ABSTRACT
Heterotrimeric G-Proteins are signal transduction complexes comprised of three subunits, Gα, Gβ and Gγ, and are involved in many aspects of plant life. The non-canonical Gα subunit XLG2 mediates PAMP-induced ROS generation and immunity downstream of PRRs. A mutant of the chitin receptor component CERK1, cerk1-4, maintains normal chitin signalling capacity, but shows excessive cell death upon infection with powdery mildews. We identified XLG2 mutants as suppressors of the cerk1-4 phenotype.
We generated stably transformed Arabidopsis lines expressing Venus-XLG2 and numerous mutated variants. These were analysed by confocal microscopy, Western blotting and pathogen infection. We also crossed cerk1-4 with several mutants involved in immunity and analysed their phenotype. Phosphorylation of XLG2 was investigated by quantitative proteomics.
Mutations in XLG2 complex partners AGB1 and AGG1 have a partial cerk1-4 suppressor effect. The cerk1-4 phenotype is independent of NADPH oxidase-generated ROS, BAK1 and SOBIR1, but requires PUB2. XLG2 mediates cerk1-4 cell death at the cell periphery. Integrity of the XLG2 N-terminal domain, but not its phosphorylation, is essential for correct XLG2 localisation and cerk1-4 signalling.
Our results suggest that XLG2 transduces signals from an unknown cell surface receptor that activates an apoplastic ROS-independent cell death pathway in Arabidopsis.
INTRODUCTION
Heterotrimeric G-proteins are signal transducing complexes which are evolutionarily conserved in eukaryotes and play important roles in many different pathways. They consist of three subunits, Gα, Gβ and Gγ, and their basic mechanism of action has been elucidated in metazoans, where they are activated by membrane bound G-protein coupled receptors (GPCRs). The Gα subunit has GTPase activity and in the inactive, GDP-bound form associates with Gβ and Gγ, which form a stable heterodimer. Ligand binding of GPCRs leads to Gα nucleotide exchange, and the GTP-bound Gα dissociates from Gβγ. Subsequently, both Gα and the Gβγ dimer can relay signals to downstream targets. Signal transduction is terminated by the GTPase activity of Gα, which can be enhanced by GTPase Activity Promoting proteins (GAPs), and leads to re-association of the ternary complex (Oldham & Hamm, 2008; Pandey, 2019).
Plants possess canonical Gα, Gβ and Gγ subunits represented by GPA1, AGB1 and the redundantly acting AGG1/AGG2 in Arabidopsis. In addition, they contain plant-specific G-proteins, such as EXTRA LARGE G-PROTEINS (XLG1, 2 and 3 in Arabidopsis), a class of alternative Gα subunits with an N-terminal domain of unknown function. Similar to metazoans, signal transduction by plant heterotrimeric G-proteins is required for many processes, including growth, development, hormone and sugar sensing as well as responses to biotic and abiotic stresses (reviewed by Pandey, 2019; Zhong et al., 2019; Ofoe, 2021; Zhang et al., 2021). However, bona fide GPCRs have not been identified in plants and relatively little is known about the regulatory mechanisms. In contrast to animal heterotrimeric G-proteins, the canonical Arabidopsis Gα subunit GPA1 is inherently GTP-bound in vitro and likely constitutively active in plants. Signalling is thought to be controlled through deactivation via GAPs (Urano & Jones, 2014). For the Extra Large G-protein XLG2, only weak GTP binding / GTPase activity has been observed in vitro (Heo et al., 2012; Lou et al., 2020) and recent studies suggest that in planta, both GPA1 and XLG2 can fulfil many of their roles independently of nucleotide exchange (Maruta et al., 2019; Maruta et al., 2021).
Mounting experimental evidence indicates that plant heterotrimeric G-proteins act downstream of receptor-like kinases (RLKs) and are regulated by phosphorylation (Aranda-Sicilia et al., 2015; Roy Choudhury & Pandey, 2016; Tunc-Ozdemir et al., 2016; Pandey, 2020). A well-studied example is the complex consisting of XLG2, AGB1 and AGG1/AGG2 that transduces signals from pattern recognition receptors (PRRs). Liu et al. (2013) showed that agb1 and agg1 agg2 mutants, but not gpa1 plants, have reduced immune responses to pathogen associated molecular patterns (PAMPs) such as flagellin, EF-Tu and chitin. XLG2 and XLG3 were identified as the Gα subunits that function in these defense pathways (Maruta et al., 2015). XLG2 was then shown to directly interact with the flagellin receptor complex components FLAGELLIN SENSING 2 (FLS2) and BOTRYTIS INDUCED KINASE 1 (BIK1) (Liang et al., 2016). The interaction with G-proteins stabilizes BIK1, presumably through inhibition of E3 ubiquitin ligases that target BIK1 for degradation (Wang et al., 2018). Consequently, xlg2 xlg3 mutants contain less BIK1, leading to a decreased PAMP-induced ROS-burst by apoplastic NADPH oxidases. Upon flagellin treatment, XLG2 is phosphorylated in the N-terminal domain (possibly by BIK1) and dissociates from the receptor. Phosphorylation of XLG2 at four specific sites is required for full flagellin-induced ROS generation, even though this modification has no impact on BIK1 protein stability. Therefore, XLG2 is thought to modulate ROS generation via a second line of action, through direct interaction with the NADPH oxidase RBOHD (Liang et al., 2016). Moreover, XLG2 is known to interact with the E3 ubiquitin ligases PLANT U-BOX 2 (PUB2) and PUB4 (Wang et al., 2017), which are also required for a full PAMP-induced ROS-burst (Desaki et al., 2019; Derkacheva et al., 2020).
Further indications for an involvement of XLG2, AGB1 and AGG1/AGG2 in immune receptor signalling come from a suppressor screen performed with a mutant of the BAK1 INTERACTING RECEPTOR-LIKE KINASE 1 (BIR1). BIR1 is a LRR-RLK that interacts with and probably negatively regulates BAK1 (BRI1-ASSOCIATED RECEPTOR KINASE 1). bir1 knock-out mutants develop spontaneous cell death and show constitutive immune responses that are dependent on BAK1 and another LRR-RLK involved in immune signalling, SUPPRESSOR OF BIR1-1 (SOBIR1) (Gao et al., 2009; Liu et al., 2013; Liu et al., 2016). Mutations in XLG2, AGB1 and AGG1/AGG2, but not GPA1, suppress the phenotype of bir1 knockout plants (Liu et al., 2013; Maruta et al., 2015).
We previously identified and characterized cerk1-4, a mutant with an amino acid exchange in the extracellular domain of the Arabidopsis chitin receptor component CHITIN ELICITOR RECEPTOR KINASE 1 (CERK1). cerk1-4 plants show normal chitin signalling, but exhibit a hyperimmunity phenotype. Upon inoculation with powdery mildews, cerk1-4 mutants display excessive cell death formation and accumulation of the plant defence hormone salicylic acid (SA). Consequently, they do not support macroscopically visible growth of adapted powdery mildew species. The phenotype of cerk1-4 plants is associated with altered receptor processing and is conferred by the N-terminal part of the CERK1-4 protein. Unlike chitin signalling, manifestation of the cerk1-4 phenotype does not require CERK1 kinase activity (Petutschnig et al., 2014).
Here we show that cerk1-4 hyperimmunity is suppressed by mutations in XLG2 and reduced in cerk1-4 agb1 and cerk1-4 agg1 plants. The formation of the cerk1-4 phenotype does not require BAK1 or SOBIR1 and is independent of NADPH-oxidase mediated ROS-generation. In depth cell biological and biochemical analyses of XLG2 mutant variants show that XLG2 fulfils its function in cerk1-4 signalling at the cell periphery. This novel role of XLG2 does not require its GTPase activity, but an intact N-terminal domain. Structural integrity of the N-terminal domain is also important for interaction with AGB1 and proper XLG2 phosphorylation. Finally, we show that the XLG2 interactor PUB2 contributes to cerk1-4 cell death.
MATERIALS AND METHODS
Plant growth conditions and pathogen infection assays
Arabidopsis thaliana (L.) Heynh. plants were grown under short day conditions (8h photoperiod) at 22°C (day) /18°C (night), 65% relative humidity and a light intensity of approximately 150 μmol m−2 s−1. Four-week-old plants were inoculated with either Blumeria graminis (DC.) Speer f.sp. hordei (Bgh) or Erysiphe cruciferarum Opiz ex L. Junell using a settling tower. Pictures and samples were taken at 5-10 dpi for Bgh and 7-14 dpi for E. cruciferarum infections.
cerk1-4 suppressor screen and mapping of les1
Seeds of cerk1-4 (Petutschnig et al., 2014) were mutagenized with ethyl methanesulfonate (EMS) and resulting M2 plants were inoculated with either Bgh or E. cruciferarum. Plants that did not show any macroscopically visible cell death were selected and confirmed as suppressors in the M3 generation. For mapping, a strategy adapted from (Hartwig et al., 2012) was used. The les1 mutant was backcrossed to cerk1-4 and approximately 250 F2 plants were scored for suppression of the powdery mildew-induced cell death phenotype. A pool of 31 clear suppressor plants and a pool of cerk1-4 plants grown in parallel were subjected to 100bp paired end Illumina sequencing. A coverage of approximately 40x was achieved for both samples. Using the CLC genomics workbench software, the reads were mapped to the Col-0 TAIR10 genome and SNPs with a frequency of 20% or higher were called. SNP tables were exported to Microsoft Excel, where SNPs present in cerk1-4 were subtracted from SNPs found in les1. For the remaining SNPs, the frequency was plotted against their chromosomal position (Fig. S1a). The obtained diagrams revealed a frequency peak on the lower arm of chromosome four, which contained a SNP leading to a premature stop codon (Q255*) in XLG2 (EXTRA-LARGE G-PROTEIN 2, AT4G34390). Sequencing of XLG2 in les2 revealed an amino acid exchange (E293K) in the same protein. Both polymorphisms were confirmed as the causative mutations by complementation (Fig. S1c).
Mutant analyses
The following previously characterized mutants were analysed and crossed with cerk1-4: agb1-2 (Ullah et al., 2003), gpa1-3 (SALK 066823) (Jones et al., 2003), agg1-1 agg2-1 (FLAG 197F06, SALK 010956) (Trusov et al., 2007), xlg1-2 (SALK 119657) (Maruta et al., 2015), xlg2-1 (SALK 062645) (Ding et al., 2008), xlg3-4 (SALK 107656) (Yu et al., 2018), bak1-4 (SALK 116202) (Kemmerling et al., 2007), bak1-5 (Schwessinger et al., 2011), sobir1-12 (SALK_050715) (Leslie et al., 2010), sobir1-14 (GABI_643F07) (Barghahn et al., 2021), rbohD, rbohF, rbohD rbohF (Torres et al., 2002), pub2-1 (SALK_086656) (Wang et al., 2017), pub4-3 (SAIL_859_H05) (Wang et al., 2013). xlg1-2, xlg2-1 and xlg3-4 were used to produce double and triple mutants.
Quantitative RT-PCR and ROS-burst assays
Quantitative RT-PCR was performed as described previously (Petutschnig et al., 2014). For ROS burst assays, 4 mm leaf discs were cut out and floated in water in a white 96-well plate for at least 4 h. Then the water was replaced with a solution containing 1μg/ml horse radish peroxidase (Sigma/Merck), 50μM L-012 (Wako Chemicals) and 100nM flg22. Chemiluminescence was immediately recorded using a Tecan Infinite M200 plate reader. Statistical significance was tested with unpaired, two-tailed t-tests.
Protein Work
Protein extraction from Arabidopsis leaves and Western blotting were performed according to previously published protocols (Petutschnig et al., 2010). The antibodies used in Western blots were anti-GFP (Chromotek), anti-FLAG M2 (Sigma), anti-CERK1 (Petutschnig et al., 2010), anti-GPA1 (Agrisera), anti-AGB1 (Agrisera) and anti-XLG2, a custom polyclonal antibody raised against the first 200 amino acids of XLG2 (Agrisera). Western Blot Signal Enhancer (Pierce / Thermo Fisher Scientific) was used for development with GFP and XLG2 antibodies. For GFP pull-downs, standard protein extracts were prepared (Petutschnig et al., 2010) and the target protein was isolated with either Chromotek GFP trap beads (mass spectrometry experiments) or Nanotag GFP selector beads (Co-IP experiments). Protein extracts were mixed with αGFP nanobody beads at a ratio of approximately 0.2-0.5mg total protein / μl bead slurry and incubated at 4°C on a rotating wheel overnight. The next day, the beads were pelleted by gentle centrifugation at 1000 × g or slower. The supernatant was discarded and the beads were washed 2-3 times with extraction buffer (Petutschnig et al., 2010) and 2-3 times with TBS-T. Bound proteins were released by boiling beads in 1x SDS loading dye or beads were directly used for trypsin digestion.
Microscopy
Confocal microscopy was performed on a Leica TCS SP8 Falcon system equipped with HyD SMD detectors, a pulsed white light laser and the LASX 3.5.5 software. Venus fluorescent protein was excited at 514nm and fluorescence was detected between 525-560nm. Since pXLG2-Venus-XLG2 and derivatives are weakly expressed, the accumulation mode was used in some experiments. A fluorescence lifetime gate was set between 0.4-6ns to exclude chloroplast autofluorescence signals from images. Excitation and detection wavelengths for subcellular marker proteins mKate2-Lti6b and mTurquoise2-N7 (Ghareeb et al., 2016) were 594nm/600-641nm and 458nm/462-485nm respectively. Stitched images were recorded and assembled with the LASX Navigator module. Z-stacks were exported as maximum projections and brightness/contrast was adjusted either with Adobe Photoshop CS5 or ImageJ (Fiji distribution).
RESULTS
A suppressor screen identifies XLG2 as an essential component of the cerk1-4 cell death pathway
In order to identify molecular components involved in the establishment of CERK1-4-dependent hyperimmunity, we performed a genetic suppressor screen. For this, we infected an EMS-mutagenized population of cerk1-4 with powdery mildews and looked for plants with wild type-like interaction phenotypes. Two recessive suppressor mutations fully inhibited the development of dysregulated cell death upon inoculation with the incompatible powdery mildew fungus Blumeria graminis f. sp. hordei (Bgh) (Fig. 1a) and the compatible Erysiphe cruciferarum (Fig. 1b). Therefore, they were named lesion suppressor 1 and 2 (les1 and les2). Neither one of the mutations led to any obvious developmental changes in non-inoculated plants (Fig. 1c). Bgh-induced PATHOGENESIS RELATED1 (PR1) expression, which is increased in cerk1-4, was reduced to levels similar to the wild type in the les1 and les2 mutants (Fig. 1d).
We identified the les1 mutation using a mapping-by-sequencing approach (Hartwig et al., 2012). Single nucleotide polymorphism (SNP) linkage analysis located the les1 mutation at the lower arm of chromosome 4 (Fig. S1a), with a mutation introducing a premature stop codon (Q255*) in XLG2 (Fig. 1e) being the most promising candidate. Interestingly, we found that our second suppressor mutant, les2, also harbours a mutation in the XLG2 gene, which results in a glutamic acid to lysine exchange (E293K) within a highly conserved region of the XLG2 N-terminal domain (Fig. 1e). Consequently, we transformed both mutants with a construct containing the wild type XLG2 promoter and coding region, and found that the cerk1-4 cell death response to powdery mildew infection was restored in les1 as well as les2 (Fig. S1b and c). Together, these complementation analyses suggest that cerk1-4-associated hyperimmunity requires functional XLG2. Since both suppressor mutations compromise XLG2, we renamed the suppressor loci in les1 and les2 to xlg2-2 and xlg2-3, respectively.
Mutations in Gα and Gβ subunits AGB1 and AGG1 partially suppress the cerk1-4 phenotype\
To investigate the role of other Arabidopsis heterotrimeric G-proteins in cerk1-4 cell death control, we crossed cerk1-4 with mutants of canonical Gα, Gβ and Gγ subunits as well as all three XLGs. The resulting double mutants were analysed for their reaction to Bgh. As expected, the xlg2-1 T-DNA insertion fully suppressed the cerk1-4 cell death response and restored expression of the SA marker PR1 to wild type levels (Fig. 2a,b). In contrast, mutations in GPA1, XLG1 and XLG3 did not suppress cerk1-4. Notably, both agb1-2 cerk1-4 and agg1-1 cerk1-4 double mutants showed an intermediate phenotype concerning cell death formation as well as PR1 expression (Fig. 2a,b). In agg1-1 cerk1-4, this might be due to functional redundancy between AGG1 and AGG2, since generation of a cerk1-4 agg1-1 agg2-1 triple mutant failed because of genetic linkage between CERK1 (At3g21630) and AGG2 (At3g22942). AGB1, however, is a single copy gene in Arabidopsis. Thus, we performed Western blotting with specific antibodies to test if CERK1 or XLG2 abundance is reduced in the G-protein mutants. CERK1 levels in these mutants were comparable to wild type (Fig. 2c), but interestingly, XLG2 had a slightly lower apparent mass in agb1 and agg1-1 agg2-1 (Fig. 2c). These findings suggest that XLG2 might require protein modifications for its function that are (partially) dependent on AGB1 and AGG1/2. GPA1 and AGB1 Western blots were included in the experiment as controls and revealed drastically reduced AGB1 abundance in agg1-1 agg2-1, probably because βγ-dimer formation is a prerequisite for AGB1 stability (Fig. 2c).
Cell death signalling in cerk1-4 is not mediated by BAK1 or SOBIR1-containing receptor complexes
Hyperimmunity phenotypes of the cerk1-4 mutant are associated with the extracellular part of the CERK1-4 protein (Petutschnig et al., 2014) and could potentially be perceived by a yet unknown RLK that signals through XLG2. Many of the RLKs and RLPs characterized to date use BAK1 as a co-receptor and/or SOBIR1 as an adaptor kinase (Escocard de Azevedo Manhaes et al., 2021). This encouraged us to test if the cerk1-4 cell death phenotype is dependent on BAK1 and SOBIR1. Phenotypic analysis of powdery mildew-infected cerk1-4 bak1 and cerk1-4 sobir1 double mutants clearly demonstrated that BAK1 and SOBIR1 are dispensable (Fig. 2d,e). PR1 expression appeared to be even higher in cerk1-4 bak1 and cerk1-4 sobir1 compared to cerk1-4 plants (Fig. 2e).
The role of XLG2 in cerk1-4 signalling is independent of its function in PAMP-induced ROS generation
XLG2 is required for full flagellin-induced ROS generation. It associates with the flagellin receptor complex and regulates NADPH oxidase activity through multiple modes of action (Liang et al., 2016; Liang et al., 2018; Wang et al., 2018). Similar to previous reports (Maruta et al., 2015), we observed a moderate reduction of flg22-induced ROS in the xlg2-1 mutant, but robustly reduced ROS generation in xlg2 xlg3 double and xlg1 xlg2 xlg3 triple mutants (Fig. 3a). This redundant action of XLG proteins argues against a role for ROS generation in cerk1-4 hyperimmunity, which can be suppressed by mutations in XLG2 alone. For further analysis, we made use of a mutated XLG2 variant lacking four in vivo phosphorylation sites (XLG2S141A, S148A, S150A, S151A, further referred to as XLG24S-A) required for full flg22-induced ROS burst (Liang et al., 2016). We introduced XLG24S-A-FLAG into the xlg2-2 cerk1-4 double mutant. Wild type XLG2-FLAG and the phosphomimic version XLG24S-D-FLAG (Liang et al., 2016) were included as controls. All three XLG2-FLAG versions were similarly able to re-establish the cerk1-4 phenotype, indicating that the XLG2 phosphorylation sites governing flg22-dependent ROS regulation do not play a role here (Fig. 3b,c, Table S2). To corroborate these findings, we crossed cerk1-4 with the NADPH oxidase mutants rbohD and rbohF and the respective double mutant (Torres et al., 2002). Neither loss of RbohD or RbohF, nor the combined loss of both, suppressed the cerk1-4 cell death phenotype upon powdery mildew infection (Fig. 3d,e). Taken together, these data indicate that XLG2 exerts a function in cerk1-4 hyperimmunity that is independent of NADPH oxidase-generated ROS.
Cell periphery localisation of XLG2 is required for cerk1-4 cell death dysregulation
When transiently expressed in N. benthamiana leaves, XLG2 localises to the cell periphery as well as the nucleus (Chakravorty et al., 2015; Maruta et al., 2015). We could confirm this finding with Venus-tagged XLG2 (Fig. S2a) and observed the same localisation upon particle bombardment in Arabidopsis epidermal cells (Fig. S2b). Interestingly, the non-functional XLG2E293K variant that we identified in the les2/xlg2-3 mutant almost exclusively localised to the nucleus in both transient systems (Fig. S2a,b). To study the subcellular localisation and its importance for XLG2 function in more detail, we analysed Venus-XLG2 and Venus-XLG2E293K expressed under the control of the XLG2 promoter in stably transformed xlg2-2 and cerk1-4 xlg2-2 plants. In these, Venus-XLG2 localised predominantly to the cell periphery and Venus-labelled nuclei were rarely found (Fig. 4a, Table S3, S4), suggesting that nuclear accumulation of XLG2 observed in transient systems may be a stress response caused by Agrobacterium infiltration or particle bombardment. Again, XLG2E293K showed almost exclusive localisation to nuclei (Fig. 4a, Table S3, S4). Localisation patterns were very similar in the CERK1 and cerk1-4 background. Since dysregulated cell death responses of cerk1-4 plants become apparent upon infection with powdery mildews, we investigated the subcellular behaviour of Venus-XLG2 and XLG2E293K after inoculation with E. cruciferarum. Fungal infection caused increased accumulation of Venus-XLG2 in the attacked epidermal cells, while distant cells were not affected. Moreover, nuclear Venus-XLG2 signals could be detected at such interaction sites (Fig. 4b, Fig. S2c). XLG2E293K also showed increased abundance in cells challenged with E. cruciferarum, but retained its predominantly nuclear localisation (Fig. 4b, Fig. S2c). Notably, Venus-XLG2 was able to re-establish the cerk1-4 phenotype in xlg2-2 cerk1-4 plants, whereas the mutant variant Venus-XLG2E293K was not (Fig. 4c, Table S4). Western blotting was employed to confirm accumulation of full-length fusion proteins in the transgenic lines (Fig. 4d). These experiments revealed that Venus-XLG2E293K migrated faster in SDS-PAGE than the wild type fusion protein, which will be discussed in more detail further below.
Since the non-functional mutant variant XLG2E293K accumulated predominantly in the nucleus, we speculated that the cell periphery localisation of wild type XLG2 might be essential for its role in cerk1-4 cell death signalling. To address this question, we attached nuclear localisation and export signals (NLS and NES) (Garcia et al., 2010) to the C-terminus of Venus-XLG2. Non-functional sequences (nls and nes) were included as controls. Venus-XLG2-NLS showed nuclear localisation without pathogen challenge, but still had considerable signals at the cell periphery. Importantly, Venus-XLG2-NES was found only at the cell periphery in unchallenged tissues and did not show the typical stained nuclei in powdery mildew infected cells (Fig. 5a). All variants accumulated to similar protein levels and were able to complement xlg2-2 cerk1-4 (Fig. 5b,c, Table S4). The data suggest that indeed the periphery-localised fraction of the XLG2 protein pool promotes cerk1-4 dependent cell death and that nuclear localisation is dispensable for this process.
Intrinsic features of the N-terminal domain control XLG2 subcellular localisation and cerk1-4 cell death signalling
In order to analyse the regulation of XLG2 subcellular localisation, we mutated an internal NLS (nls426) and NES (nes387) within the XLG2 N-terminal domain (Fig. 1e) based on a study by Chakravorty et al. (2015). Stable transgenic expression from the XLG2 promoter showed that Venus-XLG2nls426 localised to the cell periphery, had wild-type like migration properties in Western blots and could complement xlg2-2 cerk1-4. In contrast, Venus-XLG2nes387 was exclusively found in the nucleus, migrated faster in SDS-PAGE and was non-functional in cerk1-4 cell death signalling (Fig. 5a,b,d,e, Table S4). Taken together, the data corroborated that XLG2’s specific role in cerk1-4 mutants is fulfilled at the cell periphery, and indicated that subcellular localisation of XLG2 is determined by sequences within the N-terminal domain of the protein. N-terminal domains of XLG proteins contain highly conserved cysteine rich regions with four invariant CxxC motifs (Fig. 1e), which are possible zinc finger domains (Lee & Assmann, 1999). Mutation of one (C214/217A and C254/257A) or two (C229/232/237/240A) CxxC motifs (Fig.1e) resulted in non-functional Venus-XLG2 fusion proteins that resembled Venus-XLG2E293K and XLG2nes387 in confocal analyses and Western blots (Fig. S3a-d, Table S4). Thus, correct localisation of XLG2 also requires structural integrity of the cysteine rich region.
XLG proteins harbour a Gα domain and weak GTP binding and GTPase activity have been demonstrated for XLG2 in vitro (Heo et al., 2012; Lou et al., 2020). To investigate if XLG2 GTPase activity plays any role in formation of the cerk1-4 cell death phenotype, we mutated its GTP binding site (T476N) as described previously (Heo et al., 2012; Maruta et al., 2021) (Fig. 1e). In addition, we introduced a mutation that could potentially affect GTPase activity (R673L), since a mutation in this position renders GPA1 inactive (Okamoto et al., 2001; Ullah et al., 2003)(Fig. 1e). Subcellular localisation and running behaviour in Western blots were unaffected in XLG2T476N as well as XLG2R673L and both mutant proteins were able to complement xlg2-2 cerk1-4 (Fig. S4, Table S4). This indicates that GTP-binding and hydrolysis are irrelevant for XLG2 positioning and activity in cerk1-4 cell death signalling.
Gβγ modulates protein modifications and subcellular localisation of XLG2
Notably, the SDS-PAGE mobility shift of Venus-XLG2E293K (Fig. 4d), XLG2nes387 (Fig. 5d,e) and the cysteine rich region mutants (Fig. S3c,d) resembled the behaviour of wild type XLG2 in agb1 and agg1 agg2 mutants (Fig. 2c). However, direct comparison of xlg2-3 (containing untagged XLG2E293K), agb1 and agg1 agg2 showed that the XLG2 band shift in xlg2-3 is more pronounced than in the Gβγ mutants (Fig. 6a). We speculated that XLG2 might carry post-translational modifications, which are moderately reduced in agb1 and agg1 agg2, but more strongly affected or missing in XLG2E293K. To investigate this hypothesis, we transformed agb1-2 mutants with pXLG2-Venus-XLG2 (WT) as well as pXLG2-Venus-XLG2E293K (Table S5). Western blots showed that Venus-XLG2 migrated at a lower apparent mass in the AGB1-deficient background (Fig. 6b). Venus-XLG2E293K however showed no such mobility shift (Fig. 6c), suggesting that the same type of modification is affected in agb1-2 and the XLG2E293K mutant. In agreement with this notion, Venus-XLG2 localised to both, nucleus and cell periphery in unchallenged agb1-2 plants, while Venus-XLG2E293K localisation was not affected (Fig. 6d). XLG2 is known to interact with the Gβγ dimer (Zhu et al., 2009; Chakravorty et al., 2015; Maruta et al., 2015). However, this interaction capacity was strongly impaired (with and without PAMP treatment) in the XLG2E293K variant (Fig. 6e). These data favour a model where direct interaction of XLG2 with Gβγ mediates full post-translational XLG2 modification and contributes to proper XLG2 subcellular localisation.
Phosphorylation in the N-terminal region of XLG2 is not essential for cerk1-4 cell death signalling
Since XLG2 is heavily phosphorylated in the N-terminal domain (Liang et al., 2016; Chakravorty & Assmann, 2018) we purified Venus-XLG2 and Venus-XLG2E293K from transgenic Arabidopsis plants via the fluorescence tag and performed phosphatase assays. Upon de-phosphorylation, both XLG2 versions showed the same running behaviour in Western blot analyses (Fig. 7a), indicating that Venus-XLG2E293K indeed has a phosphorylation defect. For further analysis, we carried out phosphoproteomics studies on the two fusion proteins, each isolated from three independent Arabidopsis lines. Phosphopeptides were clustered into 5 non-overlapping regions (Fig. 1e, Fig. 7b), and subjected to targeted selected ion monitoring (tSIM) to quantify phosphorylation levels. These analyses revealed that XLG2E293K showed drastically reduced phosphorylation near the N-terminus (aa 1-79, Fig. 7b). Interestingly, phosphorylation of sites reported to regulate XLG2 function in PAMP-induced ROS generation (aa 141-156) (Liang et al., 2016) was increased in XLG2E293K, while more distant sites (aa 184-194) were not affected (Fig. 7b). We had already shown that phosphorylation within the aa 141-156 region does not play a role in cerk1-4 signalling (Fig. 3b,c). In order to assess the contribution of XLG2 N-terminal phosphorylation, we mutated all 27 serine and threonine residues in the aa 1-79 segment to alanines (XLG227S/T-A). XLG2 has a serine rich stretch (SGS7 at amino acid positions 124-130, Fig. 1e), which is not amenable to mass spectrometry analysis, but may still be important for XLG2 function. This was also mutated (XLG28S-A) and both constructs were transformed into xlg2-2 cerk1-4. The expressed fusion proteins showed reduced apparent masses compared to the wild type Venus-XLG2 version (Fig. 7c). Venus-XLG227S/T-A displayed a very similar motility in SDS-PAGE to Venus-XLG2E293K (Fig. 7d). However, both Venus-XLG28S-A and Venus-XLG227S/T-A showed a cell periphery localisation comparable to the wild type version (Fig. 7e) and complemented the xlg2-2 mutation (Fig. 7f). These data suggest that the mobility shift of XLG2E293K is due to reduced phosphorylation, but this is not the cause of its altered localisation and inability to signal in the cerk1-4 pathway.
XLG2 functions in cerk1-4 cell death signalling together with the E3 ubiquitin ligase PUB2
All three Arabidopsis XLGs were reported to interact with the E3 ligases PLANT U-BOX 2 (PUB2) and PUB4. Moreover, xlg1 xlg2 xlg3 triple and pub2 pub4 double mutants share similar developmental and hormone response defects (Wang et al., 2017). Indeed, in our phosphoproteomics analyses, PUB4 and PUB2 were robustly detected as interactors of Venus-XLG2 (WT) as well as Venus-XLG2E293K (Fig. 8a). Immunoblot analysis of pub4 and pub2 mutants with a XLG2-specific antibody did not reveal any obvious size differences of XLG2, suggesting that PUB4 and PUB2 do not mediate modification of XLG2 (Fig. 8b). To investigate if PUB4 and PUB2 play a role in the formation of the cerk1-4 cell death phenotype, the respective mutants were crossed with cerk1-4 and inoculated with Bgh. These experiments showed clearly reduced cell death formation in pub2-1 cerk1-4 compared to cerk1-4 alone (Fig. 8c). Similarly, Bgh-induced PR1 accumulation was lower in pub2-1 cerk1-4 than in cerk1-4 (Fig. 8d). The pub4-3 mutation however was not able to suppress the cerk1-4 phenotype (Fig. 8c,d), suggesting that PUB2 is the predominant interactor in this process.
DISCUSSION
In this study, we show that loss of function mutations in the heterotrimeric G-protein α-subunit XLG2 suppress the cerk1-4 phenotype. cerk1-4 plants harbour a single amino acid exchange in the extracellular domain of CERK1 and display excessive cell death and SA accumulation after infection with powdery mildews. Previous analyses showed that in contrast to chitin signalling, this phenotype is independent of CERK1 kinase activity, but correlates with an altered CERK1 shedding status (Petutschnig et al., 2014). This suggests a novel pathway, where CERK1-4 generates a cell death signal, but is not itself the signal transducing kinase. Identification of xlg2 mutations as suppressors confirms the idea that cerk1-4-associated cell death involves the generation of a specific molecular signal and active signal transduction by the plant.
In transient expression systems, XLG2 localises to the nucleus, cytosol and plasma membrane (Chakravorty et al., 2015; Maruta et al., 2015), while XLG2 expressed from its native promoter is found predominantly at the cell periphery in stable transgenic Arabidopsis plants. This implies that nuclear localisation of XLG2 reflects a stress response, triggered for example by Agrobacterium infiltration. Indeed, when Arabidopsis plants were inoculated with spores of the powdery mildew E. cruciferarum, (which elicits the runaway cell death phenotype in cerk1-4), attacked cells showed nuclear XLG2 signals along with overall increased XLG2 levels. From the cerk1-4 suppressor screen, we isolated a non-functional XLG2 variant with an amino acid exchange (E293K) in the N-terminal domain, which showed almost exclusive nuclear localisation. To investigate the relationship between XLG2 subcellular localisation and function in the cerk1-4 pathway, we generated XLG2 versions with altered subcellular localisation by appending additional NES and NLS signals or mutating endogenous NES and NLS sequences (XLG2nes387 and XLG2nls426, respectively). Experiments with these variants revealed that XLG2 localisation is determined by the N-terminal domain and that XLG2 performs its role in cerk1-4-associated cell death signalling at the cell periphery. These results suggest that a CERK1-4-derived signal is perceived by a PM-resident sensor, as outlined further below. The role of XLG2 in the nucleus upon fungal infection is currently unclear. (Heo et al., 2012) reported that XLG2 interacts with the DNA binding protein RELATED TO VERNALISATION 1 (RTV1) in the nucleus. RTV1 promotes flowering (Heo et al., 2012), but there is no evidence for any involvement in plant immunity to date.
The N-terminal domain of XLG2 contains a highly conserved cysteine-rich region with four invariant CxxC motifs. It is reminiscent of a zinc finger domain and therefore has been speculated to possess DNA binding activity (Lee & Assmann, 1999; Ding et al., 2008; Heo et al., 2012). If the cysteine rich region specifically mediated nuclear functions of XLG2 such as DNA binding, mutation of this segment should not affect cerk1-4 signalling. However, disruption of CxxC motifs resulted in non-functional XLG2 proteins with nuclear localisation. These resembled XLG2E293K and XLG2nls426, which carry mutations within the N-terminal domain, but not the cysteine rich region. Taken together the results suggest that the CxxC motifs, as well as other conserved regions, are important for the integrity of the entire N-terminal domain, which is in turn essential for XLG2 function.
XLG2 has weak GTP binding / GTPase activity in vitro (Heo et al., 2012; Lou et al., 2020) but it is currently not entirely clear whether this is relevant in planta. (Liang et al., 2018) reported that XLG2 variants with deleted guanine nucleotide-binding motifs display stabilized interaction with AGB1. Analysis of xlg2 xlg3 plants expressing the XLG2 deletion variants suggested that GTP binding contributed to a full flagellin-induced ROS burst and resistance to virulent Pseudomonas bacteria (Liang et al., 2018). On the other hand, a recent study showed that an XLG2 variant which is unable to bind GTP (XLG2T476N) had normal interaction with Gβγ and was able to complement PAMP-induced ROS generation and resistance to Pseudomonas in xlg2 xlg3 plants (Maruta et al., 2021). Our data indicate that the nucleotide-depleted XLG2T476N (Heo et al., 2012; Maruta et al., 2021) and the potentially GTPase-dead variant XLG2R673L are functional in cerk1-4 signalling. These findings support the idea that XLG2 can fulfil at least some of its functions independently of nucleotide exchange.
XLG2 physically interacts with AGB1 and AGG1/AGG2 at the plasma membrane (Chakravorty et al., 2015; Maruta et al., 2015) and the resulting heterotrimeric complex signals not only downstream of the flagellin receptor (Liang et al., 2016; Liang et al., 2018; Wang et al., 2018), but also mediates constitutive immune responses in the bir1 mutant (Liu et al., 2013; Maruta et al., 2015). This led us to test if AGB1 and AGG1 contribute to powdery mildew-induced excessive cell death in cerk1-4. We found that cerk1-4 agb1-2 and cerk1-4 agg1-1 had a partially suppressed cell death phenotype with weaker, but not abolished macroscopic lesions. In the case of cerk1-4 agg1-1, this could be due to an intact AGG2 protein, which we could not address in cerk1-4 agg1-1 agg2-1 mutants because of genetic linkage between AGG2 and CERK1. In contrast, AGB1 does not have any functionally redundant homologs in Arabidopsis. Consequently, the results indicate that AGB1 enhances cerk1-4 cell death signalling, but is not essential for the process. An explanation for this outcome could be that AGB1 itself does not relay any signals, but affects the status of XLG2. Indeed, Western blot analyses suggested altered phosphorylation of XLG2 in the agb1-2 and agg1 agg2 background. However, since several different multiple phosphorylation mutants of XLG2 still function normally in cerk1-4 signalling, it seems most likely that the changed subcellular localisation of XLG2 in agb1-2 is the reason for the partial cell death suppression. XLG2 is shifted towards the nucleus in agb1-2 plants, suggesting that interaction with the Gβγ dimer helps tether XLG2 to the plasma membrane. As overexpression of Gβγ can sequester XLG3 at the plasma membrane (Chakravorty et al., 2015), this seems to be a general mechanism regulating XLG localisation. If Gβγ is absent, less XLG2 might be available for signalling from membrane-associated receptors.
Knockout or mutation of immunity genes can lead to constitutive defense responses and one proposed explanation is guarding by nucleotide-binding leucine-rich repeat (NLR) proteins (Roux et al., 2011; Schwessinger et al., 2011). This hypothesis has recently gained new momentum by the discovery that bak1-4-associated autoimmunity is dependent on helper NLRs (Wu et al., 2020). However, the cerk1-4 cell death phenotype is conferred by the N-terminal part of the CERK1-4 protein and the kinase domain is not required (Petutschnig et al., 2014). This makes an intracellular perception system an unlikely scenario. It is conceivable that CERK1-4 generates a cell death signal in the apoplast, which could be sensed by a RLK or RLP-type receptor that interacts with the XLG2-AGB1-AGG1/AGG2 heterotrimeric G-protein complex. Many receptors characterized to date use BAK1 as a co-receptor and/or SOBIR1 as an adaptor kinase (Escocard de Azevedo Manhaes et al., 2021). Analysis of cerk1-4 double mutants with bak1 and sobir1 showed that the cerk1-4 signalling pathway does not rely on these two RLKs. Thus, the signal perception machinery triggering excessive immune responses in cerk1-4 and bir1 mutants (Liu et al., 2013; Liu et al., 2016) must be different. Interestingly, cerk1-4 bak1 and cerk1-4 sobir1 double mutants even had increased PR1 expression after powdery mildew infection, suggesting that BAK1 and/or SOBIR1-containing immune complexes compete for shared downstream signalling components (such as XLG2) with the receptor system that senses CERK1-4. Competitive binding of different RLKs to XLG2 was recently reported (Maruta et al., 2021).
XLG2 has been shown in multiple studies to facilitate PAMP-induced ROS bursts together with XLG3 (Maruta et al., 2015; Liang et al., 2016; Liang et al., 2018; Maruta et al., 2021). Therefore, we investigated if the role of XLG2 in ROS control is important for the formation of the cerk1-4 phenotype. We found that XLG2 mediates flagellin-triggered ROS production redundantly with not only XLG3, but also XLG1. Since the cerk1-4 phenotype is suppressed by xlg2 alone, this argued against an involvement of ROS in cerk1-4 cell death formation. Indeed, XLG2 variants with mutated phosphorylation sites that are compromised in ROS regulation (Liang et al., 2016) were fully functional in the cerk1-4 pathway. These findings were corroborated by the fact that rbohD and rbohF mutants were unable to suppress the cerk1-4 phenotype.
XLG2 in agb1-2, the XLG2E293K variant and all other non-functional XLG2 versions generated in this study showed faster migration in SDS-PAGE than the wild type version of the protein. Phosphatase assays indicated that the difference in apparent mass was caused by phosphorylation, therefore we quantitatively analysed phosphorylation patterns on XLG2 and XLG2E293K. Interestingly, sites required for proper ROS induction (Liang et al., 2016) were more phosphorylated in XLG2E293K than the wild type. Liang et al. (2016) found S141, S148, S150 and S151 to be phosphorylated in flagellin-challenged cells. Therefore, it is worth noting that we found phosphorylation in this region in the absence of any PAMP treatment. Residues closer to the N-terminus (S/T 13-79) were significantly less phosphorylated in XLG2E293K than wild type XLG2, therefore we generated corresponding XLG2 mutation variants. Analysis of Arabidopsis lines stably expressing these constructs confirmed that the lower apparent mass of XLG2E293K is caused by reduced phosphorylation, but also showed that this is not the reason why XLG2E293K is non-functional. Presumably, XLG2 is not regulated by phosphorylation in the cerk1-4 pathway. XLG2E293K is unable to interact with AGB1, which can partially explain its lack of function, but it seems very likely that the E293K mutation also disrupts interaction with other proteins.
In our mass spectrometry analyses, we robustly co-purified PUB2 and PUB4 together with XLG2. Both ubiquitin E3 ligases were previously reported to interact with XLGs and function in cytokinin signalling and floral development (Wang et al., 2017). We isolated PUB2 and PUB4 with wild type XLG2 as well as the E293K variant. This suggests that they interact with XLG2 at the cell periphery and the nucleus and that the interaction may not be dependent on an intact N-terminal domain. Western blot analysis of XLG2 in pub2 and pub4 mutants did not show any apparent mass differences to the wild type background. Thus, these E3 ligases might not modify XLG2, but might be regulated by XLG2 to modify other proteins. Mainly PUB4 is required for full chitin induced ROS generation (Desaki et al., 2019; Derkacheva et al., 2020) and it is tempting to speculate that this function depends on XLG2/3. Interestingly, pub2 but not pub4 was able to partially suppress powdery mildew induced cell death in cerk1-4. It seems that both PUBs interact with XLGs in many different pathways, but there is some level of specialisation between them.
Taken together, our results show that XLG2 functions in a novel, so far uncharacterized pathway to transduce the CERK1-4-specific cell death signal at the plasma membrane. In contrast to previously described immune signalling systems that utilize XLG2, this mechanism is independent of BAK1 and SOBIR1 and not regulated by NADPH-generated ROS or XLG2 phosphorylation in the N-terminal domain.
Supporting Information
SUPPLEMENTAL MATERIALS AND METHODS
Mass Spectrometry
After GFP pull-down, on bead trypsin digestion was performed as described in a protocol provided by the manufacturer Chromotek, which is based on (Hubner et al., 2010). Digested peptides were purified with the C18 stop and go extraction (Stage) method (Rappsilber et al., 2007). Purified peptides were vacuum dried at 45°C and then dissolved in 20 μl 2% acetonitrile, 0.1% formic acid.
3-4 μl of each sample were subjected to reversed-phase liquid chromatography for peptide separation using an RSLCnano Ultimate 3000 system (Thermo Fisher Scientific). Peptides were loaded on an Acclaim® PepMap 100 pre-column (100 μm × 2 cm, C18, 5 μm, 100 Å; Thermo Fisher Scientific) with 0.07 % trifluoroacetic acid. Analytical separation of peptides was done on an Acclaim® PepMap RSLC column (75 μm × 50 cm, C18, 2 μm, 100 Å; Thermo Fisher Scientific) running a water-acetonitrile gradient at a flow rate of 300 nl/min. All solvents and acids had Optima grade for LC-MS (Thermo Fisher Scientific). Chromatographically eluting peptides were on-line ionized by nano-electrospray (nESI) using the Nanospray Flex Ion Source (Thermo Fisher Scientific) at 1.5 kV (liquid junction) and continuously transferred into the mass spectrometer (Q Exactive HF, Thermo Fisher Scientific). Full scans in a mass range of 300 to 1650 m/z were recorded at a resolution of 30,000 followed by data-dependent top 10 fragmentation (HCD) at a resolution of 15,000 (dynamic exclusion enabled).
For targeted single-ion monitoring (tSIM) LC-MS runs, the resolution was set to 60,000. The maximum ion time for tSIM scans (AGC target 1e6) was 100 ms. The loop count equalled the number of m/ z values on the inclusion list. Dynamic exclusion was disabled. LC-MS method programming and data acquisition was performed with the XCalibur software 4.0 (Thermo Fisher Scientific).
Database searches were performed using Proteome Discoverer 2.2 software (Thermo Fisher Scientific) with the Mascot and SequestHT search algorithms against the Araport11 protein database (Cheng et al., 2017). The digestion mode was set to trypsin and the maximum number of missed cleavage sites was set to three. Carbamidomethylation of cysteines was set as fixed modification, oxidation of methionine and serine, threonine and tyrosine phosphorylation were set as variable modifications. The mass tolerance was 10 ppm for precursor ions and 0.02 Da for fragment ions. The decoy mode was rev with a false discovery rate of 0.01.
Phosphopeptides were identified in initial runs and then selected for quantitative analysis by tSIM. Phosphorylation sites within phosphopeptides were assigned with the ptmRS node of Proteome Discoverer 2.2 (Thermo Fisher Scientific). Many partially overlapping phosphopeptides were found and unambiguous identification of phosphorylation sites could be achieved for approximately 50% of obtained spectra. Therefore, phosphopeptides were grouped into distinct, non-overlapping clusters. Phosphorylation within these was quantified by counting phosphorylated peptide spectrum matches (PSMs) and normalizing them to all PSMs obtained for XLG2.
Cloning
For generation of complementation constructs, a genomic fragment containing the XLG2 coding region and 1260 bp of upstream sequence was amplified with primers EP209 and EP210 (Table S1). The PCR product was cloned into a pGreenII-0229 (Hellens et al., 2000) derivative that contains a modified MCS and the 35S terminator, via its AscI and BamHI sites. pGreenII-0229-pXLG2-Venus-XLG2 and XLG2E293K were cloned by homologous recombination in yeast. PCR products of pXLG2, Venus and the XLG2/XLG2E293K coding region were amplified with primer pairs CM82 and CM92, JE23 and CM93, CM94 and CM83 respectively (Table S1). The primers introduced overlaps between the PCR products and with the shuttle vector pRS426 (Christianson et al., 1992) as well as AscI and BamHI sites for downstream cloning. Saccharomyces cerevisiae strain S288c BY4741 (Brachmann et al., 1998) was transformed with a mix of PCR fragments to be recombined and pRS426 linearized with BamHI and KpnI. Cells were then plated on SC plates (-Ura +Gluc) and incubated at 28°C for 2 – 3 days. Transformed S. cerevisiae cells were washed from plates with water, pelleted by centrifugation (5min, 4000g, RT), resuspended in 200μl P1 buffer of the QIAGEN Plasmid Midi kit and disrupted by incubation on a vibrax with 0.3g glass beads (425-600 micron) for 10 - 15 minutes. Subsequently, glass beads were pelleted by centrifugation and plasmids were prepared according to the kit manual. The pRS426 vector construct was transformed into E. coli TOP10 cells for amplification. Next, the plasmid was isolated and cut with AscI and BamHI for cloning into a modified pGreenII-0229 vector. The mutations nls426, nes387, C214/217A, C229/232/237/240A and C254/257A were introduced to pGreenII-0229-pXLG2-Venus-XLG2 using the NEB Q5m Site Directed Mutagenesis kit and primers listed in Table S1. In order to add c-terminal NLS/NES/nls/nes sequences, an additional NotI site was introduced after the XLG2 coding region by site directed mutagenesis with primers JF17 and JF18. The resulting vector was linearized with NotI and BamHI and ligated with annealed oligonucleotide pairs JF19 and JF20, JF21 and JF22, JF30 and JF31 or JF32 and JF33 (Table S1). For the 8S-A and 27S/T-A mutations, synthetic DNA fragments containing the mutations were purchased (Thermo Fisher Scientific) and amplified with primers EP624 and EP629 (Table S1). A PCR on pGreenII-0229-pXLG2-Venus-XLG2 was carried out with primers EP625 and EP220 (Table S1) and the two fragments were joined with Gibson assembly.
All constructs were confirmed by sequencing, transformed into Agrobacterium tumefaciens GV3101 (pSoup) and used for stable transformation of Arabidopsis plants.
SUPPLEMENTAL FIGURES and FIGURE LEGEND
SUPPLEMENTAL TABLES
ACKNOWLEDGEMENTS
We thank Jian-Min Zhou (Institute of Genetics and Developmental Biology, Chinese Academy of Sciences) for providing XLG2-FLAG, XLG2 4S-A-FLAG and XLG2 4S-D-FLAG constructs. We especially thank Sabine Wolfarth and Ludmilla Heck-Hrarti for technical assistance, and Kerstin Schmitt and Oliver Valerius (Core Facility LCMS Protein Analytics of the Faculty of Biology and Psychology and the Faculty of Forest Sciences and Forest Ecology, Georg-August University Göttingen) for mass spectrometry analyses. This work was funded by the DFG grants LI 1317/10-1 and INST 186/1277-1 FUGG.