Phytophthora infestans RXLR-WY effector AVR3a associates with a Dynamin-Related Protein involved in endocytosis of a plant pattern recognition receptor

Perception of pathogen associated molecular patterns (PAMPs) by cell surface localized pattern recognition receptors (PPRs), activates plant basal defense responses in a process known as PAMP/PRR-triggered immunity (PTI). In turn, pathogens deploy effector proteins that interfere with different steps in PTI signaling. However, our knowledge of PTI suppression by filamentous plant pathogens, i.e. fungi and oomycetes, remains fragmentary. Previous work revealed that BAK1/SERK3, a regulatory receptor of several PRRs, contributes to basal immunity against the Irish potato famine pathogen Phytophthora infestans. Moreover BAK1/SERK3 is required for the cell death induced by P. infestans elicitin INF1, a protein with characteristics of PAMPs. The P. infestans host-translocated RXLR-WY effector AVR3a is known to supress INF1-mediated defense by binding the E3 ligase CMPG1. In contrast, AVR3aKl-Y147del, a deletion mutant of the C-terminal tyrosine of AVR3a, fails to bind CMPG1 and suppress INF1 cell death. Here we studied the extent to which AVR3a and its variants perturb additional BAK1/SERK3 dependent PTI responses using the plant PRR FLAGELLIN SENSING 2 (FLS2). We found that all tested variants of AVR3a, including AVR3aKl-Y147del, suppress early defense responses triggered by the bacterial flagellin-derived peptide flg22 and reduce internalization of activated FLS2 from the plasma membrane without disturbing its nonactivated localization. Consistent with this effect of AVR3a on FLS2 endocytosis, we discovered that AVR3a associates with the Dynamin-Related Protein DRP2, a plant GTPase implicated in receptor-mediated endocytosis. Interestingly, DRP2 is required for ligand-induced FLS2 internalization but does not affect internalization of the growth receptor BRASSINOSTEROID INSENSITIVE 1 (BRI1). Furthermore, overexpression of DRP2 suppressed accumulation of reactive oxygen species triggered by PAMP treatment. We conclude that AVR3a associates with a key cellular trafficking and membrane-remodeling complex involved in immune receptor-mediated endocytosis and signaling. AVR3a is a multifunctional effector that can suppress BAK1/SERK3 mediated immunity through at least two different pathways. AUTHOR SUMMARY Plants have a basal layer of immunity to mount defense responses against invading pathogens; in turn, pathogens deploy effector proteins to subvert plant immunity and manipulate host processes to enable parasitic infection. The Irish potato famine pathogen Phytophthora infestans has a large set of effectors that target multiple host cellular sites. The best-characterized P. infestans effector AVR3a supports enhanced infection and suppresses the cell death induced by the P. infestans protein INF1-elicitin. Previous work demonstrated that the plant immunity co-receptor BAK1/SERK3 contributes to basal immunity to P. infestans and that the RXLR-WY effector of P. infestans suppresses BAK1/SERK3-mediated immunity by binding the E3 ligase protein CMPG1. Here we show that AVR3a suppresses additional defense responses mediated by BAK1/SERK3 independently of CMPG1. AVR3a reduces the endocytosis of the plant receptor FLS2, which recognizes the flagellin-derived peptide flg22 in a BAK1/SERK3 dependent manner. Furthermore, we demonstrate that AVR3a associates with the Dynamin-Related Protein DRP2, a plant GTPase involved in receptor-mediated endocytosis that is required for FLS2 internalization. Our work revealed that AVR3a is a multifunctional effector that perturbs cellular trafficking initiated at the cell periphery by at least two mechanisms, and that this effector associates with a key cellular trafficking and membrane-remodeling complex involved in immune receptor-mediated endocytosis and signaling.


ABSTRACT
Perception of pathogen associated molecular patterns (PAMPs) by cell surface localized pattern recognition receptors (PPRs), activates plant basal defense responses in a process known as PAMP/PRR-triggered immunity (PTI). In turn, pathogens deploy effector proteins that interfere with different steps in PTI signaling.
However, our knowledge of PTI suppression by filamentous plant pathogens, i.e. fungi and oomycetes, remains fragmentary. Previous work revealed that BAK1/SERK3, a regulatory receptor of several PRRs, contributes to basal immunity against the Irish potato famine pathogen Phytophthora infestans. Moreover BAK1/SERK3 is required for the cell death induced by P. infestans elicitin INF1, a protein with characteristics of PAMPs. The P. infestans host-translocated RXLR-WY effector AVR3a is known to supress INF1-mediated defense by binding the E3 ligase CMPG1. In contrast, AVR3a KI-Y147del , a deletion mutant of the C-terminal tyrosine of AVR3a, fails to bind CMPG1 and suppress INF1 cell death. Here we studied the extent to which AVR3a and its variants perturb additional BAK1/SERK3 dependent PTI responses using the plant PRR FLAGELLIN SENSING 2 (FLS2). We found that all tested variants of AVR3a, including AVR3a KI-Y147del , suppress early defense responses triggered by the bacterial flagellin-derived peptide flg22 and reduce internalization of activated FLS2 from the plasma membrane without disturbing its nonactivated localization. Consistent with this effect of AVR3a on FLS2 endocytosis, we discovered that AVR3a associates with the Dynamin-Related Protein DRP2, a plant GTPase implicated in receptor-mediated endocytosis. Interestingly, DRP2 is required for ligand-induced FLS2 internalization but does not affect internalization of the growth receptor BRASSINOSTEROID INSENSITIVE 1 (BRI1). Furthermore, overexpression of DRP2 suppressed accumulation of reactive oxygen species triggered by PAMP treatment. We conclude that AVR3a associates with a key cellular trafficking and membrane-remodeling complex involved in immune receptormediated endocytosis and signaling. AVR3a is a multifunctional effector that can suppress BAK1/SERK3 mediated immunity through at least two different pathways.

AUTHOR SUMMARY
Plants have a basal layer of immunity to mount defense responses against invading pathogens; in turn, pathogens deploy effector proteins to subvert plant immunity and manipulate host processes to enable parasitic infection. The Irish potato famine pathogen Phytophthora infestans has a large set of effectors that target multiple host cellular sites. The best-characterized P. infestans effector AVR3a supports enhanced

INTRODUCTION
Plants must respond in a timely and effective manner to external cues such as biotic and abiotic stimuli. In particular, plants are associated with a huge variety of microorganisms, many of which are sophisticated parasites [1][2][3] . To keep such parasitic microbes at bay, plants have evolved a multilayered immune system, which is triggered by microbial perception via receptors located at the cell surface and the cytoplasm 3,4 . An important layer of defense relies on perception of pathogen associated molecular patterns (PAMPs) by cell surface localized pattern recognition receptors (PRRs), a basal defense response known as PAMP/PRR-triggered immunity (PTI) 5 . Early PTI signaling events include ion fluxes, reactive oxygen species (ROS) production, induction of mitogen-activated protein kinases (MAPKs) and calcium-dependent protein kinases (CDPKs) and transcriptional reprogramming [6][7][8] . Collectively, PTI contributes to immunity by delaying or arresting pathogen invasion, and in many cases PTI-deficient plant mutants became more susceptible to pathogens 9 . However, a common feature of adapted plant pathogens is the deployment of effector proteins that interfere with different steps in PTI signaling pathways 3,4,10,11 . This is particularly striking for bacterial plant pathogens, which evolved a battery of often redundant effectors to counteract plant immunity elicited by conserved patterns such as flagellin and the elongation factor thermo unstable (EF-Tu) 3,12,13 . In contrast, our understanding of how eukaryotic pathogens, such as oomycetes and fungi, suppress PTI remains fragmentary [14][15][16][17] .
BAK1/SERK3 is also required for defense responses mediated by the receptor like proteins (RLPs) Ve1, which confers resistance to race 1 strains of the fungal pathogens Verticillium dahliae and Verticillium albo-atrum, and LeEix2, which recognizes the fungal protein elicitor EIX (ETHYLENE-INDUCING XYLANASE) 34,35 .
More recently, BAK1/SERK3 emerged as a key component of basal defense against oomycetes, an important group of filamentous eukaryotic pathogens. Silencing of BAK1/SERK3 resulted in dramatically enhanced susceptibility and faster host colonization by the potato late blight pathogen Phytophthora infestans 36 . In addition, BAK1/SERK3 and its closest homolog BKK1/SERK4 are involved in resistance to the obligate biotrophic oomycete Hyaloperonospora arabidopsidis 28  Plant endocytic trafficking has emerged as a dynamic process that is diverted towards sites of pathogen infection 38,39 indicating that this plant process may play a critical role in immune responses 40 . To maintain its protein levels at the cell surface, FLS2 undergoes constitutive endocytosis and recycles between plasma membrane and trans-Golgi network, independently of BAK1 25,41 . However, FLS2 undergoes ligand-induced translocation into endosomal vesicles in a BAK1-dependent manner [41][42][43][44] . Therefore, FLS2 traffics through two different endocytic pathways depending on its activation status. Furthermore, BRI1 and BAK1/SERK3 also undergo constitutive recycling via endosomes and can localize to overlapping endosomal compartments 32 .
Receptor-mediated endocytosis was initially thought as a mechanism for attenuation of signaling through depletion of activated receptor complexes. Further studies in animal systems revealed that receptor internalization contributes to additional signaling at the endocytic compartments 45 . In animal cells, late endosomal compartments, which host internalized receptors, regulate signaling events such as pro-inflammatory signaling, growth, and development [45][46][47][48][49] . In plants, localization of BRI1 and BAK1/SERK3 as well as a brassinosteroid analog at the same endosomal compartments pointed to a possible link between internalization and signaling in growth regulation 32,50,51 . However, defense-related endocytic signaling has yet to be unambiguously demonstrated in plants.
Mechanisms of receptor-mediated endocytosis involve clathrin-mediated and clathrin-independent pathways resulting in the recruitment of plasma membrane cargo 52-54 that is later invaginated and pinched off into the cytoplasm often by the action of the large GTPase dynamin 53,55 . Dynamin is a ~100 kDa protein that selfassembles into rings and helices to promote structural reorganization to mediate membrane fission in a reaction that requires GTP hydrolysis 55 . The mechanistic details of clathrin-mediated endocytosis have been well established in animal cells and dynamin has been implicated in the internalization of the immunity-related Interleukin-2 Receptor (IL-2R) 56 . However, in plants, mechanisms of endocytosis have not been explicitly studied 57,58 and the extent to which dynamin or dynamin-like proteins play a role in receptor mediated endocytosis or plant immunity remains poorly known. P. infestans is the causal agent of potato and tomato late blight and a major threat to food security 59 . This oomycete pathogen deploys a large set of effectors that target multiple host cellular sites. Cytoplasmic effectors include the RXLR class, modular proteins that translocate inside host cells through poorly known mechanisms 60 . The biochemical activity of RXLR effectors is carried out by their C-terminal domains, which often contain variations of the conserved WY-domain fold 61,62 . One example of RXLR-WY effector is AVR3a of P. infestans. In P. infestans populations, Avr3a has two major allelic variants encoding the proteins AVR3a KI and AVR3a EM , which differ in two amino acids in the mature protein and are differentially perceived by the potato immune receptor R3a [63][64][65][66][67] . Contrary to AVR3a EM , AVR3a KI induces R3a-mediated resistance and confers avirulence to homozygous or heterozygous strains of the pathogen 63 . In host plants that do not carry R3a, AVR3a KI effectively suppresses the cell death induced by P. infestans INF1 elicitin and is thought to contribute to pathogen virulence through this and other immune suppression activities 64,65,68 .
Remarkably, AVR3a KI-Y147del , a mutant with a deleted C-terminal tyrosine residue, is not affected in activation of R3a but fails to suppress INF1-cell death demonstrating that distinct amino acids condition the two AVR3a activities 65 . Moreover, AVR3a KI-Y147del cannot bind or stabilize the plant E3 ubiquitin ligase protein CMPG1, which is required for INF1-cell death, further uncoupling the effector activities 65,67 . The current model is that AVR3a, but not AVR3a KI-Y147del , binds and stabilizes CMPG1 to suppress BAK1/SERK3-regulated immunity triggered by INF1 during the biotrophic phase of P. infestans infection 67 .
This study was prompted by our discovery that natural variants of the P. infestans effector AVR3a (AVR3a KI and AVR3a EM ) and the AVR3a KI-Y147del mutant suppress FLS2-dependent early responses. This contrasts sharply with the differential activities of these three AVR3a variants in suppressing INF1 cell death, another BAK1/SERK3-dependent pathway. The ability of the AVR3a KI-Y147del mutant to suppress FLS2-dependent responses revealed that AVR3a can suppress BAK1/SERK3-dependent responses in a CMPG1-independent manner. Furthermore, AVR3a reduced the internalization of the activated FLS2 receptor but did not interfere with its nonactivated plasma membrane localization indicating that this effector might target cellular trafficking initiated at the cell periphery. Consistent with this model, we found that AVR3a associates with a plant GTPase dynamin (NtDRP2) involved in receptor-mediated endocytosis implicating AVR3a in a cellular trafficking complex. Further, we found that NbDRP2 is required for internalization of FLS2, and that this protein plays a role in modulating the activity of immune receptors.
Overexpression of NtDRP2-1 suppressed PRR-dependent accumulation of reactive oxygen species (ROS). We conclude that AVR3a associates with a key cellular trafficking and membrane-remodeling complex required for PRR endocytic trafficking.

AVR3a suppresses PTI in a CMPG1-independent manner
To determine the degree to which AVR3a suppresses PAMP-elicited defense responses besides INF1-elicited immunity, we measured the oxidative burst The immune responses triggered by another bacterial PAMP, EF-Tu, overlap and share signaling components with those triggered by flagellin 28,69,70 . Therefore, we tested the effect of AVR3a on ROS triggered by the EF-Tu derived peptide elf18. We found that elf18 ROS production was significantly impaired to a similar extent by all variants of AVR3a ( Figure S2A). In contrast, the BAK1/SERK3 independent ROS response to chitin was not affected by any of the AVR3a variants ( Figure S2B).
These results suggest that AVR3a suppresses BAK1/SERK3-dependent pathways but not BAK1-independent immune responses.
One of the outcomes of PAMP elicitation is transcriptional reprogramming of the plant, a defense response downstream of ROS 9 . Therefore, we monitored the effect of AVR3a on gene expression of the previously characterized PTI marker genes These findings indicate that the newly identified suppression activity is CMPG1 independent and that the AVR3a effector may suppress PTI through multiple mechanisms.

AVR3a does not alter receptor levels at the cell surface or receptor complex formation
The bacterial effector AvrPtoB targets FLS2 for degradation to suppress plant immunity 72 . This prompted us to determine whether the suppression of flg22triggered responses by AVR3a involved perturbation of FLS2 or BAK1 protein accumulation or complex formation 73,74 . To address this question, we transiently coexpressed AVR3a variants with FLS2-GFP or BAK1/SERK3-YFP in N. benthamiana and assessed the fusion protein levels. We found that AVR3a did not alter FLS2 or BAK1/SERK3 protein accumulation in planta (Fig 2A). Next, we tested whether AVR3a KI perturbed heterodimerization of FLS2 with BAK1/SERK3 after flg22 treatment 25,26,28 . We used leaves of transgenic N. benthamiana plants stably expressing AVR3a KI or a vector control and transiently expressed FLS2-GFP and BAK1/SERK3-HA. In the presence of AVR3a KI a double band appeared for BAK1/SERK3 total protein extracts ( Figure 2B, WB:HA) that was not seen after immunoprecipitation. However, this observation was not consistent between experiments and probably is the result of protein degradation during protein extraction. After co-immunoprecipitation of FLS2 and BAK1/SERK3 we saw no consistent effect of AVR3a on flg22-mediated recruitment of BAK1/SERK3 into the FLS2 signaling complex ( Figure 2B). In summary, these results suggest that AVR3a does not interfere with receptor protein accumulation or complex formation and that the effector suppression of flg22 responses most likely occurs downstream of FLS2/BAK1 heterodimerization.

AVR3a interferes with FLS2 internalization
We hypothesized that AVR3a alters the subcellular distribution of FLS2 and/or BAK1/SERK3 to perturb their activities. To determine the effect of AVR3a on subcellular distribution of the receptors, we transiently co-expressed FLS2-GFP or We then examined whether AVR3a perturbs the subcellular localization of other cell surface receptors. First, AVR3a did not result in any changes in the membrane localization of the inactive immune receptors EFR and CERK1 ( Figure S3A). Next, we determined whether inhibition of receptor internalization is specific to FLS2 or a general interference with the endocytic process. For this we used BRI1, a receptor involved in development that also requires BAK1/SERK3 for its activity and shows constitutive internalization 32 . Using the same experimental procedure described above, we observed that BRI1-GFP fluorescent signal from vesicle-like structures was unaltered in the presence of AVR3a KI ( Figure S4A, S4B). These results indicate that AVR3a targets FLS2 or a host component required for FLS2 internalization to modulate early defense responses without generally affecting receptor mediated endocytosis.

AVR3a associates with a plant protein involved in cellular trafficking
AVR3a was previously shown to bind the E3 ligase CMPG1 67 . To determine what additional host proteins AVR3a associates with, we used co-immunoprecipitation of FLAG-AVR3a KI in N. benthamiana followed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis using previously described methods 75 . We identified four proteins that associated with AVR3a KI that had unique peptide counts compared to the controls in two biological replicates (Table S1). We selected a homolog of the GTPase dynamin-related protein (DRP) (NCBE_074039.1) for further studies based on the established role of dynamins in cellular trafficking and clathrinmediated endocytosis in animal systems 76 . In mammals, dynamin is a five-domain protein consisting of a GTPase domain, an unstructured middle domain, a Pleckstrinhomology domain, a GTPase effector (GED) domain and a proline-rich domain, and its functions range from vesicle scission during endocytosis to the formation of the tubular-vesicular network during cytokinesis 76 . In Arabidopsis, DRP2A and DRP2B (previously known as ADL6 and ADL3, respectively) are the only dynamin-related proteins (DRP) that have similar domain architecture to other canonical dynamins [76][77][78][79] . Database searches using tBlastN and NCBE_074039.1 as a query revealed that DRP2A and DRP2B were the closest homologs for NCBE_074039.1. Therefore, we designed primers based on the Arabidopsis DRP2A and DRP2B sequences and the partial N. benthamiana sequences available at the time of the experiment. However, these primers did not result in any amplicon from N. benthamiana cDNA but we cloned two putative DRP2 proteins from N. tabacum, termed NtDRP2-1 and NtDRP2-2, which are 99% similar to each other. Sequence analysis revealed that  Table S2). The resulting tree confirmed that NtDRP2-1/2 are most related to AtDRP2A and AtDRP2B among the 11 Arabidopsis DRPs ( Figure   S6).
We then investigated the subcellular localization of NtDRP2 to determine whether it is consistent with the presumed function of DRPs in cellular trafficking. Transient expression of GFP-NtDRP2-1 and GFP-NtDRP2-2 in N. benthamiana followed by confocal microscopy revealed that GFP-NtDRP2-1 and GFP-NtDRP2-2 mainly localized at the plasma membrane as seen by the formation of thin cytoplasmic strands at points of adhesion between the plasma membrane and the cell wall Next we validated the association of AVR3a with the cloned NtDRP2 in planta by coimmunoprecipitation analysis. All three AVR3a variants, including AVR3a KI-Y147del , coimmunoprecipitated with GFP-NtDRP2-1 when expressed in N. benthamiana as FLAG epitope tagged proteins ( Figure 4). The RXLR effector AVRblb2 was used as a negative control in these experiments and did not co-immunoprecipitate with NtDRP2-1 ( Figure 4). We also used the Phytophthora capsici effector PcAVR3a-4 61 as an additional negative control because this AVR3a homolog does not suppress the ROS burst induced by flg22 ( Figure S8B). Indeed, in side-by-side coimmunoprecipitation experiments there was only a weak NtDRP2-1 band with PcAVR3a-4 that only became visible with increased exposure times ( Figure S8A).
Overall, these results confirm that all AVR3a variants associate with NtDRP2-1, and suggest this association may be linked to the ability of AVR3a to suppress flg22responses.

NtDRP2 dynamin is required for FLS2 internalization
To further investigate the link between AVR3a effector activities and NtDRP2 dynamin, we investigated the degree to which this dynamin is required for FLS2 internalization using RNAi silencing experiments. We designed silencing constructs that target N. benthamiana DRP2 Nb05397 (NbDRP2-1) and Nb31648 (NbDRP2-2), which are the most closely related N. benthamiana DRP2 to the tobacco NtDRP2-1 and NtDRP2-2 ( Figure S6). Reports in Arabidopsis showed that members of the benthamiana DRP2 genes ( Figure S10B). We also confirmed that the silenced epidermal cells remained viable by staining with propidium iodide, which is excluded from cells with intact plasma membranes ( Figure S10C).
Next, we used the hairpin RNAi system to determine the effect of silencing NbDRP2 in N. benthamiana leaves on flg22-induced internalization of FLS2-GFP. We found that FLS2-GFP containing vesicles were absent in the NbDRP2 silenced leaves upon flg22 treatment, but were clearly visible in control silenced leaves ( Figure 5A, 5C).
Moreover, silencing NbDRP2 did not alter accumulation of FLS2 at the plasma membrane in water treated samples ( Figure 5A), indicating that DRP2 only affects FLS2 after activation. To determine whether the effect of NbDRP2-silencing on receptor internalization is specific to FLS2, we performed the same NbDRP2 RNAi experiments with the brassinosteroid receptor BRI1. Remarkably, BRI1-GFP constitutive endocytosis was not affected in leaves silenced for NbDRP2 and a similar number of BRI1-GFP labelled vesicles was observed in the NbDRP2 RNAi treatment compared to the negative control ( Figure S4C, S4E). We conclude that NbDRP2 is essential for ligand-induced endocytosis of FLS2, and that this requirement might be specific for immune-related receptors. Importantly, these findings are consistent with our earlier finding that AVR3a interferes with the internalization of the activated FLS2 receptor but not BRI1 ( Figure 3B, 3C and Figure   S4A, S4B).

Overexpression of NtDRP2 dynamin suppresses flg22-induced ROS burst
To assess the role of Solanaceous DRP2 in early defense responses, we first determined the degree to which silencing of NbDRP2 affects flg22-induced ROS production. These experiments were inconclusive as we failed to observe consistent effects on ROS burst in NbDRP2-silenced N. benthamiana leaves (data not shown).
We then determined the effect of overexpressing NtDRP2 in N. benthamiana on ROS production. Consistent with the finding that AtDRP2B is a negative regulator of flg22-triggered ROS 84 , we found that NtDRP2-1/2 decreased flg22-induced ROS burst by more than 50% compared to control plants ( Figure 6A). In addition, NtDRP2-1 significantly reduced the ROS production in response to INF1 and chitin treatments ( Figure 6B, 6C). These results indicate that the effect of DRP2 on ROS responses is not limited to BAK/SERK3-dependent pathways and that NtDRP2 dynamins modulate early signaling responses induced by several PAMPs.

DISCUSSION
Even though perception of microbe-associated molecular patterns significantly contributes to basal defense of plants against the Irish potato famine pathogen P.
In this study, we further characterized the virulence activities of AVR3a and discovered that AVR3a suppresses early defense responses mediated by the cell surface immune modulator BAK1/SERK3, which contributes to basal immunity against P. infestans 36 . More specifically, we discovered that AVR3a reduces internalization of the activated pattern-recognition receptor FLS2 but does not interfere with the plasma membrane localization of nonactivated FLS2 nor does it perturb the steady state levels of this immune receptor or other PRRs. Furthermore, we found that AVR3a associates with DRP2, a Solanaceae member of the plant GTPase dynamin family that mediate endocytosis and membrane remodeling 86 .
Interestingly, DRP2 is required for FLS2 internalization and DRP2 overexpression reduces the ROS burst induced by the flg22 flagellin peptide. Our results indicate that AVR3a associates with a key cellular trafficking and membrane-remodeling complex, and that this DRP2 dynamin complex is involved in immune receptormediated endocytosis and signaling.
Several bacterial type III secretion system effectors target cell surface immune receptor complexes to modulate their activities 13,72,74 . AVR3a is an example of a filamentous plant pathogen effector that has evolved to deregulate plant immune signaling 64,67,68 . Our work is consistent with the finding that AVR3a blocks signal transduction cascades initiated at the plasma membrane after pathogen perception 68 .
However, interference with FLS2 signaling and endocytosis does not involve the interaction with the E3 ligase CMPG1 since AVR3a KI-Y147del , a variant that is unable to bind and stabilize CMPG1, is still able to suppress FLS2 ROS burst and endocytosis ( Figure 1A; Figure 3B, 3C). Therefore, AVR3a is a multifunctional effector that can suppress BAK1/SERK3 mediated immunity through at least two different pathways, possibly by acting at different sites or at different time points during immune signaling.
The step at which AVR3a interferes with FLS2 signaling is unclear. AVR3a is unlikely to completely inhibit DRP2 activity since DRP2 silencing resulted in plant lethality unlike transient or stable in planta expression of AVR3a. AVR3a did not alter subcellular localization of FLS2 or BAK1/SERK3 ( Figure 3A) and it did not affect ligand induced complex formation between FLS2 and BAK1/SERK3 (Figure 2). However, the receptor complex somehow remained inactive in the presence of AVR3a given that downstream signaling cascades were not activated. Given that elicitor-induced internalization of FLS2 occurs in a BAK1-dependent manner 42,43 it is possible that the effect of AVR3a in FLS2 internalization occurs via BAK1/SERK3. However, BAK1/SERK3 C-terminal fusions are partially impaired in early PTI responses 87 making it difficult to assess the extent to which BAK1/SERK3 undergoes endocytosis after flg22 perception. Interestingly, AVR3a exhibits some degree of specificity in suppressing PAMP-triggered immunity. Unlike the P. syringae effector AvrPtoB, which blocks the responses elicited by bacterial flagellin and chitin [88][89][90] , none of the AVR3a variants suppressed chitin-induced ROS burst suggesting that AVR3a specifically suppresses BAK1/SERK3-dependent immune responses. At this stage, we propose that AVR3a targets BAK1/SERK3 cellular trafficking initiated at the cell periphery but additional work is needed to clarify the underlying mechanisms.
Our finding that AVR3a co-immunoprecipitates with DRP2 does not necessarily imply that these two proteins directly bind in-planta nor that DRP2 is a target of the effector. DRP2 could be a helper of AVR3a that functions as a cofactor or enables localization of AVR3a to particular subcellular compartments as defined by Win et al. 4 . Nonetheless, our results place AVR3a in proximity to a membrane-remodeling complex that is implicated in endocytosis of the cell surface pattern-recognition receptor FLS2. Remarkably, AVR3a affected FLS2 but not BRI1 endocytosis, indicating that the mechanisms for FLS2 and BRI1 internalization may be different. This is not completely unexpected, as previous reports have shown that the regulatory role of BAK1/SERK3 in BR and PTI signaling are distinct and can be mechanistically uncoupled 70,87 . Moreover, there are different pools of BAK1/SERK3 that are not interchangeable between BRI1 and FLS2 and activation by BR or flg22 does not cross activate these signaling pathways 91 . Therefore, one possibility is that FLS2 and BRI1 share a common endocytic pathway but AVR3a only blocks the FLS2 receptor complex activation and prevents the downstream endocytosis.
However, consistent with the specific activity of AVR3a on FLS2 internalization, DRP2 is specifically required for FLS2-endocytosis but not BRI1. This supports a model in which different internalization pathways for receptor endocytosis may occur following flg22 and brassinosteroid perception. For instance, AVR3a may perturb DRP2 functions only at sites where the activated PRR complexes accumulate and prevent receptor internalization. Our results indicate that BAK1/SERK3-associated membrane-bound receptors may be initially internalized via different pathways, which later on can converge at the late endosomal compartments.
In animal systems, canonical dynamin proteins mediate pinching off clathrin-coated vesicles from the membrane during constitutive endocytosis 52,55 . In Arabidopsis, the two canonical dynamins DRP2A and DRP2B have been shown to play a role in endocytosis and to be genetically and functionally redundant 78,83 . Recently, Smith et al 84 ., showed that in Arabidopsis, FLS2 endocytosis is partially dependent on DRP2B, but not DRP2A 84 . Interestingly, the Solanaceae DRP2 family has expanded relative to Arabidopsis with no obvious orthologs of DRP2A and DRP2B ( Figure S6). 84 ., our results also indicate that the DRP2 family has evolved multiple activities, with the Solanaceous DRP2 being involved in FLS2 but not BRI1 internalization ( Figure 5; Figure S4B). Possibly, plant DRPs have diversified to enable increased plasticity in response to biotic and abiotic stimuli. Further studies are clearly needed to fully understand the precise contributions of different plant DRPs to vesicle trafficking.

Consistent with Smith et al
In summary, we link the P. infestans effector AVR3a to a membrane complex that includes the vesicle trafficking protein DRP2. This complex mediates endocytosis of the classic pattern recognition receptor FLS2 following activation by the flagellinderived peptide flg22. Although the FLS2 co-receptor BAK1/SERK3 is required for basal immunity against the oomycete P. infestans, there is no evidence that FLS2, a receptor for bacterial flagellin, is activated and internalized during infection by this oomycete pathogen 39 . Therefore, AVR3a suppression of FLS2 response and endocytosis may indicate that this effector targets a common node shared by FLS2 and a yet to be discovered PRR involved in oomycete immunity. Future studies are required to address this possibility and further determine the various mechanisms by which the multifunctional RXLR-WY effector perturbs PTI signaling. Lines with different expression profiles were selected for further analysis. Transgenic lines used in this study were T4 and T3 for N. benthamiana and A. thaliana respectively.

A. tumefaciens-mediated transient gene expression assays in N. benthamiana
Agrobacterium tumefaciens (strain GV3101) carrying the desired T-DNA construct was grown overnight at 28°C in Luria-Bertani culture medium with the appropriate antibiotics. Cells were harvested by centrifugation at 8000 g and resuspended in agro-infiltration media [5mM MES, 10 mM MgCl 2 , pH 5.6] prior to syringe infiltration
tumefaciens containing TRV2 and TRV1 plasmids were grown and prepared individually for agro-infiltration assays as previously described 93,94 . A. tumefaciens

Reactive oxygen species measurement
Generation of reactive oxygen species (ROS) was measured as previously described 89  Nb11538 or Nb09838 were predicted to specifically anneal at the 3'UTR of both sequences whereas primers for Nb05397 and Nb31648 were predicted to specifically anneal along the GTPase domain of Nb05397 and Nb31648.

Elicitor preparations
Chitin (crab shell chitin) and flg22 (QRLSTGSRINSAKDDAAGLQIA) peptides were purchased from SIGMA and EzBiolab, respectively and dissolved in ultrapure water.
INF1 was purified from P. infestans 88069 by chromatography and the final working solution was dissolved in ultrapure water 36,96 .  Table 3.

Confocal microscopy
Standard            Green asterisks indicate the homologs of Arabidopsis DRP2A and DRP2B in N.
benthamiana targeted by silencing in this study.     Table S1. Plant proteins that associate with AVR3a KI in planta. Table S2. Sequence identifiers used in Figure S6.