests that these domains function in defense signaling. However, several I2-like CCs did not induce cell death, suggesting that these domains either signal in a qualitatively different manner, or induce the same response with a magnitude below a threshold required to induce cell death. Interestingly, fusing the enhanced yellow fluorescent protein (eYFP) to the C terminus of L2CC (Supplemental Fig. S2A) resulted in a fusion protein (L2CC-eYFP) that induced a strong cell-death response upon transient expression in N. benthamiana (Supplemental Fig. S2B). IB revealed that L2CC accumulated to higher levels than L2CC-eYFP (Supplemental Fig. S2C), indicating that protein stabilization by the eYFP moiety does not account for this observation. Nonetheless, this result indicates that weak I2-like CCs also have the potential to induce cell death.
Y2H Screens Identify the Chloroplastic Protein THF1 as an Interactor of the CC Domains from Several I2-Like ProteinsTo better understand the cell-death signaling function of I2-like CCs, we used Y2H screening to identify proteins with which they interact. Given the strong activity of N′CC in tobacco and N. benthamiana, we first focused on this CC domain and used it as bait to screen a cDNA library derived from stressed N. benthamiana leaves (see “Materials and Methods”). As we have found that N-terminal fusions to CC domains inhibited the activation of cell death (Supplemental Fig. S3), N′CC was cloned in a bait vector that allows C-terminal fusion of the GAL4DBD (Stellberger et al., 2010).
Following preliminary characterization of bait toxicity, expression, and autoactivation of the reporter gene His-3 in yeast (Supplemental Fig. S4), screening of the library resulted in the identification of several potential interacting partners. With 13 positive clones encoding seven fragments of the same protein (59–169 amino acids; Fig. 3A), the most frequently recovered interacting candidate was the chloroplastic protein Thylakoid Formation1 (THF1). The sequence of THF1 from N. benthamiana (NbTHF1; Nbv5.1tr6229924) was highly similar to that of Arabidopsis (AtTHF1; AT2G20890) and nearly identical (approximately 97% identity) to the THF1 protein from tobacco (NtTHF1; SGN mRNA_133291_cds; Supplemental Fig. S5). To confirm the screening data, prey vectors encoding the various THF1 fragments were purified and retransformed with N′CC as bait. Following inoculation on quadruple dropout medium (-WLHA) containing X-α-GAL, co-transformation resulted in the development of blue colonies (Fig. 3B). On the other hand, cells co-transformed with N′CC and empty prey vector (pGADT7) or THF1 fragments and empty bait vector (pGBKT7) did not form colonies on selection medium (Fig. 3B). Taken together, these results indicate that N′CC and THF1 fragments physically interact in yeast.
Figure 3.Identification of THF1 as an interacting partner of N′CC. A, On-scale depiction of the full-length THF1 protein from N. benthamiana and THF1 fragments identified during Y2H screening (F1–F7). Color-coded boxes indicate defined domains of THF1. B, For plasmid rescue experiments, various combinations of plasmids encoding THF1 fragments, N′CC as well as empty bait (pGBKT7) and prey (pGADT7) vectors, were co-transformed in yeast strain AH109. Cells were plated on quadruple dropout selection medium, supplemented with X-α-GAL and 3-AT (3-amino-1,2,4-triazole). After 5 d, colonies that turned blue were restreaked for presentation purposes onto selective-dropout medium plates lacking Trp and Leu (-WL) or Trp, Leu, His, and Ade (-WLHA) supplemented with X-α-GAL and 3-AT. On the right, a legend indicates positioning of each plasmid combination. Plates were photographed 5 d after transfer.
Sequences of the identified THF1 fragments (Fig. 3A) suggested that N′CC interacts with the C terminus of THF1, which is predicted to encode a CC domain (McDonnell et al., 2006; Trigg et al., 2011). To confirm interaction between N′CC and the CC domain of THF1, we generated THF1 deletion mutants (Supplemental Fig. S6A) and tested their abilities to interact with N′CC in yeast. First, we cloned the full-length THF1 protein, as well as a version lacking the predicted N-terminal plastid import sequence (THF1-∆SP; Emanuelsson et al., 1999). When co-transformed in yeast, full-length THF1 and THF1-∆SP did not interact with N′CC, as reflected by the absence of colony development on quadruple dropout medium (Supplemental Fig. S6B), although it cannot be ruled out that THF1 and THF1-∆SP were mislocalized or unstable in yeast. In contrast, use of a construct that mimics the longest THF1 fragment isolated by Y2H screening (THF1127-295; Supplemental Fig. S6A) allowed growth of blue colonies, suggesting interaction with N′CC (Supplemental Fig. S6B). Using additional derivatives of THF1, we showed that THF1127-295 lacking its CC domain (THF1-∆CC) did not interact with N′CC, whereas the CC domain of THF1 alone (THF1-CC) did (Supplemental Fig. S6B). Taken together, these results suggest that the CC domain of THF1 is necessary and sufficient for the interaction with N′CC. The CC domains of I2-like proteins are highly homologous and thus the CC domain of R3a (R3aCC) behaved exactly like N′CC when tested with THF1 derivatives in Y2H analyses (Supplemental Fig. S7).
In a parallel approach, amino acids 1 to 200 of the pepper I2-like protein L3 were fused to LexA and the resulting bait was used to screen a second Y2H library made of unchallenged N. benthamiana leaves (see “Materials and Methods”). A cDNA encoding the last 54 amino acids of THF1 was isolated from 26 distinct positive clones (data not shown). A construct expressing these 54 amino acids was then created to confirm interaction with L3(1-200), as well as a number of L3CC deletion constructs (Supplemental Fig. S8A). Co-transformation showed that the C terminus of THF1 interacts with all but three L3CC deletions (Supplemental Fig. S8B). From these results, it could be deduced that the THF1-interacting region of L3CC is located between amino acids 101 and 160, a region of the CC domain highly conserved in all I2-like CCs examined (Supplemental Fig. S8C).
THF1 and I2-like CCs Interact in PlantaTo confirm that THF1 and I2-like CCs interact in planta, the full-length THF1 gene from N. benthamiana was cloned into a binary expression vector fused to a C-terminal FLAG (FLG) epitope tag. THF1 was then expressed alone or in combination with HA-tagged N′CC. As a control, we used the HA-tagged CC domain of Tm-2a (Tm-2aCC), a non-I2-like NB-LRR that recognizes the movement protein (MP) of Tobamoviruses (Weber and Pfitzner, 1998; Lanfermeijer et al., 2003). Following expression by agroinfiltration in tobacco leaves, total protein extracts were prepared and IB was performed to assess the expression of input proteins (top panel, Fig. 4A). Expression of FLG-tagged THF1 was detected in all expected samples, but THF1 abundance was greatly reduced when expressed with N′CC. This suggests that this CC domain has a negative effect on the accumulation or stability of THF1 (see below). As expression of all the tagged proteins was nonetheless confirmed, immunoprecipitation (IP) of FLG-tagged THF1 was conducted using the same protein extracts. IB with anti-FLG antibodies confirmed that THF1 was immunoprecipitated (bottom panel, Fig. 4A) and IP protein levels correlated with THF1 input signals (top panel, Fig. 4A). Despite low levels of THF1 in the presence of N′CC, we still observed an interaction between THF1 and N′CC, as seen by anti-HA IB of anti-FLG IP samples (bottom panel, Fig. 4A). In contrast, N′CC was not co-immunoprecipitated when expressed with empty vector (EV) nor did THF1 co-immunoprecipitate Tm-2aCC (bottom panel, Fig. 4A). Taken together, these results indicate that N′CC and THF1 interact specifically in planta.
Figure 4.In planta interactions between THF1 and I2-like CCs. A, Co-IP experiments confirm that THF1 interacts specifically with N′CC in planta. Combinations of FLG-tagged THF1, HA-tagged N′CC, Tm-2aCC, or EV were expressed in tobacco leaves by agroinfiltration, as indicated. Approximately 30 h after infiltration, proteins extracts were prepared. Top panel (input): expression of epitope-tagged proteins was assessed by IB with anti-HA or anti-FLG antibodies, as indicated. Ponceau staining demonstrates equal loading between the various protein samples. Bottom panel (IP: α-FLG): the same protein extracts as above were used to immunoprecipitate FLG-tagged THF1 using anti-FLG agarose beads. Recovered proteins were detected by IB with anti-HA or anti-FLG antibodies, as indicated. B, BiFC experiments confirm that the interaction between THF1 and I2-like CCs take place in the cytosol. The fusion proteins THF1-YN and L3CC-YC were expressed in N. benthamiana leaves by agroinfiltration, as indicated. Binary vectors solely expressing N- or C- terminal halves of the YFP were used as controls. Approximately 42 h after infiltration, mesophyll cell protoplasts were isolated from infiltrated areas and treated with clasto-Lactacystin-β-lactone or DMSO for 2 h. Protoplasts were then observed using a confocal microscope and filtered wavelengths that depict fluorescence emitted by the YFP and chlorophyll. Overlay images are also shown. Scale bars = 25 μM.
To determine the subcellular location of the THF1 and I2-like CC interactions, we performed bimolecular fluorescence complementation (BiFC) experiments by fusing the N-terminal half of YFP (YN) to THF1 and the C-terminal half of YFP (YC) to I2-like CCs. After cloning into binary vectors, expression of the fusion proteins was conducted by agroinfiltration in N. benthamiana leaves and mesophyll cell protoplasts were isolated from infiltrated areas. As observed following transient expression of HA-tagged N′CC (Fig. 2A), agroinfiltration of N′CC-YC resulted in rapid cell death that prevented isolation and microscopic observations of intact protoplasts. As L3CC did not cause a collapse of leaf tissues (Fig. 2A) and could interact with THF1 in yeast (Supplemental Fig. S8), we performed BiFC experiments by co-expressing THF1-YN and L3CC-YC. Co-transformation of these constructs allowed detection of a weak fluorescence signal located within the cytosol of isolated protoplasts (top panel, Fig. 4B). Interestingly, we observed much stronger cytosolic signals when transformed protoplasts were treated with the proteasome inhibitor clasto-Lactacystin β-lactone (second panel, Fig. 4B). Only faint background fluorescence signals were observed when the N- or C-terminal halves of YFP were expressed together or with L3CC-YC or THF1-YN, respectively (lower panels, Fig. 4B). Taken together, these results suggest that I2-like CCs and THF1 interact in the cytosol and that these interactions result in the destabilization of THF1 before its translocation into chloroplasts.
Transient Overexpression of THF1 Suppresses the Cell Death Induced by N′CCWhile investigating the in planta interaction between N′CC and THF1, we observed that unharvested leaves infiltrated with N′CC underwent cell death and complete tissue collapse 72 h after agroinfiltration (Supplemental Fig. S9A). On the other hand, co-expression of N′CC and THF1 resulted in only faint browning of the infiltrated tissues, suggesting that overexpression of THF1 suppressed the cell death induced by N′CC. IB confirmed expression of all tagged proteins and again suggested that N′CC compromises the accumulation of THF1 (Supplemental Fig. S9B). To rule out the possibility that differences in cell-death induction were linked to leaf-to-leaf variation, similar experiments were repeated using patch infiltration within the same leaf. Again, THF1 almost completely blocked the cell death induced by transient expression of N′CC (Fig. 5A), although the accumulation of THF1 was again severely compromised when co-expressed with N′CC (Fig. 5B). Taken together, these results suggest that THF1 functions as a repressor of the cell death induced by N′CC.
Figure 5.THF1 compromises cell death induced by N′CC. Combinations of FLG-tagged THF1, HA-tagged N′CC, Tm-2aCC, or EV were expressed in tobacco leaves by agroinfiltration, as indicated. A, The tobacco leaf shown was photographed 72 h after infiltration. B, Tissues from infiltrated sectors were sampled after 30 h expression and total protein extracts were subjected to IB with anti-HA or anti-FLG antibodies. Ponceau staining demonstrates equal loading between the various protein samples.
THF1 Accumulation Is Compromised by I2-Like CCs that Induce Cell DeathWe next tested whether the CC domains of additional I2-like and non-I2-like NB-LRRs also affect the accumulation or stability of THF1. After 30 h expression in N. benthamiana, IB indicated that the highest levels of FLG-tagged THF1 were achieved when this protein was expressed alone, whereas the non-I2-like CC domains of Rx, Tm-2a, and Bs2 resulted in little or no reduction of THF1 accumulation (Fig. 6A). Interestingly, we observed a strong correlation between the propensity of the I2-like CCs to induce cell death (Fig. 2A) and their effect on THF1 accumulation (Fig. 6A). While, I2CC, N′CC, R3aCC, I2-7CC, and I2-31CC all induced strong cell death and greatly reduced THF1 levels, the CC domains of I2-5, I2-28, and the L proteins did not induce strong cell death and did not affect THF1 levels more than Tm-2aCC, which did not induce cell death. Since protein samples were collected before the appearance of the cell-death symptoms, these results suggest that destabilization of THF1 plays a role in the induction of cell death by I2-like CCs. In contrast, accumulation of the unrelated FLG-tagged protein RanGAP2 (Sacco et al., 2007) was not affected by co-expression with the CC domains from N′, R3a, or Rx (Fig. 6B), indicating that the effect on FLG-tagged THF1 accumulation is specific.
Figure 6.Multiple I2-like CC domains compromise the accumulation of THF1. A, FLG-tagged THF1, in combination with HA-tagged CC domains or EV, was expressed in N. benthamiana leaves by agroinfiltration, as indicated. Approximately 30 h after infiltration, N. benthamiana leaves were sampled and total protein extracts were subjected to IB with anti-HA or anti-FLG antibodies. Ponceau staining demonstrates equal loading between the various samples. B, FLG-tagged RanGAP2, in combination with EV, HA-tagged N′CC, R3aCC, or RxCC was expressed in N. benthamiana leaves by agroinfiltration, as indicated. Approximately 30 h after infiltration, expression of epitope-tagged proteins was assessed by IB, as above. Ponceau staining demonstrates equal loading between the various samples. C, HA-tagged N′CC was expressed in N. benthamiana leaves by agroinfiltration. At the indicated times, leaf sectors were photographed and tissues were sampled (left panels). Asterisks (*) indicate time points for which cell-death symptoms were visible in infiltrated tissues. Total protein extracts were subjected to IB (right panels) with anti-HA or anti-THF1 antibodies, as indicated. Ponceau staining was used to assess protein integrity and demonstrates equal loading between the various protein samples.
Using anti-THF1 antibodies, we also investigated the stability of endogenous THF1 upon transient expression of HA-tagged N′CC in N. benthamiana leaves (Fig. 6C). Twenty-four hours after infiltration, IB confirmed low-level accumulation of N′CC, while endogenous THF1 levels remained unaltered. At 30 h, expression of N′CC was considerably increased, whereas the accumulation of endogenous THF1 was strongly reduced. Although very weak cell-death symptoms were visible at 30 h (see asterisks in Fig. 6C), Ponceau staining confirmed that the integrity of cellular proteins was not yet altered by the onset of HR. At 40 h, Ponceau staining revealed clear protein smearing, indicating that integrity of cellular proteins was compromised. Consistent with this, infiltrated leaf sectors were characterized by obvious HR symptoms (Fig. 6C). At this time point, endogenous THF1 was barely detectable, while the level N′CC had also significantly decreased likely as a result of the induced cell death. Taken together, these results suggest that N′CC destabilizes the endogenous THF1 protein prior to the initiation of HR.
THF1 Suppresses N′-Mediated HR Following Recognition of the ToMV CPWe next assessed the effect of THF1 on the activity of full-length N′, which recognizes the CP of most Tobamoviruses, including ToMV (Sekine et al., 2012). When full-length N′ was co-expressed with the ToMV CP in N. tabacum or N. benthamiana (Fig. 7A), a strong HR was induced. However, this HR was almost completely abrogated by co-expression of THF1 (Fig. 7A). This cell-death suppression activity showed specificity in that THF1 did not suppress the HRs induced either by co-expressing Bs2 and AvrBs2 in tobacco, or Tm-2a and ToMV MP in N. benthamiana (Supplemental Fig. S10A). Together, these results confirm that THF1 not only represses the cell death induced by N′CC, but also that induced by the full-length N′ following activation by the ToMV CP.
Figure 7.THF1 compromises the HR induced by the activation of N′. A, Agroinfiltration was used to express the indicated combinations of constructs with Agrobacterium containing EV to make the total O.D.600 equal in all infiltrated patches. The N. tabacum (left-side picture) and N. benthamiana (right-side picture) leaves shown were photographed 72 h after infiltration. B, Tissues from infiltrated sectors were sampled 45 h after agroinfiltration and total protein extracts were subjected to IB with anti-FLG antibodies. Ponceau staining demonstrates equal loading between the various protein samples. C, The indicated combinations of N′, ToMV CP, or EV were expressed in N. benthamiana leaves by agroinfiltration, as indicated. Tissues from infiltrated sectors were sampled 48 h after agroinfiltration and total protein extracts were subjected to IB with anti-THF1 antibodies. Ponceau staining demonstrates equal loading between the various protein samples.
In the above-described experiment, tissues infiltrated with N′, ToMV CP, and THF1 were collected before the onset of HR (45 h after agroinfiltration) and IB analyses were conducted using anti-FLG antibodies. Expression of FLG-tagged THF1 was detected in all expected samples (Fig. 7B) but coexpression of N′ and ToMV CP resulted in markedly reduced accumulation of FLG-tagged THF1 (Fig. 7B), similar to the effect of N′CC (Figs. 4A, 5B, 6A, and S9B). For both Nicotiana species, these experiments confirm that activation of full-length N′ has a negative effect on the accumulation or stability of FLG-tagged THF1. Likewise, IB with anti-THF1 antibodies indicated that 48 h after agroinfiltration, the accumulation of endogenous THF1 was unaffected by expression of N′ or ToMV CP alone, whereas the co-expression of N′ and ToMV CP severely reduced endogenous THF1 levels prior to HR initiation (Fig. 7C). Likewise, endogenous THF1 levels were not affected upon infection of tobacco with ToMV (Supplemental Fig. S10B).
To confirm the interaction between THF1 and full-length N′, we performed Co-IP experiments, but only internally epitope-tagged versions of full-length N′ remain functional in planta (Sekine et al., 2012). We were unable to immunoprecipitate these proteins, possibly due to the epitope tag being unavailable to antibodies in the folded protein and to low expression levels. Nonetheless, our results suggest that activated N′ and the constitutively active N′CC affect THF1 similarly, and that over-production of THF1 has a similar inhibitory effect on the cell death induced by activated N′ and N′CC.
Silencing of THF1 Enhances Cell Death Mediated by N′CC or Full-Length N′To further study the cell death-inhibiting function of THF1, we used Tobacco Rattle Virus (TRV) as a Virus-Induced Gene Silencing (VIGS) vector (Ratcliff et al., 2001) to silence the expression of THF1 in N. benthamiana. Twenty-five days after infection with either pTV:00 (empty vector) or pTV:THF1, agroinfiltration was used to express N′CC and cell-death symptoms were monitored. In six-week-old pTV:00 plants, N′CC induced a relatively weak cell-death response whereas NRG1CCR induced a strong cell-death response in the same plants (Supplemental Fig. S11A). Since the plants used in this experiment are significantly older than those used in the experiments described above (i.e. four-week old plants in Fig. 2), this suggests that leaf age has a negative effect on the induction of cell death by N′CC. However, when expressed in pTV:THF1 plants, N′CC induced a more rapid and complete collapse of agroinfiltrated tissues (Supplemental Fig. S11A), suggesting that silencing THF1 enhanced cell death mediated by N′CC. IB with anti-THF1 antibodies confirmed the efficiency of THF1 silencing (Supplemental Fig. S11B).
VIGSed plants were also used to express full-length N′ and the ToMV CP. In pTV:00 plants, co-expression of N′ and ToMV CP resulted in a relatively weak HR, while co-expression of Tm-2a and the ToMV MP induced a strong HR in the same plants (Fig. 8A). As observed for N′CC, these results suggest that leaf age has a negative effect on the induction of cell death by full-length N′. However, when co-expressed in pTV:THF1 plants, N′ and the ToMV CP induced a more rapid and complete cell-death response (Fig. 8A). IB with anti-THF1 antibodies again confirmed the efficiency of THF1 silencing (Fig. 8B). Together, these results suggest that THF1 is a negative regulator of the cell death induced by the activated full-length N′, or its auto-active CC domain when expressed as a standalone protein fragment.
Figure 8.Silencing of THF1 enhances N′-induced cell death. Two-week-old N. benthamiana plants were infected with TRV EV (pTV:00), or TRV carrying a fragment of THF1 (pTV:THF1). A, Twenty-five days later, agroinfiltration was used to express N′ plus the ToMV CP or Tm-2a plus the ToMV MP. Each protein was also expressed individually, as indicated. Leaves were photographed 6 d after infiltration. The silencing experiment was conducted on 10 pTV:00 and 10 pTV:THF1 plants and a representative leaf set is shown. B, Total protein extracts from leaves of uninfected plants or plants silenced with pTV:00 or pTV:THF1 were subjected to IB with anti-THF1 antibodies. Ponceau staining demonstrates equal loading between the various protein samples. C, Agroinfiltration was used to express N′ or EV in the defined leaf areas. One hour after infiltration, the indicated leaf sectors were rub-inoculated with either sap containing ToMV:GFP virions or sap without virus (Mock). Leaves were photographed 6 d after virus infection under white light (first and third panels). Photographs taken under UV illumination (second and fourth panels) depict accumulation of ToMV:GFP (green fluorescence) or blue autofluorescing defense-related compounds. The silencing experiment was conducted on 10 pTV:00 and 10 pTV:THF1 plants, and a representative leaf set is shown.
In an additional set of silencing experiments, we combined transient expression with VIGS to investigate the role of THF1 in the interaction between the full-length N′ and a version of ToMV engineered to express GFP (ToMV:GFP; Hori and Watanabe, 2003). Plants silenced using pTV:00 or pTV:THF1 were agroinfiltrated with N′ or empty binary vector, followed by rub inoculation of ToMV:GFP. When inoculated in the absence of N′, ToMV:GFP accumulated to similar levels in both pTV:00- and pTV:THF1-silenced plants, as indicated by the visualization of green fluorescence (Fig. 8C). When ToMV:GFP was inoculated on the N′-expressing leaf sections of pTV:00 plants, small and well-defined HR lesions developed (Fig. 8C). In contrast, corresponding leaf sections of the pTV:THF1 plants were characterized by complete tissue collapse following the activation of an apparently unrestricted HR (Fig. 8C). UV illumination also indicated that the induction of HR was associated with the accumulation of blue autofluorescing compounds, as is commonly seen upon virus-induced HR in Nicotiana species (Costet et al., 2002; Chaerle et al., 2007). Consistent with the enhanced HR in THF1-silenced tissues, blue autofluorescing compounds were much more abundant in pTV:THF1 plants following activation of N′ (Fig. 8C). Together with the overexpression data (see above), these results are consistent with THF1 functioning as a repressor of cell death, whose activity regulates the HR induced by N′.
The HR Induced by N′ Is Light-DependentLight is often required for the development of the HR (Coll et al., 2011). Since THF1 is a chloroplastic protein that represses N′-induced HR, we tested whether that latter response is also light-dependent. First, we rub-inoculated ToMV:GFP on N. sylvestris plants, which express N′ endogenously (Saito et al., 1987, 1989; Culver and Dawson, 1989). Following infection, plants were either returned to their normal day and night cycle, or placed in constant darkness. After 4 d, plants under normal light conditions displayed confined HR spots (Fig. 9A). In contrast, infected leaf sectors of the dark-treated plants showed trailing necrosis that spread over the entirety of the infected areas. Defined HR spots were also visible, although these were typified by a much lighter brown coloration (Fig. 9A). Spreading cell death suggests that in the absence of light, HR was delayed and therefore less efficient at restricting the replication and spread of ToMV:GFP. Virus accumulation was monitored by UV illumination, but no green fluorescence could be observed on either the light- or dark-treated plants. This suggests that the absence of light delays the HR, but does not ultimately compromise viral resistance.
Figure 9.The HR induced by N′CC and full-length N′ is light-dependent. A, Leaf sectors from N. sylvestris plants were rub-inoculated with sap containing either ToMV:GFP virions or sap without virus (Mock). Following infection, plants were either returned to their usual light [L] cycle (16 h day/8 h night), or placed in constant darkness [D] under the same temperature and humidity conditions. Leaf sectors were photographed 4 d after viral inoculation. A thumbnail leaf diagram indicates positioning of viral and mock infections. B, Depiction of the lighting cycles used to study light-dependency of the HR induced by N′CC or full-length N′ in N. benthamiana (see C and D, respectively). Black and white rectangles correspond to light and dark conditions, respectively. Black triangles highlight time points at which leaf pictures were taken. C, N′CC, or EV were expressed in N. benthamiana leaves by agroinfiltration, as indicated. D, N′, and the ToMV CP were expressed in N. benthamiana leaves by agroinfiltration, either alone or together, as indicated. Bacteria containing EV were also used to make the total O.D.600 equal in all infiltrated sectors. Immediately after infiltration, plants were either returned to their usual light [L] cycle (16 h day/8 h night), or placed in constant darkness [D] under the same temperature and humidity conditions. Asterisks (*) indicate the time points at which cell-death symptoms were visible on the leaf sectors infiltrated with N′CC, or N′ plus ToMV CP. Thumbnail leaf diagrams indicate positioning of each expressed construct.
To further examine how light affects the timing of the HR induced by N′CC or full-length N′, we agroinfiltrated appropriate binary vector combinations in N. benthamiana leaves. After infiltration, plants were returned to their normal light cycle, or placed in constant darkness. At the indicated time points (Fig. 9B), leaves were collected and the development of cell death was monitored. When N′CC was expressed under normal light conditions, cell-death symptoms became visible around 30 h after infiltration (see asterisks in Fig. 9C). After 48 h, leaf sectors expressing N′CC had completely collapsed because of the induced cell death. When N′CC was expressed under constant darkness, cell death was obviously delayed, with the first signs of tissue collapse not detected until 48 h after infiltration (Fig. 9C). After 60 h, expression of N′CC in the dark nonetheless led to a complete collapse of infiltrated tissues, suggesting that the absence of light did not block HR, but clearly delayed its normal development. IB confirmed that accumulation of N′CC was similar under light and dark conditions (Supplemental Fig. S12).
When a similar experiment was conducted using full-length N′ and the ToMV CP, HR symptoms first became visible after 48 h expression under normal light conditions (see asterisks in Fig. 9D). For dark-treated leaves, HR was delayed, with cell-death symptoms not observed until 72 h after infiltration. At 96 h, co-expression of N′ and ToMV CP resulted in complete collapse of infiltrated tissues under light, while the HR was of much weaker intensity in dark-treated tissues (Fig. 9D).
DISCUSSION Activation of Cell Death by the CC Domains of I2-Like ProteinsExpression of the N-terminal regions from several TIR-NB-LRRs has been previously reported to induce cell death (Frost et al., 2004; Weaver et al., 2006; Swiderski et al., 2009; Krasileva et al., 2010; Bernoux et al., 2011), suggesting that TIR domains are involved in defense signaling. We have reported that the N-terminal CCR domains of NRG1 and ADR1 also induce cell death (Collier et al., 2011). However, these CC domains share little similarity with the CC domains of conventional CC-NB-LRRs and CCR-NB-LRRs are thought to function in signaling downstream of conventional NB-LRR proteins (Peart et al., 2005; Bonardi et al., 2011; Collier et al., 2011). To date, induction of cell death by the CC domains of conventional CC-NB-LRRs has only been reported for the barley (Hordeum vulgare) protein MLA10 (Maekawa et al., 2011) and the maize (Zea mays) protein Rp1 (Wang et al., 2015). Thus, the demonstration that several I2-like CCs induce cell death reinforces the idea that the N-terminal domains of at least some conventional CC-NB-LRRs are involved in defense signaling.
The CC domains of I2-like proteins do not appear to possess any obvious features distinguishing them from typical EDVID-type CC domains (Rairdan et al., 2008) that could explain why they induce cell death on their own while many others do not. However, as recent structural analyses have shown, EDVID-type CC domains can have significantly different tertiary and quaternary structures. The MLA10 CC domain forms homodimers and induces cell death (Maekawa et al., 2011), while the CC domain of Rx shows a tertiary structure that is markedly different from that of MLA10, does not form dimers, and does not induce cell death (Hao et al., 2013). As a result, signaling and/or self-interaction may represent the exception, or the rule, or may simply reflect the different propensities of CC domains to function on their own. Alternatively, these differences may reflect different mechanisms of action evolved by different NB-LRR subfamilies. Determining this will require the study of additional NB-LRR signaling complexes.
We have shown that certain I2-like CC domains induce cell death in N. benthamiana, but not in tobacco, while other I2-like CCs do not appear to trigger cell death at all (Fig. 2A). The reason for the differences in activities between plant species is unclear at this moment, although it is not without precedent. For example, in transient expression assays, the activation of the N protein induces a strong HR in tobacco, but not in N. benthamiana, whereas the Tm-2a protein shows the opposite pattern (Bhattacharjee et al., 2009). Likewise, certain CCR domains induce cell death in tobacco, but not N. benthamiana, and vice versa, for reasons that are not apparent (Collier et al., 2011). In N. benthamiana, expression levels of the different I2-like CCs do not necessarily correlate with activity (Figs. 2, 6A, S1, and S2). However, since no obvious feature differentiating I2-like CCs that do, or do not, induce cell death could be identified (Supplemental Fig. S13), we suggest that all I2-like CCs signal through the same pathway, but they do so with variable efficiencies. Signaling by these domains would normally be inhibited in the context of a nonactivated full-length protein and thus the fact that certain I2-like CCs do not show constitutive activity may stem from differences in folding when expressed as isolated protein domains. This hypothesis is supported by the observation that L2CC induces cell death when fused to eYFP, which presumably alters or stabilizes this CC domain in an active conformation (Supplemental Fig. S2).
Current models of disease resistance protein activation suggest that in the absence of pathogen, NB-LRRs fold into a signaling-competent state that retains the whole protein in an autoinhibited form through intramolecular interactions (Collier and Moffett, 2009; Takken and Goverse, 2012). Recently, random mutagenesis screening of the full-length R3a protein revealed that autoactivation of this I2-like can be achieved through a single amino-acid change within the N-terminal CC domain (I148F; Segretin et al., 2014). Although the precise molecular effects of this mutation are unknown, it is tempting to speculate that this change might interfere with the intramolecular interactions that restrain signaling by the CC domain, in agreement with our finding that R3aCC is able to induce cell death on its own.
The Interaction between I2-like CCs and THF1Bioinformatics analyses suggest that THF1 harbors a transmembrane domain such that the CC domain of the THF1 proteins present in the plastid envelope would project into the cytosol (Huang et al., 2006). Indeed, work on Arabidopsis root tissues has revealed that THF1 participates in sugar signaling through its physical interaction with the plasma membrane G protein α1 (Huang et al., 2006). Förster resonance energy transfer and biochemical analyses indicated that contacts between plastidic THF1 and plasma membrane G protein α1 occur through the cytoplasmic moieties of the two proteins at sites where the plastid membrane abuts the plasma membrane (Huang et al., 2006). Our Y2H analyses indicated that the CC domain of THF1 is necessary and sufficient for the interaction with the CC domains of N′ and R3a (Supplemental Figs. S6B and S7B). However, BiFC experiments indicated that such interactions take place in the cytosol, and no obvious fluorescence signal could be observed in or around chloroplasts, as would be expected if I2-like CC domains interacted with THF1 inside or on the surface of chloroplasts. This result is consistent with the fact that I2-like proteins lack a predicted chloroplast transit peptide and with results indicating that an I2-like protein localizes in the cytosol of leaf epidermal cells (Engelhardt et al., 2012). As a result, it appears that I2-like CCs intercept THF1 after it has been synthesized in the cytosol, but before its translocation into chloroplasts.
Is THF1 Involved in Effector Recognition or Downstream Signaling?Chloroplasts are an important source of defense signaling molecules, including ROS and defense-related hormones, and play an essential role in the development of HR (Apel and Hirt, 2004; Coll et al., 2011). Defense-related functions of chloroplasts are also highlighted by the fact that some pathogen effectors harbor chloroplast transit peptides and suppression of plant immunity requires their chloroplastic localization (Li et al., 2014). Interestingly, part of the TMV CP population localizes inside chloroplasts, where it is thought to induce chlorosis by interfering with PSII functions (Reinero and Beachy, 1986, 1989; Hodgson et al., 1989; Culver, 2002; Lehto et al., 2003). Likewise, a fraction of the ToMV CP has also been reported to localize inside chloroplasts, where it is thought to interact with a thylakoid membrane protein (Li et al., 2005; Zhang et al., 2008).
It has been previously reported that the chloroplastic protein NRIP1 is required for resistance to TMV (Caplan et al., 2008). Through its interaction with the 50 kD helicase domain of TMV replicase (p50), NRIP1 is redirected to the cytoplasm and nucleus, where it interacts with the N-terminal domain of N, a TIR-NB-LRR protein. Although taking place outside of chloroplasts, these protein-protein interactions imply that chloroplastic proteins can be involved in pathogen recognition. In light of all these findings, it was reasonable to hypothesize that THF1 corresponds to a virulence target of Tobamoviruses CP inside chloroplasts, but that it might also serve as a recognition co-factor for N′ outside the organelle, as reported for NRIP1 and N (Caplan et al., 2008). This second avenue would also be consistent with the fact that THF1 interacts with the N-terminal domain of N′, a common feature of recognition co-factors (Collier and Moffett, 2009).
As opposed to the CC domains of most NB-LRRs, I2-like CC domains are capable of inducing cell death on their own. Thus, it is equally plausible that THF1 affects signaling downstream of I2-like protein activation. This alternative hypothesis is supported by the fact that destabilization of THF1 occurs only when full-length N′ is activated following recognition of the ToMV CP (Fig. 7). In addition, silencing of a recognition co-factor would be expected to compromise the activation of its cognate NB-LRR protein, because recognition of the effector would then be severely hindered. In the present case, silencing of THF1 strongly enhanced the HR mediated by N′, suggesting that THF1 is not involved in pathogen recognition (Fig. 8). In addition, we were unable to show a direct interaction between THF1 and the ToMV CP (Supplemental Fig. S14). Although we cannot rule out that these proteins interact under conditions different from those tested, these results are consistent with the idea that THF1 affects the signaling function of N′ rather than mediating pathogen recognition. As a whole, our results thus suggest that upon recognition of the ToMV CP, N′ induces a HR at least in part by interfering with the accumulation or stability of THF1. Understanding how this destabilization is effected will require further study, although it is known that the Arabidopsis THF1 is rapidly degraded in a coronatine-insensitive protein1-dependent manner in the presence of the bacterial phytotoxin COR (Wangdi et al., 2010).
A Potential Role for THF1 in the Regulation of Plant Cell DeathIn Arabidopsis, the thf1 mutant is stunted, with variegated leaves harboring a mix of green and white sectors (Wang et al., 2004; Keren et al., 2005). Chloroplasts from green sectors have organized thylakoid membranes, whereas plastids from white sectors lack organized thylakoids and accumulate numerous membrane-bound vesicles (Wang et al., 2004). Importantly, this phenotype is more pronounced under high light and can be fully rescued under low-light conditions (Keren et al., 2005). Thus, it has been proposed that the absence of THF1 renders the thf1 mutant more susceptible to light-induced damages, resulting in a subsequent breakdown of thylakoid membranes (Keren et al., 2005).
Within thylakoid membranes, highly active photosystems require mechanisms to prevent photooxidation. A major target of photo-damage is the PSII reaction center protein D1, which is characterized by a rapid turnover via FtsH proteases (Lindahl et al., 2000; Adam and Ostersetzer, 2001; Bailey et al., 2002; Wagner et al., 2012). Diminished FtsH protease activity leads to the accumulation of photo-damaged D1 and subsequent oxidative stress caused by ROS accumulation in chloroplasts (Kato et al., 2009). Genetic and biochemical evidence now clearly indicates that THF1 and FtsH proteases are part of a common signaling pathway that regulates maintenance of the PSII (Zhang et al., 2009; Wu et al., 2013). In other words, THF1 prevents photooxidation of chloroplast components through its positive regulatory effect on FtsH protease accumulation and stability. As a result, the D1 protein is more stable in the thf1 mutant than in wild-type plants (Zhang et al., 2009).
By destabilizing THF1 before its translocation into chloroplasts, N′ likely interferes with mechanisms that protect chloroplasts from light-induced oxidation. Consistent with this, we found that light is required to ensure timely activation of cell death following recognition of the ToMV CP by N′ (Fig. 9). In addition, it has previously been shown that a chloroplast-specific FtsH protease is destabilized during N-mediated resistance against TMV (Seo et al., 2000) and light is required for the timely activation of the HR by N (Chandra-Shekara et al., 2006). Thus, we hypothesize that THF1 inhibits cell death by preventing light-induced damage in the chloroplasts and that inhibition of THF1 accumulation would in turn accelerate the negative effects of such damage, thus promoting cell death. Given that a lack of THF1 is not inherently lethal (Fig. 8 and Supplemental Fig. S11; Wang et al., 2004; Keren et al., 2005), destabilization of THF1 by I2-like proteins is likely only one of several events that lead to the HR, but which in itself induces a cellular environment that promotes the induction of cell death.
THF1 as a Hub for the Regulation of Programmed Cell Death in PlantsRecently, Arabidopsis and rice (Oryza sativa) homologs of THF1 have been reported to regulate leaf senescence, a developmentally controlled form of programmed cell death (Huang et al., 2013; Yamatani et al., 2013). In addition, at least two pathogen toxins target THF1 to promote infection, namely the necrotizing toxin ToxA from the wheat (Triticum aestivum) fungal pathogen P. tritici-repentis (Manning et al., 2007, 2010) and COR from the bacterial pathogen P. syringae (Wangdi et al., 2010). Interestingly, ToxA physically interacts with THF1 (Manning et al., 2007, 2010) and leads to cell death caused by light-dependent accumulation of ROS in chloroplasts (Manning et al., 2009). In THF1-silenced tomato plants, COR induces cell death instead of the usual chlorosis (Wangdi et al., 2010). Taken together, these findings suggest that THF1 is a critical hub for the control of programmed cell death in leaves exposed to light. Although I2-like proteins are unique to Solanaceae, given that light is often required for the HR (Chandra-Shekara et al., 2006; Negeri et al., 2013), it will be interesting to determine if additional NB-LRR proteins can interfere with the stability of THF1 to promote light-dependent cell death.
MATERIALS AND METHODS Plant MaterialNicotiana benthamiana, N. sylvestris, or tobacco (N. tabacum L. cv Samsun NN) were grown in BM6 (Berger, Sainte-Modeste, Quebec, Canada) soil (22°C/20°C day/night, 16-h d, 60% relative humidity, light intensity 100 μmol m−2 s−1), with fertilization every 7th d using a 22:11:22 (1 g L−1) solution. For Virus-Induced Gene Silencing (VIGS) experiments, plant growth conditions were the same except that a 12 h lighting cycle was used. All experiments were repeated at least three times.
Plasmid ConstructionFor the generation of coiled-coil (CC) domain expression clones, genomic DNA of the appropriate plant/cultivar or cDNAs was used as templates for PCR amplification using KOD high-fidelity DNA polymerase (Novagen, Madison, WI). Sequences of primers are listed in Supplemental Table S1. Amino-acid regions corresponding to CC domains of the studied nucleotide-binding/Leu-rich repeats (NB-LRR) proteins are as follows: NRG1 (amino acids 1–182), ADR1 (1–157), I2 (1–176), N′ (1–182), R3a/I2-5/I2-7/I2-28/I2-31/L1/L2/L3/L4 (1–183), Rx (1–147), Tm-2a (1–164), and Bs2 (1–158). Following amplification, PCR products were purified, A-tailed using Taq DNA polymerase (New England BioLabs, Ipswich, MA), and cloned into pGEM-T vectors (Promega, Madison, WI). After sequencing, inserts were cloned into the XbaI and BamHI sites of a pBIN61 binary vector containing epitope tags for the carboxy-terminal tagging of inserts in frame with the BamHI site (Moffett et al., 2002). Full-length THF1 was cloned as above, using cDNAs derived from healthy N. benthamiana leaves. Expression constructs for RanGAP2, N′, Tobacco Mosaic Virus (ToMV) coat protein, Bs2, AvrBs2, Tm-2a, and ToMV movement protein have been described elsewhere (Sacco et al., 2007; Rairdan et al., 2008; Sekine et al., 2012). For VIGS experiments, a 300-bp fragment corresponding to nucleotide positions 473–772 of the coding sequence of THF1 was amplified by PCR and cloned into the BamHI and SalI sites of pTV:00 (Ratcliff et al., 2001) to create pTV:THF1. For bimolecular fluorescence complementation (BiFC) experiments, binary vectors were constructed by amplifying the expression unit of pBI221, which comprises the Cauliflower mosaic virus 35S promoter, Gateway cassette (Invitrogen, Carlsbad, CA), partial YFP fragments (N- or C-terminal halves), and nopaline synthase gene polyadenylation signal (Kakita et al., 2007). After sequencing, expression cassettes were cloned into the NheI and SalI sites of binary vector pBA (Tomita et al., 2011) to create pBi-GWYN and pBi-GWYC. Gateway recombination was then used to introduce the coding sequence of NbTHF1 and L3CC to generate pBi-THF1-N-terminal half of YFP (YN) and pBi-l3CC-C-terminal half of YFP (YC), respectively.
Transient Expression and Protein AnalysisBinary vectors were transformed by electroporation in Agrobacterium tumefaciens strain C58C1 carrying the virulence plasmid pCH32. Agrobacterium-mediated transient expression (agroinfiltration) was carried out as previously described in Moffett (2011), using a final OD600 of 0.2 for all constructs (except for VIGS and BiFC experiments, see below). To evaluate protein expression, two foliar discs of 10 mm in diameter (approximately 20 mg of tissue) were ground in liquid nitrogen, resuspended in 75 μl 1× Laemmli loading buffer, and centrifuged at 18,000g for 30 min. Supernatants were heated at 95°C, and 25 μl of each sample was used for immunoblotting. Co-immunoprecipitation experiments were performed as previously described in Moffett (2011), using 2 g of fresh tissue that was ground in 4.5 ml ice-cold extraction buffer [GTEN: 10% (v/v) glycerol, 25 mm Tris-HCl, pH 7.5, 1 mm EDTA, 150 mm NaCl, 10 mm DTT, 0.15% (v/v) Nonidet P-40 (NP-40), and 2% (w/v) PVPP (polyvinylpolypyrrolidone) and protease inhibitors]. Epitope-tagged proteins were detected using anti-HA-HRP (1:9000; Roche, Basel Switzerland), anti-FLG-HRP (1:7000; Sigma-Aldrich, St. Louis, MO) and anticMyc-HRP (1:7000; Sigma-Aldrich). Endogenous THF1 was detected using anti-THF1 antibodies (1:9000; Agrisera, Vännäs, Sweden) followed by secondary anti-rabbit IgG-HRP antibodies (1:10,000; BioLegend, San Diego, CA). In certain cases, equal loading of proteins was confirmed with a polyclonal antibody to phosphoenol pyruvate carboxylase (anti-PEPC, 1:10,000 dilution; Rockland Immunochemicals, Limerick, PA). Proteins were visualized by enhanced chemiluminescence (Santa Cruz Biotechnology, Santa Cruz, CA).
Isolation of Protoplasts and BiFCTo perform BiFC experiments, A. tumefaciens strain GV3101 was transformed with either pBi-THF1-YN, pBi-l3CC-YC, or empty pBi-YN/YC vectors. The N. benthamiana leaves were agroinfiltrated with bacteria harboring the indicated combinations of plasmids at a final OD600 of 1.0. Using a standard protocol, protoplasts were isolated from the infiltrated areas of leaves after incubating the plants for 42 h at 24°C and using a 16 h photoperiod. Briefly, the lower epidermis was removed using tweezers and the mesophyll tissue was treated with MMC (0.5 m mannitol, 5 mm MES buffer, pH 5.7, 10 mm CaCl2) containing 1% (w/v) cellulase Onozuka R-10 (Yakult USA, Fountain Valley, CA), 0.1% (w/v) macerozyme R-10 (Yakult USA), and 0.5% (w/v) potassium dextran sulfate for 2 h. Protoplasts were then cultured in W5 solution (154 mm NaCl, 125 mm CaCl2, 5 mm KCl, 2 mm MES buffer, pH 5.7) containing 20 μm clasto-Lactacystin-β-lactone or 1% (v/v) DMSO for 2 h in the dark. The cultured protoplasts were observed under a FLUOVIEW FV10i (Olympus America, Melville, NY) confocal laser-scanning microscope for eYFP (excitation at 480 nm and emission at 527 nm) and chlorophyll (excitation at 635 nm and emission at 660–760 nm). Several Z-axis serial images at 2-μm intervals were obtained and representative images are shown.
Yeast-Two HybridThe CC domains of N′ and R3a were cloned in the bait vector pGBKc (Stellberger et al., 2010). Yeast-two hybrid (Y2H) experiments were performed using the MATCHMAKER GAL4 Two-Hybrid System 3 (BD Biosciences/Clontech, Palo Alto, CA). To prevent false-positive detection, 2.5 mm 3-AT (3-amino-1,2,4-triazole) was included in selective-dropout medium. A cDNA library was constructed using the MATCHMAKER Library Construction and Screening Kit (BD Biosciences/Clontech) according to the manufacturer’s instructions. Leaves from N. benthamiana were collected 1 h after wounding or treatment with a bacterial elicitor (10 μm flg22). Tissues were pooled, ground into powder using liquid nitrogen, and frozen at −80°C until use. Total RNA was isolated using the TRIzol reagent (Life Technologies/Invitrogen/Thermo Fisher Scientific, Waltham, MA), following the manufacturer’s instructions. Using 1 mg total RNA, the mRNA fraction was isolated using the PolyATtract mRNA Isolation System (Promega). A quantity of 350 ng mRNA was then used as a template for PCR amplification of the library. Clones from the library were inserted in the pGADT7-rec plasmid by in vivo recombination in yeast strain Y187.
Screening of the library (approximately 2.9 × 107 clones) was conducted by mating, and interacting candidates were identified by scoring growth on selective-dropout medium lacking Trp, Leu, His, and Ade (-WLHA). Positive colonies were restreaked on quadruple dropout medium supplemented with X-α-GAL and 3-AT. Inserts from positive colonies were PCR-amplified using primers targeting DNA inserts from the pGADT7-rec vector (Supplemental Table S1) and amplicons were sequenced directly. Selected colonies were liquid-cultured and plasmids were isolated using ChargeSwitch Plasmid Yeast Mini Kit (Life Technologies/Invitrogen). For directed Y2H assays, truncated versions of THF1 were cloned in pGADT7 (BD Biosciences/Clontech). Interactions with N′CC or R3aCC were assessed by co-transformation in yeast strain AH109. For all experiments, plates containing selective-dropout medium lacking Trp and Leu (−WL) were used to confirm plasmid co-transformation and viability of yeast cells.
The L3CC domain was cloned into pLexA-C (Dualsystems Biotech, Schlieren, Switzerland) and transformed into yeast strain Y187. Leaves of untreated N. benthamiana plants were used to extract total RNA and a second cDNA library was produced as specified above, except that cDNA clones were inserted in the prey vector pGADT7-rec (BD Biosciences/Clontech) using in vivo recombination in yeast strain NMY51 (Dualsystems Biotech). Library screening was conducted by mating and interacting candidates were identified on selective-dropout medium lacking Trp, Leu, and His (−WLH). For directed Y2H assays, L3CC deletions were introduced in pLexA-C (Dualsystems Biotech), while a DNA fragment encoding the last 54 amino acids of THF1 was introduced in pGAD-HA (Dualsystems Biotech). The resulting plasmids were co-transformed into yeast strain NMY51.
Virus InfectionsTo obtain infectious ToMV:GFP virions, a linearized vector carrying infectious cDNA clones of ToMV comprising the GFP coding sequence was used for in vitro transcription (Hori and Watanabe, 2003). Resulting transcripts were mixed with 1% (w/v) carborundum and rub-inoculated on susceptible four-week old N. benthamiana plants. After 5 d, inoculated leaves were homogenized in PBS buffer and homogenates were centrifuged at 10,000g for 10 min. Clarified sap was aliquoted in 1 ml fractions for long-term storage at −80°C until use. For ToMV:GFP co-expression experiments, 1% (w/v) carborundum was added to ToMV:GFP sap, and leaf sections were rub-inoculated with 10 μl of the abrasive mixture. For virus-free control leaf sections, PBS buffer containing 1% (w/v) carborundum was applied. After 6 d, GFP fluorescence was monitored with a handheld UV lamp (model no. B-100AP; UVP, Upland, CA). TRV infection and VIGS experiments were performed as previously described using A. tumefaciens strain GV3101 and a final OD600 of 0.1 (Ratcliff et al., 2001).
Supplemental DataThe following supplemental materials are available.
Supplemental Figure S1. Most I2-like CCs do not induce cell death in N. tabacum (tobacco).
Supplemental Figure S2. Fusion protein L2CC-eYFP induces cell death.
Supplemental Figure S3. N-terminal fusion of GAL4DBD compromises N′CC function.
Supplemental Figure S4. Preliminary characterization of N′CC as Y2H bait.
Supplemental Figure S5. Conservation of THF1 homologs from various plant species.
Supplemental Figure S6. The CC domain of THF1 is necessary and sufficient for interaction with N′CC.
Supplemental Figure S7. THF1 interacts with the CC domain of R3a.
Supplemental Figure S8. Identification of the THF1-interacting region within the L3CC domain.
Supplemental Figure S9. THF1 compromises cell death induced by N′CC.
Supplemental Figure S10. THF1 does not compromise the HR induced by the activation of Bs2 or Tm-2a.
Supplemental Figure S11. Silencing of THF1 enhances cell death induced by N′CC.
Supplemental Figure S12. Constant darkness does not alter accumulation of N′CC.
Supplemental Figure S13. Conservation of the CC domain from various I2-like proteins.
Supplemental Figure S14. THF1 and ToMV CP do not appear to interact in planta.
Supplemental Table S1. List of primers used in this study.
Supplemental Data
AcknowledgmentsThe authors are grateful to Yuichiro Watanabe for ToMV:GFP; to Julio Vega-Arreguin for identifying tobacco I2-like sequences; and to Geneviève Giroux, Hui Chen, Nari Takahashi, and Reiko Tomita for technical assistance. L.-P.H. is grateful to Dr. Armand Séguin (Natural Resources Canada, Canadian Forest Service) for financial support during late phases of the project.
Glossarybimolecular fluorescence complementation
coiled-coil
co-immunoprecipitation
coronatine
coat protein
empty vector
FLAG epitope tag
hypersensitive response
immunoblotting
immunoprecipitation
Leu-rich repeats
movement protein
nucleotide-binding
photosystem II
reactive oxygen species
Drosophila Toll and human IL-1 receptor
Tobacco Mosaic Virus
Tomato Mosaic Virus
Tobacco Rattle Virus
Virus-Induced Gene Silencing
yeast two-hybrid
C-terminal half of YFP
N-terminal half of YFP
Articles can be viewed without a subscription.
ReferencesThis section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary MaterialsSupplemental Data
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4
