The Arabidopsis E3 ubiquitin ligase PUB4 regulates BIK1 and is targeted by a bacterial type‐III effector

Abstract

Plant immunity is tightly controlled by a complex and dynamic regulatory network, which ensures optimal activation upon detection of potential pathogens. Accordingly, each component of this network is a potential target for manipulation by pathogens. Here, we report that RipAC, a type III‐secreted effector from the bacterial pathogen Ralstonia solanacearum, targets the plant E3 ubiquitin ligase PUB4 to inhibit pattern‐triggered immunity (PTI). PUB4 plays a positive role in PTI by regulating the homeostasis of the central immune kinase BIK1. Before PAMP perception, PUB4 promotes the degradation of non‐activated BIK1, while after PAMP perception, PUB4 contributes to the accumulation of activated BIK1. RipAC leads to BIK1 degradation, which correlates with its PTI‐inhibitory activity. RipAC causes a reduction in pathogen‐associated molecular pattern (PAMP)‐induced PUB4 accumulation and phosphorylation. Our results shed light on the role played by PUB4 in immune regulation, and illustrate an indirect targeting of the immune signalling hub BIK1 by a bacterial effector.

RipAC, a secreted plant pathogen effector, inhibits pattern‐triggered immunity by manipulating the accumulation of the central immunity protein BIK1.

Introduction

Plants constantly face pathogens, and the success of their defence depends on their ability to sense an attack and provide a fast response. The first layer of plant immunity is based on the recognition of pathogen‐associated molecular patterns (PAMPs) or damaged‐associated molecular patterns (DAMPs) via plasma membrane‐localized pattern recognition receptors (PRRs). Perception of PAMPs or DAMPs by their corresponding PRRs leads to pattern‐triggered immunity (PTI), which restricts the multiplication of most potential pathogens (Couto & Zipfel, 2016). The Arabidopsis thaliana (hereafter Arabidopsis) leucine‐rich repeat receptor kinases (LRR‐RKs) FLS2 and EFR are among the best‐characterized plant PRRs. They recognize the bacterial PAMPs flagellin (or its derived peptide flg22) and elongation factor Tu (or its derived peptide elf18), respectively (Gómez‐Gómez & Boller, 2000; Zipfel et al, 2006). Similarly, the endogenous Arabidopsis peptide AtPep1, which is released by plant cells as a DAMP to amplify the immune response, is perceived by the LRR‐RKs PEPR1/PEPR2 (Yamaguchi et al, 2006, 2010; Krol et al, 2010; Tang & Zhou, 2016). Ligand binding by these PRRs triggers the association with the LRR‐RK BAK1 (also known as SERK3) and related SERKs, which act as co‐receptors, leading to phosphorylation events and activation of downstream immune signalling (Chinchilla et al, 2007; Heese et al, 2007; Postel et al, 2010; Schulze et al, 2010; Roux et al, 2011; Schwessinger et al, 2011; Sun et al, 2013; Perraki et al, 2018). Chitin, a component of the fungal cell wall, is perceived by the LysM‐RK LYK5, which forms a complex with the co‐receptor CERK1 (another LysM‐RK) upon ligand‐binding, leading to trans‐phosphorylation and initiation of immune signalling (Miya et al, 2007; Wan et al, 2008; Petutschnig et al, 2010; Liu et al, 2012, 2018; Cao et al, 2014; Suzuki et al, 2016, 2018, 2019; Erwig et al, 2017).

The activation of PRR complexes leads to the activation of downstream receptor‐like cytoplasmic kinases (RLCKs) (Liang & Zhou, 2018). The RLCK BIK1 is a direct substrate of FLS2/EFR/PEPR1/CERK1 complexes and is thus a convergent point in PTI signalling triggered by several elicitors (Lu et al, 2010; Zhang et al, 2010; Laluk et al, 2011; Liu et al, 2013). BIK1 plays roles in plant immunity to bacterial and fungal pathogens (Veronese et al, 2006; Lu et al, 2010; Zhang et al, 2010), and is required for several PTI responses, such as production of reactive oxygen species (ROS), calcium influx, callose deposition, and stomata closure (Lu et al, 2010; Zhang et al, 2010; Kadota et al, 2014; Li et al, 2014; Ranf et al, 2014; Monaghan et al, 2015; Tian et al, 2019; Thor et al, 2020).

BIK1 accumulation is a rate‐limiting factor in PTI responses, and as such BIK1 protein levels are tightly regulated (Zhang et al, 2010, 2018; Kadota et al, 2014; Monaghan et al, 2014; Couto & Zipfel, 2016; Liang et al, 2016; Wang et al, 2018; Jiang et al, 2019). It has been suggested that two pools of BIK1 exist within a cell: “non‐activated” (non‐phosphorylated, before PRR activation) and “activated” (phosphorylated, after PRR activation; Wang et al, 2018).

The ubiquitin‐modification system, which is best known for regulating the main degradation systems in eukaryotes, is intimately connected to cellular signalling pathways. Plant U‐box proteins (PUBs) are E3 ubiquitin ligases and form regulatory modules with kinases to regulate signalling outputs and maintain homeostasis (Trenner et al, 2022). Non‐activated BIK1 is targeted for proteasomal degradation by the E3 ubiquitin ligases PUB25 and PUB26, which is promoted by the cytoplasmic calcium‐dependent kinase CPK28; this degradation is inhibited by the heteromeric G protein complex formed by XLG2‐AGB1‐AGG1/2 (Monaghan et al, 2014; Liang et al, 2016; Wang et al, 2018). After flg22 perception, the heteromeric G protein complex dissociates from FLS2, and CPK28 phosphorylates PUB25/26 to promote degradation of the non‐activated pool of BIK1, thereby adjusting the amplitude of immune response (Liang et al, 2016; Wang et al, 2018). Activated BIK1 dissociates from the FLS2‐BAK1 complex and is protected from degradation by PUB25/26 while promoting immune signalling (Wang et al, 2018).

To achieve a successful infection, pathogens secrete effector proteins that manipulate plant cellular functions, including those that target immune signalling components and manipulate PTI (Macho & Zipfel, 2015; Macho, 2016; Lee et al, 2019). Therefore, besides being important virulence factors, effector proteins constitute useful probes to identify and characterize plant proteins involved in immune signalling (Toruño et al, 2016). Ralstonia solanacearum is a soil‐borne pathogen and the causal agent of bacterial wilt disease. R. solanacearum can infect more than 250 plant species across more than 50 families, including economically important crops such as tomato, potato, banana, pepper, and eggplant (Wicker et al, 2007; Mansfield et al, 2012; Jiang et al, 2017). R. solanacearum enters the plants through the roots, reaches the vascular system, and proliferates in xylem vessels to colonize the whole plant (Genin, 2010; Xue et al, 2020). R. solanacearum relies on a type‐III secretion system to deliver over 70 type‐III effector proteins (T3Es) into the host cytoplasm (Sabbagh et al, 2019). One of these T3Es, RipAC, is able to suppress effector‐triggered immunity (ETI) by targeting SGT1, and is required for full virulence of R. solanacearum in tomato and Arabidopsis (Nakano et al, 2020; Yu et al, 2020).

Here, we demonstrate that RipAC suppresses diverse PTI responses. We also show that RipAC interacts with the E3 ubiquitin ligase PUB4 from tomato and Arabidopsis, and that pub4 mutant plants show deficient immune responses to several PAMP/DAMPs. Interestingly, we found that PUB4 has a dual impact on BIK1: before PAMP treatment, PUB4 promotes degradation of non‐activated BIK1, but, after PAMP treatment, PUB4 is required for the accumulation of activated BIK1. RipAC overexpression in Arabidopsis leads to BIK1 degradation, suggesting that RipAC exploits PUB4 to degrade BIK1 and suppress PTI responses.

Results

RipAC suppresses pattern‐triggered immunity

To test whether RipAC affects PTI in Arabidopsis, we characterized PAMP‐induced responses in Arabidopsis transgenic lines overexpressing RipAC (Yu et al, 2020). Two independent RipAC‐GFP overexpression lines showed a decreased ROS burst in response to bacterial or fungal PAMPs (flg22^Pto, elf18^Rsol, and chitin; Fig 1A–C). The level of ROS inhibition correlated with the level of RipAC accumulation in these lines (Fig 1D). Activation of PAMP‐induced signalling leads to the activation of MAPK cascades (Yu et al, 2017), and flg22‐induced MAPK activation was also decreased in RipAC‐GFP overexpression lines (Fig 1D). In agreement with the observed inhibition of early PTI responses, RipAC overexpression lines were more susceptible to Pseudomonas syringae pv. tomato (Pto) DC3000 ΔhrcC, a non‐pathogenic mutant strain unable to secrete T3Es, and therefore inducing solely PTI, but not to wild‐type Pto DC3000, which is able to suppress PTI (Fig 1E). Together, these results demonstrate that RipAC inhibits early PTI responses triggered by various PAMPs to facilitate pathogen infection.

Figure 1. RipAC inhibits PTI in Arabidopsis .

• A–C
  RipAC overexpression inhibits ROS production induced by 50 nM flg22^Pto (A), 100 nM elf18^Rsol (B), and 200 μg ml⁻¹ chitin (C) in Arabidopsis. ROS was measured as relative luminescence units (RLU) over time. Values are mean ± SE (n = 24 biological replicates).
• D
  RipAC overexpression inhibits flg22‐triggered MAPK activation in Arabidopsis. 100 nM flg22^Pto was used to treat Arabidopsis seedlings and the samples were collected at indicated time points. Immunoblots were analysed using anti‐pMAPK and anti‐RipAC antibodies. Coomassie brilliant blue (CBB) staining and a non‐specific band were used as loading control. Molecular weight (kDa) marker bands are indicated for reference.
• E
  RipAC overexpression lines display elevated susceptibility to Pto DC3000 ΔhrcC, but not to Pto DC3000. Arabidopsis plants were spray‐inoculated with indicated Pto strains (5 × 10⁷ CFU ml⁻¹) and bacterial titres were determined 3 days post‐inoculation. Values are mean ± SE (n = 6 biological replicates). Asterisks indicate significant differences compared with Col‐0 (Student's t test, **P < 0.01).

Data information: Experiments were performed three times with similar results.

Source data are available online for this figure.

RipAC interacts with PUB4 from tomato and Arabidopsis

Our previous work showed that SGT1 targeting by RipAC does not underlie PTI suppression (Yu et al, 2021), and therefore the relevant target(s) of RipAC in this context remain to be identified. As tomato (Solanum lycopersicum) is a major crop affected by R. solanacearum, we performed a yeast two‐hybrid screen using RipAC as a bait against a library of cDNA from tomato roots inoculated with R. solanacearum. We identified several clones matching the tomato ortholog of Arabidopsis PLANT U‐BOX PROTEIN 4 (AtPUB4), SlPUB4 (Appendix Fig S1A). We further confirmed the interaction between RipAC and PUB4 in planta using split‐luciferase (Split‐LUC) complementation assays in Nicotiana benthamiana; either visualized using a charge‐coupled device (CCD) imaging system (Fig 2A) or quantified using a luminometre (Fig 2B). These assays showed that RipAC interacts with both SlPUB4 and AtPUB4, but not with the aquaporin AtPIP2A used as negative control (Fig 2A and B; Appendix Fig S1B–D). Co‐expression of GFP or GFP‐tagged SlPUB4/AtPUB4 with RipAC‐nLUC in N. benthamiana followed by co‐immunoprecipitation (CoIP) revealed that RipAC associates with SlPUB4 and AtPUB4 (Fig 2C). Additionally, FRET‐FLIM assays in N. benthamiana further confirmed the direct interaction of RipAC with SlPUB4 and AtPUB4 (Fig 2D). Altogether, these results indicate that PUB4 is a novel interactor of RipAC in plant cells.

Figure 2. RipAC associates with PUB4 in plant cells.

• A, B
  Interaction between RipAC and SlPUB4/AtPUB4 was confirmed by Split‐LUC assay qualitatively (A) and quantitatively (B) in Nicotiana benthamiana. Values are mean ± SE (n = 8 biological replicates). Asterisks indicate significant differences compared with RipAC‐nLUC/cLUC‐AtPIP2A negative control (Student's t test, ****P < 0.0001). RLU, relative luminescence units.
• C
  Co‐immunoprecipitation of SlPUB4/AtPUB4‐GFP and RipAC‐nLUC after co‐expression in N. benthamiana. Two days after Agrobacterium infiltration, the plant tissues were collected and then subjected to anti‐GFP immunoprecipitation. Immunoblots were analysed using anti‐LUC and anti‐GFP antibodies. Molecular weight (kDa) marker bands are indicated for reference.
• D
  Interaction between RipAC‐RFP and SlPUB4/AtPUB4‐GFP determined by FRET‐FLIM upon transient co‐expression in N. benthamiana leaves. GFP fluorescence lifetime of each GFP fusion protein is shown as a negative control. Lines represent average values (n = 5 biological replicates) and error bars represent standard error. (Student's t‐test, ***P < 0.001).

Data information: Experiments were performed three times with similar results.

Source data are available online for this figure.

PUB4 positively regulates PTI in Arabidopsis

Arabidopsis PUB4 belongs to the family of plant U‐box proteins containing Armadillo (ARM) repeats (Mudgil et al, 2004), and has been implicated in cytokinin responses, tapetum development, meristem maintenance, cell division, oxidative stress, responses to chitin, as well as affecting growth allometry during abiotic stress and local climate (Woodson et al, 2015; Kinoshita et al, 2015a, 2015b; Wang et al, 2017; Vasseur et al, 2018; Desaki et al, 2019).

We sought to determine whether the targeting of PUB4 might underlie the suppression of PTI responses by RipAC. First, to understand the role of PUB4 in the regulation of PTI, we characterized PAMP/DAMP‐triggered responses in pub4 mutants. Two independent pub4 mutant lines showed reduced ROS burst in response to flg22^Paer, elf18^Ecol, AtPep1, and chitin (Fig 3A–D). Similar results were obtained in pub4 mutants using flg22^Pto and elf18^Rsol as elicitors (Appendix Fig S2A and B). These results are consistent with the recently reported reduced PAMP‐induced ROS production in pub4 (Desaki et al, 2019; Wang et al, 2022). The reduced PAMP‐triggered ROS in pub4 mutants resembled the phenotype of RipAC‐overexpression lines (Fig 1). MAPK activation was not affected in pub4 mutants (Fig 3E), suggesting that the effect of RipAC on MAPK activation is PUB4‐independent. The inhibition of seedling growth triggered by flg22 and elf18 was also compromised in pub4 mutants (Fig 3F and G), indicating that PUB4 is also required for late PTI responses. Moreover, pub4 mutant lines showed compromised stomatal closure triggered by flg22 or chitin (Fig 3H and I). Abscisic acid (ABA)‐triggered stomatal closure was comparable in Col‐0 and pub4 plants (Fig 3J), indicating that pub4 stomatal closure is specifically impaired in response to PAMPs. Finally, we tested the role of PUB4 in antibacterial immunity by performing surface‐inoculation with the non‐pathogenic strain Pto DC3000 ΔhrcC. Given that pub4 mutants showed deficient PAMP‐triggered stomatal closure, we also performed inoculations with a weakly virulent Pto derivative unable to produce coronatine (Pto DC3000 COR⁻), a non‐T3E virulence factor required for the suppression of stomatal defences (Melotto et al, 2006). We found that pub4 mutants were more susceptible to both strains (Fig 3K). Together, these results demonstrate that PUB4 genetically behaves as a positive regulator of PTI.

Figure 3. PUB4 positively regulates PTI in Arabidopsis .

• A–D
  ROS burst in Col‐0 WT or the indicated pub4 mutant lines induced by 100 nM flg22^Paer (A), 100 nM elf18^Ecol (B), 1 μM AtPep1 (C), and 1 mg ml⁻¹ chitin (D). ROS was measured as relative luminescence units (RLU) over time. Values are mean ± SE (n = 16 biological replicates).
• E
  flg22‐triggered MAPK activation in pub4 T‐DNA mutants. 100 nM flg22^Pto was used to treat Arabidopsis seedlings and the samples were collected at indicated time points for western blots. Immunoblots were analysed using anti‐pMAPK and anti‐FLS2 antibodies. Coomassie brilliant blue (CBB) staining was used as loading control. Molecular weight (kDa) marker bands are indicated for reference.
• F, G
  PUB4 contributes to seedling growth inhibition induced by 100 nM flg22^Paer (F) or 100 nM elf18^Ecol (G). Values represent the percentage of fresh weight of PAMP‐treated vs water‐treated seedlings, and are mean ± SE (n = 16 biological replicates). Asterisks indicate significant differences compared with Col‐0 (one‐way ANOVA, Kruskal–Wallis test, Dunn's multiple comparisons test, ***P < 0.001, ****P < 0.0001).
• H–J
  PUB4 contributes to PAMP‐induced stomatal closure. Stomatal apertures were measured as width to length ratio 2 h after treatment with mock or 10 μM flg22^Paer (H), 1 mg ml⁻¹ chitin (I) and 10 μM ABA (J). Values are mean ± SE (n > 120). Asterisks indicate significant differences between samples (one‐way ANOVA, Kruskal–Wallis test, Dunn's multiple comparisons test, ****P < 0.0001, ^ns P > 0.05).
• K
  pub4 mutant lines display elevated susceptibility to Pto DC3000 COR⁻ and Pto DC3000 ΔhrcC. Arabidopsis plants were spray‐inoculated with indicated Pto strains and bacterial titres were determined 3 days post‐inoculation. Values are mean ± SE (n = 6 biological replicates). Asterisks indicate significant differences compared with Col‐0 (Student's t test, **P < 0.01).

Data information: Experiments were performed three times with similar results.

Source data are available online for this figure.

PUB4 promotes R. solanacearum infection in Arabidopsis and tomato

To determine whether PUB4 is required for plant basal resistance to R. solanacearum, we performed soil‐drenching inoculation with R. solanacearum GM1000 in Arabidopsis wild‐type (Col‐0 WT), pub4‐1 mutant, and PUB4‐FLAG‐overexpressing plants. Surprisingly, pub4‐1 mutant plants showed weaker wilting symptoms (Fig 4A; Appendix Fig S3A and B), while overexpression of PUB4 enhanced the development of symptoms, in comparison with WT plants (Fig 4A; Appendix Fig S3I and J). Transgenic expression of RipAC in Arabidopsis plants has been shown to enhance susceptibility to R. solanacearum (Yu et al, 2020); interestingly, overexpression of PUB4 partially reduced the impact of RipAC transgenic expression on plant susceptibility (Appendix Fig S3I and J). To address the role of PUB4 in disease resistance in a natural host of R. solanacearum, we performed soil‐drenching inoculation in tomato plants with roots either expressing a SlPUB4 RNAi construct (SlPUB4‐RNAi) or overexpressing SlPUB4 (OE:SlPUB4). SlPUB4‐RNAi plants showed a slight but reproducible reduction in wilting symptoms compared with control plants carrying an empty vector (EV‐RNAi), demonstrating the same tendency as pub4 mutants in Arabidopsis (Fig 4B; Appendix Fig S3C–E). Accordingly, OE:SlPUB4 plants showed earlier wilting symptoms than control EV plants, suggesting that SlPUB4 overexpression promotes R. solanacearum infection (Fig 4C; Appendix Fig S3F–H). Together, these results suggest that PUB4 acts as a positive regulator of R. solanacearum infection and thus support the hypothesis that PUB4 is a susceptibility gene for R. solanacearum that is targeted by RipAC.

Figure 4. PUB4 promotes R. solanacearum infection in Arabidopsis and tomato.

• A
  Soil‐drenching inoculation assays in Arabidopsis Col‐0, pub4‐1 mutant, and PUB4‐overexpressing lines. Plants were inoculated with R. solanacearum GMI1000. The results are represented as disease progression, showing the average wilting symptoms in a scale from 0 to 4 (mean ± SE, n = 15 biological replicates).
• B, C
  Soil‐drenching inoculation assays in tomato plants with roots expressing a SlPUB4 RNAi construct (SlPUB4‐RNAi) (B) or overexpressing SlPUB4 (OE:SlPUB4) (C). Plants were inoculated with R. solanacearum GMI1000. The results are represented as disease progression, showing the average wilting symptoms in a scale from 0 to 4 (mean ± SE, n = 12 biological replicates).

Data information: Experiments were performed at least three times with similar results. Panels show representative results; composite data from different experiments and survival analyses are shown in Appendix Fig S3.

PUB4 associates with PRR complexes

To determine the molecular mechanism of PTI regulation by PUB4, we immuno‐purified PUB4 and its interacting proteins from mock‐ or elf18‐treated transgenic Arabidopsis plants expressing PUB4‐FLAG and analysed the resulting immunoprecipitates by liquid chromatography followed by tandem mass spectrometry (LC‐MS/MS). PUB4 was found to associate with the EFR‐BAK1 PRR complex, especially after elf18 treatment (Fig 5A). The extra‐large G protein XLG2, which is known to associate with the FLS2 complex (Liang et al, 2016), and its close homologue XLG1, constitutively associated with PUB4 (Fig 5A). Surprisingly, we did not detect peptides of BIK1, which is known to associate with PRR complexes (Lu et al, 2010; Zhang et al, 2010). However, we have previously observed that the low protein accumulation of BIK1 can make it difficult to be detected in immunoprecipitates by LC‐MS/MS. Thus, to determine whether PUB4 could also associate with BIK1, we co‐expressed BIK1‐HA with PUB4‐GFP or GFP in N. benthamiana, and performed a CoIP before or after flg22 treatment. BIK1 was found to associate constitutively with PUB4 (Fig 5B). Moreover, a CoIP assay using PUB4‐FLAG Arabidopsis plants further indicated that PUB4 associates with FLS2 and BAK1 specifically after flg22 treatment, and confirmed the constitutive association with BIK1 (Fig 5C). To determine whether PUB4 physically interacts with BIK1, we purified recombinant MBP‐PUB4 and GST‐BIK1 proteins from E. coli and performed MBP pull‐down assays. The results showed that PUB4 interacts directly with BIK1 (Fig 5D).

Figure 5. PUB4 associates with PRR complexes.

1. PUB4 associates in Arabidopsis with EFR, BAK1, and SERK2 after 1 μM elf18^Ecol treatment, and constitutively with XLG2 and XLG1. PRR complex members were identified after immunoprecipitation of PUB4‐FLAG from 35S:PUB4‐FLAG line, tryptic digestion and sample analyses by LC‐MS/MS. Untransformed Col‐0 seedlings were used as a negative control. Total spectrum count for each protein is shown.
2. Co‐immunoprecipitation of PUB4 and BIK1 in N. benthamiana. PUB4‐GFP or GFP‐LTI6b were transiently co‐expressed with BIK1‐HA in the leaves of N. benthamiana. After treatment with mock or 1 μM flg22^Paer for 10 min, total proteins (input) were extracted and subjected for immunoprecipitation with anti‐GFP beads.
3. Co‐immunoprecipitation of PUB4 with PRR complex members in Arabidopsis. PUB4 co‐immunoprecipitated with FLS2 and BAK1 specifically after 10 min of 1 μM flg22^Paer treatment, and constantly with BIK1. Total protein extracts (input) from Col‐0 and PUB4‐FLAG plants were subjected to immunoprecipitation with anti‐FLAG beads.
4. PUB4 directly interacts with BIK1 in vitro. Immunoblots were analysed using the indicated antibodies. Molecular weight (kDa) marker bands are indicated for reference.

Data information: Experiments in (B–D) were performed three times with similar results.

Source data are available online for this figure.

RipAC does not affect PUB4 association with PRRs and BIK1

To understand the effect of RipAC on PUB4, we tested whether RipAC has an impact on PUB4 interaction with PRRs and BIK1. CoIP assays, using PUB4‐FLAG or PUB4‐FLAG RipAC‐GFP plants, indicated that RipAC does not affect the constitutive association between PUB4 and BIK1, nor the flg22‐induced association between FLS2 and BAK1 (Fig 6). However, together with our previous results (Fig 5C), these assays revealed that PUB4 associates with modified BIK1, as we reproducibly observed a laddering on PUB4‐associated BIK1 (Fig 6). Interestingly, this laddering was enhanced in the presence of RipAC (Fig 6). This observation led us to consider BIK1 as a relevant PUB4‐associated protein and indirect RipAC target. Further support for this hypothesis comes from our genetic data demonstrating that both RipAC and PUB4 affect immune responses triggered by various PAMPs/DAMPs (Figs 1 and 3), and the known role of BIK1 as a convergent point downstream of multiple PRR complexes (DeFalco & Zipfel, 2021).

Figure 6. RipAC does not affect PUB4 association with PRRs and BIK1.

RipAC does not affect PUB4 association with PRR complex members. PUB4‐FLAG, PUB4‐FLAG RipAC‐GFP and Col‐0 plants were treated for 10 min with water (as mock) or 1 μM flg22^Paer and elf18^Ecol treatment. Total protein extracts (input) were subjected for immunoprecipitation with anti‐FLAG beads. Immunoblots were analysed using the indicated antibodies. Molecular weight (kDa) marker bands are indicated for reference. Data information: The experiment was performed three times with similar results. Source data are available online for this figure.

PUB4 ubiquitinates non‐active BIK1

In order to study PUB4 ubiquitination activity, we used a bacterial reconstitution system (Kowarschik et al, 2018), in which the operon containing His‐Ubiquitin, UBA (E1), HA‐UBC8 (E2), and the E3 ligase are expressed simultaneously after induction, while potential E3 ligase substrates are added via co‐transformation in E. coli. Using this system, we found that PUB4 has auto‐ubiquitination activity in vitro, while a variant carrying a mutation in one of the two loops that participate in E2 recruitment (C239A) shows a strongly impaired activity (Fig 7A). This result is consistent with previous report that PUB4‐FLAG purified from transgenic plants shows E3 ligase activity (Wang et al, 2013). We then introduced BIK1, as well as its kinase‐dead variant (K105E, BIK1‐KD), and performed His pull‐down to enrich His‐ubiquitin‐conjugated proteins. The results show that PUB4 specifically ubiquitinates BIK1‐KD, but not wild‐type BIK1, indicating that PUB4 ubiquitinates non‐active BIK1 (Fig 7B).

Figure 7. PUB4 ubiquitinates non‐active BIK1.

1. PUB4 has auto‐ubiquitination activity in vitro. PUB4 autoubiquitination assay using His‐Ubiquitin, UBA1 (E1), HA‐UBC8 (E2) and MBP‐PUB4 or MBP‐PUB4 Cys239Ala inactive mutant reconstituted in bacteria. Ubiquitinated products were purified under denaturing conditions by immobilized metal ion chromatography (IMAC). Input and precipitated proteins were resolved in 10 and 8% acrylamide gels, respectively, and analysed by immunoblot with indicated antibodies. The line indicates autoubiquitinated PUB4.
2. PUB4 ubiquitinates non‐active BIK1, and does not ubiquitinate active BIK1. Operon containing His‐Ubiquitin, UBA1 (E1), HA‐UBC8 (E2), and MBP‐PUB4 were co‐transformed with GST, GST‐BIK1, or GST‐BIK1 kinase dead (BIK1‐KD) in the presence of PUB4. Ubiquitinated products were purified under denaturing conditions by IMAC, and modified proteins further enriched by GSH chromatography. Input and precipitated proteins were resolved in 10 and 8% acrylamide gels, respectively, and analyzed by immunoblot with indicated antibodies. The line indicates the ubiquitination products.

Data information: The experiments were performed three times with similar results.

Source data are available online for this figure.

PUB4 regulates BIK1 protein homeostasis

The finding that PUB4 ubiquitinates BIK1 in vitro prompted us to test whether PUB4 affects BIK1 protein levels in planta before and after immune elicitation. To this end, we crossed the BIK1‐HA line with the pub4‐1 ^−/− mutant. We repeatedly failed to generate viable fertile pub4 homozygous plants expressing BIK1‐HA, and thus analysed BIK1‐HA accumulation in pub4‐1 ^−/− BIK1‐HA ^+/−, PUB4‐FLAG ^+/− BIK1‐HA ^+/− and BIK1‐HA ^+/− plants. In order to analyse the impact of PUB4 on BIK1 accumulation in basal conditions (before PAMP treatment), we used cycloheximide (CHX) treatment to inhibit de novo protein synthesis. Interestingly, the accumulation of non‐activated BIK1 after CHX treatment was higher in pub4‐1 ^−/− BIK1‐HA ^+/− plants compared with BIK1‐HA ^+/− (Fig 8A), which may suggest that PUB4 contributes to the degradation of non‐active BIK1, which would be consistent with our observation that PUB4 directly ubiquitinates non‐active BIK1 (Fig 7). The results also show a reduced accumulation of BIK1 after flg22 treatment in pub4‐1 ^−/− BIK1‐HA ^+/− plants compared with BIK1‐HA ^+/− plants (Fig 8A), further supporting the notion that PUB4 function is required for accumulation of wild‐type levels of activated BIK1.

Figure 8. BIK1 accumulation is dually regulated by PUB4 and negatively affected by RipAC.

• A
  Analysis of BIK1‐HA protein accumulation in BIK1‐HA ^+/−, pub4 ^−/− BIK1‐HA ^+/−, and PUB4‐FLAG ^+/− BIK1‐HA ^+/− in 4‐ to 5‐week‐old plants. Prior to protein extraction leave disks were treated for 6 h with 100 mM MG132, 50 mM CHX, and for 10 min with water (as mock) or 1 mM flg22^Paer. Proteins were extracted in SDS buffer and analysed by western blot. BAK1 was used as a loading control. Numbers correspond to the quantitation of the BIK1‐HA band, normalized to the quantitation of the BAK1 control band in the same sample.
• B
  BIK1‐HA accumulation is reduced in PUB4‐FLAG ^+/− BIK1‐HA ^+/− line compared with BIK1‐HA ^+/− line in the absence of PAMP treatment. Total protein extracts were analysed by western blot. Tubulin was used as a loading control.
• C, D
  PUB4 overexpression leads to reduction in ROS burst in response to (C) 100 nM flg22^Paer and (D) 100 nM elf18^Ecol. ROS was measured as relative luminescence units (RLU) over time. Values are mean ± SE (n = 24 biological replicates).
• E
  BIK1‐HA accumulation is reduced in RipAC‐GFP/BIK1‐HA line compared with Col‐0/BIK1‐HA line. Total protein extracts were analysed by western blot. Actin was used as a loading control.

Data information: In (A), (B), and (E), WB quantification was performed using ImageJ software. Immunoblots were analyzed using the indicated antibodies. Molecular weight (kDa) marker bands are indicated for reference. Experiments were performed at least three times with similar results.

Source data are available online for this figure.

The reduced mobility band corresponding to phosphorylated BIK1 was detectable in pub4‐1 ^−/− BIK1‐HA ^+/− plants, suggesting that the activation of BIK1 was not affected by loss of PUB4 (Fig 8A). The reduced accumulation of activated BIK1 in the pub4 mutant background could be abolished by MG132 treatment (Fig 8A), suggesting that activated BIK1 is subject to proteasomal degradation in pub4 plants. Such a requirement of PUB4 for the maintenance of activated, signalling‐competent BIK1 could explain the impaired immune responses in pub4 mutant plants in response to various PAMPs/DAMPs.

We then examined BIK1‐HA accumulation in a PUB4‐FLAG overexpression background. Interestingly, we found lower accumulation of non‐activated BIK1 in PUB4‐FLAG ^+/− BIK1‐HA ^+/− plants compared with BIK1‐HA ^+/− plants (Fig 8A and B; Appendix Fig S4A). This is consistent with the higher accumulation of non‐activated BIK1 observed in pub4 plants (Fig 8A). As it was the case in pub4 knockouts, BIK1 activation was not affected by PUB4‐FLAG overexpression (Fig 8A).

Several observations suggest that PUB4 promotes the degradation of non‐activated BIK1: (i) PUB4 associates with BIK1 in basal conditions; (ii) PUB4 ubiquitinates inactive BIK1 (BIK1‐KD); (iii) PUB4 overexpression leads to reduced accumulation of non‐activated BIK1; and (iv) pub4 mutation leads to enhanced accumulation of non‐activated BIK1. In accordance with this hypothesis, and considering that BIK1 is a rate‐limiting factor in the activation of early PTI responses, PAMP‐triggered ROS production was impaired in the PUB4‐FLAG overexpressing line (Fig 8C and D). Conversely, the accumulation of active BIK1, which is not ubiquitinated by PUB4 (Fig 7B), was lower in the pub4 mutant background. Thus, PUB4 seems to play a dual role in BIK1 stability: promoting the degradation of non‐activated BIK1, while preserving activated BIK1 from degradation.

RipAC negatively impacts BIK1 accumulation

Considering our results suggesting that PUB4 regulates BIK1 stability and is targeted by RipAC, we tested whether RipAC affects BIK1 protein accumulation. For this, we crossed RipAC‐GFP plants with BIK1‐HA plants, and analysed the resulting F1. Plants expressing RipAC showed lower BIK1 accumulation compared with control plants (Fig 8E; Appendix Fig S4B), which could explain the impaired ROS burst in these plants in response to various PAMPs. Despite the impact of RipAC on BIK1 accumulation, RipAC did not associate with BIK1 in CoIP and Split‐LUC assays in planta (Appendix Fig S5). This suggests that RipAC acts on BIK1 via PUB4, which is supported by the greater band shift of PUB4‐associated BIK1 observed after flg22 treatment in the PUB4‐FLAG RipAC‐GFP line compared with PUB4‐FLAG plants (Fig 6).

RipAC manipulates PUB4 accumulation and phosphorylation

To test if RipAC regulates PUB4, we immunoprecipitated PUB4‐FLAG from mock‐ or PAMP‐treated (using a combination of both flg22 and elf18) PUB4‐FLAG or PUB4‐FLAG RipAC‐GFP plants and analysed the samples by LC‐MS/MS and parallel reaction monitoring (PRM) LC‐MS/MS. Notably, PUB4 accumulated after flg22/elf18 treatment, and this effect was abolished by RipAC (Fig 9A), suggesting that RipAC disrupts the positive role of PUB4 in the stabilization of activated BIK1 and therefore promotion of PTI.

Figure 9. RipAC causes a reduction in PUB4 accumulation after PAMP treatment and manipulates PUB4 phosphorylation.

• A
  RipAC prevents an increase in PUB4 accumulation after PAMP treatment. PUB4 was immunoprecipitated from PUB4‐FLAG or PUB4‐FLAG RipAC‐GFP plants after treatment with PAMPs (1 μM flg22^Paer, 1 μM elf18^Ecol) or water (as mock treatment). The samples were digested with trypsin and analysed by parallel reaction monitoring (PRM) LC‐MS/MS. Values represent the sum of intensities of PUB4 corresponding peptides. Data are mean ± SE of two biological replicates, each of which contains three technical replicates (two‐way ANOVA, Tukey's multiple comparison test). Different letters indicate significantly different values at P < 0.0001).
• B–E
  RipAC affects PUB4 phosphorylation after PAMP treatment. The samples were prepared and processed as in (A). The abundance of phosphorylated peptide was calculated as the ratio of intensities of phosphorylated vs PUB4 control peptide sum. Values represent the percentage of PUB4 phosphorylated peptide in PAMP‐treated vs mock‐treated seedlings, with 100% indicating the same level of PUB4 phosphorylation in PAMP‐ and mock‐treated seedlings. Values are mean ± SE of two biological replicates, each of which contains three technical replicates (t test, asterisks indicate significant differences compared with PUB4‐FLAG, ****P < 0.0001, **P < 0.01).

Our preliminary data showed that PUB4 is a highly phosphorylated protein both before and after PAMP treatment. We thus quantified PUB4 phosphorylation by Parallel Reaction Monitoring (PRM) LC‐MS/MS in the absence and presence of RipAC, with either mock or PAMP treatment. As effectors are typically delivered once PTI has already been activated, we focused on the analyses of PUB4 phosphosites that were affected by RipAC after PAMP treatment. We identified PUB4 phosphosites that were manipulated by RipAC and quantified their relative phosphorylation after PAMP treatment compared with mock treatment (Fig 9B–E). Most PUB4 phosphosites affected by RipAC show the same tendency: PUB4 phosphosites significantly downregulated upon PAMP treatment were attenuated or upregulated in the presence of RipAC (Fig 9B and C), with two PUB4 regions demonstrating different tendencies (Fig 9D and E). Most of the affected phosphosites lie in the hinge region of PUB4, between the U‐box and ARM repeats (Appendix Fig S1A), while S762 is in the ARM domain, which is responsible for protein‐protein interactions. Hence, phosphorylation sites in the hinge region may regulate PUB4 activity, while S762 phosphorylation may affect PUB4 interaction with other proteins. Together, these data suggest that RipAC manipulates PUB4 phosphorylation to regulate its function.

Discussion

Immune signalling is tightly regulated to prevent autoimmunity while ensuring optimal strength and duration of immune responses upon detection of potential pathogens. This is achieved through negative regulators acting at the level of PRR complex formation and activation, cytoplasmic signal transduction, and activation of defence‐related genes (DeFalco & Zipfel, 2021; Trujillo, 2021). Although this provides a robust mechanism for the regulation of immune activation, it also constitutes a potential target for manipulation by invading pathogens.

In this study, we identified the U‐box E3 ubiquitin ligase PUB4 as a target of the R. solanacearum effector RipAC. Detailed genetic analysis indicated that PUB4 is a common regulator of immune signalling triggered by various PAMPs/DAMPs. We show that PUB4 constitutively associates with BIK1 and XLG1/XLG2 and is recruited to FLS2/EFR‐BAK1 complexes after PAMP treatment. Our biochemical data is consistent with previous reports showing that PUB4 associates with XLG1/2/3 in planta (Wang et al, 2017), and that XLG2 associates with PRR complexes in basal conditions (Liang et al, 2016). Interestingly, we did not detect association of PUB4 with FLS2 before PAMP treatment, and other heteromeric G protein complex subunits were not detected among PUB4‐associated proteins, suggesting that PUB4 associates with activated (GTP‐bound) XLG2 and BIK1 pools not associated with FLS2. A recent report showed that PUB4 associates with CERK1, and is a positive regulator of chitin‐triggered signalling (Desaki et al, 2019). Our data showing that PUB4 positively regulates not only chitin responses, but also flg22, elf18 and AtPep1 responses, together with the known role of BIK1 downstream of their respective PRRs, suggest that the association of PUB4 with BIK1 explains the genetic contribution of PUB4 to responses triggered by diverse elicitors. While this manuscript was under revision, another study was published showing that PUB4 regulates and connects components of PTI signalling, including FLS2, BIK1, and PBL27 (Wang et al, 2022), which is in keeping with the findings in our study.

The results of our study provide a mechanism explaining the role of PUB4 in immune responses triggered by various PAMPs/DAMPs, wherein PUB4 directly interacts with BIK1 and regulates its homeostasis. Being a hub of immune signalling and a rate‐limiting factor in PTI responses, BIK1 accumulation and activation are tightly regulated (Zhang et al, 2010, 2018; Kadota et al, 2014; Monaghan et al, 2014; Couto et al, 2016; Liang et al, 2016; Wang et al, 2018). The current model of BIK1 regulation postulates the existence of two BIK1 pools: “non‐activated” (before PAMP treatment) and “activated” (after PAMP treatment), with non‐activated BIK1 associated or not to PRR complexes. CPK28 does not associate with FLS2, but constitutively associates with BIK1 (Monaghan et al, 2014). In this scenario, PRR‐associated BIK1 is protected from degradation by PRR‐associated G protein complexes, while free BIK1 is preferentially phosphorylated by CPK28 and subsequently targeted for degradation by the E3 ubiquitin ligases PUB25/26 (Wang et al, 2018). Notably, BIK1 accumulation is higher in cpk28 mutant plants compared with pub25 pub26 mutant plants, suggesting that other E3 ubiquitin ligase(s) are involved in BIK1 degradation (Wang et al, 2018). In this study, we found that PUB4 directly ubiquitinates and promotes degradation of non‐active BIK1, which correlates with the fact that PUB4 associates with BIK1, but not with FLS2 under basal conditions. It is noteworthy that PUB25/26 are negative regulators of immune activation and promote the degradation of non‐activated BIK1, while PUB4 has a similar effect before PAMP treatment, but promotes the accumulation of active BIK1 after PAMP treatment. Plants that over‐express PUB4 accumulate less BIK1 before PAMP treatment; therefore, rapid immune responses such as ROS production would be affected, reflecting the role of PUB4 as a negative regulator before PAMP treatment. However, pub4 mutants show a reduced accumulation of activated BIK1 after PAMP treatment, which explains why immune responses are down‐regulated; this demonstrates that, after PAMP treatment, PUB4 indeed acts as a positive regulator of immunity. This represents two distinct modes of BIK1 regulation executed by PUB4 and PUB25/26.

We found that PUB4 is required for the accumulation of wild‐type levels of activated BIK1, which explains the reduction of PTI responses observed in pub4 mutant plants. Notably, BIK1 phosphorylation in response to PAMP treatment is not affected in pub4 plants, suggesting that PUB4 prevents the degradation of activated BIK1. The dual role of PUB4 in the regulation of BIK1 could be explained by the existence of two pools of PUB4: “pre‐elicitation” and “post‐elicitation”. In accordance with the fact that PUB4, as well as other E3 ligases, have been found to be rapidly phosphorylated after PAMP treatment (Benschop et al, 2007; Trujillo, 2021), we demonstrated that PUB4 is a highly phosphorylated protein, and that its phosphorylation status correlates with its function in immunity. Moreover, PAMP treatment leads to an increase in PUB4 accumulation, which promotes its role in the stabilization of activated BIK1. Both accumulation of PUB4 and phosphorylation of residues in the hinge region after immunostimulation are reminiscent of the stabilization mechanism first described for PUB22, in which phosphorylation inhibits autoubiquitination and degradation (Furlan et al, 2017), and may also contribute to regulate association to E2s (Kowarschik et al, 2018; Turek et al, 2018). While we reveal here that PUB4 positively regulates activated BIK1 accumulation, the E3 ubiquitin ligases RING‐H2 FINGER A3A (RHA3A) and RHA3B were recently shown to promote BIK1 activation (Ma et al, 2020), thus illustrating that distinct BIK1 ubiquitination events positively regulate both its accumulation and activation.

Supporting its important role in the regulation of plant immunity, we found that PUB4 is targeted by the R. solanacearum T3E RipAC. RipAC does not disrupt PUB4 association with FLS2‐BAK1‐BIK1 complex, but causes a reduction in PUB4 accumulation after PAMP treatment, thus diminishing PUB4 “post‐elicitation” pool that positively regulates BIK1 and PTI. This could explain our observation that PUB4 overexpression diminishes the impact of RipAC in plant susceptibility to R. solanacearum. Moreover, RipAC manipulates PUB4 phosphorylation state with the main tendency to promote phosphorylation of “pre‐elicitation” PUB4. We found that RipAC dramatically reduces BIK1 accumulation without interacting with BIK1. Considering the effect of PUB4 on the accumulation of non‐activated BIK1, our results suggest that RipAC exploits PUB4 and promotes accumulation of the “pre‐elicitation” form of PUB4 to reduce BIK1 accumulation. Interestingly, although pub4 mutants were more susceptible to non‐pathogenic P. syringae, PUB4 was found to promote disease symptoms caused by R. solanacearum infection. The opposite results observed in the genetic analysis of the role of PUB4 in resistance against Pto and R. solanacearum could be due to specific requirements for PUB4 in various plant tissues. This seems however unlikely considering that pub4 mutants also show reduced flg22‐triggered ROS in root tissues (Appendix Fig S6), suggesting a positive role of PUB4 in the regulation of early PTI responses also in roots. However, lack of PUB4 in leaves leads to inability of plants to close stomata in response to pathogen attack, which is an important factor contributing to virulence of Pto DC3000 COR⁻ and Pto DC3000 ΔhrcC. pub4 plants have also recently been shown to exhibit elevated levels of the defence‐related hormone salicylic acid (SA), and enhanced expression of defence genes (Desaki et al, 2019). Although one would expect such SA disbalance to affect similarly both Pto and R. solanacearum, which are both hemi‐biotrophic pathogens, we cannot fully exclude the possibility that R. solanacearum and Pto are affected differently by increased SA levels, especially considering the different tissues infected by these bacteria. Our data show that RipAC dually affects PUB4: by attenuating its positive role in immunity and by promoting its negative role in accumulation of non‐activated BIK1. The latter makes PUB4 an important susceptibility factor required for optimal R. solanacearum infection. This could explain why PUB4 promotes R. solanacearum infection, despite playing a positive role in PTI and resistance against Pto.

Future work should determine the molecular mechanisms by which PUB4 phosphorylation affects PUB4 interaction with other proteins and its E3 ubiquitin ligase activity during immunity.

Materials and Methods

Resources

All the biological and chemical materials used in this study are summarized in Appendix Table S1. Materials generated in this study are available from the corresponding authors upon request.

Arabidopsis thaliana

Arabidopsis thaliana materials in this study are derived from ecotype Columbia (Col‐0). Previously published lines include: NP::BIK1‐HA (Zhang et al, 2010), 35S::PUB4‐FLAG (Wang et al, 2013), RipAC transgenic lines (RipAC #3 and RipAC #31) (Yu et al, 2020). The T‐DNA insertion lines pub4‐1 (SALK_108269) and pub4‐3 (SAIL_859_H05) (Wang et al, 2013) were obtained from the Nottingham Arabidopsis Stock Centre (NASC), and homozygous lines were selected by genotyping using allele‐specific primers.

In the experiments with Arabidopsis seedlings in 1/2 MS media, the seedlings were kept on 1/2 MS plates in a growth chamber (22°C, 16 h light/8 h dark, 100–150 mE m⁻² s⁻¹) for germination and growth for 5 days, then transferred to 1/2 MS liquid culture for additional 7–9 days. For PAMP‐triggered ROS burst assays, Pseudomonas syringae and Ralstonia solanacearum infection assays, Arabidopsis plants were grown in either soil or jiffy pots (Jiffy International, Kristiansand, Norway) in a short day chamber (22°C, 10 h light/14 h dark photoperiod, 100–150 mE m⁻² s⁻¹, 65% humidity) for 4–5 weeks. After soil drenching inoculation, the plants were transferred to a growth chamber controlled with the following conditions: 75% humidity, 12 h light, 130 mE m⁻² s⁻¹, 27°C, and 12 h darkness, 26°C for disease symptom scoring.

For experiments in Fig 8A, pub4‐1 ^−/− BIK1‐HA ^+/−, PUB4‐FLAG ^+/− BIK1‐HA ^+/−, and BIK1‐HA ^+/− plants were initially grown on ½ MS plates with hygromycin to ensure the presence of BIK1‐HA transgene, and PUB4‐FLAG ^+/− BIK1‐HA ^+/− plants were grown additionally on ppt to ensure the presence of PUB4‐FLAG transgene. Then, plants were transferred to pots and grown for 4 weeks in short day. Pub4‐1 ^−/− BIK1‐HA ^+/− plants were screened by PCR for pub4‐1 ^−/− homozygous plants. Later the plants were screened by western blot to differentiate the BIK1‐HA hemizygous from the homozygous plants as the level of BIK1 protein expression differs significantly.

Nicotiana benthamiana

Nicotiana benthamiana plants were cultivated at 22°C in a walk‐in chamber under 16 h light/8 h dark cycle and a light intensity of 100–150 mE m⁻² s⁻¹.

Solanum lycopersicum

Tomato plants (Solanum lycopersicum cv. Moneymaker) were cultivated in jiffy pots (Jiffy International, Kristiansand, Norway) in controlled growth chambers (25°C, 16 h light/8 h dark photoperiod, 130 mE m⁻² s⁻¹, 65% humidity) for 4 weeks. After soil drenching inoculation, the plants were kept in a growth chamber under the following conditions: 75% humidity, 12 h light, 130 mE m⁻² s⁻¹, 27°C, and 12 h darkness, 26°C for disease symptom scoring. To grow tomato seedlings in vitro, tomato seeds were surface‐sterilized by soaking in 5% (v/v) sodium hypochlorite for 5 min, washed 4–5 times with distilled sterile water, and shaken slowly in sterile water overnight to facilitate germination. Then, seeds were germinated on half‐strength of Murashige and Skoog medium without sucrose (2.21 g/l MS, 0.8% w/v agar) for 3–4 days at 25°C, in darkness.

Bacterial strains

Pseudomonas syringae pv. tomato (Pto) DC3000 strains, including Pto containing an empty vector (EV), or the type‐three secretion system (T3SS)‐defective mutant ΔhrcC, or coronatine‐defective mutant COR ⁻ were cultured overnight at 28°C in LB medium containing 25 μg ml⁻¹ rifampicin, and 25 μg ml⁻¹ kanamycin.

Ralstonia solanacearum GMI1000 wild‐type strain was grown overnight at 28°C in complete BG liquid medium (Plener et al, 2012).

Agrobacterium tumefaciens GV3101 and Agrobacterium rhizogenes MSU440 with different constructs were cultured grown at 28°C on LB agar media with appropriate antibiotics. The concentration for each antibiotic is 25 μg ml⁻¹ rifampicin, 50 μg ml⁻¹ gentamicin, 50 μg ml⁻¹ kanamycin, and 50 μg ml⁻¹ spectinomycin.

Constructs and transgenic plants

The primers used to generate constructs in this work are listed in Appendix Table S2.

To generate constructs for co‐immunoprecipitation and split‐luciferase complementation (Split‐LUC) assays, coding sequences were amplified by PCR using cDNA as template and inserted into corresponding destination vectors by either gateway cloning, golden‐gate cloning system or In‐fusion cloning. The recombinant construct was transformed into A. tumefaciens GV3101.

Arabidopsis lines generated in this study include: 35S::RipAC‐GFP/NP::BIK1‐HA, 35S::RipAC‐GFP/35S::PUB4‐FLAG, 35S::PUB4‐FLAG/NP::BIK1‐HA, pub4‐1/NP::BIK1‐HA. These lines were generated by crossing, and confirmed by allele‐specific primers, antibiotics screening, and/or western blotting.

Pseudomonas syringae infection assays

For Pto inoculation, different Pto strains were resuspended in water at OD₆₀₀ = 0.1 (5 × 10⁷ CFU mL⁻¹). Before spraying, final concentration of 0.02% Silwett L‐77 was added to the inoculum. The bacterial suspensions were sprayed on 3‐ to 4‐week‐old Arabidopsis leaves. Bacterial numbers were determined 3 days post‐inoculation (dpi). The whole plants were harvested in Eppendorf tubes and weighed.

Ralstonia solanacearum infection assays

For R. solanacearum soil drenching inoculation, 15 four‐to‐five‐week‐old Arabidopsis plants per genotype or 12 tomato plants (grown in Jiffy pots) were inoculated by soil drenching with a bacterial suspension containing 10⁸ colony‐forming units per mL (CFU ml⁻¹) as previously described (Yu et al, 2020). Briefly, 300 ml of inoculum of GMI1000 strain was poured to soak each treatment. Plants were transferred from the bacterial solution to a bed of potting mixture soil in a new tray (Vailleau et al, 2007) after 20‐min incubation with the bacterial inoculum. To score the visual disease symptoms, a scale ranging from “0” (no symptoms) to “4” (complete wilting) was performed as previously described (Vailleau et al, 2007).

ROS production assays

ROS measurements were performed in Arabidopsis plants as described previously (Sang & Macho, 2017). Plant leaf discs were collected and floated on sterile water overnight in a 96‐well plate. The water was then removed and replaced with 100 μl of eliciting solution containing 17 mg ml⁻¹ luminol (Sigma Aldrich), 200 μg ml⁻¹ horseradish peroxidase (Sigma Aldrich), and an appropriate concentration of the desired PAMP: 50 nM flg22^Pto, 100–200 μg ml⁻¹ chitin, 100 nM elf18^Pto, 100 nM elf18^Rsol, 100 nM flg22^Paer or 100 nM elf18^Ecol. ROS assays in roots were performed as previously described (Wei et al, 2018). Briefly, Arabidopsis seeds were first germinated on 1/2 MS solid medium for 7 days and then transferred to 1/2 MS liquid culture for 5 days before ROS measurement. Roots of seedlings were cut into one‐centimetre‐long sections and allowed to recover for 5 h in 96‐well plates with 100 μl H₂O in each well. Sixteen root sections were analysed for each sample, using 100 nM of flg22^Pto. For root assays, luminol was replaced by the more sensitive derivative L‐012 (Wako Chemical, Japan). The luminescence was measured over 60 min using a Microplate luminescence reader (Varioskan flash, Thermo Scientific, USA) or a charge‐coupled device camera (Photek Ltd., East Sussex UK).

Seedling growth inhibition

Sterile Arabidopsis seeds were sown on MS 1% sucrose agar plates. The seeds were stratified in the dark at 4°C for 3–4 days and then transferred to light (LD). After 4 days, one seedling per well was transferred to a 48‐well plate containing 500 μl of sterile liquid MS 1% sucrose supplemented either with water (as mock treatment) or PAMP (100 nM flg22 or 100 nM elf18). Twelve seedlings per condition were used. Seedlings were transferred back to light for 10 days. Fresh weight of each seedling after blotting dry was recorded.

Stomatal closure assays

Leaf discs (two leaf discs per plant, four plants per line) were taken from 4‐ to 5‐week‐old plants grown on soil and incubated in stomatal opening buffer (10 mM MES‐KOH, pH 6.1; 50 mM KCl; 10 μM CaCl₂; 0.01% Tween‐20) for 2 h in a plant growth cabinet in the light. Subsequently, 10 μM flg22 or mock; 1 mg ml⁻¹ chitin or mock; 10 μM ABA or mock were added, and samples were incubated under the same conditions for another 2–3 h. Photographs of the abaxial leaf surface were taken using a Leica DM5500 microscope equipped with a Leica DFC450 camera. Width and length of the stomatal openings were determined using Image J software and stomatal aperture is shown as ratio of width divided by length. Values are mean ± SE (n > 120). Samples were analysed by one‐way ANOVA, Kruskal–Wallis test and Dunn's multiple comparisons test.

Transient expression in N. benthamiana

For Split‐LUC, FRET‐FLIM, and co‐immunoprecipitation assays, A. tumefaciens GV3101 or AGL1 (for epiGreen‐35S::PUB4‐GFP) carrying desired constructs were infiltrated into leaves of 5‐week‐old N. benthamiana. The OD600 used was 0.5 for each strain in all the assays, except for Split‐LUC assays, for which we used OD600 = 0.2. A. tumefaciens was incubated in the infiltration buffer (10 mM MgCl₂, 10 mM MES pH 5.6, and 150 μM acetosyringone) at room temperature for 2 h. Samples were collected 2 or 3 days after infiltration.

Ubigate cloning, bacterial proteins expression and purification

The operons used for BIK1 ubiquitination assay were generated according to Kowarschik et al (2018). Level 0 cloning of PUB4 full‐length coding sequence was generated by PCR (primers reported in Appendix Table S2). Level 0 modules were then subcloned by BsaI restriction‐ligation into positional level 1 vectors providing MBP N‐terminal tag. His‐Ub, UBA1 (E1), HA‐UBC8 (E2) level 1s were already available (Kowarschik et al, 2018). Level 2 constructs were assembled from level 1 plasmids using a BpiI restriction‐ligation reaction in the pET‐28GG destination vector generating the polycistronic operons. BIK1 and BIK1 kinase dead were cloned in pGEX‐4T‐1.

Chemically competent BL21 (DE3) pLysS E. coli cells were transformed by heat shock. 80 fmol of each construct were used for co‐transformation. Transformed cells were incubated overnight shaking at 37°C in LB‐broth supplemented with the appropriate antibiotic (80% of the normal concentration were used for co‐transformation). 50 ml of LB‐medium were inoculated with an overnight culture and grown at 37°C shaking until OD₆₀₀ = 0.8. Expression was induced with 25 μM isopropyl β‐D‐1‐thiogalactopyranoside (IPTG) and grown overnight at 18°C. Cells were harvested by centrifugation 20 min 4,000 g at 4°C, and pellets were stored at −20°C until further analysis.

Cell lysis and immobilized metal ion chromatography (IMAC) were performed according to the manufacturers' instructions. Briefly, purification buffer without imidazole (50 mM NaH₂PO₄, 300 mM NaCl pH8, and 8 M urea) was used for denaturing conditions. Cells were resuspended in the denaturing buffer supplemented with 1 mM PMSF and 2.5 mM DTT, incubated in ice for 30 min, lysed by sonication, cleared by centrifugation at 11,000 g for 30 min at 4°C and analysed directly (input), or after IMAC (1 h incubation at room temperature). Additional purification of ubiquitinated GST‐BIK1 and BIK‐KD was carried out using glutathione resin (Protino glutathione‐agarose 4B) after elution with buffer: 50 mM NaH₂PO₄, 300 mM NaCl, 250 mM imidazole pH 8.

Proteins were resolved by SDS–PAGE using 8% or 10% Tris‐glycine acrylamide gels, as indicated, and blotted onto a PVDF membrane. Proteins were detected using the following antibodies: anti‐MBP, anti‐His, and anti‐GST.

In vitro pull‐down assay

Recombinant proteins were expressed in E. coli BL21 (DE3) Rosetta pLysS cells. GST‐BIK1 was expressed in the pGEX4T1 vector (Kadota et al, 2014); 6xHis‐MBP‐PUB4 was expressed in the pOPINM vector. Approximately 6 μg of each of GST‐BIK1 and 6xHis‐MBP or 6xHis‐MBP‐PUB4 proteins were combined in buffer (25 mM Tris‐Cl pH 7.4, 100 mM NaCl, 0.2% Triton‐X, 1 mM DTT). Input samples were taken for analysis before addition of amylose resin (NEB) and incubation at room temperature for 30 min. The resin was washed three times with buffer and bound proteins eluted with SDS loading dye. Proteins were separated by SDS–PAGE, transferred to PVDF membrane, and detected using anti‐MBP (NEB #E8032) or anti‐GST (Upstate #06332) antibodies, both used at 1:10,000 dilution in TBST‐5% non‐fat milk powder.

Protein extraction and western blot assays

For protein extraction in plant tissues, 12 fourteen‐day‐old Arabidopsis seedlings or leaf discs (diameter = 18 mm) from N. benthamiana were frozen in liquid nitrogen and ground with a Tissue Lyser (QIAGEN, Hilden, Nordrhein‐Westfalen, Germany). Protein samples were extracted with buffer containing 100 mM Tris (pH 8), 150 mM NaCl, 10% Glycerol, 1% IGEPAL, (5 mM EDTA, optional), 5 mM DTT, 1% Protease inhibitor cocktail, 2 mM PMSF, 10 mM sodium molybdate, 10 mM sodium fluoride, and 2 mM sodium orthovanadate. The resulting protein samples were boiled at 70°C for 10 min in Laemmli buffer and loaded in SDS–PAGE acrylamide gels for western blot. Alternatively, proteins from leave disks of 4‐ to 5‐week‐old Arabidopsis plants were extracted by adding 2× SDS buffer and heating at 70°C for 10 min. Western blot detection was done using either Pierce™ ECL western blotting substrate or SuperSignal™ West Femto Maximum Sensitivity Substrate (ThermoFisher). All the immunoblots were probed using appropriate antibodies as indicated in the figures. Molecular weight (kDa) marker bands are indicated for reference. Western blot quantification was performed using ImageJ software (http://imagej.net/). For enrichment of membrane‐associated proteins, Minute™ Plasma Membrane Protein Isolation Kit for Plants (Invent Biotechnologies) was used.

MAPK activation assays

PAMP‐triggered MAPK activation was evaluated as previously described (Macho et al, 2012). Briefly, 2 twelve‐day‐old Arabidopsis seedlings were treated with 100 nM flg22 and samples were collected at different time points. After protein extraction, the protein samples were loaded in 10% SDS–PAGE gels and the western blots were analysed with anti‐pMAPK antibodies. Blots were stained with Coomassie Brilliant Blue (CBB) to verify equal loading.

Co‐immunoprecipitation

Co‐immunoprecipitation assays were performed as previously described (Kadota et al, 2016; Sang et al, 2018), with some changes. One gram of N. benthamiana leaf tissues was collected at 2 days after agro‐infiltration and frozen in liquid nitrogen. Sterilized Arabidopsis seeds were grown on MS 1% sucrose agar plates for one week. Then, the seedlings were transferred into 6‐well plates containing liquid MS; 5 seedlings per well. Two‐week‐old seedlings from two 6‐well plates were treated by elicitor (1 μM elf18 or/and flg22) or MS medium (as mock treatment) for 10 min, including 2 min vacuum infiltration. Seedlings were frozen and then ground in liquid nitrogen. Proteins were extracted by adding 2 volumes of the following extraction buffer to 1 volume of grounded tissue: (50 mM Tris‐HCl, pH 7.5, 150 mM NaCl, 10% glycerol, 2.5 mM NaF, 2 mM NaMo, 1.5 mM activated Na3VO4, 5 mM DTT, 1% IGEPAL CA‐630, 1× protease inhibitor cocktail 1 [Sigma Aldrich], and 1 mM PMSF) for 30 min at 4°C with rotation. Samples were centrifuged at 4°C 4,200 g for 20 min and plant extract was passed through 4× Miracloth. Inputs were taken and kept on ice. 40 μM of anti‐FLAG (or anti‐GFP) protein beads equilibrated in extraction buffer were added to the rest of the extract. Protein immuno‐precipitation was carried out for 2 h at 4°C with rotation. The beads were collected by centrifugation at 200 g for 2 min at 4°C and washed three times in extraction buffer and twice in extraction buffer with 0.5% IGEPAL CA630. Fifty microlitres of 2× SDS buffer was added to each sample; the corresponding amount of 4× SDS buffer was added to input samples and all samples were heated at 70°C for 10 min. Samples were either straight loaded on the gel or kept at −20.

LC‐MS/MS analysis

35S::PUB4‐FLAG or Col‐0 plants were grown the same way as for CoIP procedure. Ten millilitres of grounded plant tissue was used for one IP. The IP procedure was the same as described above for CoIP. Samples were run on SDS‐polyacrylamide gel, and gel was stained with Coomassie Brilliant Blue (Simply Blue™ Safe stain, Invitrogen). Afterwards, each gel line was cut in several pieces, and these pieces were kept in separate 2 ml low protein binding tubes (Eppendorf). Afterwards, the samples were processed as described previously (Bender et al, 2017). Briefly, gel slices were de‐stained in 50% acetonitrile and incubated for 45 min in 10 mM DTT. Cysteinyl residue alkylation was carried out for 30 min in the darkness in 55 mM chloroacetamide. After several washes with 25 mM ammonium bicarbonate, 50% acetonitrile gel slices were dehydrated in 100% acetonitrile. Gel pieces were rehydrated with 50 mM ammonium bicarbonate and 5% acetonitrile containing 20 ng μl⁻¹ trypsin (Pierce), and digestion was performed overnight at 37°C. Tryptic peptides were sonicated from the gel in 5% formic acid, 50% acetonitrile, and the total extracts were evaporated until dry.

LC‐MS/MS analysis was performed using an Orbitrap Fusion trihybrid mass spectrometre (Thermo Fisher Scientific) and a nanoflow‐UHPLC system (Dionex Ultimate3000, Thermo Fisher Scientific). Peptides were trapped to a reverse phase trap column (Acclaim PepMap C18, 5 μm, 100 μm × 2 cm, Thermo Fisher Scientific) connected to an analytical column (Acclaim PepMan 100, C18 3 μm, 75 μm × 50 cm, Thermo Fisher Scientific). Peptides were eluted in a gradient of 3–30% acetonitrile in 0.1% formic acid (solvent B) over 50 min followed by gradient of 30–80% B over 6 min at a flow rate of 300 nl min⁻¹ at 40°C. The mass spectrometre was operated in positive ion mode with nano‐electrospray ion source with an inner diameter of 0.02 mm fused silica emitter (New Objective). Voltage 2200 V was applied via platinum wire held in PEEK T‐shaped coupling union with transfer capillary temperature set to 275°C. The Orbitrap, MS scan resolution of 120,000 at 400 m/z, range 300–1,800 m/z was used, and automatic gain control was set to 2 × 10⁵ and maximum inject time to 50 ms. In the linear ion trap, product ion spectra were triggered with a data‐dependent acquisition method using “top speed” and “most intense ion” settings. The threshold for collision‐induced dissociation (CID) and high energy collisional dissociation (HCD) was set using the Universal Method (above 100 counts, rapid scan rate, and maximum inject time to 10 ms). The selected precursor ions were fragmented sequentially in both the ion trap using CID and in the HCD cell. Dynamic exclusion was set to 30 s. Charge state allowed between +2 and +7 charge states to be selected for MS/MS fragmentation.

Mascot generic files (.mgf files) were generated from raw data using MSConvert package (Matrix Science) and were searched on Mascot server version 2.4.1 (Matrix Science) against TAIR (version 10) database, a separate in‐house constructs database and an in‐house contaminants database. Tryptic peptides with up to two possible mis‐cleavages and charge states +2, +3, +4 were allowed in the search. The following modifications were included in the search: oxidized Met, phosphorylation on Ser, Thr, Tyr as variable modifications, and carbamidomethylated Cys as a static modification. Data were searched with a monoisotopic precursor and fragment ions mass tolerance 10 ppm and 0.6 Da, respectively. Mascot results were combined in Scaffold version 4 (Proteome Software) and exported to Excel (Microsoft Office).

Parallel reaction monitoring analyses

Parallel reaction monitoring was performed as described in Guo et al (2020). Briefly, phospho‐peptides were targeted to measure PUB4 phosphorylation at indicated residues (Appendix Table S3). The PRM assay also included a selection of non‐modified control peptides (Appendix Table S3) to measure PUB4 protein levels. These were used to normalize the measured changes in phosphorylation relative to PUB4 protein levels. For some phospho‐peptides transitions did not identify the specific site of phosphorylation and could only be narrowed down regions (Fig 8B–E; Appendix Table S3). The assay was performed three times for each of two biological replicates and results averaged ± SE.

Yeast two‐hybrid screen

Yeast two‐hybrid screening was conducted by Hybrigenics Services, S.A.S., Paris, France (http://www.hybrigenics‐services.com). The RipAC coding region from R. solanacearum GMI1000 was PCR‐amplified and inserted into pB29 as a N‐terminal fusion to LexA DNA‐binding domain (RipAC‐LexA). The construct was confirmed by sequencing the full‐length RipAC and used as a bait to screen a random‐primed Tomato Infected Roots cDNA library constructed into pP6. pB29 and pP6 derive from the original pBTM116 (Vojtek & Hollenberg, 1995) and pGADGH (Bartel et al, 1993) plasmids, respectively. 133 million clones (more than 10‐fold the complexity of the library) were screened using a mating approach with YHGX13 (Y187 ade2‐101::loxP‐kanMX‐loxP, matα) and L40ΔGal4 (matα) yeast strains as previously described (Fromont‐Racine et al, 1997). Twenty‐six His+ colonies were selected on a medium lacking tryptophan, leucine and histidine. The prey fragments of the positive clones were amplified by PCR and sequenced at their 5′ and 3′ junctions. The resulting sequences were subjected to corresponding interacting proteins analysis in the GenBank database (NCBI) using a fully automated procedure.

Split‐LUC assays

Split‐LUC assays were performed as previously described (Chen et al, 2008; Yu et al, 2020). Generally, A. tumefaciens strains containing the desired plasmids were infiltrated into N. benthamiana leaves. Split‐LUC assays were conducted both qualitatively and quantitatively after 2 dpi. For the CCD imaging, the leaves were infiltrated with 0.5 mM luciferin in water and kept in the dark for 5 min before CCD imaging. The images were taken with either Lumazone 1300B (Scientific Instrument, West Palm Beach, FL, USA). To perform the quantification of the luciferase signal, leaf discs (diameter = 4 mm) were taken into a 96‐well microplate (PerkinElmer, Waltham, MA, US) with 100 μl H₂O. Then the leaf discs were incubated with 100 μl water with 0.5 mM luciferin in a 96‐well plate wrapped with foil paper to remove the background luminescence for 5 min, and the luminescence was recorded with a Microplate luminescence reader (Varioskan flash, Thermo Scientific, USA). Each data point contains at least eight replicates. The protein accumulation was determined by immunoblot as described above.

FRET‐FLIM

Förster resonance energy transfer – fluorescence lifetime imaging (FRET‐FLIM) experiments were performed as previously described (Rosas‐Diaz et al, 2018; Xian et al, 2020) with several modifications. Briefly, AtPUB4 and SlPUB4 (fused to GFP) were expressed from epiGreen‐35S and pGWB505, respectively; and RipAC (fused to RFP) was expressed from pGWB554. FRET‐FLIM experiments were performed on a Leica TCS SMD FLCS confocal microscope excitation with WLL (white light laser) and emission collected by a SMD SPAD (single photon‐sensitive avalanche photodiodes) detector. Two days after infiltration, N. benthamiana plants transiently coexpressing donor and acceptor proteins were visualized under the microscope. Accumulation of the GFP‐ and RFP‐tagged proteins was estimated before measuring lifetime. The tuneable WLL set at 488 nm with a pulsed frequency of 40 MHz was used for excitation, and emission was detected using SMD GFP/RFP Filter Cube (with GFP: 500–550 nm). The fluorescence lifetime shown in the figures corresponding to the average fluorescence lifetime of the donor was collected and analysed by PicoQuant SymphoTime software. Lifetime is normally amplitude‐weighted mean value using the data from the single (GFP‐fused donor protein only or GFP‐fused donor protein with free RFP acceptor or with non‐interacting RFP‐fused acceptor protein) or biexponential fit (GFP‐fused donor protein interacting with RFP‐fused acceptor protein). Mean lifetimes are presented as mean ± SEM based on eight images from three independent experiments.

Tomato root transformation

Tomato root transformation was performed as previously described (Morcillo et al, 2020). Briefly, the radicle and bottom part of the hypocotyl of four‐days‐old tomato seedlings were removed and cut seedlings were immersed on Agrobacterium rhizogenes MSU440 bacterial mass containing pUBIcGFP‐DR::SlPUB4 for overexpression and pK7GWIWG2_II‐RedRoot::SlPUB4 for RNAi silencing (pUBIcGFP‐DR and pK7GWIWG2_II‐RedRoot empty vector were used as control). Seedlings inoculated with A. rhizogenes were kept on half‐strength MS medium (0.8% agar) and covered with filter paper to maintain humidity. During the following weeks, roots were screened and selected by DsRed fluorescence visualization, using In Vivo Plant Imaging System NightShade LB 985 (Berthold Technologies). Subsequent plant handling and R. solanacearum inoculation was performed as previously described (Morcillo et al, 2020).

Quantification and statistical analysis

Statistical analyses were performed with Prism 7 software (GraphPad). The data are presented as mean ± SE. The statistical analysis methods are described in the figure legends.

Author contributions

Gang Yu: Conceptualization; data curation; formal analysis; investigation; methodology; writing—review and editing. Maria Derkacheva: Conceptualization; data curation; formal analysis; funding acquisition; investigation; methodology; writing—original draft; project administration; writing—review and editing. Jose S Rufian: Conceptualization; data curation; formal analysis; investigation; methodology. Carla Brillada: Investigation; methodology. Kathrin Kowarschik: Investigation; methodology. Shushu Jiang: Investigation; methodology. Paul Derbyshire: Resources; data curation; investigation; methodology. Miaomiao Ma: Resources. Thomas A DeFalco: Investigation; methodology; writing—review and editing. Rafael J L Morcillo: Investigation; methodology. Lena Stransfeld: Investigation. Yali Wei: Investigation; methodology. Jian‐Min Zhou: Resources; supervision. Frank L H Menke: Resources; data curation; formal analysis; supervision; visualization; methodology. Marco Trujillo: Conceptualization; resources; data curation; formal analysis; supervision. Cyril Zipfel: Conceptualization; data curation; supervision; funding acquisition; writing—original draft; project administration; writing—review and editing. Alberto P Macho: Conceptualization; data curation; supervision; funding acquisition; writing—original draft; project administration; writing—review and editing.

Disclosure and competing interests statement

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Source Data for Expanded View

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 5

Source Data for Figure 6

Source Data for Figure 7

Source Data for Figure 8

The EMBO Journal (2022) 41: e107257

Data availability

The mass spectrometry proteomics data produced in this study are available in the following databases:

• Protein interaction AP‐MS data: PRIDE PXD036739 (http://www.ebi.ac.uk/pride/archive/projects/PXD036739).

• Phosphoproteomic PRM data: Panorama Public PXD036761 (https://panoramaweb.org/LbxfZG.url).