Abstract
Abstract Rho GTPases are master regulators of cell signaling, but how they are regulated depending on the cellular context is unclear. Here, we show that the phospholipid phosphatidylserine acts as a developmentally-controlled lipid rheostat that tunes Rho GTPase signaling in Arabidopsis. Live super-resolution single molecule imaging revealed that RHO-OF-PLANT6 (ROP6) is stabilized by phosphatidylserine into plasma membrane (PM) nanodomains, which is required for auxin signaling. Furthermore, we uncovered that the PM phosphatidylserine content varies during plant root development and that the level of phosphatidylserine modulates the quantity of ROP6 nanoclusters induced by auxin and hence downstream signaling, including regulation of endocytosis and gravitropism. Our work reveals that variations in phosphatidylserine levels are a physiological process that may be leveraged to regulate small GTPase signaling during development.
One Sentence Summary Phosphatidylserine acts as a developmentally-controlled lipid rheostat that regulates cellular auxin sensitivity and plant development.
Main Text
Proteins from the Rho/Ras superfamily are small GTPases that regulate fundamental eukaryotic functions, including cell signaling, cell polarity, intracellular trafficking and cytoskeleton dynamics (1, 2). Furthermore, they control the morphology and behavior of cells and organisms by integrating signaling pathways at the cell surface into various cellular outputs. The small GTPase paradigm stipulates that they are in an “inactive” form when bound to GDP, and in an “active” form when bound to GTP. However, emerging evidence suggest that this view is likely oversimplified, since their membrane environment also dictates the signaling capacity of these GTPases (2, 3). In particular, Ras/Rho signaling is intimately linked with membrane lipids in all eukaryotes. Interaction with anionic lipids is important for their plasma membrane (PM) targeting (4, 5), but also mediates the clustering of these small GTPases at the cell surface into nanometer scale membrane domains (6-8). Phosphoinositides are low abundant anionic phospholipids that can be acutely produced or metabolized by dedicated enzymes with exquisite subcellular precision, and as such often function as signaling lipids (9). Moreover, they mediate the recruitment of some Ras/Rho proteins to the cell surface and into nanoclusters (4, 7, 10). Phosphatidylserine (PS) is also involved in the nanoclustering and signaling of some GTPase, such as K-Ras in human and Cdc42 in yeast (3, 6, 8, 11-13). However, by contrast to phosphoinositides, PS is a relatively abundant anionic phospholipid, representing up to 10-20% of the total phospholipids at the PM inner leaflet (14). In addition, PS is not constantly modified by specialized metabolizing enzymes and the subcellular PS repartition is thought to be relatively stable across cell types (14). Therefore, PS appears to be a structural component of the membrane, which is required for K-Ras/Cdc42 nanoclustering. It is unknown, however, whether PS also has a regulatory role in vivo in modulating nanocluster formation and subsequent signaling. In other words, is PS function in GTPase nanoclustering rate limiting? And if yes, are PS levels regulated during development and what are the consequences of such changes on small GTPases signaling capacity? Here, we addressed these questions using the Arabidopsis thaliana root as a model system, because it is a genetically tractable multicellular organ, with a variety of cell types and cell differentiation states and amenable to live imaging, including super-resolution microscopy (15).
In plants, there is a single protein family in the Ras/Rho GTPase superclade, called ROP for RHO-OF-PLANT(16). ROPs are master regulators of cell polarity and cell morphogenesis, but they also sit at the nexus of plant hormone signaling (including auxin and abscisic acid), cell wall sensing pathways and receptor-like kinase signaling (involved in development, reproduction and immunity) (16-27). Here, we show that auxin triggers ROP6 nanoclustering within minutes, in a PS dependent manner. Furthermore, we found that PS is required for ROP6 signaling, and variations in the cellular PS content directly impact the quantity of ROP6 nanoclusters and thereby subsequent downstream auxin signaling, including the regulation of endocytosis and root gravitropism. Therefore, PS is not a mere structural component of the membrane, it is a bona fide signaling lipid that acts as a developmentally-controlled lipid rheostat to regulate small GTPases in a cell-context dependent manner
Results and discussion
Plasma membrane phosphatidylserine levels vary during root cell differentiation
Phosphatidylserine (PS) is an anionic phospholipid that partitions between the cytosolic leaflets of the PM and endosomes (28). Bulk PS measurement in Arabidopsis thaliana suggested that the relative PS concentration can vary in vivo depending on the organ (29). In order to get tissue and cellular resolution on the PS distribution, we recently validated the use of two PS reporters in Arabidopsis (28, 30), the PS-binding C2 domain of Lactadherin (C2Lact) (5) and the PH domain of EVECTIN2 (2xPHEVCT2) (31). In both cases, the proportion of PS sensors was markedly more pronounced at the PM than endosomes in the root basal meristem compared to cells in the elongation zone (Fig. 1A, Fig. S1A-B). This developmental gradient appeared to be in part regulated by the plant hormone auxin as relatively short treatment (60min) with the synthetic auxin naphthalene-1-acetic acid (NAA) increased the level of both PS sensors at the PM at the expense of their endosomal localization in the elongation zone (Fig. 1B, Fig. S1C). Therefore, not only the overall PS level vary depending on the organ but there are also local variations of the PS content at the PM within an organ, during cell differentiation and in response to hormonal cues.
Graded phosphatidylserine levels tune ROP6 signaling
In order to test the potential impact of PS variations during development, we experimentally manipulated the plant PS content, from no PS biosynthesis in the phosphatidylserine synthase1 (pss1) mutant (28), to mild PS levels in transgenic lines expressing artificial microRNAs against PSS1 (amiPSS1), and high PS levels in transgenic lines overexpressing PSS1 (PSS1-OX) (Fig. S1D-E). The changes in PS content measured in amiPSS1 and PSS1-OX lines of about ±2-fold fell well into the physiological range, since PS levels in Arabidopsis vary about 5-fold between roots and leaves tissues (29). The pss1 mutant showed defects in root gravitropism (Fig. S1F-G). Quantitative analyses of root bending following gravistimulation (Fig. S1H) revealed that the pss1-3 mutant had no gravitropic response (Fig. 2A), amiPSS1 lines had an attenuated response, while PSS1-OX lines were hypergravitropic (Fig. 2B). These opposite gravitropic phenotypes of PSS1 loss- and gain-of-function resembled those of ROP6, a Rho-Of-Plant (ROP) GTPase, which is activated by auxin and regulates root gravitropism (21, 22). Like PSS1-OX lines, lines overexpressing either ROP6 (ROP6-OX) or constitutive active GTP-lock ROP6 (ROP6CA) showed a hypergravitropic phenotypes, which were abolished in a pss1-3 background (Fig 2A). During root gravitropism, ROP6 acts downstream of auxin to inhibit endocytosis and regulate microtubule orientation (21, 22, 32, 33). Similar to rop6 (21, 22, 32), we observed that in pss1-3: i) FM4-64 and PIN2-GFP uptake in the presence of BrefeldinA (BFA) was increased (Fig. 2C, Fig. S2A-D), ii) auxin failed to inhibit FM4-64 and PIN2-GFP endocytosis (Fig. 2C, Fig. S2A-D), iii) CLATHRIN-LIGHT-CHAIN2 (CLC2)-GFP PM association was insensitive to auxin treatment (Fig. S2E), and iv) auxin-triggered microtubule reorientation was abolished (Fig. S2F). FM4-64 uptake in pss1-3xROP6CA was identical to that of pss1-3 single mutant and opposite to ROP6CA(Fig. 2C), showing that PSS1 is required for ROP6-mediated inhibition of endocytosis. Furthermore, transgenic lines with low PS content (amiPSS1) had decreased auxin-mediated inhibition of endocytosis, while lines with heightened-PS content (PSS1-OX) mimicked ROP6CA phenotypes with pronounced inhibition of endocytosis upon auxin treatment (Fig. 2D, Fig. S2G-I). Together, our analyses suggest that i) PS is required for auxin-mediated ROP6 signaling during root gravitropism and ii) PS levels impact the strength of ROP6 signaling output in a dosedependent manner.
Auxin triggers ROP6 nanoclustering at the plasma membrane
PS and ROP6 both accumulate at the PM, we therefore reasoned that PS may contribute to ROP6 localization. Surprisingly, GFP-ROP6 localization, as seen by confocal microscopy, was almost identical in pss1-3 and WT, being mainly at the PM and only faintly delocalized in intracellular compartments in pss1-3 (Fig. S3A). In leaves, ROP6CA was previously shown to be confined in membrane domains (34), raising the possibility that PS could contribute to ROP6 signaling by regulating its lateral segregation at the PM. To analyze ROP6 PM partitioning in root cells and in the context of auxin response, we developed several microscopy-based assays, including Fluorescence Recovery After Photobleaching (FRAP), Total Internal Reflection Fluorescence Microscopy (TIRFM) (35) and PhotoActivated Localization Microscopy (PALM) (15) (Fig. S4). As shown for ROP6CA in leaf (34), activation of ROP6 (here using auxin treatment) delayed GFP-ROP6 fluorescence recovery after photobleaching (Fig. 3A-B and Fig. S3B-D). TIRFM on root tip epidermal cells allowed to focus only on the plane of the PM with a 100nm axial resolution (35) (Fig. S4B) and revealed that GFP-ROP6 mostly localized uniformly at the PM (Fig. 3C). By contrast, in auxin-treated plants, GFP-ROP6 additionally resided in diffraction-limited spots present in the plane of the PM (Fig. 3C), suggesting that auxin treatment triggers the clustering of ROP6 in membrane domains. By using stochastic photoswitching on live roots, single particle tracking PALM (sptPALM) experiments provided tracks of single molecule localization through time, and therefore allowed us to analyze the diffusion behavior of single ROP6 molecule in response to auxin (Fig. S4D). While mEos-ROP6 molecules in the untreated condition were almost exclusively diffusing, mEos-ROP6 molecules in plants treated for 5 minutes with auxin (or mEos-ROP6CA molecules) existed in two states at the PM of epidermal cells: immobile or diffusing (Fig. 3D-E, Fig. S5A-C and Supplementary Video 1). Clustering analyses on live PALM images (36, 37) showed that auxin triggered the clustering of mEos-ROP6 in PM-nanodomains of about 50 to 70 nm wide (Fig. 3F-G and Fig. S6). Together, our data indicate that activation, either genetically (i.e. ROP6CA) or by an endogenous activator (i.e. auxin), triggers ROP6 recruitment, immobilization and stabilization into PM-nanodomains and that these events happen minutes following auxin treatment.
PS regulates auxin-induced ROP6 nanoclustering in a dose-dependent manner
Next, we tested the impact of PS on ROP6 PM dynamics. In FRAP experiments, GFP-ROP6 sensitivity to auxin was reduced in pss1-3 (Fig. 4A and Fig. S3B-E), suggesting that PS is critical for the immobilization of ROP6 by auxin. In WT plants, NAA-induced GFP-ROP6 presence in PM-nanodomains was more pronounced in the basal meristem than in the elongation zone in TIRFM experiments (Fig. 4B), which correlated with the observed differential presence of PS content at the PM in these regions (Fig. 1A). To analyze whether this differential auxin sensitivity was dependent on the amount of PS present in these cells, we performed PS loss- and gain-of-function experiments. First, auxin failed to induce GFP-ROP6 nanodomains in both region of the root in pss1-3 (Fig. 4C), suggesting that PS is indeed required for auxin-triggered ROP6 nanoclustering. Second, exogenous treatment with lysophosphatidylserine (lyso-PS), a more soluble lipid than PS but with an identical head group (28), boosted the number of auxin-induced GFP-ROP6 nanodomains observed in TIRFM in WT plants (Fig. 4D). Together these data suggest that the quantity of PS at the PM impacts ROP6 nanoclustering. While PS was required for auxin-triggered ROP6 nanoclustering, a certain amount of ROP6 was still found in PM domains in pss1, independent of the presence of auxin (Fig. 4C). Kymograph analyses revealed that ROP6-containing PM-nanodomains observed by TIRFM were immobile in both WT and pss1-3 (Fig. 4E). Photobleaching experiments showed that GFP-ROP6 was highly stable in these PM-nanodomains in the WT (i.e. no fluorescence recovery of GFP-ROP6 in PM-nanodomains, by contrast to a fast recovery of fluorescence outside of these domains) (Fig. 4E and Supplementary Video 2). By contrast, GFP-ROP6 fluorescence in PM-nanodomains was rapidly recovered in pss1-3, suggesting that ROP6 was not stabilized into PM-nanodomains in the absence of PS (Fig. 4E and Supplementary Video 3). Together, our results show that PS is necessary for both ROP6 stabilization into PM-nanodomains and downstream ROP6 signaling, including regulation of endocytosis and root gravitropism.
Immobile phosphatidylserine molecules accumulate in PM-nanodomains together with ROP6
Next, we addressed whether the regulation of ROP6 clustering and signaling by PS was direct. If so, we should expect PS, like ROP6, to also localize in PM-nanodomains. Using sptPALM and clustering analyses, we found that i) the PS reporter mEos-2xPHEVCT2 segregated into nanodomains at the PM of root epidermal cells (Fig. 5A) and ii) about 35% of mEos-2xPHEVCT2 molecules were present as a slow-diffusible population (Fig. 5B and Fig S5D), showing an apparent diffusion coefficient similar to that of immobile mEos-ROP6 (Fig. 3E) and suggesting that PS and ROP6 may co-exist in the same PM-nanodomains. Accordingly, co-visualization of GFP-ROP6 and the PS sensor 2xmCHERRY-C2Lact in TIRFM confirmed that they at least partially reside in the same PM-nanodomains in response to auxin (Fig. 5C).
ROP6 interaction with anionic phospholipids is required for nanoclustering and downstream signaling
ROP6 possess in its C-terminus a polybasic region adjacent to a prenylation site (Fig. S7A). Such polybasic region is anticipated to bind to anionic phospholipids, including PS, via non-specific electrostatic interactions (4, 5), which we confirmed in protein-lipid overlay experiments (Fig. S7B). Substitution of seven lysines into neutral glutamines in ROP6 C-terminus (ROP67Q) abolished in vitro interactions with all anionic lipids (Fig. S7B). In planta, diminishing the net positive charges of ROP6 C-terminus (ROP67Q) or the net negative charge of the PM gradually induced ROP6 mislocalization into intracellular compartments (Fig. S7C-D). To test the functionality of ROP67Q at the PM, we selected transgenic lines that had strong expression level to compensate for their intracellular localization and therefore have comparable levels of ROP67Q and ROP6WT at the PM (Fig. S8A and D-F). ROP67Q mutants were not functional in planta (Fig. 6A-B, Fig. S8B-C), even though the 7Q mutations had no impact on ROP6 intrinsic GTPase activity in vitro and ROP6-GTP conformation in vivo (Fig. S9). We next analyzed the dynamics of mEos-ROP67Q at the PM of wild-type roots by sptPALM experiments and found that it had the same proportion of immobile molecules than mEos-ROP6WT in pss1-3, and that in both cases they were insensitive to auxin (Fig. 6C, Fig. S5E-H). Therefore, impairing PS/ROP6 interaction by either removing PS from the membranes (pss1 mutant), or by mutating the anionic lipid-binding site on ROP6 (ROP67Q) similarly impacted ROP6 signaling and its auxin-induced nanoclustering.
Conclusions
Here, we showed that in root tip epidermal cells: i) ROP6 is immobilized in PM-nanodomains upon activation by auxin, ii) PS is necessary for both ROP6 stabilization into PM-nanodomains and signaling, iii) ROP6 directly interacts with anionic lipids, including PS, and iv) PS itself is present and immobile in these PM-nanodomains, suggesting that stabilized ROP6 in PS-containing nanoclusters constitutes the functional signaling unit of this GTPase. Our imaging pipeline revealed that ROP6 nano-organization is rapidly remodeled by auxin and as such will provide a quantitative in vivo read-out to re-evaluate how auxin may be perceived upstream of ROP6 activation. Given that plants have 11 ROPs, which can respond to a wide range of signals (16), it will be intriguing to address whether nanoclustering is specific to auxin response or common to other signals and to various ROPs, and to what extent it may contribute to signal integration by plant Rho GTPases. All ROP proteins have polybasic clusters at their C-terminus (Fig. S10A), and PS could therefore potentially regulate additional member of this family. Interestingly, in addition to root gravitropic defects, pss1 had many developmental phenotypes that may be linked to altered ROP function (e.g. pavement cell and root hair morphology, planar polarity defects, see Fig. S10B-F) but that are not found in rop6 single mutant and could therefore involve additional ROP proteins. Furthermore, nanoclustering seems to be a shared feature of several yeast and animal small GTPases, including K-Ras, Rac1 and Cdc42 (6-8), and both K-Ras and Cdc42 require PS for nanoclustering (3, 6, 8, 11, 13). Here, we found that in vivo variations of the PS concentration at the PM act like a rheostat to adjust the sensitivity of ROP6-nanoclustering and hence auxin signaling in a cell-context dependent manner. Our results open the possibility that variations of the PS concentration at the PM in animal systems could also control the signaling capacity of these small GTPases during either normal or pathological development.
Funding
Y.J. is funded by ERC no. 3363360-APPL under FP/2007-2013; Y.J and A.M. by an INRA innovative project (iRhobot).
Author contributions
M.P.P. generated all transgenic material, and was responsible for all experiments. V.B., M.P.P. and A.M. conceived, performed and analyzed super-resolution imaging. V.B. performed and analyzed TIRFM and FRAP imaging. L.M-P and P.M. performed lipid measurements. A.M. imaged Raichu-ROP6 sensors. J.B. produced recombinant ROP6 and performed GTPase assays. M.P.P and L.A. performed lipid overlay experiments. M.M.M-B., and C.M. assisted with phenotyping and cloning, A.C. performed qRT-PCR analyses, J-B.F. and M.N. designed the sptPALM analyses pipeline, M.P.P., V.B. and Y.J. conceived the study, designed experiments and wrote the manuscript and all the authors discussed the results and commented on the manuscript. Correspondence and requests for materials should be addressed to Y.J.
Competing interests
Authors declare no competing interests.
Supplementary Video 1: Example of video showing mEos-ROP6 single molecule imaging used for sptPALM and livePALM analysis.
Supplementary Video 2: Video showing GFP-ROP6 localization (10μM NAA, 20 min) in TIRFM before and after bleaching a region of interest containing GFP-ROP6 nanoclusters. Note that GFP-ROP6 nanoclusters are immobile and that they do not recover fluorescence after photobleaching (by contrast to surrounding “non-clustered” GFP-ROP6 signal at the PM, for which recovery of fluorescence is fast)
Supplementary Video 3: Video showing GFP-ROP6 localization (10μM NAA, 20 min) in pss1-3 mutant in TIRFM before and after bleaching a region of interest containing GFP-ROP6 nanoclusters. Note that, like in the WT, GFP-ROP6 nanoclusters are immobile. However, by contrast to the WT, GFP-ROP6 nanoclusters rapidly recover fluorescence after photobleaching.
Acknowledgments
We thank M. Dreux, E. Bayer, O. Hamant, S. Mongrand, Y. Boutté, J. Gronnier, J. Reed, T. Vernoux and the SiCE group for discussions and comments, T. Stanislas for help with root hair phenotyping, S. Bednarek, S. Yalovsky, B. Scheres and the NASC collection for providing transgenic Arabidopsis lines, A. Lacroix, J. Berger and P. Bolland for plant care, J.C. Mulatier for help in preparing lipids. We acknowledge the contribution of SFR Biosciences (UMS3444/CNRS, US8/Inserm, ENS de Lyon, UCBL) facilities: C. Lionet, E. Chattre, and C. Chamot at the LBI-PLATIM-MICROSCOPY for assistance with imaging and V. Guegen-Chaignon at the Protein Science Facility for assistance with protein purification. We thank the PHIV and MRI platform for access to microscopes.