CRK2 and C-terminal Phosphorylation of NADPH Oxidase RBOHD Regulate Reactive Oxygen Species Production in Arabidopsis[OPEN]

CRK2 associates with and activates RBOHD to trigger MAMP-induced ROS production, revealing a regulatory mechanism for plant NADPH oxidases through phosphorylation of the C terminus. Reactive oxygen species (ROS) are important messengers in eukaryotic organisms, and their production is tightly controlled. Active extracellular ROS production by NADPH oxidases in plants is triggered by receptor-like protein kinase-dependent signaling networks. Here, we show that CYSTEINE-RICH RLK2 (CRK2) kinase activity is required for plant growth and CRK2 exists in a preformed complex with the NADPH oxidase RESPIRATORY BURST OXIDASE HOMOLOG D (RBOHD) in Arabidopsis (Arabidopsis thaliana). Functional CRK2 is required for the full elicitor-induced ROS burst, and consequently the crk2 mutant is impaired in defense against the bacterial pathogen Pseudomonas syringae pv tomato DC3000. Our work demonstrates that CRK2 regulates plant innate immunity. We identified in vitro CRK2-dependent phosphorylation sites in the C-terminal region of RBOHD. Phosphorylation of S703 RBOHD is enhanced upon flg22 treatment, and substitution of S703 with Ala reduced ROS production in Arabidopsis. Phylogenetic analysis suggests that phospho-sites in the C-terminal region of RBOHD are conserved throughout the plant lineage and between animals and plants. We propose that regulation of NADPH oxidase activity by phosphorylation of the C-terminal region might be an ancient mechanism and that CRK2 is an important element in regulating microbe-associated molecular pattern-triggered ROS production.


INTRODUCTION
Plants are continuously confronted with stimuli from the surrounding environment, including abiotic cues and invading pathogens. Plant cells also perceive a plethora of signals from neighboring cells and distant tissues. Numerous plasma membrane proteins are involved in the meticulous monitoring and transduction of signals for inter-and intracellular communication.
A common early feature of many cellular responses to various environmental changes involves the production of reactive oxygen species (ROS; Kimura et al., 2017;Waszczak et al., 2018). While ROS are an inevitable byproduct of aerobic metabolism and their unrestricted accumulation can have deleterious consequences (Waszczak et al., 2018), ROS are also ubiquitous signaling molecules in plants and animals (Suzuki et al., 2011;Waszczak et al., 2018). Eukaryotic cells produce ROS in several subcellular compartments as well as the extracellular space, in plants referred to as the apoplast (Kimura et al., 2017;Waszczak et al., 2018). A major component in the production of extracellular ROS is the evolutionarily conserved NADPH oxidase (NOX) family (Meitzler et al., 2014;Kimura et al., 2017). NOX-dependent ROS production is involved in regulation of immune functions, cell growth, and apoptosis in animals and plants (Jiménez-Quesada et al., 2016;Waszczak et al., 2018).
Plant NOXs, referred to as respiratory burst oxidase homologs (RBOHs), have been identified as homologs of phagocyte gp91 phox /NOX2, which contains six transmembrane helices and a C-terminal NADPH-and FAD-binding cytoplasmic region (Torres et al., 2002). Unlike gp91 phox /NOX2, RBOHs contain an additional N-terminal region with Ca 21 binding EF hands, similar to nonphagocytic NOXs, such as NOX5 (Suzuki et al., 2011). RBOH activity is strictly controlled to avoid damaging consequences of unrestricted ROS production (Suzuki et al., 2011). Arabidopsis (Arabidopsis thaliana) RBOHD is the best-characterized RBOH and is involved in biotic and abiotic stress responses (Torres et al., 2002;Lee et al., 2013Lee et al., , 2018. The N-terminal region of RBOHD is phosphorylated by a variety of protein kinases, including receptorlike cytoplasmic kinases (Ogasawara et al., 2008;Kimura et al., 2012;Dubiella et al., 2013;Kadota et al., 2014;Li et al., 2014;Zhang et al., 2018;Han et al., 2019;Kaya et al., 2019), for example BOTRYTIS-INDUCED KINASE 1 (BIK1; Kadota et al., 2014;Li et al., 2014). While previous research has suggested a predominant role of phosphorylation of the N-terminal region for regulation of RBOH, phosphorylation of the C-terminal region is important for the regulation of human gp91 phox /NOX2 and NOX5 (Jagnandan et al., 2007;Raad et al., 2009). NADPH-and FADbinding sites in C terminus are highly conserved in NOXs and RBOHs, but it is unclear whether the C terminus of plant RBOHs could also be a target for regulation of the ROS producing activity.
Here, we characterize the role of CRK2 in immune signaling in response to MAMP perception. CRK2 exists in a preformed complex with RBOHD. CRK2 controls the activity of RBOHD, and functional CRK2 is required for full MAMP-induced ROS production. Importantly, we show that CRK2 phosphorylates the C-terminal region of RBOHD and modulates the ROS-production activity of RBOHD in vivo. Our results lead us to propose a novel mechanism for the regulation of RBOHD activity through phosphorylation of the C-terminal region and highlight a critical role for CRK2 in the precise control of the ROS burst in response to biotic stress.

CRK2 Kinase Activity Is Important for Plant Development
CRK2 has been previously implicated in stress responses and development in Arabidopsis (Bourdais et al., 2015). CRK2 is a typical CRK with an N-terminal signal peptide, extracellular region containing two DUF26 domains, transmembrane region, and intracellular protein kinase domain ( Figure 1A). The crk2 mutant was smaller than wild-type (Columbia-0 [Col-0]) plants ( Figure 1B; Bourdais et al., 2015) and displayed significantly reduced fresh ( Figure 1C) and dry weight (Supplemental Figure 1A). Overaccumulation of the plant hormone salicylic acid (SA) often leads to a reduction of plant size, but SA levels were not significantly different between crk2 and wild-type plants (Supplemental Figure 1B). Expression of yellow fluorescent protein (YFP)-tagged CRK2 under the control of the CRK2 promoter (CRK2pro:CRK2-YFP) in crk2 background-restored plant growth (Figures 1B,1C,and Supplemental Figure 1A). Substitution of the ATP binding Lys (K) at position 353 with Glu (E; CRK2 K353E ) or the Asp (D) at position 450 in the catalytic domain VIb (Stone and Walker, 1995) with Asn (N; CRK2 D450N ) abated the kinase activity of CRK2 in vitro (Hunter et al., 2019). Kinase-dead CRK2 K353E -YFP or CRK2 D450N -YFP under control of the CRK2 promoter was expressed in leaves and roots and displayed the same subcellular localization as wild-type CRK2-YFP (Supplemental Figures 1C and  1D) but failed to restore the growth defect of crk2 ( Figures 1B, 1C, and Supplemental Figure 1A). The amino acid substitutions did not reduce the abundance of kinase-dead CRK2-YFP (Supplemental Figure 1E). In summary, our results show that CRK2 is important for proper plant growth, and its kinase activity is crucial for this function.
CRK2 Is Required for MAMP-Triggered Responses and Resistance to Pseudomonas syringae pv tomato DC3000 Previous results suggested that ROS production triggered by flg22, a MAMP derived from bacterial flagella, is reduced in crk2 (Bourdais et al., 2015). Therefore, we tested the role of CRK2 in MAMP-induced ROS production in detail. ROS production triggered by flg22 was reduced in crk2 and reintroduction of CRK2-YFP into the mutant background restored ROS production to the same levels as in Col-0 ( Figure 2A). The flg22-induced ROS production in plants expressing CRK2 D450N -YFP was comparable to crk2 ( Figure 2B). Transcriptional upregulation of flg22 responsive genes (FLG22-INDUCED RECEPTOR-LIKE KINASE 1 [FRK1] and NDR1/HIN1-LIKE 10 [NHL10]) showed that MAMP perception was not impaired in crk2 (Supplemental Figures 2A and B). To test whether the reduced response of crk2 to flg22 was accompanied by altered pathogen susceptibility, we measured growth of the hemibiotrophic bacterial pathogen Pseudomonas syringae pv tomato DC3000 (Pto DC3000). The crk2 mutant was significantly more susceptible to the virulent pathogen compared to Col-0 ( Figure 2C). CRK2-YFP but not the kinase-dead CRK2 D450N -YFP rescued the pathogen susceptibility of crk2 ( Figure 2C). ROS production induced by chitin (Supplemental Figure 2C) or pep1 (Supplemental Figure 2D) was also reduced in crk2 compared to Col-0, suggesting that the reduced MAMP-or DAMP-triggered ROS production in crk2 is a general response and not specific to flg22. MAMP-triggered ROS production is also an important element in pathogen-triggered stomatal closure. Therefore, we tested whether the reduced flg22-triggered ROS production in crk2 also coincided with altered stomatal closure. The crk2 mutant displayed slightly reduced stomatal aperture compared to wild-type plants under control conditions. However, treatment with flg22-triggered stomatal closure in wild-type but not in crk2 plants (Supplemental Figure 2E). (C) Box plot shows the fresh weight of 21-day-old plants (n 5 10). Differences between Col-0 and transgenic lines were evaluated with one-way ANOVA with Tukey-Kramer honestly significant difference (HSD) post-hoc test, *** P < 0.001; ns, not statistically significant. The experiment was repeated three times with similar results.
To further investigate the role of CRK2 in flg22-triggered responses, we assessed Ca 21 signaling, MAPK activation, and callose deposition in crk2. Application of flg22 resulted in a rapid increase in cytosolic Ca 21 ([Ca 21 ] cyto ) levels in wild-type plants, which express the Förster resonance energy transfer (FRET)based Ca 21 -sensor YCNano-65 (Choi et al., 2014;Lenglet et al., 2017;Toyota et al., 2018). This response was reduced in the YCNano65/crk2 background (Supplemental Figures 2F and G). Interestingly, flg22-dependent MAPK activation (Supplemental Figure 2H) was more pronounced in crk2 compared to Col-0. Callose deposition was enhanced in crk2 compared to Col-0 after 30 min of treatment with flg22 (Supplemental Figures 2I and 2J). However, flg22-triggered callose deposition was similar in Col-0 and crk2 after 12 h (Supplemental Figure 2I). Taken together, CRK2 is an essential component for mounting immune responses against Pto DC3000 in Arabidopsis, modulating extracellular ROS production, stomatal closure, Ca 21 influx, MAPK activation and early callose deposition.

CRK2 Interacts with RBOHD and Controls ROS Production
RBOHD is the main source of MAMP/DAMP-induced extracellular ROS production (Couto and Zipfel, 2016;Kimura et al., 2017) and flg22-, pep1-, and chitin-induced ROS production was significantly reduced in crk2 (Figure 2; Supplemental Figure 2). RBOH proteins, including RBOHD, are synergistically activated by protein phosphorylation and Ca 21 binding to EF-hand motifs in the N-terminal region (Kaya et al., 2019). Given the importance of CRK2 kinase activity in MAMP-induced ROS production, we investigated whether CRK2 could activate RBOHD. To test this, we used human embryonic kidney 293T (HEK293T) cells, a human cell culture that produces minimal amounts of extracellular ROS due to a lack of expression of endogenous NOXs (Ogasawara et al., 2008). HEK293T cells were cotransfected with 3FLAG-RBOHD and CRK2-3Myc or 3Myc-GFP as a control. Subsequently, RBOHD-mediated extracellular ROS production was measured by luminol-amplified chemiluminescence. Despite equal 3FLAG-RBOHD protein levels (Supplemental Figure 3A), ROS production in cells cotransfected with CRK2-3Myc and 3FLAG-RBOHD was strongly elevated compared to cells cotransfected with (A) Values represent mean 6 SEM of n $ 16. Differences compared with Col-0 were evaluated with one-way ANOVA with Tukey-Kramer HSD, ***P < 0.001; ns, not statistically significant. (B) Values represent the mean 6 SEM of n $ 19. Differences compared with Col-0 were evaluated with one-way ANOVA with Tukey-Kramer HSD, * P < 0.05, *** P < 0.001. (C) Quantitative analysis of bacterial growth in Col-0, crk2, and CRK2pro:CRK2-YFP/crk2 or CRK2pro:CRK2 D450N -YFP/crk2 following syringe infiltration with Pto DC3000 (1 3 10 5 CFU/mL). Box plot shows the numbers of bacteria in leaf discs after infiltration (0 day postinfiltration [0 DPI], n 5 3) or 2 DPI (n 5 6). Each data point contains four leaf discs. Letters indicate significant differences at P < 0.05 (one-way ANOVA with Tukey-Kramer HSD). (A) to (C) The experiment was repeated three times with similar results.
Since RBOHD can also be activated by Ca 21 , HEK293T cells were treated with ionomycin, a Ca 21 ionophore that induces a rise in cytosolic Ca 21 levels. Ionomycin-induced transient ROS production (D delta ROS, ROS T530 to ROS T531 ) in CRK2-3Myc and 3FLAG-RBOHD cotransfected cells was not different from 3Myc-GFP and 3FLAG-RBOHD cotransfected cells ( Figure 3A). Enhancement of RBOHD activity by CRK2 was not dependent on Ca 21 influx as the elevated basal ROS production activity of RBOHD cotransfected with CRK2-3Myc (ROS T50 to ROS T530 in Figure 3A) was also observed when using Ca 21 -free assay buffer (Supplemental Figures 3B and C). These results suggest that CRK2-3Myc enhanced the basal ROS-producing activity of 3FLAG-RBOHD in HEK293T cells, uncoupling it from Ca 21 dependence.
To investigate whether CRK2 and RBOHD interact in planta, we performed coimmunoprecipitation assays using rbohD plants expressing 35S:CRK2-YFP and 35S:FLAG-RBOHD. CRK2-YFP was immunoprecipitated using an anti-GFP antibody coupled to magnetic beads, and copurified RBOHD was detected using a RBOHD-specific antibody. RBOHD copurified with CRK2 ( Figure 3B) and treatment of plants with flg22 did not alter the interaction of CRK2 with RBOHD ( Figure 3C). To analyze this interaction in more detail, we performed in vitro interaction assays between the cytosolic region of CRK2 ( Figure 1A) and the cytosolic N-terminal and C-terminal regions of RBOHD ( Figure 3D). Recombinant RBOHD/N or RBOHD/C tagged with 6His and maltose binding protein (MBP; 6His-MBP-RBOHD/N, 6His-MBP-RBOHD/ C) or MBP were incubated with the cytosolic region of CRK2, which contains the kinase domain (CRK2cyto), tagged with 6His and glutathione S-transferase (GST; 6His-GST-CRK2cyto) and glutathione sepharose beads. GST pull-down assay showed that 6His-GST-CRK2cyto interacted in vitro with 6His-MBP-RBOHD/ N but intriguingly also with 6His-MBP-RBOHD/C ( Figure 3E). In summary, our results suggest that CRK2 is capable of direct interaction with RBOHD. CRK2 and RBOHD form a complex, which exists independently of flg22 perception in planta, in contrast to many other RLK-containing complexes that are formed in response to ligand binding.

CRK2 Phosphorylates RBOHD In Vitro
The kinase activity of CRK2 was essential for the full flg22triggered ROS burst in planta as well as for enhancing ROS production by RBOHD in HEK293T cells. Therefore, we tested whether CRK2 could phosphorylate RBOHD in vitro. Recombinant 6His-GST-CRK2cyto and 6His-MBP tagged RBOHD cytosolic regions (Figures 1A and 3D) were produced in Escherichia coli and affinity purified. The 6His-GST-CRK2cyto phosphorylated 6His-MBP-RBOHD/N but not MBP ( Figure 4A). Because of the similar molecular weight of 6His-GST-CRK2cyto (68.5 kD) and 6His-MBP-RBOHD/C (78.4 kD), RBOHD/C was divided into three overlapping fragments (C1, C2, and C3; Figure 3D). The results showed that the C1 and C3 fragments of 6His-MBP-RBOHD were preferentially phosphorylated by 6His-GST-CRK2cyto, while C2 displayed considerably weaker phosphorylation ( Figure 4B). Mass spectrometric analysis of in-gel trypsin-or Lys-C-digested peptides identified in vitro RBOHD phosphorylation sites targeted by CRK2cyto (Supplemental Table 1). In the N-terminal region of RBOHD, two sites targeted by CRK2 were identified (S8 and S39), while three sites (S611, S703, S862) were identified in the C-terminal region. Taken together, our results show that the Nand C-terminal regions of RBOHD are phosphorylated by CRK2 in vitro.

CRK2 Regulates RBOHD via Phosphorylation of S703 and S862
Among the RBOHD phospho-sites targeted by CRK2 in vitro, S8 and S39 have been previously described to be phosphorylated by SIK1  and BIK1 (Kadota et al., 2014;Li et al., 2014). S703 has been reported to be phosphorylated upon xylanase treatment, but no responsible kinase was identified (Benschop et al., 2007), while phosphorylation of S611 and S862 has not been described so far. In order to test whether the identified phospho-sites in RBOHD were important for the regulation of RBOHD activity, the residues S8, S39, S611, S703, and S862 were substituted with Ala to make them nonphosphorylatable. Wild-type RBOHD and phospho-site mutant constructs were transfected into HEK293T cells together with CRK2. Amino acid substitutions did not affect RBOHD protein levels (Supplemental Figures 4A and B). Substitution of S8 or S39 in the N-terminal cytoplasmic region of RBOHD did not impact ROS-producing activity compared to the wild-type protein when cotransfected with CRK2 ( Figure 5A). The 3FLAG-RBOHD S703A and CRK2-3Myc cotransfected cells showed reduced basal ROS production as compared to 3FLAG-RBOHD and CRK2-3Myc ( Figure 5B), suggesting that S703 could be a positive regulatory site for RBOHD activity. In contrast to 3FLAG-RBOHD S703A , HEK293T cells expressing 3FLAG-RBOHD S862A and CRK2-3Myc exhibited higher basal ROS production compared to 3FLAG-RBOHD and CRK2-3Myc ( Figure 5B), suggesting that S862 could act as a negative regulatory site. ROS production of 3FLAG-RBOHD S611A cotransfected with CRK2-3Myc was similar to 3FLAG-RBOHD, suggesting no regulatory role of this single site. Mutation of S703 or S862 of RBOHD did not impair Ca 21 -dependent activation of ROS production of RBOHD itself without CRK2 cotransfection (Supplemental Figures 4C and D). Taken together, our results suggest that the phosphosites in the C-terminal cytoplasmic region of RBOHD could be crucial for fine-tuning ROS production activity in HEK293T cells. (A) ROS production of RBOHD-expressing HEK293T cells. 3FLAG-RBOHD was transiently coexpressed with either 3Myc-GFP or CRK2 (wild type or D450N)-3Myc. After 30 min, 1 mM ionomycin was added to the medium (black arrow). Values represent mean 6 SEM of n 5 3. E.V., Empty vector. The experiment was repeated three times with similar results. (B) and (C) Coimmunoprecipitation analysis of interaction between RBOHD and CRK2. CRK2-YFP was immuno-precipitated using anti-GFP beads followed by immunoblotting with anti-RBOHD and anti-GFP antibodies. FLAG-RBOHD, 105 kD; CRK2-YFP, 99.9 kD; and YFP-6Myc, 36.7 kD.

Phosphorylation of S703 of RBOHD Modulates flg22-Induced ROS Production In Planta
To investigate whether RBOHD phosphorylation sites targeted by CRK2 in vitro were also phosphorylated upon flg22-treatment in planta, we performed targeted phosphoproteomic analyses of Col-0 seedlings treated with flg22 for 5 min. Phosphorylation of S8 was not significantly induced by flg22-treatment ( Figure 6A; Supplemental Figures 5A and S5B), while S39 phosphorylation was strongly enhanced ( Figure 6B and Supplemental Figure S5C). Phosphorylation of S611 and S862 could not be evaluated, as trypsin or Lys-C digestion resulted in phospho-site-containing peptides of inappropriate length for liquid chromatography-mass spectrometry (LC-MS)-based targeted analyses. However, phosphorylation of S703, which was targeted by CRK2 in vitro, and mutation of which to Ala reduced ROS production in HEK293T cells, was significantly enhanced upon flg22 treatment ( Figure 6C; Supplemental Figure 5D). In agreement with previous studies, phosphorylation of S163, S343, and S347 in RBOHD (Supplemental Figures 5E to 5G), as well as dual phosphorylation of the TEY-motif in the MAPKs MPK3, MPK6, and MPK11 (Supplemental Figures 5H to 5J), were enhanced by flg22-treatment (Kadota et al., 2015).
To test whether phosphorylation of S703 in the C-terminal region of RBOHD was mediated by CRK2 in planta, we used leaf discs of CRK2pro:CRK2 (wild type or D450N)-YFP/crk2. The basal and flg22-induced phosphorylation of S8 and S39 were not significantly altered in CRK2pro:CRK2-YFP/crk2 or the kinase-dead variant (Supplemental Figures 5K to 5M), suggesting that CRK2 might not be the main kinase phosphorylating RBOHD S8 and S39 in planta. However, flg22-induced phosphorylation of S703 in RBOHD was significantly reduced in kinase-dead CRK2pro: CRK2(D450N)-YFP/crk2 plants compared CRK2pro:CRK2-YFP/ crk2 ( Figure 6D). These results suggest that CRK2 is responsible, at least partially, for flg22-induced phosphorylation of S703 in planta.
To investigate whether phosphorylation of S703 in the C-terminal region of RBOHD also impacts RBOHD-dependent ROS production in planta, we generated transgenic plants expressing RBOHD or RBOHDS703A under the control of the RBOHD promoter (RBOHDpro:3FLAG-RBOHD) in the rbohD background. The phospho-site mutations did not alter growth or development compared with the 3FLAG-RBOHD-expressing plants (Supplemental Figure 6A). Protein amounts of 3FLAG-RBOHD or 3FLAG-RBOHDS703A were comparable in the transgenic plant lines (Supplemental Figures 6B and 6C). ROS production triggered by flg22 was significantly reduced in 3FLAG-RBOHDS703A plants compared with 3FLAG-RBOHD plants ( Figure 6E; Supplemental Figure 6D). The results suggest that phosphorylation of S703 in the C terminus of RBOHD is important for full flg22-triggered ROS production also in planta. Notably, expression of 3FLAG-RBOHDS703A in rbohD mutant background impaired flg22-triggered stomatal closure (Supplemental Figure 6E) and consequently enhanced susceptibility to PtoDC3000 after spray infection ( Figure 6F). In summary, our results suggest phosphorylation of RBOHD S703 is an important regulatory element of plant immune responses.

C-Terminal Phosphorylation Sites Are Conserved in Plant and Animal NOXs
Since little is known about control of RBOH activity via its C terminus, we investigated whether regulation through S703 was unique to RBOHD or conserved in other RBOHs. We constructed a phylogenetic tree of RBOHs from plant genomes representing major branches of the plant lineages (Figure 7). Plant RBOHs form a monophyletic group, which is separated from human NOX2 and NOX5. The C-terminal region of plant NOXs contains a high number of serines and threonines (Supplemental Table 2). Furthermore, conservation of the amino acid sequence surrounding a phosphorylation site may be required for recognition of the site by the kinase (Miller and Turk, 2016;Trost et al., 2016). Accordingly, the sites corresponding to RBOHD S611, S703, and S862 exhibited conservation of amino acid motifs surrounding the phosphorylation site.
The phospho-sites in the C-terminal region displayed strong conservation throughout the plant RBOH clade. S703 was conserved in eight of the 10 RBOHs from Arabidopsis, but not in RBOHH and RBOHJ (Figure 7). In the clade containing AtRBOHE, the sequence motif harboring S703 was far less conserved compared to other clades. Intriguingly, AtRBOHE itself as well as its homolog from the Brassicaceae Capsella rubella contains a Ser corresponding to S703 of RBOHD, while the other members of the clade do not possess a Ser or Thr at this position. The phosphosites S862, which may be involved in negative regulation based on experiments in HEK293T cells ( Figure 5B), as well as S611 were strongly conserved in all plant RBOHs. Intriguingly, the sequence motifs harboring S611 and S862 are conserved even in human NOX2 and NOX5 (Figure 7). The C-terminal region binds FAD and NADPH. Therefore, it may underlie strong evolutionary constraints to conserve these binding properties. This may be also reflected in the strong conservation of C-terminal phospho-sites and their sequence context in plant and animal NOXs alike.

DISCUSSION
CRKs are a large group of RLKs involved in biotic and abiotic stress signaling in Arabidopsis (Bourdais et al., 2015). We have previously shown that flg22-triggered extracellular ROS production is altered in several crk mutants (Bourdais et al., 2015). In particular CRK2, a member of the basal clade of CRKs (Vaattovaara et al., 2019), has been highlighted since crk2 displays striking phenotypes (Bourdais et al., 2015), including reduced rosette size and diminished flg22-induced ROS production. Functional CRK2 restored the reduced rosette size (Figure 1) as well as the MAMP-triggered ROS burst ( Figure 2). In addition to their roles in stress responses, extracellular ROS have also been implicated in leaf cell expansion (Schmidt et al., 2016). For example, the rbohD rbohF double mutant displays reduced rosette size (Torres et al., 2002). Overexpression of CRKs has been associated with increased SA accumulation (Chen et al., 2004;Acharya et al., 2007), which could potentially explain the growth phenotype of crk2. However, since SA levels were unaltered in the loss-of-function mutant crk2 (Supplemental Figure 1B), its smaller size may rather be a consequence of impaired ROS production ( Figure 2). This is supported by the observation that CRK2 enhanced the activity of RBOHD and RBOHF in HEK293T cells ( Figure 3A; Supplemental Figure 3F). Alternatively, other proteins interacting with or substrates of CRK2 might be involved in the regulation of plant growth.
reports suggest a bifurcation to ROS burst and MAPK activation in defense signaling following MAMP perception (Zhang et al., 2007;Yeh et al., 2016). CRK2 could be involved in balancing MAMPinduced ROS signaling pathways and MAPK signaling, but the mechanisms are still unclear.
Our results suggest that CRK2 participates in the control of ROS production via interaction with RBOHD. Intriguingly, CRK2 existed in a preformed complex with RBOHD in planta independent of MAMP treatment ( Figure 3C), while many other RLK protein complexes are formed upon signal perception. The observation that, similar to flg22-triggered ROS production, also chitininduced ROS production was reduced in crk2 suggests that CRK2 might not act on MAMP receptor complexes but rather on the regulation of ROS production itself. Phosphorylation of the C terminus is critical for the regulation of human NOXs. Phosphorylation of the NOX2 C terminus by protein kinase C enhances assembly of the multimeric NOX2 complex and its activity, whereas phosphorylation by ATM kinase inhibits NOX2 activity (Raad et al., 2009;Beaumel et al., 2017). NOX5 activity is regulated by Ca 21 binding to EF hands in the N terminus (Bánfi et al., 2004), but NOX5 is also activated by phosphorylation of the C terminus by protein kinase C a or calcium/calmodulin-dependent kinase II (Pandey et al., 2011;Chen et al., 2014). Although the C-terminal catalytic domain of RBOHs is highly conserved in plants and animals, the N terminus has been considered as important for activation of the ROS-production activity. Accordingly, multiple phospho-sites (S8, S39, S133, S148, S163, S339, S343, and S347) in the N-terminal region have been reported. Intriguingly, CRK2cyto interacted with and phosphorylated the RBOHD C-terminal region at S611, S703, and S862 ( Figures 3E and 4B; Supplemental Table 1). Phosphorylation of S703 upon xylanase treatment has been reported (Benschop et al., 2007) but not linked with other MAMPs or modulation of ROS production. Treatment with flg22 enhanced phosphorylation of S703 in Arabidopsis Phylogenetic tree showing that plant RBOHs form a single clade that is parallel to the NADPH oxidases NOX2 and NOX5b from Homo sapiens. The tree was constructed using RAxML from a MAFFT alignment, 1000 bootstraps were calculated with RAxML. The full sequence alignment can be found in Wasabi at http://was.bi?id5p_nwi1. Plant species included were Arabidopsis (At), Capsella rubella (Carub), Prunus persica (Prupe), Solanum lycopersicum (Solyc), Aquilegia coerulea (Aqcoe), Oryza sativa (LOC_Os), Sorghum bicolor (Sobic), Amborella trichopoda (Atr), and Marchantia polymorpha (Mapoly). Numbers of phospho-sites in the meme figures represent the position of the amino acid in RBOHD from Arabidopsis. Arrows indicate the position of the phospho-site (S or T) or corresponding amino acid.
( Figures 6C and 6D; Supplemental Figure 5D). Mutation S703A in RBOHD led to reduced CRK2-dependent RBOHD activity in HEK293T cells ( Figure 5B) and decreased flg22-induced ROS production in Arabidopsis ( Figure 6E; Supplemental Figure 6D). The flg22-induced phosphorylation of S703 was diminished in CRK2 D450N -YFP/crk2 plants compared to the wild type ( Figure 6D). These results suggest that phosphorylation of the RBOHD C terminus at S703 by CRK2 contributes to the regulation of MAMP-induced ROS production. However, other kinases than CRK2 might also target S703 in RBOHD since flg22-induced phosphorylation of S703 in CRK2 D450N -YFP/crk2 plants was significantly reduced but not completely abolished. Phosphorylation sites in the C terminus are highly conserved among RBOHs (Figure 7), suggesting that phosphorylation of the C-terminal region could be a general feature of plant NOXs. Remarkably, two putative RBOHD phospho-sites, S611 and S862, were identified even in the human NOXs NOX2 and NOX5 (Figure 7).
Substitution of RBOHD S862 to Ala resulted in enhanced ROSproducing activity in HEK293T cells. However, substitution of RBOHD S611 to Ala, similarly to RBOHD S39A, did not alter ROS production in HEK293T cells ( Figure 5). RBOHD S39A also did not affect flg22-induced ROS production but phospho-mimic S39D enhanced the ROS production and phosphorylation of S39 is enhanced by MAMP treatment in planta (Kadota et al., 2014; Figure 6B;Supplemental Figure 5C). These results suggest the importance to determine the phosphorylation status of S611 and S862 in planta in the future. RBOHD can also be regulated by Cys S-nitrosylation in the C terminus (Yun et al., 2011), but it is unclear how this modification is integrated with other regulatory mechanisms. Taken together, our results suggest that phosphorylation of the C-terminal region of plant NOXs is strongly conserved and important for controlling ROS production.
Several protein kinases phosphorylate RBOHD N terminus and regulate the activity including receptor-like cytoplasmic kinases (Kadota et al., 2014;Li et al., 2014;Lin et al., 2015), MAP4Ks , CPKs (Dubiella et al., 2013), and RLKs (Chen et al., 2017), but how is regulation by phosphorylation of the N-and C-terminal regions coordinated? BIK1 activates RBOHD by phosphorylation (Kadota et al., 2014;Li et al., 2014), and ROS production in bik1 is reduced to a similar extent as in crk2. However, decreased flg22-induced ROS production in crk2 was not due to lower BIK1 transcript abundance (Supplemental Figure 7). BIK1 homologs, AVRPPHB SUSCEPTIBLE1 (PBS1) and AVRPPHB SUSCEPTIBLE1-LIKE (PBL) kinases, contribute to the regulation of RBOH activity and ROS production is progressively reduced in double mutants of bik1 with its homologs (Lin et al., 2015;Zhang et al., 2018). CRK2 and BIK1 might synergistically regulate ROS production, but we were unable to obtain a double mutant between bik1 and crk2 (Supplemental Table 3). Therefore, we propose that at least one of these components is essentially required. BIK1 has previously been shown to interact with other kinases including, CRKs (Lee et al., 2017), but interaction with CRK2 has not been investigated. BIK1 and CRK2 are likely highly coordinated in order to precisely control ROS production in response to environmental stimuli (Figure 8). Like CRKs, RBOHs are involved in diverse processes in stress responses and also plant development, and it is conceivable that different CRKs regulate the diverse set of RBOH proteins in various cellular contexts potentially via phosphorylation of the C-terminal region.
In summary, we propose that CRK2 is a central element in orchestrating the extracellular ROS burst and in mediating the balance between different defense responses. The full complexity and integration of the regulatory components controlling RBOH activity is still a topic of much speculation (Kimura et al., 2017). The MAMPs are recognized by MAMP receptor complexes. RBOHD N terminus is phosphorylated by BIK1 and SIK1, and apoplastic ROS production is induced. Apoplastic ROS production by RBOHD leads to Ca 21 influx into the cytosol. Ca 21 binding to RBOHD N terminus and to CPKs leads to Ca 21 -dependent activation of RBOHD. We found that CRK2 also contributes to the activation of RBOHD via phosphorylation of its C terminus at S703. CRK2 can also mediate inhibition of MAPK activation and callose deposition via CALS after MAMP perception. MPK, mitogen-activated protein kinase; MP2K, MPKK; MP3K, MPKKK. diversity of regulators converging at RBOHs reflects the prominent role of apoplastic ROS in signal transduction, while strict control of their activity is required to circumnavigate oxidative damage. We suggest that RBOHD is regulated by phosphorylation of the C-terminal region to complement regulatory mechanisms targeting the N terminus (Figure 8). Based on the conservation of amino acid motifs in the C terminus of NOXs harboring Ser and Thr residues (Figure 7), we propose that this mode of regulation could be evolutionarily conserved in plants and animals. In the future, it will be interesting to investigate how CRK-mediated phosphorylation of the RBOH C terminus is integrated in the diverse processes that incorporate extracellular ROS.

Subcellular Protein Localization
Fluorescent images were obtained using a TCS SP5 II HCS confocal microscope (Leica). For investigation of CRK2-YFP localization, 514 nm excitation and 525 to 590 nm detection range were used.
The ROS-producing activity of RBOHs in HEK293T cells was measured as previously described by Kimura et al. (2012). Two days after transfection, medium was removed and cells were gently washed with 13 Hanks' Balanced Salt Solution (Gibco, catalog nos. 14025-092 or 14175-095).

Bacterial Growth Assay
To quantify bacterial growth on 4-week-old plants infected with the virulent Pto DC3000 (Whalen et al., 1991), growth curve assays were performed as previously described by Wrzaczek et al. (2007). Spray infections were performed as previously described (Yao et al., 2013).

Stomatal Aperture
Measurements of stomatal aperture were performed as previously described by Kadota et al. (2014) with minor modifications. Leaves were cut at the petiole from 3-week-old plants and incubated in water for 2 h. Samples were then incubated overnight in stomatal opening buffer (50 mM KCl, 10 mM CaCl 2 , 0.01% [v/v] Tween20, 10 mM MES, pH 6.15) to induce stomatal opening. For treatments, 5 mM flg22 was added to the buffer for 2 h prior to imaging. Brightfield images were obtained using a TCS SP5 II HCS confocal microscope (Leica).

Ca 21 Imaging
Calcium imaging with YCNano-65-expressing plants was performed as previously described by Choi et al. (2014), Lenglet et al. (2017), and Toyota et al. (2018). In brief, YCNano-65 was visualized by a fluorescence stereo microscope (Nikon) with a 13 objective lens (Nikon), image-splitting optics (Hamamatsu Photonics), and a sCMOS camera (Hamamatsu Photonics). To excite YCNano-65, a mercury lamp (Nikon), a 436/20-nm excitation filter (Chroma) and a 455 nm dichroic mirror (Chroma) were used. The fluorescent signal from YCNano-65 was separated by a 515 nm dichroic mirror (Chroma) equipped in the image splitting optics. The resultant cyan fluorescent protein (CFP) and YFP (FRET) signals passed independently through a 480/40 nm and 535/30 nm emission filters, respectively (Chroma). A pair of the CFP and FRET images was simultaneously acquired every 4 s with the sCMOS camera using NIS-Elements imaging software (Nikon). A 7-d-old seedling was filled with ;50 mL of 1 mM flg22 or 13 MS 1% (w/v) Suc liquid media. A region of interest was placed on the adaxial surface of cotyledons to analyze both CFP and FRET signals. The FRET/ CFP ratio was calculated by the 6D imaging plug-in modules in NIS-Elements imaging software (Nikon).

Callose Staining
Callose staining was performed as previously described by Hunter et al. (2019).

RT-qPCR
Col-0, crk2, and fls2 seedlings were grown on MS 1% (w/v) Suc agar plate for 5 d and were transferred into MS 1% (w/v) Suc liquid media and grown for 5 d. Plants were incubated with 1 mM flg22 for 30 min, 1 h, and 3 h, respectively. Plants were ground in liquid nitrogen and total RNA was extracted using the GeneJET Plant RNA purification kit (Thermo Fisher Scientific, catalog no. K0802). Total RNA was treated with DNase I (Thermo Fisher Scientific, #EN0525), and cDNA was synthesized with Maxima H Minus Reverse Transcriptase (Thermo Fisher Scientific, catalog no. EP0751). qPCR analysis was performed with CFX real-time PCR (Bio-Rad) using 53 HOT FIREPol EvaGreen qPCR mix plus ROX (Soils Biodyne). SAND, TIP41, and YLS8 were used as reference genes for normalization. Relative expression was calculated with qBase1 (Biogazelle; https://www. qbaseplus.com/). Primers are listed in Supplemental Table 4.

Phytohormone Analysis
SA was analyzed from adult rosettes as previously described by Forcat et al. (2008) with minor modifications. Rosettes were flash-frozen in liquid nitrogen and freeze-dried for 24 h. Rosettes were ground in a mortar before weighing in to 2-mL tubes. About 6 mg aliquots of freeze-dried material were further homogenized by shaking with 5-mm stainless steel beads in a Tissue Lyser II (Qiagen) for 2 min at 25 Hz. Shaking was repeated after addition of 400 mL extraction buffer (10% [v/v] methanol, 1% [v/v] acetic acid) with internal standard (28 ng Salicylic-d4 Acid; CDN Isotopes). Samples were then incubated on ice for 30 min and centrifuged for 10 min at 16,000g and 4°C. Supernatants were centrifuged three times to remove all debris before liquid chromatography tandem mass spectrometry (LC-MS/ MS) analysis. The chromatographic separation was performed using an Acquity UHPLC thermo system (Waters) equipped with a Cortecs C18 column (2.7 mm, 2.1 3 100 mm; Waters). The solvent gradient (acetonitrile [ACN] /water with 0.1% [v/v] formic acid each) was adapted to a total run time of 7 min: 0 to 4 min 20% (v/v) to 95% (v/v) ACN, 4 to 5 min 95% (v/v) ACN, 5 to 7 min 95% (v/v) to 20% (v/v) ACN; flow rate 0.4 mL/min. For hormone identification and quantification, a tandem mass spectrometric system Xevo TQ-XS, triple quadrupole mass analyzer with a ZSpray ESI function (Waters) was used. Mass transitions were SA 137 > 93, D4-SA 141 > 97.

Protein Purification from E. coli
Cytosolic regions of CRK2 were expressed in Escherichia coli Lemo21. Cytosolic regions of RBOHD and MBP were expressed in E. coli BL21. GST-tagged recombinant proteins were purified using glutathione sepharose 4B (GE Healthcare, catalog no. 17-0756-01) according to manufacturer's instructions. MBP-tagged proteins were purified using amylose resin (New England Biolabs, catalog no. E8021S) according to manufacturer's instructions.

In Vitro Kinase Assay
Purified recombinant proteins were incubated with [g-32 P] ATP for 30 min at room temperature in the kinase assay buffer (50 mM HEPES [pH 7.4], 1 mM DTT, 10 mM MgCl 2 , 0.6 mM unlabeled ATP). The mixture was subsequently separated by SDS-PAGE, and autoradiography was detected by FLA-5100 image analyzer (Fujifilm, Japan). For identification of in vitro phosphorylation sites by LC-electrospray ionization (ESI)-MS/MS, 1.5 mM unlabeled ATP was used in the kinase buffer. The proteins were separated by SDS-PAGE, followed by Coomassie Brilliant Blue staining and were digested by trypsin (Thermo Fisher Scientific, catalog no. 90057) or Lys-C (Thermo Fisher Scientific, catalog no. 90051).

Identification of In Vitro Phosphorylation Sites of RBOHD by LC-Electrospray Ionization-MS/MS
Trypsin or Lys-C-digested protein samples were analyzed by a Q Exactive mass spectrometer (Thermo Fisher Scientific) connected to Easy NanoLC 1000 (Thermo Fisher Scientific). Peptides were first loaded on a trapping column and subsequently separated inline on a 15-cm C18 column (75 mm 3 15 cm, ReproSil-Pur 5 mm 200 Å C18-AQ, Dr. Maisch HPLC). The mobile phase consisted of water with 0.1% (v/v) formic acid (solvent A) or acetonitrile/water (80:20 [v/v]) with 0.1% (v/v) formic acid (solvent B). A linear 10-min gradient from 8 to 42% (v/v) B was used to elute peptides.
Mass spectrometry data was acquired automatically by using Xcalibur 3.1 software (Thermo Fisher Scientific). An information-dependent acquisition method consisted of an Orbitrap mass spectrometry survey scan of mass range 300 to 2000 m/z (mass-to-charge ratio) followed by higherenergy collisional dissociation (HCD) fragmentation for 10 most intense peptide ions. Raw data was searched for protein identification by Proteome Discoverer (version 2.2; http://www.matrixscience.com/blog/mascotworkflows-in-proteome-discoverer.html) connected to in-house Mascot (v. 2.6.1; https://www.thermofisher.com/fi/en/home/industrial/massspectrometry/liquid-chromatography-mass-spectrometry-lc-ms/lc-ms-software/multi-omics-data-analysis/proteome-discoverer-software.html) server. Phosphorylation site locations were validated using phos-phoRS algorithm (Taus et al., 2011). A SwissProt database (https:// www.uniprot.org/) with a taxonomy filter Arabidopsis was used. Two missed cleavages were allowed. Peptide mass tolerance 6 10 ppm and fragment mass tolerance 6 0.02 D were used. Carbamidomethyl (C) was set as a fixed modification and Met oxidation, acetylation of protein N terminus, and phosphorylation of Ser and Thr were included as variable modifications. Only peptides with a false discovery rate of 0.01 were used.

Plant Treatment
Arabidopsis seeds were sterilized by incubating with 1.5% (w/v) NaClO 0.02% (v/v) Triton X-100 solution for 5 min and vernalized at 4°C for 2 d. Sterilized seeds were germinated and grown in liquid culture on 6-well plates (30 seeds/well) in MGRL medium with 0.1% (w/v) Suc (2 mL/well; Fujiwara et al., 1992) at 23°C under continuous light (100 mmol m 22 s 21 , LED) in a Percival growth chamber. Plates with 11-d-old seedlings were transferred from the growth chamber to a workbench and kept overnight for acclimatization before treatments. Seedlings were treated with either 1 mM flg22 or sterile water for 5 min, after which seedlings were immediately collected and flash-frozen in liquid nitrogen and stored at 280°C. Leaf discs were collected using a cork borer from 4-week-old Arabidopsis plants and floated overnight in sterile distilled water in Petri dishes under continuous light (17 mmol m 22 s 21 , LED) at room temperature. On the following day, water was replaced with new sterile distilled water or 200 nM flg22. After 5 min incubation, leaf discs were immediately collected and flash-frozen in liquid nitrogen and stored at 280°C.

LC-MS/MS Data Acquisition
Samples were analyzed using an EASY-nLC 1200 (Thermo Fisher) coupled to a Q Exactive Plus mass spectrometer (Thermo Fisher Scientific). Peptides were separated on 16-cm frit-less silica emitters (New Objective, 0.75-mm inner diameter), packed in-house with reversed-phase ReproSil-Pur C18 AQ 1.9 mm resin (Dr. Maisch). Peptides were loaded on the column and eluted for 115 min using a segmented linear gradient of 5 to 95% (v/v) solvent B (0 min, 5% B; 0 to 5 min, 5% B; 5 to 65 min, 20% B; 65 to 90 min, 35% B; 90 to 100 min, 55% B; 100 to 105 min, 95% B; 105 to 115 min, 95% B; solvent A [0% ACN, 0.1% (v/v) formic acid]; solvent B [80% (v/v) ACN, 0.1% (v/v) formic acid]) at a flow rate of 300 nL/min. Mass spectra were acquired using a targeted (parallel reaction monitoring [PRM]) approach. The acquisition method consisted of a full-scan method combined with a nonscheduled PRM method. The 16 targeted precursor ions were selected based on the results of a data-dependent acquisition peptide search of phospho-enriched samples in Skyline (version 4.2.0.x; https://skyline.ms; MacLean et al., 2010). Mass spectrometry spectra were acquired in the Orbitrap analyzer with a mass range of 300 to 1750 m/z at a resolution of 70,000 full width at half maximum (FWHM) and a target value of 3 3 10 6 ions, followed by MS/MS acquisition for the 16 targeted precursors. Precursors were selected with an isolation window of 2.0 m/z. HCD fragmentation was performed at a normalized collision energy of 27. MS/ MS spectra were acquired with a target value of 2 3 10 5 ions at a resolution of 17,500 FWHM, a maximum injection time of 120 ms, and a fixed first mass of m/z 100.

Mass Spectrometry Data Analysis
Raw data from PRM acquisition were processed using MaxQuant software (version 1.5.7.4; http://www.maxquant.org/; Cox and Mann, 2008). MS/MS spectra were searched by the Andromeda search engine against a combined database containing the sequences from Arabidopsis (TAIR10_pep_ 20101214; ftp://ftp.arabidopsis.org/home/tair/Proteins/TAIR10_protein_ lists/) and sequences of 248 common contaminant proteins and decoy sequences. Trypsin specificity was required, and a maximum of two missed cleavages allowed. Minimal peptide length was set to seven amino acids. Carbamidomethylation of Cys residues was set as fixed, and phosphorylation of Ser, Thr and Tyr, oxidation of Met, and protein N-terminal acetylation was set as variable modifications. The match between runs option was disabled. Peptide-spectrum matches and proteins were retained if they were below a false discovery rate of 1% in both cases. The "msms.txt" output from MaxQuant was further analyzed using Skyline in PRM mode. Trypsin specificity was required and a maximum of two missed cleavages allowed. Minimal and maximum peptide lengths were set to seven and 25 amino acids, respectively. Carbamidomethylation of Cys, phosphorylation of Ser, Thr and Tyr, oxidation of Met, and protein N-terminal acetylation were set as modifications. Results were filtered for precursor charges of 2 and 3, and b-and y-ions with ion charges of 11 and 12. Product ions were set to "from ion 1 to last ion." All chromatograms were inspected manually and peak integration was corrected for best representation of mass spectrometry 2 signals. Peak area data was exported and further processed. The Skyline documents containing the data for the targeted phophoproteomics experiments have been uploaded to Panorama Public and can be obtained from https://panoramaweb.org/ RBOHDphosphorylation.url. Raw data have been deposited to the Pro-teomeXchange Consortium via the Panorama partner repository with the data set identifier PXD013525 (http://proteomecentral.proteomexchange. org/cgi/GetData set?ID5PXD013525).

Phylogenetic Analysis
Sequences for plant RBOH genes were extracted from public genome databases and manually curated. Sequences were aligned using MAFFT version 7.407 (Katoh and Standley, 2013) using iterative refinement (L-INSi). This alignment was used to infer a phylogenetic maximum likelihood tree using RAxML version 8.1.3 (Stamatakis, 2014) with the JTT or DAYHOFF substitution models (which resulted in identical phylogenetic trees). Thousand bootstrap replicates were calculated using RAxML. Human NOX2 and NOX5b were used as outgroup to root the phylogenetic tree of plant NOXs. The sequence alignment of plant RBOHs, human NOX2 and NOX5b can be viewed on the Veidenberg et al. (2016) web server (http:// was.bi?id5p_nwi1) and the tree as a nexus file is available as the Supplemental File. Sequence motifs were analyzed using the MEME suite (Bailey et al., 2009).

Accession Numbers
Phylogenetic tree of human and plant NOXs with bootstrap information for 1000 replicates and corresponding sequence alignment has been deposited on Wasabi (http://was.bi?id5p_nwi1). Data for the targeted phophoproteomics experiments have been uploaded to Panorama Public (https://panoramaweb.org/RBOHDphosphorylation.url). Raw data have been deposited to the ProteomeXchange Consortium via the Panorama partner repository with the dataset identifier PXD013525 (http:// proteomecentral.proteomexchange.org/cgi/GetDataset?ID5PXD013525). Materials used in the experimental work are available from the authors upon request. Supplemental Table 2. The numbers of Ser (S) and Thre (T) residues in C terminus of NOXs in plants and animals Supplemental Table 3. Progeny of CRK2/crk2 BIK1/bik1 parent and CRK2/crk2 bik1/bik1 parent Supplemental Table 4. Primer sequences Supplemental Table 5. One-way ANOVA results Supplemental File. Nexus file of the alignment corresponding to the phylogenetic tree in Figure 7