PC (phosphatidylcholine, catalog number 850375), PE (phosphatidylethanolamine, catalog number 850725), PG (phosphatidylglycerol, catalog number 841148), PA (phosphatidic acid, catalog number 840875), and DG (diglyceride, catalog number 800811) were purchased from Avanti Polar Lipids, Inc.; PI (phosphatidylinositol), PS (phosphatidylserine), DGDG (digalactosyldiacylglycerol), MGDG (monogalactosyldiacylglycerol), and SQDG (sulfoquinovosyl diacylglycerol) were purchased from Lipid Products, United Kingdom. LPA (lysophosphatidic acid, catalog number L7260), MG (monoglyceride, catalog number M7765), and NAA (1-naphthylacetic acid) were ordered from Sigma-Aldrich. FM4-64 was ordered from Molecular Probes, Inc.

All Arabidopsis thaliana lines used in this study are as follows: Columbia-0 (Col-0), 35S::GFP-ROP6 was kindly provided by Dr. Ying Fu (China Agricultural University), PIN2-GFP, rop6^CA and rop6-2 were kindly provided by Dr. Deshu Lin (Fujian Agriculture and Forestry University). Arabidopsis seeds were sterilized, sown on 0.443% (w/v) Murashige and Skoog salts (MS, Sigma-Aldrich) medium plus 20 g/L sucrose (pH 5.8), and grown vertically or horizontally in controlled growth chamber at 22°C under 16-h light/8-h dark cycle for 5–10 days (light intensity of 50 μmol m^-2s^-1) unless indicated otherwise. For soil growth, the seedlings were then transferred to soil under a 16-h light (22°C)/8-h dark (20°C) cycle.

The coding sequence of AtROP6 was amplified with the R6-Bf/R6-Hr primer and cloned into the pET-30a vector between the BamHI and Hind III sites to generate the recombinant plasmid pET-30a-AtROP6. The coding sequences of AtROP1 and AtROP3 were amplified with the R1-Ef/R1-Sr, R3-Ef/R3-Sr primers and cloned into the pGEX-6p1 vector between the EcoRI and SalI sites to generate the recombinant plasmid 6p1-AtROP1 and 6p1-AtROP3, respectively. The coding sequence of AtROP6 was amplified with the R6-Bf/R6-Er primer and cloned into the pGEX-6p1 vector between the BamHI and EcoRI sites to generate the recombinant plasmid 6p1-AtROP6. The coding sequence of 14-3-3λ was amplified with the λ-Bf/λ-Sr primer and cloned into the pGEX-6p1 vector between the BamHI and SalI sites to generate the recombinant plasmid 6p1-14-3-3λ. Mutation of Thr-20 to Asn was generated by first amplification with the R6TN-f/R6-Er, R6-Bf/R6TN-r primers and second amplification with the R6-Bf/R6-Er primers, and then cloned into the pGEX-6p1 vector to generate the recombinant plasmid 6p1-AtROP6^DN, or cloned into the pCAMBIA1390-GFP vector to generate the recombinant plasmid 1390-GFP-AtROP6^DN. Mutation of Gly-15 to Val was generated by first amplification with the R6GV-f/R6-Er, R6-Bf/R6GV-r primers and second amplification with the R6-Bf/R6-Er primers, and then cloned into the pGEX-6p1 vector to generate the recombinant plasmid 6p1-AtROP6^CA, or cloned into the pCAMBIA1390-GFP vector to generate the recombinant plasmid 1390-GFP-AtROP6^CA. The plasmid was verified by sequencing and was then transformed into the bacterial strain BL21 or protoplast, respectively. Primer sequences are listed in Supplementary Table 1.

Bacterial cells were grown in Luria-Bertani medium supplemented with 50 mg/L kanamycin or 100 mg/L ampicillin at 37°C until cells reached an optical density OD_{600 nm} from 0.7 to 0.9. The recombinant protein was expressed at 37°overnight after induction with 0.4 mM of IPTG. Proteins were purified using Ni-beads as described in the manufacturer’s instructions.

The bacterial strain expressing MBP-RIC1 in pMAL21 construct was kindly provided by Dr. Ying Fu (China Agricultural University). MBP-RIC1-Conjugated Amylose Beads was prepared as described previously (Xu, 2012).

Stocks of phospholipids were first prepared in organic solvents according to the manufacturer’s instructions or in the lipid-soluble solvent: DCM:MeOH:H₂O 65:35:8 (v/v/v). Lipid-protein overlay assay was performed as described previously (Sun et al., 2013). Briefly, lipid test strip was prepared by spotting the indicated lipids with the amount of 5 nmol onto a PVDF membrane and kept dry for 1 h at room temperature. The lipid test strip was incubated with 1 μg/mL of His fusion ROP6 protein in 3% BSA-20 mM Tris-HCl (pH 8.0) for 2 h at room temperature for blocking and incubation. After washing with PBST (0.1% tween 20) buffer, the presence of bound ROP6 protein was detected using mouse anti-His-tag antibody as the primary antibody and goat anti-mouse antibody conjugated to HRP as the secondary antibody.

Stocks of phospholipids were first prepared in organic solvents according to the manufacturer’s instructions or in the lipid-soluble solvent: DCM:MeOH:H₂O 65:35:8 (v/v/v). Liposome sedimentation assay was performed as described previously (Sun et al., 2013). Briefly, liposome mixture was prepared with a weight ratio of 1:1 DOPC/DOPE and a serial weight ratio of PG. The total lipid weight was 200 μg. The solvent in the mixtures was first removed using nitrogen gas for 30 min without heating and kept dry in a desiccator overnight at room temperature. The lipids were then hydrated in 200 μL buffer containing 100 mM NaCl, 1 mM NaN₃ and 20 mM Tris-HCl (pH 6.8) at 67°C for 1 h, during which time the lipids were mixed by pipetting up and down with a micropipette every 15 min. After being freeze-thawed three times, the liposomes were formed by sonication. The liposomes were further pelleted, re-suspended in 40 μL buffer containing 100 mM NaCl, 1 mM NaN₃ and 20 mM Tris-HCl (pH 6.8), and protein (10 μg) was then added. Incubation was performed with the liposome for 30 min at room temperature. Centrifugation at 12,000 rpm gets liposome-bound pellet fraction and free-protein supernatant fraction, which were further analyzed by SDS-PAGE and Coomassie Blue Staining. Quantitation of protein content was performed with ImageJ software.

MST assay was carried out using Monolith NT.115 instrument (NanoTemper Technologies) as described previously (Lin C.C. et al., 2012; Entzian and Schubert, 2016). The buffer with Tris salt in the purified recombinant proteins was first replaced with PBST buffer (0.005% tween 20, pH = 7.5) using column A supplied by the manufacturer. Then the proteins at a final concentration of 5 μM were labeled with excess NHS NT-647 dye at a molar ratio of 1:5 at room temperature for 30 min in the dark according to the manufacturer’s instructions. Free unlabeled dye was removed using column B pre-equilibrated with PBST buffer (0.005% tween 20, pH = 7.5). PG (1 mg) was first dissolved in 50 μL organic solvent DCM:MeOH:H₂O 65:35:8 (v/v/v), and then dried with nitrogen gas for 30 min and kept dry in a desiccator overnight at room temperature. The dried PG was dissolved in 1 mL PBST buffer (0.005% tween 20, pH = 7.5) by hydration for 1 h at room temperature until no lipid could be seen on the Eppendorf tube wall, and was then centrifuged at 12,000 rpm for 10 min at room temperature to remove the possible existing pellet. The PG solution was serially diluted with PBST buffer and mixed with the same amount of labeled protein. The samples were loaded into capillaries (NanoTemper Technologies) and analyzed. The assay was carried out with 20% LED power and 20% MST power. Signal Thermophoresis + T-Jump Data were used for calculating dissociation constant (Kd). Data was analyzed using software NT Analysis and Origin9.

FM4-64 was dissolved in distilled water, BFA and cycloheximide (CHX) were dissolved in dimethyl sulfoxide, and NAA was dissolved in methanol. For phenotypic analysis, PG was dissolved in methanol. For other experiments, the solvent to dissolve PG was indicated in the experiment.

ROP6 activity assay was performed as described previously (Xu, 2012). ROP6 with activity could be pulled down by MBP-RIC1-conjugated agarose beads, while ROP6 without activity could not be, so MBP-RIC1-conjugated agarose beads and GFP-tagged ROP6 were both need to be prepared.

For preparation of MBP-RIC1-conjugated agarose beads, MBP-RIC in pMAL21 construct was transformed into Escherichia coli. Bacterial cells BL21 and the cells were cultured in Luria-Bertani medium with 100 mg/L ampicillin at 37°C. When the optical density of the cells reached about 0.8, the cells were cooled to 16°C and 0.4 mM of IPTG was added into the medium. The expression of MBP-RIC fusion protein was induced overnight at 16°C and then purified according to the manufacturer’s instructions.

For preparation of GFP-tagged ROP6, 10-day-old 35S::GFP-ROP6 transgenic seedlings were treated with indicated amounts of PG or without PG in liquid 1/2 MS for 48 h, and then the seedlings were grounded with liquid nitrogen and extracted at 4°C for 1 h in extraction buffer: 25 mM Hepes, 10 mM MgCl₂, 1 mM EDTA, 100 mM KCl, 5 mM DTT, 5 mM Na₃VO₄, 5 mM NaF, 1 mM PMSF, 1% Triton X-100, pH 7.4, in which, DTT, Na₃VO₄, PMSF were added before use. After extraction, 50 μL of the total protein was transferred out for the final total protein analysis, and for the reserved solution, the centrifugation of 10,000 g at 4°C was used to remove the debris and the same volume of extraction buffer without Triton X-100 was added into the total protein in the supernatant.

For preparation of GFP-tagged ROP6^CA and GFP-tagged ROP6^DN, the plasmids were purified by CsCl gradient centrifugation. Then the Col-0 protoplasts were prepared and the plasmids were transformed into the protoplasts as described previously (Sheen, 2001). The GFP-tagged ROP6^CA and GFP-tagged ROP6^DN were prepared as described previously (Xu, 2012).

Then the MBP-RIC1-conjugated agarose beads were added into the supernatant and incubated at 4°C for 3 h. After incubation, the beads were washed five times with washing buffer: 25 mM Hepes, 5 mM MgCl₂, 1 mM EDTA, 1 mM DTT. The MBP-RIC1 bound GFP-ROP6 was separated by SDS-PAGE and detected using mouse anti-GFP antibody as the primary antibody and goat anti-mouse antibody conjugated to HRP as the secondary antibody.

FM4-64 staining was performed as described previously (Chen et al., 2012). Three-day-old seedlings were transferred to MS medium supplemented with or without indicated amounts of PG. After growing for 48 h, the seedlings were first labeled with 2 μM FM 4-64 for 5 min, washed out for three times, and further incubated in liquid 1/2 MS medium for 20 min at room temperature. PIN2 internalization observation was performed as described previously (Chen et al., 2012). Three-day-old seedlings were transferred to MS medium supplemented with or without indicated amounts of PG. After growing for 48 h, the seedlings were treated with 50 μM BFA or 50 μM BFA plus 5 μM NAA for 2 h. For confocal microscopy observation, the seedlings were mounted on glass slides in 10% glycerol for observation under confocal microscope (Andor Dradonfly spinning disk confocal, Nikon TiE microscope, plan APO 60×, NA1.4 objective, and andor zyla4.2plus sCMOS camera) with excitation wavelength 561 nm and emission wavelength 620–650 nm for FM4-64 observation and excitation wavelength 488 nm and emission wavelength 507 nm for GFP observation. Quantitation of the fluorescence signal was performed with ImageJ software.

Five-day-old seedlings were transferred to the indicated amounts of PG-containing MS medium and applied recovered growth for 12 h, and were then gravity stimulated for 12 h (at 90° rotation). The root tips were labeled at indicated time periods and the angles formed were measured using ImageJ software.

Quantitation of PG content in Arabidopsis was performed as described previously with minor modification (Milne et al., 2005; Hsu et al., 2007). Five-day-old seedlings were transferred to MS medium supplemented with or without 0.1 μM NAA. After growing for 5 days, the root part and leaf part were collected separately using a scissor, and only 2/3 root part near the root tip was collected to avoid hypocotyl contamination.

The samples were grounded under liquid nitrogen, and 200 μL ice-cold 2:1 CH₂Cl₂:CH₃OH (containing 0.25% 12 M HCl, v/v) was added into each sample. After a vortex for 2 min, 40 μL of 1 M HCl was added and vortexed for 30 s. Then the samples were centrifuged for 5 min at 12,000 g at 4°C, and the lower organic phase was transferred to a new tube and dried using a vacuum centrifuge (Eppendorf Concentrator plus). The extracted lipids were re-dissolved in 100 μL CH₂Cl₂-CH₃OH-H₂O-300 mM piperidine (1:1:0.2:0.1, v/v/v/v) for further HRESIMS (high-resolution electrospray ionization-mass spectrometry) analysis (Thermo Fisher, Q-Exactive).

AtROP6, a plasma membrane-localized protein in Arabidopsis, functions as a molecular switch in many cellular signaling responses (Sorek et al., 2010; Craddock et al., 2012). To investigate whether lipids in the plasma membrane are involved in the regulation of AtROP6 function, we performed a protein-lipid overlay screen assay to look for the possible lipids that AtROP6 physically interacts with.

Glycerophospholipids including PA, PI, PS, PE, LPA, PG, PC and glycerolipids including MG, DG, MGDG, DGDG, SQDG are lipids that were reported to be located or possibly located at the plasma membrane (Manoharan et al., 1985; Ritter and Yopp, 1993; Drøbak et al., 1994; Moreau et al., 1998; Andersson et al., 2003; Kim et al., 2004; Jouhet et al., 2007; Li et al., 2007, 2016; Yang et al., 2010; Dong et al., 2012). We spotted these lipids onto the PVDF membrane and performed the binding assay between these lipids and AtROP6. The recombinant protein His-ROP6 was purified from E. coli strain BL21 expressing recombinant plasmid pET-30a-AtROP6. The result showed that His-AtROP6 specially bound to PG, but not other lipids (Figure 1A). Further interaction assay between PG and His-casein kinase1-like protein2 (CKL2) (Zhao et al., 2016), a negative control, revealed that PG specially bound to His-AtROP6, not His-CKL2 (Supplementary Figure 1A).

To further verify the interaction between PG and AtROP6, we performed a liposome sedimentation assay. We prepared control liposome containing only PC and PE, and also prepared PG-containing liposomes. PG-containing liposomes were prepared with the same total lipid content as the control liposome by replacing equal amounts of PC and PE with PG. After incubation with the recombinant protein His-AtROP6, the liposomes were separated into the pellet and the supernatant by centrifugation, in which liposome-bound His-AtROP6 existed in the pellet and liposome-unbound His-AtROP6 existed in the supernatant. After analysis by SDS-PAGE and Coomassie Blue Staining, we found that the protein level of His-AtROP6 increased dramatically in the pellet with the elevated PG content in the liposomes (Figures 1B,C). This result supports the evidence that AtROP6 specially binds to PG, not PC and PE.

We also performed MST assay to investigate the binding affinity between AtROP6 and PG. To prepare the PG solution, we used PBST (0.005% tween 20) to hydrate PG to get the protein-compatible PG solution. His-AtROP6 was labeled with fluorescent dye and mixed with a series dilution of PG. The dissociation constant (Kd) between His-AtROP6 and PG was determined to be 4.8 ± 0.4 μM, while PG and fluorescent dye, His-AtROP6 and solvent (PBST, 0.005% tween 20) had no interaction and could not be fitted (Figure 1D). These results suggest that AtROP6 specially binds to PG, and the dissociation constant is 4.8 ± 0.4 μM.

To further investigate whether PG binds to other ROP protein family members, we performed liposome sedimentation assay and MST assay between PG and GST-AtROP1, GST-AtROP3 or GST-AtROP6. In the liposome sedimentation assay, after liposome preparation and protein incubation, we found that the protein level of GST-AtROP1, GST-AtROP3 and GST-AtROP6 increased in the pellet with the elevated PG content (Supplementary Figure 1B). By contrast, the negative control protein levels of GST-tag and GST-14-3-3λ in the pellet did not change with the increase of PG content in the liposomes (Supplementary Figure 1C). In the MST assay, the recombinant proteins were labeled with fluorescent dye and mixed with a series dilution of PG. The binding affinities between PG and GST-AtROP1, GST-AtROP3, or GST-AtROP6 were calculated as 22.4 ± 0.7 μM, 34.5 ± 0.9 μM, or 11.6 ± 0.3 μM (Supplementary Figure 1D). The negative controls between GST tag and PG, and GST-14-3-3λ and PG had no interaction (Supplementary Figure 1D). These results indicate that other ROP protein family members AtROP1 and AtROP3 also bind to PG but with a little weaker interaction compared with AtROP6. The differences in the binding affinity between His-ROP6 and PG, and GST-ROP6 and PG might due to the influence of tag on ROP6 structure. We also performed the interaction test between PG and AtROP6^CA (mutation of Gly-15 to Val, a constitutively active form of ROP6) or AtROP6^DN (mutation of Thr-20 to Asn, a dominant negative form of ROP6), which showed that both GST-AtROP6^CA and GST-AtROP6^DN bound to PG in the liposome sedimentation assay (Supplementary Figure 2).

These results suggest that PG not only binds to AtROP6, but also binds to other ROP protein family members, such as AtROP1 and AtROP3.

PG is widely studied as a thylakoid lipid in plants, which plays essential roles in oxygenic photosynthesis (Babiychuk et al., 2003; Kobayashi et al., 2016). PG is also reported to localize in the oat root with decreased content when plant is under phosphate-limitation condition (Andersson et al., 2003). However, the function of PG in the root has never been reported. In our study, we found that PG binds to AtROP6. To investigate what role PG binding to AtROP6 plays, we performed and AtRop6 activity assay. ROP6 activity could be analyzed by an effector binding-based assay using RIC1, an effector of AtROP6, which specifically binds to the active form of ROP6 but not to the inactive form of ROP6 (Xu, 2012). To test the effect of PG on ROP6 activity, Arabidopsis seedlings expressing 35S::GFP-ROP6 were treated with indicated amounts of PG, followed by a pulled-down assay using MBP-RIC1-conjugated agarose beads and active ROP6 was detected with anti-GFP antibody. As shown in Figures 2A,B, we found that ROP6 activity was increased with the addition of PG and the activity displayed a dose-dependent manner with the increase of PG concentration. PC and PE were also applied on the seedlings expressing 35S::GFP-ROP6; however, no stimulation of ROP6 activity was seen (Supplementary Figure 3). In the interaction assay, PG also bound to AtROP6^CA and AtROP6^DN, so we further performed a ROP6 activity assay both in the protoplasts expressing 35S::GFP-ROP6^CA and protoplasts expressing 35S::GFP-ROP6^DN, and no stimulation effect was seen in either of them after PG treatment (Figure 2C). These results suggest that the interaction between PG and AtROP6 plays a role in stimulating AtROP6 activity.

Since PG activates AtROP6 activity, we want to investigate the distribution of PG species in Arabidopsis. The lipids from Arabidopsis leaves and roots were extracted separately, and then analyzed by mass spectrometry. By evaluating PG species, we found that PG existed both in leaves and roots but with distinctive fatty acyl chain patterns (Figures 3A–C). The PG from leaves showed nine species with the major species being C34:4-PG, which is similar to the work previously reported (Hsu et al., 2007). However, the PG from roots showed the major species being C34:3-PG. Besides C34:3-PG and C34:4-PG, all other PG species showed dramatic difference between roots and leaves (Figures 3A–C). These results suggest that PG exists in Arabidopsis but with distinctive fatty acyl chain patterns between roots and leaves.

AtROP6 was reported to function in membrane trafficking in plants (Chen et al., 2012). Amphiphilic styryl dye FM4-64 is a marker widely used to monitor endocytosis in plants, which stains plasma membrane and is integrated into vesicles accompanied by endocytosis process (Rigal et al., 2015). As AtROP6 regulates endocytosis in Arabidopsis in previous report (Chen et al., 2012), and PG bound to and activated AtROP6 in our study, we performed an endocytosis experiment to investigate whether PG is involved in this process. Arabidopsis seedlings of Col-0 were first applied treatment with indicated amounts of PG or solvent, and then the root was stained with FM4-64 and endocytosis was observed. We found that, with PG treatment, the uptake of FM4-64 was reduced in wildtype Col-0 in a dose-dependent manner (Figures 4A,B), which is consistent with the PG concentration on ROP6 activity. When PG was applied at 20 μM, the endocytosis process in Col-0 seedlings was almost completely inhibited, which is similar to that in rop6^CA (a constitutively active mutant of ROP6) (Figures 4A,B). When we also applied treatment with PC and PE on the seedlings of Col-0, no inhibition of FM4-64 uptake was seen (Supplementary Figures 4A,B). To further investigate whether PG inhibits endocytosis process through ROP6, we applied the same PG treatment on the rop6-2 mutant, which had been reported as a knockout mutant of ROP6 and showed increased endocytosis process in previous report (Chen et al., 2012). The endocytosis process in rop6-2 also displayed a dose-dependent inhibition with PG treatment and showed complete inhibition at 20 μM (Figures 4A,B); however, the endocytosis process in rop6-2 showed an obvious slower rate in the reduction of FM4-64 uptake compared with Col-0 at lower PG concentrations of 2 and 5 μM (Figure 4C). These results indicate that the endocytosis process in plants could be inhibited by PG, and the inhibition is partially through ROP6.

Auxin inhibits endocytosis process in plants, thereby internalization of PIN-FORMED (PIN) proteins of auxin transporters is inhibited and further auxin efflux is promoted (Paciorek et al., 2005; Chen et al., 2012). As ROP6 is required for auxin inhibition of endocytosis in the previous report (Chen et al., 2012), and PG is involved in AtROP6-mediated endocytosis regulation in our study, we applied NAA treatment to investigate whether PG is involved in the auxin inhibition of endocytosis process. Brefeldin A (BFA) is a fungal toxin repressing the endosomal recycling of plasma membrane proteins in plants, which could induce PIN protein aggregation and form BFA bodies in the cytoplasm (Irani and Russinova, 2009; Chen et al., 2012). We first applied BFA treatment on the seedlings of Arabidopsis Col-0 in PIN2-GFP background to investigate the internalization of PIN2 signals, which showed BFA-induced PIN2 aggregates in the cytoplasm of epidermal cells (Figures 5A,B). To exclude the possibility that the accumulation of BFA bodies came from de novo PIN2-GFP synthesis, we applied a protein synthesis inhibitor CHX to investigate this process. CHX blocks translational elongation step in protein synthesis and is widely used as a protein synthesis inhibitor (Paciorek et al., 2005; Pan et al., 2009; Wang et al., 2016). Consistent with previous report (Pan et al., 2009), PIN2-GFP accumulation in BFA bodies was still observed in the presence of CHX (Supplementary Figures 5A,B). When PG was applied on the treatment, the accumulation of BFA bodies was decreased (Figures 5A,B), which is consistent with the PG inhibition on the uptake of FM4-64. When we applied treatment with PC and PE on the seedlings, no inhibition of BFA bodies was seen (Supplementary Figures 6A,B). Then we performed this BFA-induced PIN2 internalization experiment with NAA treatment, which showed that auxin inhibits PIN2 internalization as previously reported (Paciorek et al., 2005; Pan et al., 2009; Chen et al., 2012) however, PIN2-GFP accumulation in the BFA bodies upon NAA treatment (Figures 5C,D) was much less than that in the mock treatment (Figures 5A,B) in the seedlings pre-treated with PG, indicating that PG enhances NAA effect on the inhibition of endocytosis. To investigate whether PG affects exocytosis, we performed a BFA washout experiment to restore PIN2-GFP signals at the plasma membrane, which showed that the BFA bodies have no significant difference in all the seedlings with or without PG treatment (Supplementary Figures 7A,B). This indicates that PG inhibits endocytosis process, but not exocytosis process.

To investigate whether the content of PG changes with NAA treatment, we performed PG content analysis from Arabidopsis roots with or without NAA treatment using mass spectrometry. With 0.1 μM of NAA treatment, C34:3-PG, the major species of PG in the root, increased with NAA treatment, and the other PG species showed no significant difference compared with seedlings without NAA treatment (Figure 5E).

It was reported that ROP6 pathway acts downstream of auxin in the regulation of endocytosis process in Arabidopsis roots (Chen et al., 2012). Root gravity response in plants is an auxin-mediated developmental process beneficial for plant adaption to its environment and ROP6 is required for this seedling development (Chen et al., 2012). Previous reports showed that the roots in rop6^CA exhibit a hypergravitropic response, whereas display an attenuated gravitropic response in loss-of-function mutant rop6-2 (Chen et al., 2012; Lin D. et al., 2012). To determine whether PG regulates root gravity response in Arabidopsis, seedlings of Col-0 were applied treatment with indicated amounts of PG or solvent, and reoriented by 90° for gravity stimulation. We found that root gravitropic bending curvatures increased with the treatment of PG in a dose-dependent manner at all time points after gravity stimulation (Figures 6A–C), which is consistent with the PG concentration on the ROP6 activity and endocytosis phenotype except that the treatment of PG at 20 μM did not show regular pattern on root gravity response and 50 μM showed a waved pattern on root growth (Supplementary Figure 8). As negative control, PC and PE did not change root gravitropic bending curvatures in Col-0 seedlings (Supplementary Figures 9A,B). As root gravity response is dependent on root growth, we applied PG treatment to investigate whether PG influences root growth. After measuring the root length of the seedlings, we found that PG did not enhance root growth (Supplementary Figures 10A,B). We further investigated the PG effect on the rop6-2 mutant seedlings, which also exhibited an increase in gravitropic bending curvatures after seedling re-orientation, but was much less compared with Col-0 seedlings at PG concentrations of 2 and 5 μM (Figures 6A,B,D), suggesting that PG elevates gravity response partially through ROP6. Taken together, these results indicate that PG regulates the root gravity response in Arabidopsis and this regulation is partially through ROP6.