Rare earth elements induce cytoskeleton-dependent and PI4P-associated rearrangement of SYT1/SYT5 endoplasmic reticulum–plasma membrane contact site complexes in Arabidopsis

Rare earth elements induce ER membrane remodeling and increase ER–PM connectivity in a process that involves phosphoinositide-associated reorganization of synaptotagmin-tethering complexes.

Recent studies have partially elucidated SYT1's mechanism of action by showing it increases ER-PM connectivity by promoting the cytoskeleton-independent and phosphatidylinositol 4,5-bisphosphate [PI(4,5)P 2 ]-associated EPCS expansion (Lee et al., 2019), and that SYT1-labeled ER tubules can be disrupted by pharmacologically decreasing the intracellular Ca 2+ concentration (Ishikawa et al., 2018). Despite these advances, many aspects including the specificity of the PI(4,5)P 2 signal as a trigger for EPCS expansion, and the dynamics of EPCS organization in response to extracellular Ca 2+ depletion remain largely unexplored.
In this study, we corroborate a recent report describing the establishment of a putative tethering complex between the synaptotagmins 1 and 5, and the Ca 2+ -dependent lipid binding protein CLB1/SYT7 (hereafter CLB1) at EPCSs (Ishikawa et al., 2020), and we expand their analysis by showing that SYT1 and SYT5 can form homo-and heterodimers in vivo. We also show that changes in extracellular Ca 2+ have a limited effect in EPCS organization with the exception of treatments with salts of the rare earth elements (REEs) lanthanum (La 3+ ) and gadolinium (Gd 3+ ). Short-term treatments with REEs (minutes) have been classically used to block non-selective cation channels (Biagi and Enyeart, 1990;Lansman, 1990;Elinder and Arhem, 1994) and/or stretch-activated Ca 2+ -permeable channels at the PM (Yang and Sachs, 1989;Franco et al., 1991;Hamill and McBride, 1996;Ermakov et al., 2010), but recent studies have shown that long-term treatments with REEs promote their internalization and activate endocytosis in plant cells (L. . Here we show that the dynamics of the REE-induced EPCS reorganization are not consistent with the Ca 2+ channel-blocking activity of REEs at the PM but rather is a consequence of their slow internalization to the cytosol. We also show that the EPCS-remodeling process is associated with the activation of the Ca 2+ signaling in the cytosol, and the accumulation of phosphatidylinositol-4phosphate (PI4P) at the PM.
Our results highlight commonalities between the EPCS remodeling triggered by REEs (this study) and NaCl (Lee et al., 2019), such as the slow dynamics of the remodeling process and the concomitant accumulation of negatively charged phosphoinositides at the PM. These findings also uncover key differences such as the identity of the phosphoinositide species that are accumulated, PI4P for REEs (this study), and PI(4,5) P 2 for NaCl (Lee et al., 2019), and the differential requirement for a functional cortical cytoskeleton for REE-and NaCl-induced EPCS remodeling. In a broader context, our study shows that the direct manipulation of extracellular Ca 2+ levels has limited effects on plant EPCS organization, and supports a model where the slow accumulation of stress-specific phosphoinositide species at the PM acts as a general adaptive mechanism governing cortical ER-PM communication during sustained stress conditions.

Plant materials and growth conditions
Arabidopsis thaliana Columbia (Col-0) was used as the wild type and the background for transgenes. Seeds of the mutants syt5-1 (SALK_036961) and clb1-2 (SALK_006298) were obtained from the Arabidopsis Biological Resource Center (Ohio State University). Previously published lines in this study are SYT1-green fluorescent protein (GFP) and MAPPER-GFP (Lee et al., 2019); GFP-HDEL (Batoko et al., 2000); 35S::C2AB (Pérez-Sancho et al., 2015); GCaMP3 (DeFalco et al., 2017); CITRINE-2×PH PLC ; and CITRINE 1×PH FAPP (Simon et al., 2016). Plants were grown on half-strength Murashige and Skoog (MS) medium (Caisson Labs) or soil (Sunshine mix #4, Sun Gro Horticulture Canada Ltd) at 22 °C with a 16 h light/8 h dark cycle. For the NaCl assays, Arabidopsis seedlings were grown vertically for 4 d on 1/10th strength MS medium, and similar sized seedlings were transferred to the same medium supplemented with different NaCl concentrations. The root elongation and root hair phenotypes were scored after 9 d.

Chemical applications
Chemicals were exogenously applied by incubating 5-day-old seedlings in liquid 1/10th strength MS medium and supplementing them with 500 µM LaCl 3 (Sigma-Aldrich), 500 µM GdCl 3 (Sigma-Aldrich), 5 mM EGTA (Sigma-Aldrich), or 25 μM oryzalin (Sigma-Aldrich) for 2 h or 16 h, or with 250 µM bis-(o-aminophenoxy) ethane-N,N,N',N'-tetraacetic acid (BAPTA) (Sigma-Aldrich) or 1 µM latrunculin B (Abcam) for 2 h or 16 h. The duration of the treatments was based on the general toxicity caused by the different chemical compounds in plants. To visualize Hechtian strands, 5-day-old cotyledon epidermal cells expressing the different markers were plasmolyzed for 4 h using 0.4 M mannitol. The images are an overlay of propidium iodide-stained cell walls with the localization of the GFP fusion proteins in green.

Image acquisition and quantitative analyses
Living cell images were obtained using a Nikon C1 confocal laser scanning microscope, a Perkin-Elmer spinning disk confocal microscope, and an Olympus FV1000 multiphoton confocal laser scanning microscope. The Nikon C1 confocal laser scanning microscope was equipped with 488 nm and 515/30 nm emission filters and Nikon Plan Apochromat oil immersion objectives (×40, 1.0 NA and ×60, 1.4 NA, respectively). The Perkin-Elmer spinning disk confocal microscope was equipped with 488 nm and 561 nm lasers. The Olympus FV1000 was equipped with 405, 473, and 559 nm lasers and a ×60 oil Planon (×60, 1.4 NA). Images were captured using Nikon-EZ C1, Olympus FV1000, and Volocity software, respectively. To quantify the number of 'beads' configuration, 5-day-old Arabidopsis seedlings harboring the SYT1-GFP or SYT5-GFP marker were incubated for 16 h in liquid 1/10th strength MS medium (Mock) or liquid 1/10th strength MS medium supplemented with the different chemicals. For each treatment, the number of 'beads' labeled by SYT1-GFP or SYT5-GFP in the cortex of cotyledon epidermal cells was scored in at least 50 (15 μm×15 μm) regions of interest (ROIs) using the cell counter tool of Fiji (ImageJ) (National Institutes of Health, http://imagej.nih.gov/ij/) (Schindelin et al., 2012). To compare the fluorescent intensity of the ratiometric CITRINE-1×PH FAPP between control and treated samples, confocal laser scanning images of 5-day-old epidermal cotyledon cells were acquired from at least 10 individual seedlings. For each data point, the fluorescence intensity data were scored from at least 100 (15 μm×15 μm) ROIs using Fiji's integrated density measurement tool (Schindelin et al., 2012). In this analysis, stomatal lineage cells were excluded from the quantification. To compare the fluorescent intensity of the ratiometric GCaMP3 sensor, images of 5-day-old seedlings were acquired using a Nikon SMZ18 stereo microscope equipped with a 480/40 nm excitation filter, a Nikon P2-SHR Plan Apo ×0.5 objective, and a Nikon DS-Ri2 camera. The images were captured using NIS-Elements BR software version 4.60. For each data point, the fluorescence intensity data were scored from at least 50 seedlings. In the ratiometric analyses, the fluorescent data were normalized using the equation: ∆F/F=(F-F 0 )/F 0 , where F 0 is the mean intensity of background fluorescence. The data were subject to one-way ANOVA to identify statistically significant differences among treatments. All statistical analyses were performed using the GraphPad Prism 5.0b software.

SYT5 and CLB1 are EPCS-localized proteins that interact with SYT1 in vitro
SYT1 is a protein tether implicated in the establishment, organization, and function of plant EPCSs (Pérez-Sancho et al., 2016a;Tilsner et al., 2016;Bayer et al., 2017;Wang et al., 2017). Because the SYT1 orthologs in mammals [extended synaptotagmins (E-Syts)] and yeast [tricalbins (Tcbs)] establish tethering complexes in vivo (Creutz et al., 2004;Giordano et al., 2013), we searched for additional proteins physically associated with Arabidopsis SYT1. For this purpose, we used a SYT1-GFP line in the syt1-2 background (Lee et al., 2019) and performed IP assays using agarose beads coupled to an anti-GFP nano-body (GFP-Trap beads). The IP results from three independent biological replicates provided a large number of proteins physically associated with SYT1 that we identified using LC-MS/MS. We filtered the results using the following criteria: (i) presence in all three biological replicates; (ii) detection of two or more exclusive unique peptides; and (iii) absence in the negative IP control (IP using a transgenic line expressing free GFP). After applying these filters, we identified two putative SYT1 interactors: Arabidopsis SYT5 (At1g05500; Ishikawa et al., 2020) Table S1).
To assess the subcellular localization of the SYT5 and CLB1 proteins, we generated fluorescent SYT5-GFP and CLB1-GFP marker lines driven by their respective endogenous promoters. We used confocal microscopy and compared the SYT5-GFP and CLB1-GFP subcellular localization with that of the SYT1-GFP marker (Lee et al., 2019). Figure 1F-Q shows that the SYT5-GFP and CLB1-GFP localization strongly resembles that of the SYT1-GFP marker in all tissues analyzed. These localizations include a 'beads and strings' arrangement in cotyledon epidermal cells (Fig. 1F-H), perinuclear labeling consistent with the ER in root meristematic cells (Fig. 1I-K), associations with the cell wall through Hechtian strands ( Fig. 1L-N), and strong signal accumulation at root hair initiation sites (Fig. 1O-Q). The latter localization is consistent with a putative function for the SYT1/SYT5/CLB1 complex in root hair polarity maintenance, as indicated by the root hair phenotypes in the presence of NaCl of the syt1/syt5/clb1 triple mutant ( Supplementary Fig. S2). Although confocal microscopy alone is not sufficient to establish unequivocally whether the observed subcellular localizations represent EPCSs, the protein interaction data, the shared structural and functional features, and the common localization patterns strongly suggest that, like SYT1, SYT5 and CLB1 are enriched at EPCSs.

SYT1 and SYT5 establish homotypic and heterotypic interactions in vivo at EPCSs
To validate the interactions between SYT1, SYT5, and CLB1, we used a targeted co-IP assay using a previously reported anti-SYT1 polyclonal antibody (Pérez-Sancho et al., 2015). For this experiment, we used the SYT1-GFP line in the syt1-2 background (Lee et al., 2019), a SYT1-GFP line in the Col background (Pérez-Sancho et al., 2015), and we generated a transgenic line expressing SYT5-GFP under its native promoter (SYT5::SYT5-GFP) and a transgenic line expressing CLB1-GFP under a constitutive ubiquitin 10 promoter (pUB10::CLB1-GFP). Figure 2A and Supplementary  Fig. S3 show that the affinity-purified SYT1-GFP was able to pull-down the native SYT1 (lane 1, 61.7 kDa band) from protein extracts in vitro, and that this interaction was not present when the SYT1-GFP line in the syt1-2 background was used (lane 2). SYT5-GFP and CLB1-GFP were also able to pull-down the native SYT1 from protein extracts (lanes 4 and 5, 61.7 kDa band). Next, we assessed the putative interaction of these proteins in vivo using BiFC assays. Figure 2B-E shows that transient co-expression of different SYT1 and SYT5 BiFC constructs in Nicotiana benthamiana leaves render BiFC signals consistent with SYT1 and SYT5 interacting and forming homo-and heterodimers at PM subdomains. Despite multiple attempts, we failed to observe a BiFC signal between SYT1 and CLB1, and focused our subsequent analyses on SYT1 and SYT5. During the revision of this study, Ishikawa et al. (2020) reported the interaction between SYT1 and CLB1 in vivo using BiFC.

The putative SYT1/SYT5 EPCS complex is largely insensitive to extracellular Ca 2+ depletion but relocalizes in response to internalization of REEs
The SYTs orthologs in yeast and mammals are Ca 2+ -responsive proteins that sense changes in [Ca 2+ ] cyt and regulate the nonvesicular transfer of signaling molecules between the cortical ER and the PM (Creutz et al., 2004;Giordano et al., 2013;Helle et al., 2013;Prinz, 2014). Because SYT1, SYT5, and CLB1 contain a putatively conserved Ca 2+ -binding site in their 3D structure, we asked whether Ca 2+ signals could influence the localization and dynamics of the SYT1/SYT5 tethering complex. To address this question, we first analyzed the effect of extracellular Ca 2+ depletion on SYT1-GFP and SYT5-GFP localization using the extracellular Ca 2+ -chelating agents EGTA and BAPTA (Brault et al., 2004;Nakagawa et al., 2007) at different time points. In 2 h treatments, the depletion of free apoplastic Ca 2+ induced by either EGTA or BAPTA does not have a significant effect on the number of SYT1-GFP-and SYT5-GFP-labeled 'beads' at the cell cortex ( Supplementary  Fig. S4). In 16 h treatments, EGTA and BAPTA induced an ~1.5-to 1.8-fold increase in the number of SYT1-GFP-and SYT5-GFP-labeled 'beads' and a reduction in the average reticule size of the cortical ER network (Fig. 3A-F, R). We also tested the effect of La 3+ and Gd 3+ REEs on SYT1-GFP and SYT5-GFP localization at different time points. Supplementary  Fig. 1. The Ca 2+ -dependent phospholipid-binding proteins SYT5 and CLB1 interact with SYT1. (A) Peptide counts detected upon GFP immunoprecipitation followed by LC-MS/MS analysis using Arabidopsis plants expressing GFP (control) and SYT1-GFP. Numbers indicate the total spectrum counts corresponding to the indicated proteins, and the exclusive unique peptides represented within them. The best Mascot ion score among these peptides is indicated. The number of peptides corresponding to GFP is shown for reference. This result is representative of three independent experiments (for details on the replicates, see Supplementary Table S1 S5 shows that the Ca 2+ channel-blocking activity of REEs in 30 min treatments did not induce changes in SYT1-GFP and SYT5-GFP localization. Remarkably, 2 h treatments induced a variable 2-to 4-fold increase in the number of SYT1-GFP-and SYT5-GFP-labeled 'beads' in individual epidermal cells ( Supplementary Fig. S4), and 16 h treatments induced a generalized 4-to 5-fold increase in the number of SYT1-GFP-and SYT5-GFP-labeled 'beads' in all epidermal cells (Fig. 3J-Q). The treatment was associated with a significant reduction of the average reticule size of the cortical ER network (Fig. 3R), and with an increase in the number of 'beads' labeled by the artificial EPCS marker MAPPER-GFP (Lee et al., 2019) (Supplementary Fig. S6). Remarkably, a 16 h La 3+ treatment did not induce major changes in the localization of a truncated SYT1-GFP marker harboring the C2 phospholipidbinding domains (C2AB-GFP; Pérez-Sancho et al., 2015) that is still homogeneously distributed at the PM ( Supplementary  Fig. S7).
The dynamics of SYT1-GFP and SYT5-GFP relocalization are consistent with EPCS remodeling being triggered by the slow internalization of REEs to the cytosol (L. . In this scenario, the addition of EGTA, a polydentate chelator that forms stable complexes with both Ca 2+ and REEs (Tei et al., 2010), should maintain REEs in the extracellular space and prevent the REE-induced SYT1-GFP and SYT5-GFP relocalization. Figure 4A-E shows that the supplementation of the REE treatments with 5 mM EGTA was sufficient to abolish the REE-induced SYT1-GFP relocalization in 16 h treatments, and reduce the long-term toxicity of the REEs in seedlings treated for 14 d ( Supplementary  Fig. S8). These results suggest that REE internalization, and not the blockage of extracellular Ca 2+ entry induced by either REEs or Ca 2+ chelators, underlies EPCS remodeling.

Internalized REEs can act as Ca 2+ signaling surrogates in the cytosol
Biochemical studies have shown that REEs act as allosteric regulators of multiple Ca 2+ -binding proteins in vitro (Mills and Johnson, 1985;Bertini et al., 2003;Ye et al., 2005;L. Wang et al., 2016), so we asked whether internalized REEs could replace Ca 2+ and mimic its effect in vivo. To answer this question, we analyzed the effect of REEs on the cytosolic activity of the calmodulin-based ratiometric Ca 2+ sensor GCaMP3 (Tian et al., 2009). Figure 5A-G shows that 16 h REE treatments that promote their internalization also induce a 2-to 3-fold increase in the GCaMP3 fluorescent signal in a process that is abolished by 5 mM EGTA supplementation. These results are consistent with internalized REEs acting as Ca 2+ surrogates and binding proteins containing Ca 2+ /calmodulin-like binding domains.

The La 3+ -induced relocalization is cytoskeleton dependent and it is associated with PI4P accumulation at the PM
In the final experiment, we used 500 μM La 3+ and 100 mM NaCl treatments for 16 h to explore whether the REEinduced EPCS reorganization is mechanistically similar to that previously reported for NaCl stress (Lee et al., 2019). First, we tested whether the cortical cytoskeleton plays a role in the La 3+ -induced EPCS remodeling. Our results show a differential behavior between the treatments as, compared with NaCl stress (Lee et al., 2019), La 3+ does not cause visible disruption of the cortical cytoskeleton network ( Supplementary Fig. S9), and its effect on EPCS organization is partially abolished by pre-treatments with the microtubule-depolymerizing drug oryzalin or the actin polymerization inhibitor latrunculin B (Fig. 6A-N). Next, we tested whether the La 3+ -induced EPCS remodeling was associated with the accumulation of phosphoinositides at the PM using the ratiometric sensors citrine 1×PH FAPP1 (for PI4P) and 2×PH PLC [for PI(4,5)P 2 ] (Simon et al., 2016). Our results show that 16 h La 3+ treatments do not induce accumulations of the PI(4,5)P 2 sensor ( Supplementary  Fig. S10), but did induce an ~2-fold increase of the PI4P fluorescent signal at the PM (Fig. 7A, B, E). Remarkably, the 16 h La 3+ treatment also induced the formation of PI4P-labeled vesicle-like structures closely associated with the PM (Fig. 7B  asterisks). Consistent with previous findings, the addition of 5 mM EGTA to the extracellular medium was sufficient to inhibit the La 3+ -induced PI4P accumulation and the formation of PI4P vesicles (Fig. 7C, D). The results support a model In (P-R), the number of puncta or closed reticules was scored using 50-60 arbitrary 225 µm 2 ROIs from at least 15 cells from five independent seedlings. In the box and whiskers plots, the center line represents the median number of puncta or closed reticules per 225 µm 2 , the top and bottom edges are the 25th and 75th percentiles of the distribution, and the ends of the whiskers are set at 1.5 times the interquartile range (IQR). All values outside the IQR are shown as outliers (dots). Letters indicate statistically significant differences using Tukey multiple pairwise comparisons P<0.05. Scale bars=20 μm.  5. REEs induce the activation of the cytosolic GCaMP3 Ca 2+ sensor. Fluorescence images of seedlings expressing the GCaMP3 Ca 2+ sensor. Fiveday-old seedlings were treated in liquid 1/10th strength MS medium supplemented with Mock, 16 h (A), LaCl 3 (500 µM/16 h) (B), GdCl 3 (500 µM/16 h) (E), or the same medium supplemented with 5 mM EGTA (D-F). The activity of the Ca 2+ sensor is shown as color-coded pixel intensity following the LUT scale shown in (F). (G) Quantification of the GCaMP3 signal relative to mock conditions. The center line represents the median fluorescence intensity fold increase relative to mock, the cross represents the mean fluorescent intensity, the top and bottom edges are the 25th and 75th percentiles of the distribution, and the ends of the whiskers are set at 1.5 times the interquartile range (IQR). All values outside the IQR are shown as outliers. The intensity of the signal was measured for at least 50 seedlings per treatment. Letters indicate statistically significant differences using Tukey multiple pairwise comparisons P<0.05. Scale bar=5 mm. Fig. 4. The addition of EGTA to the growth medium reduces the SYT1-GFP and SYT5-GFP localization changes associated with REE internalization. Five-day-old SYT1-GFP seedlings were treated in liquid 1/10th strength MS medium supplemented with Mock (A), LaCl 3 (500 µM/16 h), (B) GdCl 3 (500 µM/16 h) (C), or the same medium supplemented with 5 mM EGTA (D-F) before imaging. (G) Quantification of the SYT1-GFP cortical signal. For each treatment, the number of puncta was scored using 50-60 arbitrary 225 µm 2 ROIs from at least 15 cells from five independent seedlings. In the box and whiskers plots, the center line represents the median number of puncta per 225 µm 2 , the top and bottom edges are the 25th and 75th percentiles of the distribution, and the ends of the whiskers are set at 1.5 times the interquartile range (IQR). When present, the minimum/maximum values outside the IQR are shown as outliers (dots). Letters indicate statistically significant differences using Tukey multiple pairwise comparisons P<0.05. Scale bar=20 μm.
where stress-induced accumulations of specific phosphoinositides is associated with either cytoskeleton-dependent or cytoskeleton-independent rearrangements of EPCS-localized protein complexes (Fig. 8).

Discussion
The plant EPCS responses to [Ca 2+ ] cyt are unique among eukaryotes EPCSs are ubiquitous structures in eukaryotes, and they adopt distinct shapes and architectures in response to environmental and developmental cues. In mammals and yeast, EPCSs have a well-known role in the control of Ca 2+ dynamics but, in plants, the presence of a cell wall that maintains a high extracellular Ca 2+ concentration, and the presence of a complex suite of Ca 2+ channels, transporters, and signaling components (Wheeler and Brownlee, 2008;De Vriese et al., 2018) has limited our understanding of the role of Ca 2+ in cortical ER-PM communication. Mammalian E-Syts and plant SYT1/SYT5 EPCS complexes share a common basic organization as both establish homotypic and heterotypic protein-tethering complexes with their N-terminal domains anchored to the ER, and their C-terminal C2 domains establishing Ca 2+ -dependent interactions with the PM (Giordano et al., 2013;Pérez-Sancho et al., 2015; this study). Intriguingly, the mammalian E-Syts aggregate and concentrate at membrane junctions following a rise in [Ca 2+ ] cyt (Giordano et al., 2013), and this behavior is replicated by SYT1 in response to La 3+ treatments. To explain this observation, Ishikawa et al. (2018) proposed that SYT1 responds to a decrease, instead of an increase, in [Ca 2+ ] cyt due to the activity of La 3+ as a Ca 2+ channel blocker at the PM. Our results reconcile this seemingly contrasting behavior in plants and mammals by showing that, in long-term treatments, internalized REEs are capable of triggering intracellular Ca 2+ signals, effectively offsetting their effect as PM Ca 2+ channel blockers. Given that REEs can act as allosteric regulators of the activity of calmodulins (Mills and Johnson, 1985;L. Wang et al., 2016), and C2-containing proteins in vitro (Essen et al., 1997), we propose that internalized REEs could facilitate slow changes in ER-PM communication either by activating calmodulin signaling or through direct binding to the SYT1, SYT5, and/or CLB1 Ca 2+ -binding domains in vivo.
The results discussed above highlight a clear difference in the temporal regulation of the Ca 2+ -mediated responses between mammalian E-Syts and plant SYTs. In non-excitable mammalian cells, EPCSs control intracellular Ca 2+ levels using storeoperated Ca 2+ entry (SOCE), a fast process that couples the Ca 2+ influx from the extracellular space to the cytosolic Ca 2+ release from the ER within seconds (Orci et al., 2009). These mammalian cells can also sense high [Ca 2+ ] cyt and trigger the recruitment of E-Syt1 tethers to SOCE-independent EPCSs within minutes (Wu et al., 2006). In contrast, the depletion of extracellular Ca 2+ by chelating agents in Arabidopsis has limited effect on SYT1-GFP and SYT5-GFP localization, and EPCS remodeling in response to REEs and NaCl takes place within hours (this study; Lee et al., 2019). Based on these observations, we hypothesize that the plant SYT1/SYT5 complexes are involved neither in the fast coupling of the extracellular and ER-lumen Ca 2+ stores, nor in the fast response to [Ca 2+ ] cyt changes induced by stress. Instead, we propose that the observed EPCS remodeling in response to REEs is a consequence of the sensing and transduction of stress signals that promote long-term cellular adaptive responses, such as the slow changes in the PM lipid composition discussed in the next section.

Stress-specific regulatory mechanisms controlling EPCS organization in Arabidopsis
The cortical ER is a complex arrangement of tubules and small cisternae distributed towards the PM (Stefano et al., 2014;Griffing et al., 2017). EPCSs are important substructures within the cortical ER that can be defined as 200-300 nm long and 30 nm wide cortical ER nanodomains, which anchor to the PM using specialized tethering complexes (McFarlane et al., 2017). In a differentiated plant cell, EPCSs can be localized in immobile ER tubules (Ishikawa et al., 2018), and are associated with the cortical cytoskeleton (Peña and Heinlein, 2013;P. Wang et al., 2014Lee et al., 2019). Currently, two functions of the cortical cytoskeleton array in EPCS establishment have been proposed. On the one hand, the actin and microtubule networks physically interact with VAP27/NET3C tethering complexes, fixing them on specific positions within the cell cortex (P. . This interaction might be required for cargo exchange during endocytic and exocytic trafficking (Peña and Heinlein, 2013;P. Wang et al., 2014). On the other hand, the cortical cytoskeleton is required for the delivery of SYT1 tethers to EPCSs, and could also generate spatial incompatibility for EPCS establishment in regions where 'thick' cortical microtubules (25 nm in diameter) are closely associated with the PM (Pérez-Sancho et al., McFarlane et al., 2017;Lee et al., 2019).
Given that SYT1/SYT5 complexes require a functional cortical cytoskeleton for proper reorganization in response to REEs, we hypothesize that their activity could be coordinated with that of the VAP27/NET3C EPCS complexes. In response to REE stress, the VAP27/NET3C and SYT1/ SYT5 complexes could integrate cytoskeleton dynamics, . The presence of PI4P-labeled vesicles is shown as bright dots at the PM (yellow asterisks). PI4P accumulation is shown as color-coded pixel intensity following the LUT scale shown in (D). (E) Quantification of the CITRINE-tagged 1×PH FAPP signal relative to mock conditions. In the box plots, the center line represents the median fluorescence intensity fold increase relative to mock, the cross represents the mean fluorescent intensity, the top and bottom edges are the 25th and 75th percentiles of the distribution, and the ends of the whiskers are set at 1.5 times the interquartile range (IQR). All values outside the IQR are shown as outliers. At least 100 regions of interest (ROIs) were measured for each treatment. The letters indicate statistically significant differences using Tukey multiple pairwise comparisons P<0.05. Scale bars=20 μm.
cortical ER stability, and EPCS positioning, effectively controlling cortical ER-PM communication. In this context, the REE-induced PI4P accumulation at the PM could influence the electrostatic surface of the PM (Simon et al., 2016), and regulate the docking affinity of the EPCS tethering complexes. Whether these changes in lipid composition could also activate endocytic and/or autophagic processes at EPCSs as proposed in Wang and Hussey (2019) has not been established and it is an area of active research in our laboratory. Remarkably, in plants subject to stress conditions that induce cytoskeleton disassembly (e.g. NaCl), an alternative SYT1dependent mechanism promotes cytoskeleton-independent EPCS remodeling (Lee et al., 2019). In these conditions, the NaCl-induced accumulation of PI(4,5)P 2 at the PM would have a minor influence on the PM electrostatic field, as PI4P still acts as the main contributor in this process (Simon et al., 2016), but could fine-tune EPCS-associated signaling pathways acting as a substrate of PM-localized phospholipases (e.g. PI-PLCs) (Singh et al., 2015). Together, these mechanisms illustrate the specificity and plasticity that govern EPCS rearrangements as an adaptive response to environmental stresses (Fig. 8).

Supplementary data
Supplementary data are available at JXB online. Table S1. Unique peptide counts in three independent immunoprecipitation experiments. Table S2. Primers used in this study. Fig. S1. Multiple sequence alignment of the SYT1/SYT5/ CLB1 C2 domains. Fig. S2. Root hair polarization defects in the triple syt1/ syt5/clb1 mutant.  Fig. S4. Effect of 2 h extracellular Ca 2+ depletion and REE treatments on EPCS number. Fig. S5. Effect of short-term REE treatments on EPCS number. Fig. S6. Effect of REE treatments on MAPPER-GFP localization. Fig. S7. Effect of La 3+ treatments on the localization of the PM marker C2AB-GFP. Fig. S8. Effect EGTA supplementation on REE-induced seedling growth defects. Fig. S9. Effect of NaCl and LaCl 3 treatments on cortical cytoskeleton organization. Fig. S10. Effect of La 3+ treatments on the accumulation of PI(4,5)P 2 at the PM.