A PDLP-NHL3 complex integrates plasmodesmal immune signaling cascades

Significance Plants close plasmodesmata to restrict cell-to-cell communication after pathogen perception and in response to a range of stresses. All these stimuli trigger callose deposition at the plasmodesmal neck, suggesting a convergence of signaling. We have defined the hierarchy of molecular components and signals required to mediate plasmodesmal closure in immune responses, identifying a PDLP–NHL3 complex as a critical node that integrates multiple signaling cascades to regulate plasmodesmata.

The innate immune system of plants relies in part on the perception of invading pathogens at the cell surface. Cell-surface perception initiates signal transduction to induce cellular responses via molecular machinery that includes receptor kinase complexes, secondary messengers, and enzymes. This machinery can function across different subcellular structures, cells, and tissues, and be co-opted to mediate diverse defense responses. A single stimulus can initiate multiple signaling cascades: For example, receptor activation can trigger a range of responses that include a burst in production of reactive oxygen species (ROS) and changes to gene expression. Further, different stimuli can independently initiate cascades that converge on common responses such as the activation of callose synthesis in different cellular locations triggered by a range of microbial elicitors. By understanding the hierarchy and relationship of different signaling cascades, we can reveal the immune signaling network and identify critical nodes.
Plasmodesmal closure in response to a range of pathogen and stress elicitors is ultimately executed by callose deposition at the plasmodesmata neck, produced by callose synthases (4). PLASMODESMATA-LOCATED PROTEIN 5 (PDLP5) positively regulates callose deposition via CALLOSE SYNTHASE 1 (CALS1) in response to the defense hormone salicylic acid (SA) (5), but the components that specifically mediate plasmodesmal callose deposition in response to chitin and flg22 are not yet known. Further, the observations that SA (6) and ROS produced by the NADPH oxidase RESPIRATORY BURST OXIDASE HOMOLOGUE D (RBOHD) trigger plasmodesmal responses (3) identify that secondary messengers common to a range of immune responses also play specific roles in the activation of callose synthesis at plasmodesmata. Therefore, characterization of the signaling pathways triggered by different elicitors at plasmodesmata will define nodes that integrate distinct signals to enable a common response and establish how specific responses are executed at plasmodesmata.
We sought to leverage our knowledge of plasmodesmal machinery that functions in specific responses to determine the relationships between signaling cascades triggered by different immune elicitors. We found that flg22, like chitin, elicits a cascade that signals via RBOHD, and that PDLP1, PDPL5, and CALS1 are common to microbe elicitor-and SA-triggered responses. To dissect the possibility that PDLPs act as a signaling nexus and understand how they activate callose synthesis, we identified NDR1 (NON-RACE SPECIFIC DISEASE RESISTANCE)/HIN1 (HAIRPIN-INDUCED)-LIKE protein 3 (NHL3) as a PDLP5 interactor. Specifically, NHL3 is a key plasmodesmal component required for PDLP-mediated activation of callose synthesis. The data support a model in which NHL3 interacts with PDLPs to form a central integrator of plasmodesmal immune signaling, transmitting information to activate CALS1 and plasmodesmata closure.

Results
Chitin-and flg22-Triggered Plasmodesmata Closure Signal via RBOHD to a Downstream Nexus Requiring PDLP5. To investigate whether and where plasmodesmal signaling pathways converge upstream of callose deposition, we surveyed known plasmodesmal signaling components for their role in response to specific elicitors. Focusing initially on chitin-and flg22-triggered plasmodesmal responses, we hypothesized that as chitin-triggered plasmodesmal closure is mediated by activation of RBOHD (3), and that flg22 triggers RBOHD activation (7), RBOHD might also mediate flg22-triggered plasmodesmal signaling. Thus, we assayed for flg22-triggered plasmodesmal closure in rbohD mutants using microprojectile bombardment assays and found that rbohD cannot close plasmodesmata in response to flg22 ( Fig. 1A and Dataset S1A), demonstrating that flg22-triggered plasmodesmal closure signals via RBOHD. For chitin, RBOHD dependence is linked to phosphorylation of S133 and S347 (3), and we therefore assayed for flg22-triggered plasmodesmal closure in rbohD mutants complemented with alleles of RBOHD in which active phosphorylation sites (7) are mutated: RBOHD S163A , RBOHD S133A , RBOHD S343A/S347A , or RBOHD S39A/S339A/S343A . Chitin and flg22 reduce GFP movement to neighboring cells in Col-0 but not in pdlp5 (n ≥ 58 bombardment sites). (C) Chitin and flg22 reduce GFP movement to neighboring cells in Col-0 but not in cals1 (n ≥ 120 bombardment sites). SA reduces the movement of GFP to neighboring cells in Col-0, cml41 (D, n ≥ 119 bombardment sites) and lym2 (E, n ≥ 117 bombardment sites). (F) Chitin and flg22 reduce GFP movement to neighboring cells in Col-0 and ics1 (n ≥ 114 bombardment sites). (G) flg22 and SA reduce GFP movement into neighboring cells in Col-0, but not in pdlp1 (n ≥ 107 bombardment sites). Asterisks indicate statistical significance compared to the mock treatment within a genotype: *P < 0.05, ***P < 0.001. This revealed that neither RBOHD S133A nor RBOHD S343A/ S347A complemented the loss of rbohD-dependent flg22-induced plasmodesmal closure (SI Appendix, Fig. S1 and Dataset S1B) and that S133 and S347 of RBOHD are required for both flg22-and chitin-triggered plasmodesmal response.
SA-triggered plasmodesmata closure is independent of RBOHD (5) but, like chitin and flg22, induces callose deposition. Chitinand flg22-induced plasmodesmal closure occurs via callose deposition within 30 min (2, 3) and so to determine if SA might plausibly exploit a similar signaling mechanism, we assayed for SA-induced callose deposition at plasmodesmata. Quantifying aniline blue-stained plasmodesmata-associated callose, we observed that SA induces plasmodesmal callose deposition within 30 min of treatment (SI Appendix, Fig. S2). Therefore, SA responses have a similar temporal profile to chitin and flg22 responses, allowing for the possibility that SA, chitin, and flg22 induce plasmodesmal closure via a similar mechanism.
To explore whether flg22, chitin, and SA signaling converge on the same pathway upstream of callose deposition, we tested the requirement for known mediators of SA-triggered plasmodesmata closure, PDLP5 and CALS1, in chitin and flg22 responses. Using microprojectile bombardment assays, we found that neither pdlp5 ( Fig. 1B and Dataset S1C) nor cals1 ( Fig. 1C and Dataset S1D) mutants closed their plasmodesmata in response to flg22 or chitin. As CALS1 produces callose and must lie at the endpoint of the signaling pathways, we conclude that flg22, chitin, and SA signaling at plasmodesmata converge downstream of RBOHD, and at, or upstream of, PDLP5.
Immune-Associated SA Production Is Not Required for Elicitor-Induced Plasmodesmal Closure. As SA is known to be synthesized in response to flg22 treatment (8,9), we aimed to determine whether SA mediates flg22-and chitin-triggered plasmodesmal closure. Given that rbohD exhibits normal plasmodesmal closure response to SA (5), and that RBOHD is required for both chitinand flg22-triggered plasmodesmal closure, we reasoned that a role for SA-triggered and/or mediated responses would lie downstream of LYM2 and CML41. To test this, we first generated a null cml41 mutant (SI Appendix, Fig. S3) by CRISPR/Cas9 gene editing that is perturbed in flg22-induced plasmodesmal closure (SI Appendix, Fig. S3D and Dataset S1E) and found that this mutant can close its plasmodesmata in response to SA ( Fig. 1D and Dataset S1F). Likewise, lym2 ( Fig. 1E and Dataset S1G) shows wild-type-like SA-induced plasmodesmata closure.
To assay for a direct role of SA in chitin-and flg22-triggered plasmodesmal closure, we asked whether these elicitors trigger plasmodesmal closure if SA synthesis is perturbed. Thus, we assayed for plasmodesmal responses in the ISOCHORISMATE SYNTHASE 1 (ics1) mutant, in which pathogen-induced SA synthesis and accumulation is significantly reduced (10). We observed that ics1 plasmodesmata can close in response to chitin and flg22, and therefore chitin-or flg22-triggered SA production is not necessary for plasmodesmal signaling ( Fig. 1F and Dataset S1H). Collectively, these data suggest SA itself does not act as a secondary messenger within microbe-triggered plasmodesmal cascades, but rather initiates a signaling pathway that converges with other responses.

PDLP1 Is Required for Immune Elicitor-Triggered Plasmodesmata
Closure. PDLP1 positively regulates callose deposition at plasmodesmata and at the downy mildew haustorial membrane (11) but has not been associated with immune-associated plasmodesmal responses. However, it has been shown that PDLP5 and PDLP1 can interact (12), suggesting PDLP1 might cooperatively function in the same responses as PDLP5. To test this, we assayed the pdlp1 mutant for its ability to respond to flg22 and SA and found neither induced plasmodesmal closure ( Fig. 1G and Dataset S1I). Therefore, PDLP1 is also required for elicitor-triggered signaling cascades in plasmodesmata, supporting a model in which PDLPs are common to multiple signaling pathways and function downstream of, or at, the point at which these signals converge.

The NDR1/HIN1-Like Protein NHL3 Interacts with PDLP5 at
Plasmodesmata. Having established that PDLP5 functions in MAMP-and SA-triggered plasmodesmal signaling, and considering it is a positive regulator of callose deposition (5, 13), we hypothesized that PDLP5 is a component of a central node in immune-triggered plasmodesmal signaling cascades. To characterize its function in this context, we aimed to identify PDLP5 interactors that might link PDLPs to callose synthase activity. Thus, we commissioned a split-ubiquitin screen of an Arabidopsis thaliana (Arabidopsis) cDNA library with a PDLP5 bait (DualSystems Biotech, Switzerland; now Hybrigenics Services, France). The screen identified 65 prey hits, and we refined this list based on their presence in plasmodesmal proteomes of Arabidopsis suspension culture cells (14,15), producing 13 prey proteins (SI Appendix, Table S1). NHL3 (AT5G06320) is present in both proteomes and is a membrane-associated protein like PDLP5. Thus, we chose NHL3 for further characterization as a likely PDLP5 interactor.
We first examined the subcellular localization of NHL3 to determine whether a bona fide interaction with PDLP5 is plausible. We C-terminally tagged NHL3 with mCherry and observed , and in the presence of acceptors NHL3-mCherry (NHL3) or LTI6b-mCherry (LTI6b). Independent factor "genes expressed" was determined as significant (ANOVA; F = 6.7, df = 2, P < 0.01), with significant differences denoted by a and b, P < 0.05. Each sample is represented by six measurements from five biological replicates. that NHL3-mCherry is located in the plasma membrane, as previously reported [(16, 17), Fig. 2A]. However, NHL3-mCherry fluorescence also accumulated in puncta along the plasma membrane that colocalized with both PDLP5-eGFP and PDLP1-eGFP, indicating that it accumulates in plasmodesmata ( Fig. 2A).
To further test this association, we used Förster resonance energy transfer-fluorescence lifetime imaging (FRET-FLIM) analysis. As NHL3 has been differentially predicted to carry one or two transmembrane domains (16), and a strategy for epitope tagging to allow FRET was unclear, we first resolved the membrane topology of NHL3 with bimolecular fluorescence complementation by transient expression in N. benthamiana. NHL3 was either N-or C-terminally tagged with the N terminus of yellow fluorescent protein (YFP N ) and coexpressed with a cytosolic (YFP C ) (SI Appendix, Fig. S4 A and B) or a secreted (SP-YFP C ) C terminus of YFP (SI Appendix, Fig. S4 C and D). YFP fluorescence was visible in the positive control combinations: cytosolic YFP C coexpressed with cytosolic YFP N (SI Appendix, Fig. S4E), and SP-YFP C with SP-YFP N , (SI Appendix, Fig. S4F). YFP fluorescence was similarly detected when NHL3-YFP N or YFP N -NHL3 was coexpressed with cytosolic YFP C (SI Appendix, Fig. S4 A and B), but not with SP-YFP C (SI Appendix, Fig. S4 C and D). Therefore, both termini of NHL3 are cytosolic facing.
We assayed for interaction between NHL3-mCherry and PDLP5-eGFP fusion proteins using FRET-FLIM. As a negative control, we fused LTI6b to mCherry and an ACT2 promoter to drive expression (pACT2::LTI6b-mCherry) and found that, unlike previous reports (18), LTI6b localizes to the plasma membrane and was visible in puncta in the plasma membrane coincident with PDLP5-eGFP (SI Appendix, Fig. S5). This allowed us to resolve our FRET-FLIM analysis to plasmodesmata within the plasma membrane. We found that the fluorescence lifetime of PDLP5-eGFP was reduced at plasmodesmata in the presence of NHL3-mCherry, but not LTI6b-mCherry ( Fig. 2C and Dataset S2), indicating that FRET occurs between PDLP5 and NHL3. Alongside co-IP data and the initial split ubiquitin screen, this supports a model in which PDLP5 and NHL3 form a complex for plasmodesmal signaling.

NHL3 Is Required for Immune-Associated Plasmodesmata
Closure. The NHL gene family has homology to NDR1 and the TOBACCO HIN1 genes (19). However, while HIN1 and NHL members have been associated with various aspects of plant defense and stress responses (16,(20)(21)(22)(23), with NHL3 interacting with OZF1, a zinc-finger protein that positively influences SA signaling in defense against pathogens (24), the specific molecular functions of NHLs are unknown. Given the plasmodesmal localization of NHL3, and its interaction with PDLP5, we investigated the plasmodesmal responses of nhl3 mutants to chitin, flg22, H 2 O 2 , and SA. Plants carrying the nhl3 mutant alleles, nhl3-1 (25) and nhl3-2 (24), could not close their plasmodesmata in response to chitin or flg22 (Fig. 3A and Dataset S3A). Furthermore, while nhl3 mutants were able to close their plasmodesmata in response to H 2 O 2 , they could not in response to SA (Fig. 3B and Dataset S3B), phenocopying the pdlp5 and cals1 mutants (5,6). Thus, NHL3 is required for elicitor-triggered plasmodesmal responses.

NHL3 Positively Regulates Callose Deposition at Plasmodesmata.
Reasoning that NHL3 likely cooperates with PDLP5 to positively regulate callose deposition at plasmodesmata, we assayed for the impact of NHL3 overexpression on callose deposition at plasmodesmata using quantitative imaging of aniline bluestained callose. When NHL3 and PDLP5 were transiently expressed individually or together in N. benthamiana, there was a significant increase in total aniline blue fluorescence in a plasmodesmal callose deposit compared to the infiltration control (Fig. 4A), independent of whether NHL3 was N-or C-terminally tagged (SI Appendix, Fig. S6 and Dataset S4A). When NHL3 and PDLP5 were coexpressed, we did not detect a further increase in callose per plasmodesmal deposit, but we did detect an increase in the number of deposits detected per field of view compared to when both proteins were individually expressed (SI Appendix, Fig. S7). This might arise from more plasmodesmata producing callose above the detection threshold and indicate increased callose synthesis. These data support a function of NHL3 in cooperating with PDLP5 to increase the activity of callose synthases, thereby positively regulating callose deposition at plasmodesmata. PDLP Function Requires NHL3. To dissect the hierarchy of the relationship between PDLP5 and NHL3, we asked whether PDLPs require NHL3 to function. For this, we used plants that overexpress PDLP5 [PDLP5OE, (13)] and PDLP1-GFP [PDLP1OE (4)], which are both phenotypically dwarfed in comparison to wildtype plants. First, we crossed PDLP5OE with nhl3-2 (PDLP5OE × nhl3-2), and found the mutation in nhl3-2 reverted the dwarf phenotype of PDLP5 overexpression (Fig. 4B, Dataset S4B, and SI Appendix, Fig. S8). In light of this dwarf reversion, and considering PDLP5 and NHL3 are associated with positive regulation of callose at plasmodesmata [(13), Fig. 4A], we assayed for callose deposition at plasmodesmata in PDLP5OE × nhl3-2 lines, relative to PDLP5OE. We confirmed that PDLP5OE plants exhibit enhanced callose deposition at plasmodesmata in our growth conditions and found that this was reverted to wild-type levels in PDLP5OE × nhl3-2 plants (Fig. 4C). Given it has previously been shown that PDLP5OE plants show increased resistance to bacterial and fungal pathogens (13,26), we tested whether the absence of NHL3 would counteract this using the fungal pathogen Botrytis cinerea. PDLP5OE plants showed a significant reduction in lesion size in comparison to Columbia-0 (Col-0) (Fig. 4D, SI Appendix,  Fig. S10, and Dataset S4 C and D), indicating that overexpression of PDLP5 enhances resistance. By contrast, there was no significant difference in the size of the disease lesions that developed on Col-0 and PDLP5OE × nhl3-2 plants, indicating the conferred resistance of PDLP5OE is lost with the absence of NHL3.
To investigate whether PDLP1 function also depends on NHL3, we first crossed PDLP1OE with nhl3-1 (PDLP1OE × nhl3-1) and found that the absence of NHL3 also reverts the dwarf phenotype of PDLP1OE (Fig. 4E and Dataset S4E). PDLP1 is also involved with callose regulation at cell locations other than plasmodesmata (11), and we observed an overall increase in the deposition of callose across the cell in PDLP1OE that interfered with plasmodesmal callose quantification [SI Appendix, Fig. S9, (11)]. Therefore, we assayed for GFP flux through plasmodesmata using microprojectile bombardment. While PDLP1OE showed increased plasmodesmatal closure, this was reverted to wild-type levels in PDLP1OE × nhl3-1 (Fig. 4F and Dataset S4F). Therefore, NHL3 is required to mediate PDLP overexpression-induced plasmodesmal regulation. Overall, these data suggest that NHL3 is common to downstream functions of both PDLP1 and PDLP5, integrating a range of signals for plasmodesmal function.

Discussion
Plants deploy a suite of signaling cascades with specific molecular components to initiate and execute defense responses against invading pathogens. These cascades, and the machinery they require, can be specific to subcellular structures such as is observed in plasmodesmata, where microbe elicitors trigger discrete responses. Despite the specificity in signaling observed in plasmodesmata, pathways initiated by different elicitors converge on the common response of callose synthesis to mediate plasmodesmal closure. Thus, plasmodesmal responses likely converge at a node that integrates multiple independent inputs.
determine that immune-associated signaling cascades progressively converge on a single node involving PDLPs and NHL3, upstream of CALS1 activation (Fig. 4G). Our data indicate that at the top of the signaling cascade, chitin and flg22 both relay signals via specific receptors and molecules to RBOHD [(3), Fig. 1A and SI Appendix, Fig. S1]. As an NADPH oxidase, RBOHD produces ROS that act as small molecule transmitters of plasmodesmal responses. How ROS are perceived and transmit information within this cascade remains an open question and the observation that hydrogen peroxide can induce plasmodesmata closure via an independent pathway (5) complicates experimental dissection of this signaling step without a candidate receiver. However, increasing knowledge of proteins present in plasmodesmata and the role which they play will likely reveal candidate components to fill this gap.
A previous study suggested that SA-triggered plasmodesmal responses occur independently of RBOHD (5), and our work here further demonstrates that SA-elicits a signaling cascade that converges on microbial-triggered responses downstream of RBOHD. Our observation that SA-induced plasmodesmal closure occurs within 30 min (SI Appendix, Fig. S2), as well as its dependence on PDLP1 (Fig. 1G) which is not transcriptionally induced by SA (27), challenges the hypothesis that SA-triggered plasmodesmal closure occurs only via SA-induced expression of PDLP5. However, how SA might elicit plasmodesmal closure via another mechanism is not currently clear.
Leveraging our understanding of PDLPs as a common component of different immune-associated plasmodesmal responses, we identified NHL3 as a cooperative interactor of PDLP5. We found that NHL3 positively regulates PDLP5-and PDLP1-mediated callose deposition and functions in elicitor-triggered plasmodesmal responses. Microprojectile bombardment assays produce datasets with high variance, and we interpret the loss of elicitor-induced responses in both pdlp1 and pdlp5 mutants as an indication that both PDLPs are required and essential components of the signaling cascade. Thus, we propose PDLP1 and PDLP5 function in complex with NHL3, forming an integrator of immune signal inputs (Fig. 4G). However, our bombardment data might represent a partial response and allow for a model in which PDLP1 and PDLP5 each form a complex with NHL3 and that signaling might pass via either to trigger callose synthesis (SI Appendix, Fig. S11). Irrespective of the working model, PDLPs are only found in seed plants (28)(29)(30)(31)(32), while NHL-related proteins have orthologues the earlier land plant Physcomitrium patens (30,33). This allows speculation that PDLPs were co-opted into a preexisting complex of NHLs that activates callose synthases to enable an extra regulatory mechanism for plasmodesmata responses.
How NHL3 mediates signaling is not clear as its only annotated domain is a LATE EMBRYOGENESIS ABUNDANT (LEA) protein LEA_2 domain found close to the C terminus, which has no known function. LEA_2 domain proteins are produced during responses to dehydration or hyperosmotic stress, and have been postulated to stabilize macromolecules against freezing, dehydration, ionic or osmotic stress (34). They have a hydrophobic (35), glycine-rich chemical composition (36), and are generally predicted to be intrinsically unstructured proteins with a left-handed polyproline II helix conformation (35,(37)(38)(39). This disordered conformation is thought to form tight hydrogen-bond networks in concert with carbohydrates, providing stability to macromolecular structures (34,40). Thus, any role for NHL3 in plasmodesmal signaling might relate to recruitment, aggregation, or stabilization of signaling or response components. Given that the NHL protein family has many members in Arabidopsis, and that another member has also been found to have plasmodesmata-associated localization and functions (41), the mode of action of NHLs might regulate plasmodesmata in diverse response contexts.
PDLPs have two extracellular DOMAIN OF UNKNOWN FUNCTION (DUF) 26 domains. Structural analysis of the PDLP5 DUF26 domains suggests these are lectin dimers, which could form an extended binding cleft for a carbohydrate polymer (28). While the PDLP5 DUF26 domain did not bind to a range of cell wall-derived carbohydrates (28), callose binding has not been tested. Considering the LEA_2 domain sits on the cytosolic C terminus of NHL3, it seems possible that NHL3 might serve as a scaffold for PDLP5 and CALS1, thereby stabilizing the machinery required for callose deposition in plasmodesmata. Indeed, such a model might challenge the notion of a linear signaling pathway, in that PDLP1/5 and NHL3 might not only activate callose synthesis but could also bind and stabilize callose at the site of synthesis via the DUF26 domains of PDLPs. The observation that PDLP1 and PDLP5 overexpression phenotypes are reverted to wild-type levels in the absence of NHL3 (Fig. 4 B-F) suggests NHL3 either acts downstream of PDLPs or that they are codependent. A phloem-specific PDLP-CALS interaction has recently been identified (42); resolving whether PDLP5 or NHL3 directly activates CALS1, or whether they transmit a signal to an intermediary component, will further dissect how the PDLP5/NHL3 complex acts to regulate callose synthases.
How plasmodesmata specifically respond to a range of immunity-and stress-associated elicitors is a critical question that underpins how organism-level responses are regulated in a multicellular context. Further, how immune signals are localized, specialized, and integrated is key to mapping the molecular signaling networks that orchestrate cellular responses to pathogens. Characterization of plasmodesmal signaling has revealed that a PDLP-NHL3 complex integrates different signaling cascades to activate callose synthesis and close plasmodesmata during immune responses. Thus, we have uncovered a mechanism by which diverse signals are integrated to produce a single response, further characterizing the signaling cascades that regulate cell-to-cell connectivity during immune responses. How NHL3 functions and the full range of signals it can integrate, and resolving how small molecules such as ROS and SA are received in plasmodesmal responses, remain questions to be addressed to understand the molecular dynamics of plasmodesmal closure and callose synthesis across plant responses.

Materials and Methods
Extended Methods and Materials are available in SI Appendix. Microprojectile Bombardment Assays. Bombardment assays were performed as described in ref. 45. Bombardment sites were imaged 16 h after bombardment using a confocal microscope (LSM Zeiss 800) with a 20× water dipping objective (W N-Achroplan 20×/0.5).
Plasmodesmata Callose Staining and Quantification. The 7th or 8th Arabidopsis leaf or N. benthamiana leaf was infiltrated with 1% or 0.1% aniline blue in PBS buffer (pH 7.4), respectively. Plasmodesmata callose deposits were imaged from the abaxial side by a confocal microscope (LSM Zeiss 800) using a 63× water immersion objective (Plan-Apochromat 63×/1.0 M27). Aniline blue was excited with a 405-nm UV laser and collected at 430 to 550 nm. Z-stacks from multiple areas (specified in figure legends) per leaf were collected. We used an automated image analysis pipeline to quantify aniline blue-stained plasmodesmata, available at ref. 46.
Split-Ubiquitin Screen. The split-ubiquitin screen was performed by DualSystems Biotech, Switzerland (now known as Hybrigenics Services, France).
Confocal Microscopy. Transiently expressed constructs for colocalization and split-YFP analysis were imaged by confocal microscopy (LSM Zeiss 800) 2 d postinfiltration using a water immersion 63× objective (C-Apocromat 63×/1.20WKorr UV VIS IR-water). GFP/YFP was excited with a 488-nm argon laser and collected at 505 to 530 nm while mCherry was excited with a 561-nm DPSS laser and collected at 600 to 640 nm.
FRET-FLIM. Leaves of N. benthamiana transiently expressing constructs of interest were imaged using a STELLARIS 8 FALCON microscope (Leica Microsystems). The abaxial side of the leaf sample was imaged using a 63× objective (Leica, HC PL APO CS2 5 63×/1.20 Water). GFP was excited with a 488-nm WLL laser and collected at 503 to 530 nm. Acquisition was set at a mode of xyz, format size of 512 × 512, scan speed of 40 Hz, single-line repetitions, and a pixel dwell time of 3.16 µs. To reduce the risk of bleaching the sample, laser power was adjusted to ensure no more than 1.0 photon per pulse occurred. Datasets were acquired by scanning each image until a suitable number of photon counts per pixel were reached in the brightest channel, up to a maximum of 1,000 counts per pixel. Data were analyzed using the LAS X software. Plasmodesmal regions of interest were specified using the intensity threshold and manual selection. Using a two-exponential decay for GFP, a fit was generated and the mean τ amplitude-weighted GFP lifetime (in ns) for all plasmodesmata per image was calculated.
Rosette Area Phenotyping. For rosette area phenotyping and subsequent callose staining, Arabidopsis seeds were surface sterilized with 70% ethanol and 2% bleach with 0.02% Triton X-100, then washed with sterile dH 2 O and sown onto Murashige and Skoog medium 1% sucrose 0.8% agar plates 16-h light at 22 °C. Photos were taken 21 d postgermination for determination of rosette area. Individual plants were traced by hand, and area of the rosette was measured in FIJI.
B. Cinerea Infection Assay. B. cinerea spores were harvested in 0.25× potato dextrose broth and adjusted to a concentration of 2.5 × 10 5 spores/mL. The spore suspension was incubated at room temperature for 2 h with continual shaking to germinate the spores. Expanded leaves of 5-wk-old Arabidopsis plants adhered in a 1.5% agar H 2 O plate, were inoculated with 2 µL of the spore suspension, with a droplet placed in between the mid vein and the leaf edge of each leaf side (i.e., two droplets per leaf). Plates were sealed and left in a cabinet at 22 °C, with 10-h light. Developing disease lesions were photographed 2 d postinoculation and measured using FIJI. Each prepared plate had three to four leaves per genotype, and the average lesion area was taken for each genotype as a single biological replicate. Data, Materials, and Software Availability. All data measurements used in this paper are available in SI Appendix. Raw image data are available on the BioImage Archive (https://www.ebi.ac.uk/biostudies/bioimages/studies/S-BIAD626) (47).