The GYF domain protein PSIG1 dampens the induction of cell death during plant-pathogen interactions

The induction of rapid cell death is an effective strategy for plants to restrict biotrophic and hemi-biotrophic pathogens at the infection site. However, activation of cell death comes at a high cost, as dead cells will no longer be available for defense responses nor general metabolic processes. In addition, necrotrophic pathogens that thrive on dead tissue, take advantage of cell death-triggering mechanisms. Mechanisms by which plants solve this conundrum remain described. Here, we identify PLANT SMY2-TYPE ILE-GYF DOMAIN-CONTAINING PROTEIN 1 (PSIG1) and show that PSIG1 helps to restrict cell death induction during pathogen infection. Inactivation of PSIG1 does not result in spontaneous lesions, and enhanced cell death in psig1 mutants is independent of salicylic acid (SA) biosynthesis or reactive oxygen species (ROS) production. Moreover, PSIG1 interacts with SMG7, which plays a role in nonsense-mediated RNA decay (NMD), and the smg7-4 mutant allele mimics the cell death phenotype of the psig1 mutants. Intriguingly, the psig1 mutants display enhanced susceptibility to the hemi-biotrophic bacterial pathogen. These findings point to the existence and importance of the SA- and ROS-independent cell death constraining mechanism as a part of the plant immune system.


Introduction
Programmed cell death (PCD) has crucial roles in development and immunity in multicellular organisms [1].In plants, hypersensitive response (HR) cell death is one of most studied forms of PCD.The HR is a part of effector-triggered immunity (ETI), the second layer of the plant immune system, and plays an important role in restricting pathogen growth.ETI is primarily effective against biotrophic and hemi-biotrophic pathogens [2], which obtain nutrients from live host cells and actively suppress the first layer of the plant immune system, pathogen-associated molecular pattern (PAMP)-triggered immunity (PTI).By contrast, ETI-associated HR cell death may benefit necrotrophic pathogens, which often secret toxic compounds to kill host cells and obtain nutrients from dead cells [3].Indeed, some necrotrophic pathogens promote virulence by hijacking the plant's HR machinery [3].The contribution of HR cell death to resistance against hemi-biotrophic pathogens, which switch from a biotrophic phase to a necrotrophic one [4], is still under debate [5][6][7].Under this premise, minimizing the induction of cell death, as part of a defense response, would result in an advantage for plants against pathogens that can benefit from dead cells.
Identification of so called 'lesion mimic mutants (LMMs)' that display spontaneous HRlike cell death, has greatly advanced our understanding of HR cell death regulation [8].The phytohormone SA promotes HR cell death induction, and LMM phenotypes are often compromised in SA-deficient mutants background such as the sid2 mutant, which are not able to accumulate SA upon immune activation [9][10][11][12].Several LMMs initiate lesion formation under specific growth conditions and/or upon chemical treatments [1].Lesion formation of the lsd1 mutant can be triggered by shifting plants from short day conditions to long day conditions [13].Upon pathogen inoculation, the lsd1 mutant displays runaway cell death (RCD) phenotype that forms lesions beyond the inoculation site [14].LSD1 encodes a zinc finger protein, and negatively regulate initiation of PCD and RCD, partly via maintenance of ROS homeostasis [11,15,16].The basic region leucine zipper (bZIP) transcription factor, bZIP10, and the type I metacaspase, MC1, interact with LSD1 and regulate the PCD [6,17].HR cell death can be regulated both positively and negatively by ROS [18][19][20][21].Likewise, autophagy can act as both a positive and negative regulator of HR cell death, which was proposed to be dependent on plant age [22][23][24][25][26]. Infection of the avirulent bacterial strain, Pseudomonas syringae pv.tomato DC3000 (Pto) carrying AvrRPM1 or AvrRPS4, triggers the autophagic activity [23].Application of an SA agonist, benzo(1,2,3)thiadiazole-7-carbothioic acid (BTH), induces autophagosome formation [26].The atg5 mutant displays an RCD phenotype that depends on SA accumulation and signaling [26].Similarly, the autophagic component BECLIN1 is required to prevent RCD [27,28], suggesting that autophagy negatively regulates RCD.In contrast, the autophagic components positively regulates HR cell death induction upon Pto AvrRPS4 or the avirulent oomycete Hyaloperonospora arabidopsidis (Hpa) infection [23].
A mechanistic link between PCD and immune system is also suggested by the observation that the activation of PTI suppresses Fumonisin B1 (FB1)-triggered PCD.Fumonisin B1, is a mycotoxin produced by the necrotrophic fungal pathogen Fusarium moniliforme that induces host PCD and promotes fungal virulence [35].This suggesting the existence of a pathway capable of restricting cell death induction [35].Recently, a signaling sector that mediates ETI and is inhibited by PTI, namely ETI-mediating and PTI-inhibited sector (EMPIS), has been identified [36].However, the molecular mechanisms underlying the restriction of cell death remain obscure.
The glycine-tyrosine-phenylalanine (GYF) motif represents the conserved signature of the GYF domain, which was first identified in human CD2-binding protein 2 (CD2BP2), where it was required for binding to the cytoplasmic tail of CD2 [37].CD2BP2 plays a role in CD2-triggered T lymphocyte activation and spliceosomal protein functions [37,38].The GYF domain is highly conserved across eukaryotic species and can be roughly classified into two subfamilies, the CD2BP2-type GYF domain and the suppressor of myosin 2 (SMY2)-type GYF domain [38,39].Structures of the GYF domains have been determined [39,40], and the recognition motifs have been characterized in detail [41].The GYF domain of PSIG1 (At5g42950) recognizes proline-rich sequences as do other GYF domains found in human and yeast proteins [41].Recently, PSIG1 was reported to be indispensable for plantago asiatica mosaic virus infection and designated as Essential for poteXvirus Accumulation 1 (EXA1) [42].Furthermore, MUTANT, SNC1-ENHANCING 11 (MUSE11) was found to be EXA1/PSIG1 [43].EXA1/ MUSE11/PSIG1 was proposed to regulate levels of plant immune receptors via translational repression, and thereby negatively regulate plant immunity [43].Nevertheless, significance of the GYF domain for functions of PSIG1 and other plant GYF domain proteins has not been addressed yet.
Here, we identify PSIG1, a plant-specific protein with the GYF domain, as a key player in the restriction of pathogen-induced PCD.Analyses of psig1 mutants suggested that the PCDrestriction system is crucial for resistance against hemi-biotrophic bacterial pathogens.

PSIG1 is an early PAMP-responsive phosphoprotein and negatively regulates PAMP responses in an SA-independent manner
To explore for novel components in PTI, we performed a differential phosphoproteome analysis and found that the phosphorylation status of the N-terminal region of PSIG1 (At5g42950) was modulated within 10 minutes after treatment of Arabidopsis seedlings with PAMP flg22,

PSIG1 is required for resistance against virulent hemi-biotrophic bacterial pathogen
Because the psig1 mutants displayed enhanced responsiveness to flg22, we hypothesized that this would result in increased pathogen resistance.To our surprise, the psig1 mutants displayed enhanced susceptibility against the virulent hemi-biotrophic bacterial pathogen Pseudomonas syringae pv.tomato DC3000 (Pto) (Fig 3A and 3B and S5A,S5B and S6D Figs).The enhanced susceptible phenotype may reflect stomatal defects, as plants were spray inoculated.To test this possibility, we analyzed stomata density and flg22-induced stomatal closure.However, psig1-1 responses were comparable to that of WT plants (S7 Fig).

Atypical induction of cell death appears in the psig1 mutants
To investigate the role of PSIG1 in plant immunity, we challenged the mutant plants with different types of pathogens.Bacterial growth was unaffected in the psig1 mutants spray inoculated with the avirulent strain Pto carrying AvrRPM1 (Pto AvrRPM1) (Fig 3D).Instead, we observed the development of visible cell death symptoms 4 days after Pto AvrRPM1 inoculation in the psig1 mutants, which were absent in WT plants (Fig 3C).Infected leaves were stained with trypan blue to visualize dead cells, revealing that cell death induction was significantly increased in the psig1 mutants compared to WT plants (S8A and S8B Fig) The Col-8 ecotype is susceptible to obligate biotrophic oomycete Hyaloperonospora arabidopsidis (Hpa) isolate Noco2, meaning that penetrated cells generally do not trigger HR cell death (Figs 3E and 4E).However, infection with the virulent Hpa Noco2, resulted in the induction of cell death in the psig1 mutants (Figs 3E and 4E and S6E Fig).In spite of the cell death,  This observation contrasts to incompatible interactions, where penetrated cells trigger a fast HR cell death response, thereby limiting further pathogen growth [47].Of note, RCD was not observed in the psig1 mutants, and cell death induction was restricted to cells in contact with the Hpa Noco2 hyphae, and therefore, potentially penetrated (Figs 3E and 4E).Our observations that Hpa Noco2 grew in spite of HR-like cell death may be reconciled by the possibility that pathogen growth was faster than the cell death.Together, these results suggest that PSIG1 negatively regulates the induction of cell death in pathogen-targeted cells.

PSIG1 suppresses the induction of ETI-triggered cell death in an SAand ROS-independent manner
We found that the frequency of cell death after inoculation with the avirulent bacteria Pto AvrRPS4 markedly increased in the psig1 mutants compared to WT plants (Fig 4A -4D).It was not clear, however, whether the observed cell death in the psig1 mutants was associated with the activation of AvrRPS4-dependent ETI.AvrRPS4 is recognized by the TIR-NB-LRR-type receptor pair RRS1/RPS4 [48], and EDS1 is required for RRS1/RPS4-triggered ETI [49].Therefore, we generated psig1-1 eds1-2 double mutant and inoculated it with Pto AvrRPS4.As expected, AvrRPS4-triggered cell death was suppressed in the psig1-1 eds1-2 mutant (Fig 4A and 4B).In addition, AvrRPM1-triggered cell death, which does not require EDS1, was unaffected in the psig1-1 eds1-2 mutant (S9 Fig) .These results indicate that the induction of ETItriggered cell death is increased in the psig1 mutants.
Since SA potentiates the HR [9], increased SA levels in the psig1 mutants could result in the enhanced induction of ETI-triggered cell death.To examine this possibility, the psig1-1 sid2-2 mutant was challenged with Pto AvrRPS4 or Pto AvrRPM1.The enhanced cell death phenotype of psig1-1 was retained in the sid2-2 background (Fig 4A and 4B), indicating that PSIG1 negatively regulates the induction of ETI-triggered cell death in an SA-independent manner.
Similarly, enhanced ROS production in response to flg22 treatment in the psig1 mutants occurred in an SA-independent manner (Fig 2F and S5C Fig).ROS accumulation is also known to regulate the induction of cell death in both positive and negative manners [21].Therefore, we crossed psig1-1 with the rbohD mutant, which has a defect in pathogen-induced ROS production [19], and investigated the contribution of ROS accumulation to the psig1-1 phenotypes.There was no detectable flg22-induced ROS production in the psig1-1 rbohD double mutant (Fig 2G  indicating that PSIG1 negatively regulates the induction of ETI-triggered cell death in a ROS-independent manner.Collectively, these results suggest that the induction of cell death by avirulent bacterial pathogen infection is limited to a certain level in WT plants, and PSIG1 is involved in a mechanism required for the restriction of cell death in an SA-and ROS-independent manner. We additionally investigated the contribution of SA and ROS to the Hpa Noco2-induced cell death.Similarly, the induction of cell death observed in the psig1-1 mutant for cells in contact with Hpa Noco2 hyphae was still present in both genetic backgrounds containing sid2-2 or rbohD mutations (Fig 4E).However, the induction of cell death was clearly compromised in the sid2-2 background (Fig 4E ), indicating that elevated SA levels in psig1-1 partly contributed to the Hpa Noco2-induced cell death.Interestingly, we also observed a clear reduction in cell death in the eds1-2 background (Fig 4E).This result may suggest that Hpa Noco2 effectors are weakly recognized by TIR-NB-LRR-type receptor proteins, and the remaining cell death is triggered by alternative NB-LRR-type receptors, or the result of reduced SA level.

PAMP-signaling mutants display the psig1 mutant-like cell death phenotype
As PSIG1 was originally identified as an early PAMP-responsive phosphoprotein, it is possible that PSIG1 is a PAMP-signaling component that regulates the plant immune system.If this is the case, there may be other PAMP-signaling mutants that phenocopy the psig1 mutant cell death phenotype.BAK1 is a co-receptor of PAMP and BR receptors, and the bak1-5 mutant is only impaired in PAMP-signaling [34].BIK1 and PBL1 are highly homologous receptor-like cytoplasmic kinases that directly interact with PAMP receptors, and the double mutant is impaired in PAMP-induced resistance [50].We found that flg22-induced phosphoregulation of PSIG1 was compromised in the bak1-4, bak1-5, and bik1 pbl1 mutants (Fig 5A).In addition, the bak1-4, bak1-5, and bik1 pbl1 mutants displayed the psig1-like cell death phenotype upon Pto AvrRPS4 or Hpa Noco2 inoculation (Fig 5B and 5C and S10 Fig).These results suggest that activation of the PAMP-signaling pathway suppresses the induction of cell death through an elusive mechanism that is independent of the reported BAK1/SERK4-regulated cell death mechanism [30,32,33].

PAMP-induced restriction of avirulence effector injection by bacterial pathogens is not affected in the psig1 mutant
Activation of the PAMP-signaling pathway has been reported to inhibit HR cell death induced by avirulent bacterial pathogens by restricting the ability of type III protein secretion system to inject effectors [51].Therefore, impaired PAMP-signaling may result in enhanced effector injection and thus excess HR cell death.To test whether PSIG1 is involved in the PAMP-induced restriction of effector injection by Pto, we inoculated mutant plants with Pto AvrRPM1.As reported, WT plants displayed inhibition of cell death by the flg22 pretreatment, in contrast to the fls2 mutant (Fig 5D and S11 Fig).As in WT plants, the induction of HR cell death was also inhibited in the psig1-1 mutant by flg22 pretreatment, suggesting that the PAMP-induced restriction of effector injection by Pto is intact in the psig1 mutants.These results suggest that PAMPsignaling pathways suppress effector-triggered HR cell death through at least two different mechanisms at different stages of infection.

PSIG1 is required for PAMP-induced suppression of necrotrophic fungal pathogen-derived mycotoxin-triggered cell death
Next, we investigated whether PSIG1 suppresses cell death induced by different mechanisms involved in plant-pathogen interactions.Necrotrophic pathogens actively induce cell death of host cells by secreting toxins that support infections.Fumonisin B1 (FB1) is a mycotoxin produced by the necrotrophic fungal plant pathogen Fusarium moniliforme.FB1 induces a type of PCD with some similarities to HR cell death [47].Of note, FB1-induced PCD can be suppressed by PAMP pretreatment in Arabidopsis [35].FB1 treatment resulted in leaf chlorosis and cell death, which was confirmed by trypan blue staining, in both psig1-1 and WT plants (Fig 5E).The pronounced induction of cell death in the psig1-1 mutant is most likely the result of elevated SA levels [9].Pretreatment with flg22 less effectively suppressed the cell death in the psig1-1 mutant, irrespective of the enhanced FB1-triggered cell death, as compared to WT plants (Fig 5E).These results indicate that PSIG1 participates in the PAMP signaling-dependent suppression of FB1-induced PCD.

The GYF domain is required for the cell death regulation
To investigate the role of the GYF domain in the function of PSIG1, we complemented the psig1-1 mutant with a construct driven by its native promoter and carrying Y575A mutated version of the GYF domain (Fig 6B).The Y575A mutation was shown to abrogate interaction with proline-rich sequences of interacting proteins [41,52].Even though expression of PSI-G1 Y575A complemented the dwarf phenotype (Fig 6A ), it failed to suppress the enhanced cell death phenotype triggered by Pto AvrRPS4 inoculation (Fig 6C and 6D).These results uncouple the GYF domain function in the regulation of cell death from other functions responsible for the dwarf phenotype.The GYF domain of PSIG1 was shown to recognize the proline-rich sequence proline-proline-glycine-phenylalanine (PPGF) [41].Therefore, proteins that have the PPGF sequence expected to regulate the cell death along with PSIG1.

PSIG1 localizes to P-bodies and interacts with SMG7
To explore the molecular function of PSIG1, we assessed the subcellular localization of PSIG1.The C-terminus of PSIG1 was tagged with GFP, and the labelled protein was transiently expressed in Nicotiana benthamiana leaves.Interestingly, PSIG1-GFP localized to cytoplasmic foci and co-localized with DCP1-mCherry, a processing body (P-body) marker (Fig 7D  Importantly, mutations in SMG7 were previously shown to cause autoimmune phenotypes, including the induction of spontaneous cell death in Arabidopsis [53].We therefore investigated whether PSIG1 interacts with SMG7.GST-tagged PSIG1 and HisMBP-tagged SMG7 recombinant proteins were prepared, and the interaction was assessed by a GST-pull down assay.We found that GST-PSIG1 interacts with HisMBP-SMG7 in vitro (Fig 7B).The C-terminal half of SMG7 (SMG7-C), which contains the PPGF sequence, was required for the interaction, while the N-terminal half of SMG7 (SMG7-N) was dispensable (Fig 7B).Interestingly, PSIG1 possesses the PPGF sequence (Fig 7A ), opening the possibility of an self-inhibitory intramolecular interaction [41].Although the full-length PSIG1 interacted with SMG7 in vitro, their interaction could be fine-tuned through the intramolecular interaction in planta.In spite of interacting in vitro, the PPGF sequence in PSIG1 could nevertheless affect the interaction with SMG7 and complicate interpretation of the results.Therefore, we decided to use the N-terminal region of PSIG1 that contains the GYF domain but not the PPGF sequence for the further analyses.To investigate the significance of the GYF domain in PSIG1 and the PPGF sequence in SMG7, we mutated residues important for the interaction [41].Collectively, these results suggest that PSIG1 directly interacts with SMG7 through its GYF domain and functions within P-bodies.

The smg7-4 mutant allele displays a psig1 mutant-like cell death phenotype
As in the case of the psig1 mutants, two smg7 mutant alleles, smg7-4 and smg7-6, that have T-DNA insertions in the C-terminal region, do not show autoimmune phenotypes and are indistinguishable from WT plants (Fig 7E) [53].The PPGF sequence that mediates interaction with the GYF domain is located downstream of the T-DNA insertion sites of smg7-4 and smg7-6 (Fig 7E).Therefore, if SMG7 plays a role in the regulation of PSIG1-dependent cell death, the smg7-4 allele is expected to display the enhanced cell death phenotype as seen in the psig1 mutants.To test this, we challenged the smg7-4 allele with Pto AvrRPS4 or Pto AvrRPM1 and found that the smg7-4 allele phenocopied the psig1's enhanced cell death (Fig 7F

Discussion
Activation of cell death during the interaction of plants with biotrophic or necrotrophic pathogens, can lead to antagonistic results [4,54].The ability to fine-tune PCD activation during plant-pathogen interactions is critical for plants to effectively fend off pathogens with different infection strategies [3,50].However, the biological significance of the systems regulating cell death activation remains unclear because a general framework and the molecular components of the system are poorly understood.With the initial aim of understanding PAMP-signaling pathways, we identified PSIG1 as an early PAMP-responsive protein by differential phosphoproteome analysis (S4 Table ).Through our approach, we were also able to detected phosphoregulation of MAPK cascade components, SERK2, CDPKs, MPK1, RbohD, and WRKY1 (S4 HisMBP-SMG7, HisMBP-SMG7-N or HisMBP-SMG7-C were incubated with GST or GST-PSIG1 and the conjugates were pulled down with Glutathione-Sepharose beads.HisMBP-SMG7, HisMBP-SMG7-N, or HisMBP-SMG7-C were detected by immunoblotting using anti-His antibody.GST-PSIG1 was stained with Coomassie brilliant blue (CBB).Arrowheads indicate the position of each proteins.c, Tyr-575 of PSIG1 is required for interaction with SMG7 in vitro.HisMBP-SMG7 was incubated with GST, GST-PSIG1 1-606 , GST-PSIG1 1-606 Y575A , or GST-PSIG1 1-606 W570A/Y575A , and the conjugates were pulled down with Glutathione-Sepharose beads.HisMBP-SMG7 and GST-PSIG1 were detected by immunoblotting using anti-His antibody or anti-GST antibody.Table ).By isolating T-DNA insertion mutants of PSIG1, we found that PSIG1 was required for resistance against the hemi-biotrophic bacterial pathogen Pto (Fig 3B).However, detailed analyses with well-described PAMP responses in the psig1 mutants, did not reveal major differences to WT plants.Instead, we discovered that the induction of cell death was accelerated in the psig1 mutants upon inoculation with avirulent bacterial pathogens (Fig 4  The significance of SA and ROS pathways in the regulation of cell death has been well documented.However, our genetic analyses with SA biosynthesis-and ROS production-deficient mutants indicated that PSIG1 regulates the induction of cell death in an SA-and ROS-independent manner (Fig 4A -4D).In this respect, the psig1 mutants differ strikingly from the lsd1 and autophagy mutants that require SA accumulation for their uncontrolled PCD phenotypes [11,26,55], suggesting that PSIG1 controls cell death through an alternative pathway.
Moreover, we addressed whether PSIG1 regulates the induction of cell death in a PAMP signal-dependent manner.Up to now, there has been no experimental system available to evaluate effect of PTI on the induction of HR cell death [35].Therefore, we focused on the inhibitory effect of PTI on PCD induction by the fungal toxin FB1 [35], and reveal that the PTI-induced inhibition was compromised in the psig1 mutant (Fig 5E).Collectively, these results suggest that plants possess a pathway to restrict cell death induction during PTI.A potential reason may be to override cell death activation during an effective PTI response to avoid unnecessary tissue damage and infection by necrotrophic pathogens.It will be important to address how PSIG1 suppresses the cell death triggered by ETI activation and the fungal toxin FB1.To that purpose, however, molecular mechanisms how ETI and FB1 lead to the cell death must be unraveled.
With regards to the enhanced cell death phenotype of the psig1 mutants, they expectedly displayed enhanced resistance against the biotrophic pathogen Hpa Noco2, and enhanced susceptibility to the necrotrophic pathogen Alternaria brassiciola O-264 (Fig 3G).Necrotrophic pathogens could kill host cells in a variety of ways, some of which may not utilize PCD machinery of host cells.Therefore, the psig1 mutants may not necessarily display susceptible phenotype against all necrotrophic pathogens.Again, an intriguing observation was that the psig1 mutants were more susceptible to the hemi-biotrophic pathogen Pto compared to WT plants.These results suggest that the PTI-dependent PCD regulation system may be crucial for coping with hemi-biotrophic pathogens that can switch their infection strategies during colonization.In addition, we found that resistance against the avirulent bacteria strains was not enhanced in the psig1 mutants (Fig 3D and S17 Fig), supporting the idea that the host plants do not benefit much by maximizing the induction of HR cell death.Taken together, these results have allowed us to identify PSIG1 as a unique component in plant immunity that connects PTI and PCD, highlighting the biological significance of the PTI-dependent PCD regulation.Our findings may also offer an explanation why toxin-producing necrotrophic pathogens potentially suppress PTI for their successful infection [56].
The psig1 mutants, especially the psig1-1 and psig1-2 alleles, seem to be sensitive to environmental conditions.Recently, exa1-1/muse11-1/psig1-1 was reported to display auto-immune phenotype, which was characterized by elevated PR1 gene expression and reduced plant size [43].However, these phenotypes were not observed in another study [36].Under our experimental conditions, the psig1-1 allele displays intermediate phenotype based on the PR1 gene expression and plant size.Importantly, the psig1-3 allele displays the cell death phenotype but is less sensitive to growth conditions compared to the psig1-1 and psig1-2 alleles.The difference between these alleles may result from the expression of a C-terminally truncated PSIG1 in  [41].PSIG1 gene family is well conserved in land plants but not in other organisms.No additional domains with known function are predicted on PSIG1, which would help elucidate further molecular function(s).In many cases, the GYF domain proteins modulate signaling pathways through their binding to key regulatory components.Therefore, identification of PSIG1 interactors are expected to facilitate an understanding of the molecular mechanisms by which PSIG1 inhibits PCD.A previous study identified the PPGF sequence containing fragments of RNA binding proteins in a Y2H screen using the GYF domain of PSIG1 as a bait [41].Interestingly, PSIG1 itself has the PPGF sequence (Fig 7A ), and the intramolecular interaction could happen and affect interaction with other PPGF sequence containing proteins [41].In another study, PSIG1 was identified in 5 0 -cap complexes [57].These findings suggest that PSIG1 might be involved in regulating RNA metabolism and/or translation.
Indeed, we found that PSIG1 localizes to P-bodies that are involved in post-transcriptional regulation [58] (Fig 7D).Moreover, we found that PSIG1 interacts with the NMD factor SMG7 through the GYF domain (Fig 7C).In general, SMG7 is functionally conserved among different organisms [53,59,60].However, the PPGF sequence recognized by the GYF domain was only found in the less conserved C-terminal region of plant SMG7 homologs (S13A Fig) .Additionally, the GYF domain of yeast Smy2 was shown to interact with P-body components other than the SMG7 homolog [61].These results suggest that the GYF domain proteins function within P-bodies in general but cooperate with different components in different organisms.
Disruption of the NMD-related genes often results in autoimmune phenotypes [62].In addition, PAMP treatment was shown to modulate NMD efficiency [63].Moreover, the C-terminal region of SMG7 was suggested to play a role in the target degradation step of plant NMD [64].The observations that PSIG1 interacts with the C-terminal region of SMG7 (Fig 7B ) and that the C-terminal truncated smg7-4 allele phenocopies the psig1 cell death phenotype (Fig 7F ) suggest that PSIG1 regulates the induction of cell death by modulating an SMG7-dependent NMD.It will be important to identify PSIG1-and SMG7-dependent NMD targets to understand how NMD contributes to the induction of cell death.

Phylogenetic analysis
Phylogenetic analysis was performed using MEGA6.06software (www.megasoftware.net).Full-length amino acid sequences were used to generate the phylogenetic tree.The bootstrap value was set at 10,000 replicate samples.

Phosphoproteome analysis
Arabidopsis seedlings were grown in liquid MGRL medium with 0.1% (w/v) sucrose [66] at 22˚C under continuous light for 10 days.Phosphoproteome analysis was performed as described previously with minor modifications [67].MS peaks were detected using an inhouse 2DICAL software package [68] that identified all MS/MS spectra using Mascot software, adjusted the retention times of each LC-MS data point utilizing the similarity index of the mass spectrum pattern, and grouped peaks from different samples that were derived from same peptides in the direction of acquiring time.The peak intensities of MS chromatograms were used for quantitative values.Three biological replicates were analyzed for each condition, and the significance of differences was tested.Statistical relevance was determined using a two-tailed Student's t-test (p-value < 0.05).

RNA isolation and qRT-PCR analysis
Total RNA was isolated using the RNeasy Plant Mini Kit (Qiagen, Netherlands), and cDNA was prepared using the ReverTra Ace Reverse Transcription Kit (Toyobo, Japan).Quantitative reverse transcription PCR (qRT-PCR) was performed using the Mx3000P QPCR system (Agilent Technologies, USA) with the Thunderbird SYBR qPCR Mix (Toyobo, Japan).Data were analyzed using an in-house script written in the R language (S1 Information) as described previously [69].

Callose deposition assay
The callose deposition assay was performed as described previously with minor modifications [70].Leaves from 6-week-old plants were syringe infiltrated with 100 nM flg22 or 10 mM MgCl 2 .Leaf discs were removed 20 h post-infiltration using an 8-mm-diameter cork borer and fixed in an acetic acid:ethanol (1:3) solution for several h.Leaves were rehydrated in 70% (v/v) ethanol for 2 h, 50% (v/v) ethanol for 2 h, and overnight in water.Leaves were stained for 1 h in the dark in 150 mM sodium phosphate buffer (pH 9.0) containing 0.05% (w/v) aniline blue.Leaves were mounted in 10% glycerol (v/v) and the localization of fluorescently stained callose was determined by fluorescence microscopy.Callose deposition was quantified by Image-J software (http://imagej.nih.gov/ij/).At least three leaves from three independent plants were used as biological replicates for the analysis.

Pathogen infection assays
Pseudomonas syringae pv.tomato DC3000 (Pto), Pto AvrRPM1 or Pto AvrRPS4 were grown on King's B medium at 28˚C for 2 to 3 days.Six to 7-week-old plants grown in the 8 h light /16 h photoperiod were spray inoculated with bacterial suspensions of 5 x 10 7 to 5 x 10 8 c.f.u.(colony forming unit) ml -1 containing 0.04% silwet-L77 (Bio medical science, Japan).For the infiltration assay, 5 to 6-weeks-old plants grown in the 12 h light /12 h photoperiod were syringe infiltrated with bacterial suspension of 5 x 10 4 c.f.u.ml -1 .For bacterial growth experiments, inoculated plants were maintained at high humidity at 22˚C in the 8 h light /16 h dark or 12 h light /12 h photoperiods.Hyaloperonospora arabidopsidis isolate Noco2 was maintained weekly by transferring spores onto WT Col-8 plants growing at 16˚C in an 8 h light /16 h dark photoperiod.Fourteen-day-old plants were spray inoculated with 5 x 10 4 spores ml -1 of Hpa Noco2.Infected plants were maintained at high humidity at 16˚C in an 8 h light /16 h dark photoperiod.Infection development was scored 6 days after infection by counting sporangiophores on true leaves.Alternaria brassicicola isolate O-264 was maintained on potato dextrose agar medium.The third to fourth leaves of 27-day-old plants were inoculated with 5 μl drops of a spore suspension (5 x 10 5 spores ml -1 in distilled water).Infected plants were maintained at high humidity at 16˚C in an 8 h light /16 h dark photoperiod.Lesion sizes were measured at 6 days after inoculation using Image-J software (http://imagej.nih.gov/ij/).

Trypan blue staining
Trypan blue staining was performed as described previously with minor modifications [71].Leaves were stained in lactophenol-trypan blue solution (10 mg trypan blue dissolved in 80 ml of lactic acid:glycerol:phenol:distilled water:ethanol 1:1:1:1:4, v/v) and cleared in chloral hydrate solution (2.5 g chloral hydrate in 1 ml distilled water).Leaves were mounted in chloral hydrate solution and dead cells were observed using a BX51 microscope (Olympus, Japan).Trypan blue staining area was measured using Photoshop CC software (Adobe, USA).Statistical analysis was performed using the Tukey HSD test (S2 Information) with R software version 3. 2. 3 (https://www.r-project.org/).

Fumonisin B1 assay
The fumonisin B1 assay was performed as described previously with minor modifications [35].Leaves from 6 to 7-week-old plants were syringe infiltrated with 100 nM flg22 or distilled water (mock) 24 h prior to FB1 treatment.Pretreated leaves were then syringe infiltrated with 50 μM FB1 (Sigma-Aldrich, USA) or a mock solution (distilled water containing 1% (v/v) methanol) and plants were kept at 22˚C in a 16 h light /8 h dark photoperiod.Photographs were taken 4 days after FB1 treatment, and leaves were stained with trypan blue to visualize dead cells.

Constructs and transformation
The open reading frame (ORF) of PSIG1 was cloned into pENTR4 dual-selection vector (Thermo Fisher SCIENTIFIC, USA) between SalI and NotI sites using an IN-FUSION HD Cloning Kit (Clontech Laboratories, USA).The 1.5 kb promoter fragment of PSIG1 was cloned into the SalI site of the pENTR4-PSIG1 construct using IN-FUSION HD Cloning Kit.PSIG1pro::PSIG1 stop codon deletion and PSIG1pro::PSIG1 Y570A were generated by PCR-based mutagenesis using the PSIG1pro::PSIG1 construct as a template.PSIG1pro::PSIG1 and PSIG1pro::PSIG1 Y570A were subcloned into binary vector pGWB1 [72], and the stop codon-mutated PSIG1pro::PSIG1 was subcloned into pGWB4 using LR clonase II enzyme mix (Thermo Fisher SCIENTIFIC, USA) [64].The binary vectors were introduced into Agrobacterium tumefaciens C58C1.The psig1-1 mutant was transformed using a floral dip method [73].T3 or T4 homozygous transformants were used in this study.

GST-pull down assay
The purified recombinant PSIG1 and SMG7 proteins were mixed in a binding buffer [50mM Tris-HCl pH7.5, 150 mM NaCl, 1 mM dithiothreitol (DTT), 0.1% (v/v) Triton X-100 and one Complete tablet, EDTA-free per 50 mL (Roche)].The Glutathione-Sepharose B beads were mixed with the protein mixtures and incubated for 2 h at 4˚C.The beads were washed three times with a washing buffer [50mM Tris-HCl pH7.5, 300 mM NaCl, 1 mM dithiothreitol (DTT), 0.1% (v/v) Triton X-100 and one Complete tablet, EDTA-free per 50 mL (Roche)].Proteins were eluted from the beads with 2 x SDS buffer, and were detected by SDS-PAGE or immunoblotting.

Localization analysis
A Nicotiana benthamiana transient expression system was used for subcellular localization analysis.The stop codon-mutated PSIG1 ORF (PSIG1sm) was amplified by PCR and cloned into the pENTR4m between SalI and NotI sites using the IN-FUSION HD Cloning Kit.The PSIG1 and PSIG1sm fragments were subcloned into pGWB6 and pGWB5 binary vectors, respectively, using the LR clonase II enzyme mix.The SMG7 fragment was subcloned into pGWB45 or pGWB661 binary vectors [75].The P-body marker construct pTA7002::DCP1-mCherry was described previously [76].The binary vectors were introduced into A. tumefaciens C58C1 strain carrying pCH32 plasmid.Agrobacterium cultures harboring each construct were incubated overnight at 28˚C and subsequently resuspended in 10 mM MES pH 5.6, 10 mM MgCl 2 buffer with 100 μM acetosyringone at an optical density of 0.3 at 600 nm.Agrobacterium cultures were mixed and syringe infiltrated into N. benthamiana leaves.Agrobacterium carrying pJL3-p19 was co-infiltrated [69].Two days after inoculation, agroinfiltrated leaves were sprayed with 30 μM dexamethasone (Dex).One day after Dex treatment, localization was observed using Leica TCS SP5II (Leica Microsystems, Germany).
the conserved immunogenic epitope of bacterial flagellin (Fig 1A and 1B, S19 Fig and S4 Table).PSIG1 contains the GYF domain that is highly conserved among a variety of proteins from diverse eukaryotic species (Fig 1C and S1A Fig).However, the predicted PSIG1 sequence was found to be plant specific (Fig 1D and S1B Fig).The functions of plants proteins containing the GYF-domain are largely unknown.We therefore isolated three independent T-DNA insertion mutants to study PSIG1's functions (S2 Fig).Treatment with flg22 enhanced oxidative bursts for all three psig1 mutant alleles when compared to wild-type (WT) Col-8 plants (Fig 2A and S3A Fig).Similarly, mitogen-activated protein kinase activation and callose deposition were also enhanced in the psig1-1 and psig1-2 alleles, but were similar to WT for psig1-3 (Fig 2B and 2C and S3B-S3D Fig).Both the psig1-1 and psig1-2 alleles displayed a weak dwarf phenotype mostly absent in psig1-3 (S2A Fig).Dwarf phenotypes are often associated with the inappropriate activation of the SA pathway [44].Indeed, expression of the SA-related marker gene PATHOGENESIS-RELATED 1 (PR1) was upregulated in the psig1 mutants under normal growth conditions (Fig 2E and S4A and S4B Fig).Therefore, to investigate the contribution of SA signaling to the phenotypes, the mutant having the strongest phenotype, psig1-1, was crossed with the SA biosynthesis-deficient sid2-2 mutant [45].Unexpectedly, the enhanced oxidative burst phenotype was not suppressed in the psig1-1 sid2-2 double mutant.Instead, the oxidative burst was delayed (Fig 2F and S5C Fig), and in addition, the dwarf phenotype was unaffected by the sid2-2 introgression (Fig 2D).The expected decrease in SA levels caused by sid2-2 in the double mutant, was confirmed by analyzing PR1 expression (Fig 2E).Phenotypes of psig1-1 were complemented by the expression of PSIG1 driven by its native promoter (S6A-S6C Fig), indicating that PSIG1 negatively regulates PAMP responses and positively plant growth in an SA-independent manner.
Similar levels of cell death induction were observed in the psig1-1 mutant after inoculation with Pto carrying AvrRPS4 (Pto AvrRPS4) that activates effectual ETI in Col-0 without macroscopically visible HR cell death (Fig 4A-4D and S8F and S8G Fig) [46].Cell death was not observed in the psig1 mutants under normal growth conditions without pathogen inoculation (Fig 4A-4D and S8-S11 Figs), in difference to the typical LMMs that develop spontaneous cell death without pathogen attack [8].It suggests that PSIG1 function is dedicated to the restriction of cell death after activation.

Fig 1 .
Fig 1. GYF domain proteins.a, Schematic structure and the phosphorylation site of PSIG1.Ser-39 was found to be the phosphorylation site.The red box indicates the GYF domain.b, Relative abundance of the 'DIQGSDNAIPLpSPQW LLSKPGENK' phosphopeptide upon flg22 treatment.Arabidopsis seedlings were treated with 1 μM flg22 or received a mock treatment (dH 2 O) prior to phosphoproteome analysis.Data are shown as the mean ± SD from three independent experiments.c, Aligned amino acid sequences of the GYF domains from diverse eukaryotic species.Key residues for the GYF domain are delineated as white text on a black background.At, Os, Smo, Phpat, Cre, Kfl, Hs and Sc stand for following species: Arabidopsis thaliana, Oryza sativa, Selaginella moellendorffii, Physcomitrella patens, Chlamydomonas reinhardtii, Klebsormidium flaccidum, Homo sapiens and Saccharomyces cerevisiae, respectively.d, Phylogenetic tree and schematic structures of GYF-domain proteins from diverse eukaryotes.Species abbreviations are defined in Fig 1C.Numbers on the phylogenetic tree indicate the bootstrap values.Red boxes indicate the GYF domain.https://doi.org/10.1371/journal.pgen.1007037.g001

Fig 2 .
Fig 2. PTI responses in the psig1 mutants.a, Flg22-induced ROS production in the psig1 mutants.Data are shown as the mean ± SE. b, Flg22-induced MAPK activation in the psig1 mutants.c, Flg22-induced callose deposition in the psig1 mutants.Callose deposition was quantified with Image J software.Data are shown as the mean ± SE.Statistical groups were determined using the Tukey HSD test.Statistically significant differences are indicated by different letters (p < 0.05).The scale bar represents 200 μm.d, The psig1-1 mutant has a slight dwarf phenotype.Photograph of 6-week-old plants grown under short day conditions.e, PR1 gene expression in 10-day-old seedlings.Data are shown as the mean ± SE.Statistical groups were determined using the Tukey HSD test.Statistically significant differences are indicated by different letters (p < 0.01).f, Flg22-induced ROS production in the psig1-1 sid2-2 mutants.Data are shown as the mean ± SE. g, Flg22-induced ROS production in the psig1-1 rbohD mutants.Data are shown as the mean ± SE. https://doi.org/10.1371/journal.pgen.1007037.g002

Fig 3 .
Fig 3.The psig1 mutant phenotypes against pathogens.a, Photograph of Pto-infected plants.Six-week-old plants were spray inoculated with 1 x 10 8 c.f.u.ml -1 of Pto and photographed 3 days after inoculation.b, The psig1 mutants were more susceptible to Pto.Plants were spray inoculated with Pto, and bacterial growth was determined at 0 and 3 dpi.Data are shown as the mean ± SE.Statistical groups were determined using the Tukey HSD test.Statistically significant differences are indicated by different letters (p < 0.05).c, Photograph of Pto AvrRPM1-infected leaves.Six-week-old plants were spray inoculated with 5 x 10 8 c.f.u.ml -1 of Pto AvrRPM1 and photographed 4 days after inoculation.d, The psig1 mutants did not display reduced or enhanced susceptibility against Pto AvrRPM1.Plants were spray inoculated with Pto AvrRPM1, and bacterial growth was determined at 0 and 4 dpi.Data are shown as the mean ± SE.Statistical groups were determined using the Tukey HSD test.Statistically significant differences are indicated by different letters (p < 0.05).e, Photograph of Hpa Noco2-infected leaves.Plants were inoculated with Hpa Noco2, and true leaves were stained with trypan blue 6 days after inoculation.White arrowheads indicate infection hyphae of Hpa Noco2 and a red arrowhead indicates dead cell.Scale bars in upper panels and lower panels indicate 1 mm and 100 μm, respectively.f, The psig1 mutants were less susceptible to Hpa Noco2.Fourteen-day-old seedlings were inoculated with spores of Hpa Noco2, and the number of sporangiophores on true leaves was scored (0 = 1, 1-10 = 2, 11-20 = 3, >20 = 4) 6 days after inoculation.Bars show the percentage of leaves for each score (n = 25).g, Boxplots represent lesion size (n = 63 to 66).Boxes show upper and lower quartiles of the data, and black lines represent the medians.Statistical groups were determined using the Tukey HSD test.Statistically significant differences are indicated by different letters (p < 0.05).https://doi.org/10.1371/journal.pgen.1007037.g003

Fig 4 .
Fig 4. PSIG1 negatively regulates the induction of cell death during pathogen infection.a and c, Induction of RPS4-triggered cell death was pronounced in the psig1-1 mutant in an SA and ROS-independent manner.Plants were spray inoculated with 1 x 10 8 c.f.u.ml -1 of Pto AvrRPS4, and dead cells were visualized by trypan blue staining 2 days after inoculation.The scale bar represents 200 μm.b and d, Trypan blue stained area.Plants were spray inoculated with 1 x 10 8 c.f.u.ml -1 of Pto AvrRPS4, and dead cells were visualized by trypan blue staining 2 days after inoculation.The stained area was measured using an imaging software.Two to 3 leaves were taken from each of at least 5 individual plants for b.Three leaves were taken from each of 3 individual plants for d.The box plot indicates the area of trypan blue stained cells.Boxes show upper and lower quartiles of the data, and black lines represent the medians.Statistical groups were determined using the Tukey HSD test.Statistically significant differences are indicated by different letters (p < 0.05).e, The psig1-1 mutant induces cell death upon Hpa Noco2 infection.Plants were inoculated with spores of Hpa Noco2, and dead cells on true leaves were visualized by trypan blue staining 5 days after inoculation.White arrowheads indicate infection hyphae of Hpa Noco2 and red arrowheads indicate dead cells.Scale bars in the upper and lower panels indicate 200 μm and 100 μm, respectively.https://doi.org/10.1371/journal.pgen.1007037.g004 and S5D Fig), and the dwarf phenotype of psig1-1 was retained in the rbohD background (Fig 2D).In addition, enhanced ETI-triggered cell death phenotype of psig1-1 was also unaffected by the rbohD introgression (Fig 4C and 4D and S8C Fig),

Fig 5 .
Fig 5. PSIG1 is required for flg22-induced cell death suppression.a, Phosphoregulation of PSIG1 in the PAMP-signaling mutants.Relative abundance of the 'DIQGSDNAIPLpSPQWLLSKPGENK' phosphopeptide upon flg22 treatment.Arabidopsis seedlings were treated with 1 μM flg22 for 10 min or received a mock treatment (dH 2 O) prior to phosphoproteome analysis.Data are shown as the mean ± SD from three independent experiments.b, The bak1-5 and bik1 pbl1 mutants induce cell death upon Hpa Noco2 infection.Plants were inoculated with spores of Hpa Noco2, and dead cells on true leaves were visualized by trypan blue staining 5 days after inoculation.The scale bar represents 200 μm.c, Induction of RPS4-triggered cell death is pronounced in the bak1-5 and bik1 pbl1 mutants.Plants were spray inoculated with 1 x 10 8 c.f.u.ml -1 of Pto AvrRPS4, and dead cells were visualized by trypan blue staining 2 days after inoculation.The scale bar represents 200 μm.d, Flg22-induced restriction of effector injection by Pto is intact in the psig1-1 mutant.Leaves were infiltrated with 100 nM flg22 or received a mock treatment (dH 2 O).Twenty-four h after the pretreatments, plants were spray inoculated with 1 x 10 8 c.f.u.ml -1 of Pto AvrRPM1, and dead cells were visualized by trypan blue staining 24 h after inoculation.The scale bar represents 200 μm.e, Suppression of flg22-induced FB1-triggered cell death is compromised in the psig1-1 mutant.Leaves were infiltrated with FB1 after mock (dH 2 O) or flg22 pretreatments.Control leaves were infiltrated with dH 2 O (mock) after mock (dH 2 O) or flg22 pretreatments.Photographs were taken 4 days after FB1 infiltration.Dead cells were visualized by trypan blue staining.The scale bar represents 200μm.https://doi.org/10.1371/journal.pgen.1007037.g005

Fig 6 .
Fig 6.The GYF domain is required for the cell death but not growth regulation.a, Expression of PSIG1 Y575A complements the psig1-1 growth phenotype.Photograph of 5-week-old plants grown under long day conditions (12 h light/ 12 h dark).b, PSIG1 gene expression in 5-week-old plants.Data are shown as the mean ± SE.Statistical groups were determined using the Tukey HSD test.Statistically significant differences are indicated by different letters (p < 0.05).c, Plants were spray inoculated with 1 x 10 8 c.f. u. ml -1 of Pto AvrRPS4 under long day condition (12 h light/ 12 h dark), and dead cells were visualized by trypan blue staining 1 day after inoculation.The scale bar represents 200 μm.d, Trypan blue stained area.Plants were spray inoculated with 1 x 10 8 c.f.u.ml - 1 of Pto AvrRPS4, and dead cells were visualized by trypan blue staining 1 day after inoculation.The stained area was measured using an imaging software.Two leaves were taken from each of 4 individual plants.The box plot indicates the area of trypan blue stained cells.Boxes show upper and lower quartiles of the data, and black lines represent the medians.Statistical groups were determined using the Tukey HSD test.Statistically significant differences are indicated by different letters (p < 0.05).https://doi.org/10.1371/journal.pgen.1007037.g006 and S12 Fig).The GYF domain of PSIG1 was shown to recognize the proline-rich sequence PPGF [41].Therefore, we searched for PPGF sequences in known plant P-body components and found the PPGF sequence at the C-terminal region of SMG7 (At5g19400) (Fig 7A and S13 Fig).
and S15 Fig).These results suggest that PSIG1 regulates the induction of cell death by interacting via the C-terminal region of SMG7.

Fig 7 .
Fig 7. PSIG1 interacts with SMG7.a, Schematic structure of PSIG1 and SMG7.b, PSIG1 physically interacts with the C-terminus of SMG7 in vitro.HisMBP-SMG7, HisMBP-SMG7-N or HisMBP-SMG7-C were incubated with GST or GST-PSIG1 and the conjugates were pulled down with Glutathione-Sepharose beads.HisMBP-SMG7, HisMBP-SMG7-N, or HisMBP-SMG7-C were detected by immunoblotting using anti-His antibody.GST-PSIG1 was stained with Coomassie brilliant blue (CBB).Arrowheads indicate the position of each proteins.c, Tyr-575 of PSIG1 is required for interaction with SMG7 in vitro.HisMBP-SMG7 was incubated with GST, GST-PSIG11-606 , GST-PSIG1 1-606 Y575A , or GST-PSIG1 1-606 W570A/Y575A , and the conjugates were pulled down with Glutathione-Sepharose beads.HisMBP-SMG7 and GST-PSIG1 were detected by immunoblotting using anti-His antibody or anti-GST antibody.d, PSIG1 and SMG7 co-localize with DCP1, a P-body marker.The images show the CFP signal in cyan, the GFP signal in green, and the mCherry signal in magenta.The merged images indicate the overlay of two signals in yellow or purple and the overlay of three signals in white.White arrowheads indicate overlayed signals.Cellular localization was analyzed at 4 days after inoculation in agroinfiltrated Nicotiana benthamiana.The scale bars represent 10 μm.e, Genomic structure of the SMG7 gene.Black boxes indicate the coding region, and white boxes indicate the non-coding region.The PPGF sequence resides downstream of the T-DNA insertion site.f, Induction of RPS4-triggered cell death is pronounced in the smg7-4 mutant allele.Plants were spray inoculated with 1 x 10 8 c.f.u.ml -1 of Pto AvrRPS4, and dead cells were visualized by trypan blue staining 2 days after inoculation.The scale bar represents 200 μm.
Fig 7. PSIG1 interacts with SMG7.a, Schematic structure of PSIG1 and SMG7.b, PSIG1 physically interacts with the C-terminus of SMG7 in vitro.HisMBP-SMG7, HisMBP-SMG7-N or HisMBP-SMG7-C were incubated with GST or GST-PSIG1 and the conjugates were pulled down with Glutathione-Sepharose beads.HisMBP-SMG7, HisMBP-SMG7-N, or HisMBP-SMG7-C were detected by immunoblotting using anti-His antibody.GST-PSIG1 was stained with Coomassie brilliant blue (CBB).Arrowheads indicate the position of each proteins.c, Tyr-575 of PSIG1 is required for interaction with SMG7 in vitro.HisMBP-SMG7 was incubated with GST, GST-PSIG11-606 , GST-PSIG1 1-606 Y575A , or GST-PSIG1 1-606 W570A/Y575A , and the conjugates were pulled down with Glutathione-Sepharose beads.HisMBP-SMG7 and GST-PSIG1 were detected by immunoblotting using anti-His antibody or anti-GST antibody.d, PSIG1 and SMG7 co-localize with DCP1, a P-body marker.The images show the CFP signal in cyan, the GFP signal in green, and the mCherry signal in magenta.The merged images indicate the overlay of two signals in yellow or purple and the overlay of three signals in white.White arrowheads indicate overlayed signals.Cellular localization was analyzed at 4 days after inoculation in agroinfiltrated Nicotiana benthamiana.The scale bars represent 10 μm.e, Genomic structure of the SMG7 gene.Black boxes indicate the coding region, and white boxes indicate the non-coding region.The PPGF sequence resides downstream of the T-DNA insertion site.f, Induction of RPS4-triggered cell death is pronounced in the smg7-4 mutant allele.Plants were spray inoculated with 1 x 10 8 c.f.u.ml -1 of Pto AvrRPS4, and dead cells were visualized by trypan blue staining 2 days after inoculation.The scale bar represents 200 μm.https://doi.org/10.1371/journal.pgen.1007037.g007 and S8 Fig).This observation led us to hypothesize that PSIG1 has a role in adjusting the induction of ETI-triggered cell death by potentially monitoring the active state of PTI.
psig1-3 allele (S2C Fig).The psig1-3 allele is a useful tool to assess significance of the cell death regulation in disease resistance.Indeed, the psig1-3 allele clearly displayed susceptible phenotype against Pto also when it was inoculated by syringe infiltration, while the psig1-1 and psig1-2 alleles gave inconsistent results (S16 Fig).The psig1-1 and psig1-2 alleles are more likely to be affected by stress responses caused by the syringe infiltration.Interestingly, the psig1-3 allele became susceptible 3 days post inoculation (dpi) but displayed no difference 2 dpi compared to WT plants, which is different from the eds1-2 mutant that displayed susceptible phenotype 2 dpi (S16B Fig).A similar trend was observed by spray inoculation (S5B Fig).In contrast, growth of the avirulent strain Pto AvrRPS4 was unaffected in the psig1 mutants (S17 Fig).These observations, especially the Pto growth kinetics in the psig1-3 allele, support our hypothesis that the cell death regulation becomes crucial when hemi-biotrophic pathogens switch their infection strategies.The GYF domains found in plant proteins are all classified into the SMY2-type GYF domain family (S1 Fig).Although the structure and recognition properties of the GYF domains in plant proteins are highly conserved, the overall structure of GYF domain-containing proteins in plants is rather plant specific (Fig 1D, S1B Fig and S1 and S2 Tables) seeds were kindly provided by Cyril Zipfel (The Sainsbury Laboratory, UK) and Jian-Min Zhou (National Institute of Biological Sciences, Beijing, China).Double mutants were generated by crossing individual mutants and were identified by PCR.Primer sets used in this study are listed in S3 Table.