Synergy among Exocyst and SNARE Interactions Identifies a Functional Hierarchy in Secretion during Vegetative Growth[CC-BY]

Emily R. Larson,a,1,2 Jitka Ortmannová,b,1,3 Naomi A. Donald,a Jonas Alvim,aMichael R. Blatt,a,4 and Viktor Žárskýb,c a Laboratory of Plant Physiology and Biophysics, Bower Building, University of Glasgow, Glasgow G12 8QQ, United Kingdom b Institute of Experimental Botany, Academy of Sciences of the Czech Republic, 165 02 Prague 6, Czech Republic c Department of Experimental Plant Biology, Faculty of Science, Charles University, 128 44 Prague 2, Czech Republic


INTRODUCTION
Exocytosis depends on the spatiotemporal formation, delivery, and fusion of membrane vesicles that distribute their cargos to the plasma membrane and apoplast. Proteins that regulate exocytotic mechanics include the exocyst complex, which tethers and docks vesicles to target membranes (TerBush et al., 1996;Zarský et al., 2009;Vuka sinović andŽárský, 2016), and soluble NSF attachment protein receptor (SNARE) protein complexes that drive the final steps of vesicle fusion (Jahn and Scheller, 2006;Bassham and Blatt, 2008;Südhof and Rothman, 2009). Structural protein models suggest the exocyst assembles as a set of helical bundles (Picco et al., 2017;Mei et al., 2018) that are evolutionary conserved and comprise eight subunits (TerBush et al., 1996;Zarský et al., 2013): SEC3, SEC5, SEC6, SEC8, SEC10, SEC15, EXO70, and EXO84. The EXO70 family of subunits in land plants shows substantial gene expansion, with 23 paralogues in Arabidopsis (Arabidopsis thaliana) that may reflect specialization in the maintenance of different tethering domains (Synek et al., 2006;Cvrčková et al., 2012). SNARE proteins are classified by the Q or R amino acid residue found at the core of the interacting assembly and localize to vesicle (R-SNAREs or vesicle-associated membrane proteins [VAMPs]) and target (Qa-, Qb-, and Qc-SNAREs) membranes (Fasshauer et al., 1998;Blatt and Thiel, 2003;Jahn and Scheller, 2006). SNARE function is conserved within eukaryotes, but these gene families are similarly expanded in land plants consistent with evolutionary specialization (Sanderfoot et al., 2000;Blatt and Thiel, 2003;Pratelli et al., 2004). The Qa-SNAREs, SYP121 and SYP122, have overlapping functions at the plasma membrane and are associated with accelerated traffic during growth as well as specialized functions in abiotic and biotic stress resistance Karnik et al., 2017;Waghmare et al., 2018). SYP121 and SYP122 assemble with cognate R-SNAREs VAMP721 and VAMP722 as functional complexes that fuse vesicles with the plasma membrane (Kwon et al., , 2013Karnik et al., 2015).
Current models differentiate temporally and mechanistically between vesicle tethering mediated by the exocyst and fusion mediated by the SNAREs to highlight the sequential functions of these protein complexes (Cai et al., 2007;Mei and Guo, 2019). Nonetheless, with the spatial and functional proximities of the two complexes, increasing evidence for interactions between subunits of the separate complexes has surfaced. Several exocyst subunits in yeast interact with SNAREs in vitro or affect their localization within the cell (Sivaram et al., 2005;Morgera et al., 2012;Shen et al., 2013;Yue et al., 2017), leading to the idea of a coordination between exocyst and SNARE complexes at the plasma membrane (Picco et al., 2017;Mei and Guo, 2019). In Arabidopsis, the exocyst EXO70B1 and EXO70B2 subunits interact in vitro with the SNAP33 subunit of the plasma membrane SNARE complex (Pecenková et al., 2011), although the significance of these interactions in vivo is unknown.
One difficulty is the paucity of in vivo data that might speak to the physiological consequences of exocyst and SNARE subunit interactions, especially in plants. Inhibiting SNARE function has been shown to disrupt exocytosis in Arabidopsis and other plant models (Geelen et al., 2002;Tyrrell et al., 2007;Karnik et al., 2013;Grefen et al., 2015), and impairing exocyst function affects growth and, in some instances, SNARE distributions within cells Synek et al., 2017). These observations are consistent with the concept of serial processing from tethering through to vesicle fusion. What has been missing is evidence arising from an interaction between one or more exocyst and SNARE subunits that surpasses any functional consequences of the subunits in isolation.
We report here the interactions of members of the EXO70 exocyst subfamily with SNARE proteins that drive vesicle fusion at the plasma membrane and the consequences for secretory vesicle traffic. We show that the omnipresent EXO70A1 exocyst subunit requires the so-called longin regulatory domain of the R-SNARE VAMP721 for interaction. Most important, we report that the loss of a subset of interactors among exocyst and SNARE proteins affecting vesicle traffic leads to synergistic impacts on cell expansion and vegetative growth that are unique to EXO70 interactions with the R-SNARE. The findings demonstrate a functional hierarchy in interactions between EXO70A1 with VAMP721 that suggest a handover of vesicles between exocyst and SNARE complexes that leads to vesicle fusion and establishes a landmark for further exploration of secretory mechanics in plants.

RESULTS
The Exocyst and SNARE Proteins Physically Interact EXO70A1 associates dynamically with the plasma membrane, where it has been suggested to provide a landmark for vesicle docking. Indeed, VAMP721 localization is altered in the exo70A1 mutant , consistent with this idea. To assess whether direct interactions between the exocyst and one or more SNAREs may underpin these processes, we used a yeast matingbased split-ubiquitin system (mbSUS; Grefen et al., 2015;Zhang et al., 2018). The mbSUS approach offers a number of advantages over yeast two-hybrid methods in allowing for analysis of interactions between full-length proteins, especially with integral membrane and membrane-associated proteins, and in screening weighted against false-positive interactions that facilitates comparisons of relative binding specificities (Zhang et al., , 2018Xing et al., 2016). We tested for interactions with four EXO70 subunits, including EXO70A1 that is widely expressed in the vegetative plant (Synek et al., 2006), EXO70H1, and two EXO70B subunits that have been associated with pathogen defense and autophagy (Pecenková et al., 2011;Kulich et al., 2013;Wang et al., 2020), challenging these with several plasma membraneassociated SNAREs (Figure 1; Supplemental Figure 1). Strong yeast growth was recovered, notably with EXO70A1 when paired with the plasma membrane Qa-SNAREs SYP121 and SYP122; the plasma membrane R-SNAREs VAMP721, VAMP722, VAMP724; and VAMP727, and to a lesser extent, with the cognate Qbc-SNARE SNAP33.
Increasing Met concentrations in the medium, to suppress bait expression and test the strength of interactions, showed that EXO70A1 has a modest preference for SYP121 over SYP122 and for VAMP721 over VAMP722. It also showed no appreciable interaction with the endomembrane R-SNARE VAMP723, even though the R-SNARE was strongly expressed in each of the assays (Supplemental Figure 2), consistent with a specificity for components assembling in plasma membrane SNARE core complexes. Like EXO70A1, the cognate SNARES SYP121 and VAMP721 are widely expressed throughout the plant (Bassham and Blatt, 2008;Karnik et al., 2017;Waghmare et al., 2018). As these two SNAREs generally interacted most strongly with the EXO70 subunits, especially with the widely expressed EXO70A1 subunit of the exocyst, we focused on these interactors.
To assess the characteristics of exocyst binding with the R-SNARE, we compared interactions of EXO70A1 with VAMP721 and VAMP723. These R-SNAREs are 89% sequence identical and exhibit >94% similarity at the amino acid level (for R-SNARE protein alignment, see supplemental Figure 1 of Zhang et al. [2015]). Previous studies showed that domain exchanges between the longin domains of these R-SNAREs is sufficient to alter their binding with the plasma membrane K 1 channels KAT1 and KC1 and affect channel gating, effects that were traced to a single amino acid residue that differs between the two R-SNAREs . Furthermore, a functional mapping of the putative binding surface exposed a secondary binding site for the K 1 channels that is common to both R-SNAREs , thus arguing against substantial conformational differences between the two R-SNAREs and their mutations.
We used the same constructs of VAMP721 and VAMP723 incorporating the respective domain exchanges  to explore the R-SNARE domains required for interaction with the EXO70A1 subunit bait. We observed yeast growth in mbSUS assays with VAMP723 only when the VAMP721 longin domain was substituted for that of VAMP723; conversely, yeast growth was lost when the longin domain of VAMP723 was substituted for the longin domain in VAMP721. A similar loss of interaction was also observed with the other EXO70 subunits, indicating that the VAMP721 longin domain incorporates an essential structural component determining EXO70 binding ( Figure 1B; Supplemental Figures 3 and 4) much as it does for VAMP721 binding with the K 1 channels. The finding is of interest, because both R-SNARE partners have been proposed as plasma membrane landmarks affecting the final stages of vesicle fusion Karnik et al., 2017;Zhang et al., 2017).
To validate the EXO70 interactions in vivo, we used Förster resonance energy transfer (FRET) after transiently transforming Nicotiana benthamiana leaves with mCherry-EXO70A1 and several GFP-VAMP protein fusions under control of the 35S promoter. The constructs were incorporated within the pFRETgc-2in1 multicistronic vector to ensure equal genetic loads as well as provide markers for expression on a cell-by-cell basis (Hecker et al., 2015). These experiments were combined with mCherry-SYP121 and GFP-VAMP721 and with mCherry-EXO70A1 with free GFP (2xGFP) as positive and negative controls, respectively. To accommodate differences in fluorescence intensities of the protein fusions, we used the PixFRET method (Feige et al., 2005) that corrects for spectral bleed-through relative to fluorophore intensity and for bias arising from differences in fluorophore emission that would otherwise affect comparisons of FRET across (A) Yeast mbSUS assay for interaction of the OST4-anchored EXO70A1-Cub bait and Nub prey fusions with the SNARE proteins as indicated. Yeast dropped as 1.03 and 0.13 dilutions of 0.6 OD 600 yeast suspensions and grown on selective synthetic medium lacking Leu, Trp, Ura, Met (CSM -LTUM ) as a control for mating and on the same medium additionally lacking Ala and His (CSM -LTUMAH ) as a test for bait and prey interaction. NubG and NubI were included as negative and positive prey controls, respectively. Growth on 5 and 50 mM Met to suppress bait expression tests for interaction specificity. Immunoblots confirming bait and prey expression are included in Supplemental Figure 3. interacting partner sets. Analysis of the resulting FRET images showed a strong signal with the SYP121-VAMP721 positive control and a lower, but highly significant, signal when the R-SNARE was paired with EXO70A1, but no appreciable signal between EXO70A1 and VAMP723 or the negative 2xGFP control, despite the comparable levels of donor and acceptor fluorescence ( Figure 2). These data support the capacity for EXO70A1 and VAMP721 to interact in planta and are consistent with the mbSUS findings on heterologous expression in yeast.

Impaired Exocyst Function Inhibits Secretion at the Plasma Membrane
To explore the contributions of EXO70A1 subunit to secretory vesicle traffic, we used the tetracistronic expression vector pTecG-2in1-CC (Karnik et al., 2013;Grefen et al., 2015;Zhang et al., 2015). The pTecG-2in1-CC vector incorporates, within a single backbone, expression cassettes for secreted yellow fluorescent protein (secYFP) to monitor traffic, GFP-HDEL as a marker for transformation and for ratiometric quantification of secretion, and two additional cassettes for expressing proteins of interest. Using this vector ensured equal genetic loads for the encoded proteins, enabling the essential controls for quantification of their actions on a cell-by-cell basis, which would not be possible by other means . Previous studies showed that blocking secretory traffic yielded a substantial increase in the cellular YFP:GFP fluorescence ratio recorded by confocal fluorescence imaging (Karnik et al., 2013) and concomitant and quantitative loss in secYFP protein collected from the apoplast .
We transiently expressed the pTecG-2in1-CC vector with secYFP and GFP-HDEL alone in the wild-type and exo70A1 Arabidopsis seedling roots to visualize and quantify secretion at the plasma membrane. As a positive control for secretory block, seedlings were transiently transformed with the same vector carrying the cytoplasmic domain of SYP121 (SYP121 DC ) that (A) FRET analysis on coexpressing mCherry-EXO70A1 and GFP-VAMP721, GFP-VAMP723 and 2xGFP in transiently transformed N. benthamiana leaf epidermis. Coexpression used the pFRETgc-2in1-NN multicistronic vector and quantification was with the PixFRET tool (Feige et al., 2005;Schindelin et al., 2012). The combinations of mCherry-SYP121 with GFP-VAMP721 and mCherry-SYP121 with 2xGFP were included as positive and negative controls, respectively. Images in the panel rows (top to bottom) are projections of image stacks analyzed for FRET, the GFP (donor) and mCherry (acceptor) fluorescence signals. The FRET analysis is pseudo-color-coded to the scale shown (right). A FRET signal was evident between EXO70A1 and VAMP721, but not VAMP723 or 2xGFP. Bar 5 50 mm. (B) Quantification of FRET efficiency (top) calculated from PixFRET grayscale outputs with correction for the relative intensities of GFP (donor) and mCherry (acceptor) fluorescence signals. Data are means 6 SE of n $ 8 independent experiments, with measurements in each experiment taken from the entire image planes of three to five randomly selected image sets. Quantification of GFP (donor) and mCherry (acceptor) fluorescence signals (middle and bottom, respectively) are for the same data sets and show no significant difference in donor fluorescence. No significant differences between mCherry-EXO70A1 fluorescence signals were observed, and only the mCherry-SYP121 fluorescence was stronger. Note that PixFRET (Feige et al., 2005;Schindelin et al., 2012) corrects for variation in donor and acceptor between data sets as the product of their fluorescence signals. Lowercase letters indicate significance at P < 0.01; uppercase letters indicate significance at P < 0.05. functions as a dominant-negative fragment and suppresses secretion at the plasma membrane (Tyrrell et al., 2007;Grefen et al., 2015). Expressing SYP121 DC in the wild-type and exo70A1 mutant seedlings provided a basis for comparing the impact on secretion in each genetic background.
We quantified a secretory block as the ratio of YFP-to-GFP fluorescence in live imaging of the transiently transformed seedlings by confocal microscopy (Figure 3). In the absence of SYP121 DC , we observed a significant block of secretion in the exo70A1 mutant compared with the wild-type seedlings, evidenced by the elevated mean in the YFP:GFP ratio. Seedlings transformed to express the SYP121 DC fragment in addition to the fluorescent proteins yielded a similar mean YFP:GFP ratio in the exo70A1 mutant. By contrast, the mean YFP:GFP ratio was statistically different and much higher on expressing the SYP121 DC fragment in the wild type than in its absence. These results showed that the impaired exocyst function in the exo70A1 mutant inhibits secretion to the plasma membrane with an efficacy comparable with that of the SYP121 DC fragment. Furthermore, given the absence of further secretory suppression in the presence of the SYP121 DC fragment, the results suggest that this inhibition functionally overlaps with that introduced by the Qa-SNARE fragment.

SNARE Protein Localization Is Altered in the exo70A1 Mutant
Both the exocyst and a subset of SNARE proteins associate dynamically with the plasma membrane. The exocyst forms a peripheral membrane complex Synek et al., 2014;Kulich et al., 2018). The Qa-and R-SNAREs are integral membrane proteins; whereas SYP121 localizes to the plasma membrane (Lipka et al., 2007;Honsbein et al., 2009;Grefen et al., 2015;Waghmare et al., 2019), the R-SNAREs, including VAMP721, localize to the vesicle membrane and associate transiently with the plasma membrane (Uemura et al., 2005;Zhang et al., 2011Zhang et al., , 2015. We therefore asked whether the loss of one interacting partner could affect the localization of the other as a possible mechanism to explain secretory block. For this purpose, SNARE and exocyst GFP protein fusion constructs were stably transformed into the reciprocal exocyst and SNARE mutant backgrounds. We quantified their distributions as ratios between the fluorescence signal present at the cell periphery, delineated as the first 2 mm from the cell surface, and in the interior of cells within the hypocotyl and roots of transgenic plants, normalizing the fluorescence distribution to the total fluorescence signal in each case. In the exo70A1 mutant, we observed a pronounced shift in VAMP721-GFP away from the cell periphery ( Figures 4A and 4C), much as has been reported also for the sec84b exocyst mutant . Similarly, VAMP722-GFP and SYP121-GFP distributions were reduced at the cell periphery in exo70A1 mutants compared with the wild type ( Figures 4A and 4B). However, the GFP-EXO70A1 signal, which localized primarily to the cell periphery, was not significantly affected in each of the SNARE mutant backgrounds (Figures 4C and 4D). These results suggest that EXO70A1 is important either for localization of the SNAREs or their maintenance at the cell periphery, especially of the R-SNAREs. This interpretation is consistent with the idea that EXO70A1 is important for exocyst targeting Kalmbach et al., 2017) and its loss delocalizes and destabilizes the exocyst complex, thereby affecting SNARE complex assembly, SNARE protein delivery, or SNARE recycling. The interpretation is also supported by evidence from yeast that suggests exocyst subunits can bind both vesicle-and plasma membrane-associated proteins (Shen et al., 2013;Yue et al., 2017;Mei and Guo, 2019;Rossi et al., 2020).

Differential Effects of Exocyst-SNARE Interactions on Plant Growth
Single mutants of several plasma membrane SNAREs normally show subtle defects in Arabidopsis; these weak phenotypes are generally thought to reflect functional redundancies. For example, the single R-SNARE mutants vamp721 and vamp722 are largely indistinguishable from the wild type and only the double vamp721vamp722 mutant is lethal (Zhang et al., 2011). Similarly, Transient expression of secYFP in the hypocotyl and root of the 5-d-old wild-type and exo70A1 Arabidopsis seedlings. To ensure equal genetic loads, seedlings were transformed with the pTecG-2in1-CC multicistronic vector (Karnik et al., 2013) carrying cassettes to express secYFP, GFP-HDEL as an expression and ratiometric marker. The lower set of images show the results of including the dominant-negative SYP121 DC fragment as a test for secretory block. When expressed with GFP-HDEL alone, secYFP yielded a modest signal relative, consistent with its loss to the medium on secretion . When expressed additionally with SYP121 DC an increase in secYFP signal relative to GFP-HDEL was evident, indicating secretory block and secYFP retention in the tissue (Tyrrell et al., 2007;Grefen et al., 2015). Note the retention of secYFP in the exo70A1 mutant, even in the absence of SYP121 DC expression. YFP-to-GFP fluorescence ratios calculated after subtracting the autofluorescence signals from untransformed seedlings. Data are means 6 SE, each of n $ 18 independent transformants, with letters indicating statistical significance on post hoc Holm-Sidak analysis (P < 0.001). Bar 5 50 mm.
the single Qa-SNARE mutants syp121 and syp122 exhibit strong phenotypes that surface only when challenged with abiotic and biotic stress (Zhang et al., 2008;Eisenach et al., 2012;Waghmare et al., 2018), again reflecting functional overlaps as well as potential specificities among noncanonical protein partners. The single mutant exo70A1 also exhibits a range of phenotypes, including reduced vegetative growth and cell-specific patterning (Synek et al., 2006;Fendrych et al., 2010;Kalmbach et al., 2017;Vuka sinović et al., 2017). These characteristics are consistent with the effects of the exo70A1 mutant in secretory traffic and SNARE localization, despite the apparent lack of any additive effect over that of the SYP121 DC dominant-negative fragment (Figures 3  and 4).
Vesicle tethering by the exocyst is thought to occur before the SNAREs assemble to drive vesicle fusion, but both are necessary for cargo secretion (Bassham and Blatt, 2008;Ravikumar et al., 2017). We therefore hypothesized that loss of both EXO70A1 and single SNARE gene expression should exhibit defects synergistic to those of the single mutant parents if the functions of the gene products are closely connected. The exo70A1 null mutant is male lethal and is impaired in stigma receptivity (Synek et al., 2006). Therefore, we crossed exo70A1 1/2 heterozygote with the homozygotic mutants vamp721 2/2 and syp121 2/2 and examined the segregation of the double mutant in the F2 generations. Growth of the syp121 mutant is normally only marginally retarded compared with the wild type (Bassham and Blatt, 2008;Eisenach et al., 2012). Like the exo70A1 single mutant, we found that growth of the double exo70A1 syp121 mutant showed a significant but modest reduction in vegetative growth in the F2 generation, suggesting that the exo70A1 mutation is epistatic to that of syp121.
By contrast, the F2 population of the exo70A1 1/2 and vamp721 2/2 cross yielded a segregating population of extremely dwarfed plants, each of which was identified as a exo70A1 vamp721 double mutant ( Figures 5A and 5B). The exo70A1 vamp721 double mutant plants could be maintained for up to 10 weeks when grown under a short-day light cycle, but the plants remained severely dwarfed, showed late lethality, and failed to flower. We therefore maintained the double mutant as a exo70A1 1/2 vamp721 2/2 segregating population.
To examine the mutant lines in greater detail, we grew seedlings on agar plates for 10 d and examined the characteristics of early development following germination. The exo70A1 vamp721 double mutants showed severely stunted growth in the root and cotyledon expansion was much reduced compared with the wildtype seedlings and the single mutant parents ( Figure 5C). These phenotypes were evident 7 to 10 d after germination. Transient transformation by cocultivation must be started with younger seedlings, precluding direct assays of traffic using the pTecG-2 in1-CC vector with secYFP and GFP-HDEL (Figure 3). Nonetheless, we noted that the exo70A1 vamp721 double mutants produced small, delicate leaves with morphologically similar structure. Quantitative analysis of the epidermal cells, which are important drivers in leaf expansion (Walter et al., 2009), showed that in the exo70A1 vamp721 double mutants these cells were significantly smaller than those from leaves of the wild-type and single mutant plants (Figures 5D and 5E). These observations are consistent with suppressed cell expansion arising from the (B) SYP121, VAMP721, and VAMP722 localization in the wild-type (black) and exo70A1 mutant (gray) Arabidopsis, including data in (A), as the ratio of fluorescence between the periphery and cytosol, indicating a shift of SNARE distribution to the cytosol in the absence of EXO70A1. Data are means 6 SE, each of n $ 36 randomly selected images from n $ 6 independent experiments, with letters indicating pairwise statistical significance from post hoc Holm-Sidak analysis (P < 0.001).
(C) Representative images of GFP-tagged EXO70A1 in the wild-type and syp121, vamp721, and vamp722 mutant Arabidopsis, showing no appreciable change in its peripheral localization. (D) EXO70A1 localization in the wild-type and syp121, vamp721, and vamp722 mutant Arabidopsis as the ratio of fluorescence between the periphery and cytosol. Data are means 6 SE, each of n $ 36 randomly selected images from n $ 6 independent experiments. No statistically significant differences from the wild type were evident. Bars in (A) and (C) 5 10 mm. impaired traffic of the exo70A1 vamp721 double mutant, and they implicate specific overlaps in the progression of exocyst-SNARE interactions that occur in the events leading up to secretion.

DISCUSSION
Models of exocytosis have generally divided the process into three stages, vesicle tethering, docking, and fusion, each thought to occur as part of a serial progression with distinct sets of proteins that regulate each step (Cai et al., 2007;Bassham and Blatt, 2008;Zarský et al., 2009). Major protein assemblies contributing to this progression include exocyst and SNARE complexes, each of which relies on interactions between a discrete set of protein subunits and on their correct cellular location for function. Even so, there is growing evidence that these assemblies are not isolated interaction networks (Shen et al., 2013;Mei and Guo, 2019;Rossi et al., 2020) nor, as we now show, are their functional impacts. Here, we present evidence in vitro and in vivo that point to direct interactions in Arabidopsis between members of the exocyst complex EXO70 subunit family that are important for vesicle tethering and recruitment to the plasma membrane, and the SNAREs that drive the final stages of membrane fusion. Furthermore, the functional bias evident among these interactions leads to a clear hierarchy between exocyst-SNARE binding partners with differential consequences for SNARE protein localization, secretion, cellular expansion, and vegetative growth. Thus, our findings point both to an overlap between the exocyst and SNAREs and to a temporal "handshake" between subunits of these protein complexes that is necessary for vesicles to progress through the final stages of tethering and fusion.

Exocyst-R-SNARE Interactions Engender a Functional Synergy in Membrane Traffic
A singular feature of the EXO70-SNARE interactions we uncovered through yeast mbSUS is its R-SNARE bias. Each of the major plasma membrane-associated SNAREs-the Qa-SNARE SYP121; the R-SNAREs VAMP721, VAMP722, VAMP724, and VAMP727; and to lesser extents, the Qa-SNARE SYP122 and Qbc-SNARE SNAP33-bound differentially with several EXO70 subunits, but not with the endomembrane R-SNARE VAMP723 (Figure 1; Supplemental Figure 1). These data also highlighted an overall pattern among the EXO70 subunits that favored the plasma membrane R-SNAREs, as demonstrated by VAMP721. In secretory traffic assays, we found no evidence for an additional impact of the cognate Qa-SNARE SYP121 in secretory trafficking assays (Figure 3) when combined with the exo70a1 mutant. The block of traffic in the the exo70a1 mutant was comparable with that of the dominant-negative SYP121 DC fragment that competes with the full-length Qa-SNAREs SYP121 and SYP122 to suppress (A) Representative Arabidopsis plants grown over 4 weeks under 16-hlight/8-h-dark from F2 seed germinated on soil to isolate the double mutants from crosses of the SNARE mutants with exo70A1. syp121 and vamp721 single mutants were visually indistinguishable from the wild type, while the exo70A1 mutant was 40 to 50% smaller than the wild type. exo70A1 syp121 double mutants were visually indistinguishable from the exo70A1 single mutant, while the exo70A1 vamp721 double mutant showed a severe suppression of growth unique from the wild-type and single mutant parents. Plants in every case were genotyped as in (C). Bar 5 2 cm. (B) Rosette diameter of Arabidopsis grown over 4 (light gray bars) and 8 weeks (dark gray bars) as in (A). Data are means 6 SE of n $ 10 plants for each line. Letters indicate significant differences with post hoc Holm-Sidak analysis (P < 0.01). (C) (Left) Seedling and root development of the wild type, exo70a1 and vamp721 single mutants, and exo70a1 vamp721 double mutants grown on 0.5 MS and 1% agar medium for 10 d. Note the modest reduction in growth and smaller epinastic cotyledons of the exo70A1 mutant and much stronger growth suppression of the exo70A1 vamp721 double mutant. (Right) Seedling genotypes confirmed by PCR using gene-specific primers (Supplemental Table) and a T-DNA left boarder primer to amplify the wildtype and mutant allele products. V1 and A1 indicate the wild-type alleles of VAMP721 and EXO70A1, respectively; v1 and a1 indicate the T-DNA mutant alleles of vamp721 and exo70A1, respectively. Bar 5 1 cm. were analyzed for epidermal cell surface area (green) and data assembled (E) in a box and whisker plot. Cell area of exo70a1 vamp721 plants were significantly smaller than those of the wild-type and parental lines. Data are means 6 SE of n $ 150 cells chosen at random from n $ 8 plants for each line. Note the logarithmic y axis scale. Letters indicate significant differences with post hoc Holm-Sidak analysis (P < 0.001). Bar 5 50 mm.
secretion (Tyrrell et al., 2007;Grefen et al., 2015), and their combination yielded no evidence of additive interaction. Similarly, the strongest effect in complementary protein localization assays was of the exo70a1 mutant on VAMP721 (Figure 4).
Most telling, however, we uncovered a functional synergy of EXO70A1 with the R-SNARE. Crossings to yield the exo70a1 syp121 double mutant resulted in mutant plants with characteristics similar to the effects of the exo70a1 mutant; by contrast, the exo70a1 vamp721 double mutant was strongly impaired in cell expansion, exhibited severe dwarfing in vegetative growth, and failed to transit to flowering, even though the vamp721 mutant showed no impact on cell expansion or vegetative growth Cartoon of EXO70A1 and VAMP721 synergy in vesicle fusion. Peripherally localized EXO70A1 is indicated by the exposed (orange) face of an oval exocyst complex (purple). SYP121 and SYP122 are shown in green and blue, respectively, with membrane anchor, longer H3 (SNARE) and shorter Habc domains. VAMP721 and VAMP722 are shown in red and yellow, respectively, with only a membrane anchor and SNARE domain. Other SNARE and associated proteins are omitted for clarity. (A) Normal (wild-type) secretion engages EXO70A1 in sequentially recruiting VAMP721 (a) and SYP121 from Qa-SNARE "icebergs" (b) leading to vesicle fusion (c). (B) exo70a1 mutant impairs secretion by slowing trans-SNARE assembly (d).
(C) exo70a1 syp121 double mutant similarly slows secretion with the remaining excess of Qa-SNARE SYP122 available for trans-SNARE assembly (e). (D) exo70a1 vamp721 double mutant strongly suppresses secretion by reducing R-SNARE availability as well as eliminating exocyst recruitment for trans-SNARE assembly.
( Figure 5; Zhang et al., 2011Zhang et al., , 2015Zhang et al., , 2017. The effects of EXO70A1 with the Qa-SNARE can be understood as the result of the sequential actions of the exocyst and Qa-SNARE, so that the exocyst subunit mutation dominates over events engaging the Qa-SNARE. Such is not true for the effects of the EXO70 subunit mutation with that of the R-SNARE. In this case, the strong synergy evident in vivo is best understood in context of a mutual interdependence in the roles of the two proteins. Most important, this synergy implies a clear hierarchy among interactions between the Qa-SNARE SYP121 and the R-SNARE VAMP721 with EXO70A1. How is the spatiotemporal overlap of EXO70 and R-SNARE binding enfolded within the respective roles of these subunits within the functions of their respective complexes? In vivo and cryo-electron microscopy analyses of the exocyst complex indicate a topologically optimal space between the exocyst complex, plasma membrane, and tethered vesicle membrane in yeast that could accommodate exocyst and SNARE proteins in association (Picco et al., 2017;Mei et al., 2018). This proximity of exocyst and SNARE complexes fits well with the evidence of direct physical interaction, the requirement for EXO70A1 in SNARE distributions (Figures 2 and 4), and with previous evidence for a close complementarity in protein localizations in yeast and plants Shen et al., 2013). The exocyst topology suggests that the EXO70 subunit is exposed to the cytosol when the exocyst complex is anchored to the plasma membrane. Such exposure would certainly offer a surface for interaction with the R-SNARE and its cytosolic longin domain as the vesicle membrane approaches its target. Indeed, recent findings implicate vesicle tethering in yeast that depends on R-SNARE binding with Sec6 associated in complex with EXO70: Rossi et al. (2020) developed an in vitro assay using vesicle clustering to monitor tethering, incorporating components of the exocyst, to show that clustering is enhanced by the vesiclelocalized R-SNARE Snc2 and strongly promoted with EXO70 mutants that bypass Rho GTPase activation. These results, and reconstructions from single-particle electron microscopy, indicate a conformational change of the holocomplex on activation by the Rho GTPase that exposes a binding site within the Sec6-EXO70 complex for the R-SNARE. Although interactions in yeast between the R-SNARE and exocyst appear dominated by binding to the exocyst Sec6 subunit, in vitro assays have indicated a direct interaction between the EXO70 subunit and Snc2 (Shen et al., 2013). Assuming similar topological relationships for Arabidopsis, the functional synergy uncovered for EXO70A1 and VAMP721 leads us to posit that these binding events precede SYP121 introduction and the assembly of VAMP721 with the Qa-SNARE leading to vesicle fusion.

EXO70 and SNARE Redundancies
It is perhaps surprising that, unlike the parental exo70A1 and vamp721 mutants, the exo70A1 vamp721 double mutant exhibited so profound a set of phenotypes allied to secretory traffic. The EXO70 subfamily in Arabidopsis incorporates 23 members and although the majority of these differ in developmental and tissue-specific expression, there are nonetheless overlaps (Cvrčková et al., 2012). Similarly, VAMP721 and VAMP722 are near sequence identical, and with the exception of the associated interactions with the Sec1/Munc18 protein SEC11 that distinguish their binding between the plasma membrane Qa-SNAREs SYP121 and SYP122 , the two R-SNAREs appear functionally redundant. Nonetheless, the several plasma membrane Qa-and R-SNAREs, although cross-interactive and seemingly redundant in function, show very different phenotypes. Consider SYP121 and SYP122 that both bind the R-SNAREs VAMP721 and VAMP722. The single syp121 and syp122 mutant plants appear similar to the wild type when grown under permissive conditions (Zhang et al., 2007), and the dominant-negative SYP121 DC and SYP122 DC fragments are equally effective in secretory traffic block in vivo (Tyrrell et al., 2007;Grefen et al., 2015). Yet, the two Qa-SNARE null mutants show very different phenotypes in pathogen defense and abiotic stress resistance (Collins et al., 2003;Zhang et al., 2007;Eisenach et al., 2012) and the Qa-SNAREs mediate the traffic of different subsets of secretory cargo (Rehman et al., 2008;Waghmare et al., 2018). These, and associated studies (Reichardt et al., 2011;Hashimoto-Sugimoto et al., 2013;Xia et al., 2019;Zhang et al., 2019), implicate a functional network of trafficking and related regulatory protein partners that is substantially more dynamic than may be deduced from an analysis of gene families.
Our present data do not speak directly to the functional specificities among the several plasma membrane SNAREs with EXO70A1. Even so, when compared with VAMP722, we note that the peripheral localization of VAMP721 was most strongly affected in the exo70a1 mutant (Figure 4), which may indicate more subtle functional differences between the two R-SNAREs that have yet to be resolved. Furthermore, the strong phenotypic synergy in the exo70A1 vamp721 double mutant that we identified indicates that the presence of VAMP722 is not sufficient to compensate for the loss of VAMP721. One simple explanation is that EXO70A1 and VAMP721 may be specialized as a functional unit, analogous to that implicated in yeast (Shen et al., 2013;Rossi et al., 2020), for the delivery of a unique set of cargos critical to the growth of the plant, much as is now evident from analysis of the distribution of cargos between the Qa-SNAREs SYP121 and SYP122 (Waghmare et al., 2018).
Finally, our data focus attention to the longin domain of the R-SNARE. The longin domain of VAMP721 was previously shown to be a key regulator in SNARE protein interactions (Uemura et al., 2005). It is essential for selective binding with noncanonical K 1 channel partners that have been suggested to coordinate secretion with K 1 uptake  and, as we demonstrate, it is important for the interactions with several EXO70 subunits (Figure 1; Supplemental Figures 1 and 3). Thus, the VAMP721 longin domain may function as the nexus in a progression of hand-offs between multiple binding partners. We speculate that the exocyst complex could participate in targeting VAMP721-loaded vesicles to the plasma membrane through direct interactions between SNAREs and the exocyst during vesicle tethering and/or recruitment to the plasma membrane ( Figure 6). Important questions now will be to identify the domains within the plant EXO70 and SNARE proteins that are critical for their interactions and to determine their dynamics in context of vesicle delivery to the plasma membrane.

Plant Growth Conditions, Crosses, and Genotyping
Seeds from the wild-type (Columbia-0) and mutant Arabidopsis (Arabidopsis thaliana) were stratified in the dark at 4°C and then germinated in long-day conditions in growth chambers (60% relative humidity, 16-hlight/8-h-dark cycle, 22°C) on half-strength Murashige and Skoog (MS) medium solidified with 1% (w/v) agar. Sealed plates were placed in longday light conditions (150 mmol m 22 s 21 photosynthetically active radiation) before seedlings were evaluated for mutant phenotypes and collected for genotyping or were transplanted to single pots or 15-holder trays for further growth and observation.
To generate the exo70A1 vamp721 double mutant, the heterozygous exo70A1 1/2 (SALK_135462) mutant was used as a maternal parent in a cross with the homozygous vamp721 mutant (SALK_037273). The F1 generation was allowed to self-fertilize, and the F2 seed was collected for further analysis. Tissue was harvested from individual F1 and F2 plants of the ex70A1 1/2 3 vamp721 cross for PCR analysis using gene-specific and T-DNA primers (Supplemental Table). Double heterozygous F1 mutants were identified and allowed to set seed. Since the exo70A1 single mutant is infertile, a segregating population of exo70A1 1/2 vamp721 2/2 was maintained for isolating double mutants. For the exo70A1 syp121 double mutant, the exo70A1 1/2 (SALK_135462) line was crossed with the syp121 null mutant carrying a point mutation that encodes a premature stop codon and does not produce a transcript but introduces a MluI restriction site that is identifiable as a band shift on digestion (Pajonk et al., 2008).The double mutant was propagated as an exo70A1 1/2 syp121 2/2 line.

Yeast and Plant Vector Construction
All plasmids were constructed using the Gateway cloning system (Thermo Fisher Scientific). Details for all yeast and plant expression vectors are included in publications as cited in the text and can be found at www.psrg. org.uk.

mbSUS
Haploid yeast strains for bait and prey, THY.AP4 and THY.AP5, respectively, were used to express pMetYOst-Dest or pNX35-Dest constructs of exocyst subunits or SNARE proteins and analyze protein-protein interactions as described previously . Ten to 15 yeast colonies were selected and inoculated into selective media (CSM -LM for THY.AP4 and CSM -MTU for THY.AP5) for overnight growth at 180 rpm and 28°C. Liquid cultures were harvested and resuspended in yeast extract peptone dextrose (YPD) medium. Yeast mating was performed in sterile PCR tubes by mixing equal aliquots of yeast containing bait and prey construct. Aliquots of 5 mL were dropped on YPD plates and incubated at 28°C overnight. Colonies were transferred from YPD onto CSM -LTUM plates and incubated at 28°C for 2 to 3 d. Diploid colonies were selected and inoculated in liquid CSM -LTUM media and grown at 180 rpm 28°C overnight before harvesting by centrifugation and resuspension in sterile water. Serial dilutions at OD 600 1.0 and 0.1 in water were dropped, 5 mL per spot, on CSM -LTUMAH plates with added Met. Plates were incubated at 28°C, and images were taken after 3 d. Yeast were also dropped on CSM -LTUM control plates to confirm mating efficiency and cell density, and growth was imaged after 24 h at 28°C. To confirm construct expression, mated yeast was grown overnight in 5 mL of selective medium and harvested by centrifugation at 5000g. Pelleted yeast was resuspended in diluted lysis buffer (1/1 [v/v], 10% SDS, 4 mM EDTA, 0.2% Triton-X 100, 0.01% bromophenol blue, 20 mM DTT, 20% glycerol, and 100 mM Tris, pH 6.8), sonicated two times for 30 s, and boiled at 100°C for 5 min. Prepared protein samples were loaded on 10% polyacrylamide gels and blotted onto a nitrocellulose membrane. The membrane was blocked overnight (13 PBS, 0.25% [v/v] Tween, and 5% [w/v] low fat milk) and then incubated with primary a-HA or a-VP16 antibodies at 1:2000 dilution. Proteins were detected using the secondary a-mouse or a-rabbit antibodies (Promega).

Transformations for FRET and Secretion Analysis
Nicotiana benthamiana plants with fully expanded true leaves were used for transient transformation with Agrobacterium tumefaciens strain GV3101 carrying the multicistronic vector pFRETgc-2in1-NN that enabled simultaneous expression of GFP and mCherry protein fusions (Hecker et al., 2015). Overnight cultures of GV3101 expressing these constructs were diluted in infiltration buffer (10 mM MgCl 2 and 100 mM acetosyringone) to OD 600 0.07 and infiltrated with a syringe into the abaxial surface of N. benthamiana leaves. A Zeiss LSM 880 confocal microscope with 403 water immersion objective (Zeiss) was used for image acquisition of transformed N. benthamiana leaves. For FRET, the GFP constructs were excited at 488 nm and detected at 505 to 530 nm. mCherry constructs were detected at 600 to 620 nm after excitation at 552 nm to assess fluorophore expression and at 488 nm to collect the FRET signal. FRET image analysis was performed using the PixFRET tool (Feige et al., 2005) in ImageJ (Schindelin et al., 2012) applied to image stacks across the entire fields of view. PixFRET corrects for spectral bleed-through and for variations in donor and acceptor signals using the product of their fluorescence signals on a pixel-by-pixel basis. FRET analysis of each experiment was typically applied to three to five randomly selected images for each construct pair.
Secretion assays used the tetracistronic vector pTecG-2in1-CC (Karnik et al., 2013) that incorporates four expression cassettes, each driven by a 35S promoter, two carrying the coding sequences for secYFP and GFP-HDEL, and the remaining two cassettes providing Gateway recombination sites for proteins of interest. Assays were performed without and with SYP121 DC incorporated in one of the latter cassettes as described previously (Karnik et al., 2013). Arabidopsis seed was sterilized and grown in 0.05 MS liquid medium for 2 to 3 d until root and cotyledons emerged. Seedlings were then cocultivated with Agrobacterium carrying the multicistronic constructs for an additional 3 to 4 d before imaging. Live root epidermal tissue was imaged under a SP8-SMD confocal microscope with a 203/0.85 numerical aperture objective (Leica Microsystems) using the 470-nm and 514-nm laser lines for excitation and fluorescence was collected over the bandwidths 490 to 530 and 520 to 565 nm for the GFP and YFP protein fusions, respectively. Roots and roothairs were routinely assayed near the root base to ensure imaging of live cells, and cell viability was checked by monitoring of cytoplasmic streaming in root hairs. GFP and YFP signals were corrected for background postcollection using the mean signals from untransformed seedlings before calculating YFP:GFP fluorescence ratios.

Protein Localization
The dynamic study of the lateral membranes of root epidermal cells of the elongation zone was performed on a Nikon TE200e with a Yokogawa Andor spinning disc unit with 403 and 633 oil immersion objectives. The images were processed by Fiji/ImageJ software, and the periphery:internal ratio was calculated as previously described after demarking a 2-mm-wide perimeter at the cell surface corresponding to the peripheral zone (Fendrych et al., 2010).

Statistics
Statistical analyses were conducted with the SigmaPlot (Systat Software) utility for analysis of variance with post hoc Holm-Sidak analysis.

Supplemental Data
Supplemental Figure 1. EXO70 homologs interact differentially with several SNAREs.
Supplemental Figure 2. Immunoblot analysis for EXO70 interaction with several SNAREs.
Supplemental Figure 3. The longin domain of VAMP721 is essential for EXO70 interaction.
Supplemental Figure 4. Immunoblot analysis for EXO70 interaction with VAMP721, VAMP723 and their chimeras.
Supplemental Table. Gene-specific primers for PCR analysis of parental lines and crosses.

ACKNOWLEDGMENTS
We thank Jana S tovíčková (Laboratory of Cell Biology, Institute of Experimental Botany of the Czech Academy of Sciences) for technical help with plant harvesting, and Martin Potocký and Tamara Pečenková for critical reading of this article. This work was supported by the Ministry of Education, Youth and Sports of the Czech Republic from the European Regional Development