To gain insight into the behavior of the AtFlot1 and AtHIR1 in vivo, we used VA-TIRFM to monitor the dynamics of individual AtFlot1 and AtHIR1 particles at the PM with high resolution in the epidermal cells of different tissues. For both proteins, we observed distinct spots with almost constant fluorescence; in addition, patchy localization patterns were observed in leaf and elongating hypocotyl epidermal cells (Figures 1A,B). In addition, we found that the moving AtFlot1 and AtHIR1 particles appeared as well-dispersed diffraction-limited fluorescent spots that displayed high-speed or low-speed dynamics at the PM. Moreover, some GFP-AtFlot1-associated spots disappeared from the PM after a long period of residence, whereas other spots appeared to rise up from within the cell. On the other hand, most of the AtHIR1-GFP fluorescent spots stayed at the PM during the entire observation period. By contrast, neither protein exhibited a fluorescence pattern consistent with localization to distinct membrane domains in root epidermal cells (Figures 1A,B, bottom panels). Together, these results indicate that AtFlot1 and AtHIR1 are distributed differently in different tissues.

We next quantified the densities and sizes of AtFlot1- and AtHIR1-containing particles. The AtFlot1 particle density, expressed as mean value ± SD, was 0.15 ± 0.014 counts/area in the Arabidopsis leaf, while the AtHIR1 particle density in the same tissue was 0.16 ± 0.009 counts/area (Figure 1C). The density increased slightly in Arabidopsis, but the difference was statistically significant. The mean size of AtFlot1-containing particles in leaf tissues was 3.95 × 3.95 ± 0.59 pixels, whereas the mean size of AtHIR1 particles was 3.62 × 3.62 ± 0.48 pixels (Figure 1D). Thus, AtFlot1 and AtHIR1 exist in different types of nanodomains within the PM.

Our result showed that the sizes and densities of particles containing the two proteins were different, and individual AtFlot1 and AtHIR1 particles appeared at the membrane as isolated fluorescent spots and demonstrated dramatically different dynamics. Therefore, in order to further evaluate the dynamics of AtFlot1 and AtHIR1 particles at the PM, we used SPT to investigate particle trajectories. We found that AtFlot1-associated particles have a wider range of motion than do AtHIR1-associated particles; only a few AtFlot1-labeled foci displayed motion patterns similar to those of AtHIR1 particles (Figures 2A–C).

We generated transgenic Arabidopsis coexpressing GFP-AtFlot1 and AtHIR1-mCherry. VA-TIRFM, coupled with the use of Imaris image analysis software (Bitplane), revealed that AtFlot1 and AtHIR1 signals showed only a 2.25 ± 0.9% overlap (n = 39 images from five seedlings) (Figures 2D,E). Furthermore, we also analyzed the fluorescence signals of AtFlot1 and AtHIR1 on the surfaces of cells. Here, the results showed that AtFlot1 foci displayed a higher fluorescence intensity than did AtHIR1 foci. More importantly, the fluorescent signals of GFP-AtFlot1 showed little co-localization with those of AtHIR1-mCherry (Figures 2F,G).

We further measured the dynamic properties of AtFlot1 and AtHIR1 particles by calculating their diffusion coefficients and velocities. We calculated the diffusion coefficients by linear fitting of mean square displacement vs. time (MSD-t) and plotted these diffusion coefficients in histograms with logarithmically spaced bins. We considered the position of the peak (Ĝ) to be the characteristic diffusion coefficient. We found that diffusion coefficients of AtFlot1 could be distributed into two subpopulations: a small Ĝ, 1.8 × 10^–3 μm²/s, and a large Ĝ, 3.5 × 10^–2 μm²/s. In contrast, the Ĝ of AtHIR1 particles was distributed into one subpopulation of 2.2 × 10^–3 μm²/s (Figure 3A).

Upon calculating the velocities of the signals of AtFlot1 and AtHIR1, we found that AtFlot1 exhibited a bimodal distribution of velocities and that AtHIR1 exhibited a unimodal distribution. The Ĝ values of AtFlot1 were 0.25 and 1.15 μm/s, respectively, and the Ĝ of AtHIR1 was 0.45 μm/s (Figure 3B). AtFlot1 was confined to a relatively larger area than was AtHIR1.

We compared the intensities of GFP-AtFlot1 and AtHIR1-GFP particles as shown in Figure 3C. Quantitative image analyses of the TIRFM images indicated that the GFP-AtFlot1 particle fluorescence intensity, expressed as mean ± SD, was 3170.14 ± 459.86 counts/pixel, whereas the fluorescence intensity of AtHIR1-GFP particles was only 1156.33 ± 68.64 counts/pixel (Figure 3C and Supplementary Figure 1). While it should be noted that minor fluorescence intensity variations may appear because of slight differences in the angle of VA-TIRFM between measurements, the clear significant differences that we identified are not likely to be explained in this way. However, to further investigate the underlying differences, we further analyzed the lifetimes of GFP-AtFlot1 and AtHIR1-GFP particles, in terms of the dwell time (τ value), which is defined as the duration in which a protein remains at the cell surface prior to endocytosis (Flores-Otero et al., 2014). When the frequency distribution was fitted with an exponential function, the τ value was 1.11 ± 0.13 s in cells expressing GFP-AtFlot1 (Figure 3D), which was not significantly different than the τ value of AtHIR1-GFP particles, 1.09 ± 0.07 s (Figure 3D). These results suggest that both AtFlot1 and AtHIR1 form particles that are non-randomly distributed across the PM. Thus, the two proteins were comparable in dynamics, intensity, and dwell time despite their significant differences in dynamic features.

The distributions of fluorescence intensities of nanodomains are typically affected by multiple stimuli, implying that nanodomains may play a role in signal transduction. We investigated whether the fluorescence intensity distributions of AtFlot1 and AtHIR1 are associated with biotic and abiotic stresses. We first mimicked biotic stress by the application of exogenous peptides, including the conserved 22-amino-acid epitope flagellin 22 (flg22), which is recognized by most land plants through the leucine-rich repeat receptor kinase FLS2 (Li et al., 2016; Zipfel and Oldroyd, 2017), and a 23-amino acid endogenous peptide Pep1, which triggers defenses through the leucine-rich repeat receptor kinase PEPR1 (Zhou and Yang, 2016; Zhou and Zhang, 2020).

We analyzed the fluorescence intensity distribution of AtFlot1 and AtHIR1 spots before and after peptide treatment. Under conditions that reflected the steady state configuration prior to pathogen attack, AtFlot1 and AtHIR1 foci at the PM showed heterogeneous fluorescence intensities (Figure 4A) with average values of 3170.1 ± 459.9 counts/pixel and 1162.4 ± 68.6 counts/pixel, respectively (Figure 4B). After the seedlings of AtFlot1 were treated with flg22 or Pep1 for 15 min, the fluorescence intensities of most AtFlot1 foci markedly decreased to average values of 2672.3 ± 426.5 counts/pixel for flg22 and 2708.9 ± 426.2 counts/pixel for Pep1 (Figure 4B and Supplementary Figure 1). By contrast, flg22 or Pep1 treatment significantly increased the fluorescence intensity of AtHIR1 foci to average values of 1652.99 ± 14.97 counts/pixel for flg22 and 1412.3 ± 79.3 counts/pixel for Pep1 (Figure 4B and Supplementary Figure 1).

We applied FCS to precisely calculate the densities of AtFlot1 and AtHIR1 within different plant immune system environments as well. FCS measurement was established by focusing an excitation laser beam onto the membrane and then monitoring fluorescence fluctuations within the focal volume of the laser beam (Li et al., 2016; Zhu et al., 2020). As shown in Figures 4C,D, flg22 and Pep1 treatment decreased AtFlot1 density as per FCS. By contrast, the abundance of AtHIR1 molecules at the plasma membrane further increased after treatment with flg22 and Pep1 (Figure 4D). Similar results were found in a parallel experiment using SPT (Figure 5C).

In addition to biotic stress, plants and animals can perceive extracellular abiotic signals by cell surface receptors (Bucherl et al., 2017). To investigate whether abiotic stress affects AtFlot1 and AtHIR1 distributions, we treated Arabidopsis with abscisic acid (ABA) and NaCl, which are commonly used to simulate drought conditions. It is now well accepted that ABA plays important roles in plant adaptation to environmental stresses including drought or high salinity (Gong et al., 2020). In cells treated with ABA or NaCl, the fluorescence intensities of GFP-AtFlot1 and AtHIR1-GFP significantly decreased compared with the control (AtFlot1: 3262 ± 595.3 counts/pixel for control, 2940.2 ± 588.3 counts/pixel for ABA and 2847.4 ± 565.3 counts/pixel for NaCl; AtHIR1: 1407.8 ± 223.3 counts/pixel for control, 1323 ± 181.4 counts/pixel for ABA and 1296.4 ± 137.8 counts/pixel for NaCl) (Figure 4E and Supplementary Figures 2, 3). The fluorescence intensities of GFP-AtFlot1 and AtHIR1-GFP significantly decreased as well (AtFlot1: 8.8% for ABA and 8% NaCl; AtHIR1: 17.4% for ABA and 15.6% NaCl) (Figure 4F).

To further analyze these densities, we used Imaris image analysis software to segment the cells. The results showed that treatment with ABA or NaCl significantly changed the densities of AtFlot1, but the densities of AtHIR1 barely changed (Figures 5A,B). Similar results were observed using SPT (Figure 5D). These data indicate that membrane nanodomains have dynamic distributions under different environmental conditions.

Previous studies have shown that nanodomains are associated with plant-microbe interactions (El Yahyaoui et al., 2004; Cui et al., 2018a). Accordingly, we tested whether the presence of exogenous peptides influences the dynamic behavior of AtFlot1 and AtHIR1. We compared the lateral mobilities of AtFlot1 and AtHIR1 particles under different conditions.

First, FRAP assays were performed to examine the dynamics of AtFlot1 and AtHIR1. The initial bleaching region was divided into three parts, designated (op, middle, and bottom) (Supplementary Figure 4A). These FRAP experiments showed that for AtFlot1 and AtHIR1, recovery of fluorescence in the middle of the bleached area was slower than that in the periphery (Supplementary Figure 4). This suggests that AtFlot1 and AtHIR1 undergo lateral diffusion in the membrane. In addition, further FRAP analysis showed that in cells treated with flg22 or Pep1, AtFlot1 the percentage of recovery increased relative to untreated cells (control, 35.5%; flg22 treated, 48.6%; Pep1 treated, 40.7%) (Figures 6A,B). By contrast, in the AtHIR1-GFP cell lines, kinetic profiles were similar before and after flg22 or Pep1 treatment (Figure 6C).

We used TIRFM to further verify the effects of flg22 and Pep1 on the lateral diffusion of AtFlot1 and AtHIR1. The diffusion coefficient and velocity of AtFlot1 particles were divisible into two subpopulations (Supplementary Figure 5). There were small, similar Ĝ values of diffusion coefficients for AtFlot1, but there were also large Ĝ values of diffusion coefficients at 2.46 × 10^–2 μm²/s (SE: 2.31–2.61 × 10^–2 μm²/s) in the control seedlings (Figure 6D and Supplementary Figure 5). After treatment with flg22 or Pep1, the large Ĝ of diffusion coefficients was 3.41 × 10^–2 μm²/s (SE: 2.76–4.06 × 10^–2 μm²/s) and 3.31 × 10^–2 μm²/s (SE: 2.92–3.70 × 10^–2 μm²/s), respectively (Figure 6D and Supplementary Figure 5), indicating that the diffusion coefficients of AtFlot1 increased in the presence of flg22 or Pep1. We also found that the Ĝ of velocity was 0.31 ± 0.067 μm/s and 1.20 ± 0.18 μm/s in control seedlings (Figure 6F and Supplementary Figure 6). After flg22 or Pep1 treatment, the large Ĝ of velocity was 1.52 ± 0.17 μm/s and 1.41 ± 0.15 μm/s, respectively, marking a significant increase over the control seedlings (Figure 6F and Supplementary Figure 6).

We found that the diffusion coefficients and velocity of AtHIR1 were distributed into a single population (Supplementary Figure 5). The diffusion coefficient and velocity appeared to change slightly in response to flg22 or Pep1 treatment, but these changes were not significant (diffusion coefficient: control: 2.73 × 10^–3 μm²/s, SE: 2.61–2.85 × 10^–3μm²/s; flg22: 3.07 × 10^–3μm²/s, SE: 2.80–3.34 × 10^–3μm²/s; Pep1: 2.85 × 10^–3μm²/s, SE: 2.65–3.05 × 10^–3μm²/s; Velocity: control: 0.55 ± 0.034 μm/s; flg22: 0.54 ± 0.039 μm/s; Pep1: 0.54 ± 0.033 μm/s) (Figures 6E,G and Supplementary Figures 5, 6). These data indicate that the lateral movements of AtFlot1 and AtHIR1 particles changed to varying extents in response to flg22 or Pep1.

In addition to lateral diffusion, some vertical movement was noted, as well. Specifically, while some fluorescent spots remained at the cell surface, others gradually moved out of the focal plane. We analyzed the dwell time of AtFlot1 and AtHIR1 particles, accordingly. The τ values of GFP-AtFlot1 under flg22 treatment (0.96 ± 0.11 s) and under Pep1 treatment (0.84 ± 0.13 s) were significantly lower than the τ value of AtFlot1 particles under control conditions (1.11 ± 0.13 s; Figure 6H). In contrast, the τ value of AtHIR1 was not significantly affected by flg22 or Pep1 treatment (control: 1.09 ± 0.07 s; flg22: 1.10 ± 0.10 s; Pep1: 1.11 ± 0.11 s; Figure 6H). Thus, the dwell time of AtFlot1-containing particles on the PM was shorter than that of AtHIR1-containing particles, and the dwell time of AtFlot1-containing particles was uniquely responsive to flg22 and Pep1 treatment. These results suggest that AtFlot1 and AtHIR1 participate in plant immunity via different dynamic behaviors.

Nanodomain marker proteins play crucial roles in plant resistance to abiotic stress (Checker and Khurana, 2013). To investigate whether abiotic stress affects nanodomain marker protein diffusion, we observed the dynamics of GFP-AtFlot1 and AtHIR1-GFP in response to ABA and NaCl treatment. When seedlings were treated with ABA, the large Ĝ of the diffusion coefficient of GFP-AtFlot1 particles was 2.18 × 10^–2 μm²/s (SE: 1.58–2.78 × 10^–2 μm²/s) (Figure 7A and Supplementary Figure 7) and that of velocity was 0.86 ± 0.23 μm/s (Figure 7C and Supplementary Figure 8), indicating that the dynamics of AtFlot1 significantly decreased. Unexpectedly, in AtHIR1-GFP Arabidopsis plants, the diffusion coefficient and velocity values of AtHIR1-GFP significantly increased in response to ABA treatment (Figures 7B,D).

We also analyzed GFP-AtFlot1 and AtHIR1-GFP dynamics in seedlings treated with NaCl. We found that the large Ĝ of diffusion coefficient (2.08 × 10^–2 μm²/s; SE: 1.38–2.78 × 10^–2 μm²/s) and large Ĝ of velocity (0.95 ± 0.019 μm/s) of AtFlot1 particles significantly decreased compared with the untreated control value (Figures 7A,C and Supplementary Figures 7, 8). By contrast, the diffusion coefficient (4.22 × 10^–3 μm²/s; SE: 3.32–5.12 × 10^–3 μm²/s) and velocity (0.58 ± 0.08 μm/s) of AtHIR1 significantly increased (Figures 7B,D). When we photobleached GFP signals on the cell surface and measured FRAP signals from AtFlot1 and AtHIR1 over time, we found that after ABA and NaCl treatment, the lateral diffusion of GFP-AtFlot1 and AtHIR1-GFP was markedly altered compared to that in normal lines. Thus, abiotic stress changed the dynamic behavior of AtFlot1 and AtHIR1 (Figures 7E,F). Abiotic stress appears to strongly affect the trajectory and speed of PM protein diffusion.

To further test whether abiotic stress plays a role in the dynamic behavior of AtFlot1 and AtHIR1 in Arabidopsis, we conducted a kymograph analysis based on space-time segmentation. This type of analysis more effectively exploits spatiotemporal information than common frame-by-frame tracking methods (Smal et al., 2010). As shown in Figure 7G, vertical lines in the kymographs represent the lateral stability of AtFlot1 and AtHIR1 particles. Upon ABA and NaCl treatment, the vertical lines became curved (Figure 7G).

We further compared the dwell times of GFP-AtFlot1 and AtHIR1-GFP before and after ABA and NaCl treatment. The τ value of GFP-AtFlot1 was 1.20 ± 0.14 s following ABA treatment and was 1.28 ± 0.13 s after NaCl treatment, indicating that these treatments significantly decreased dwell times compared to the control seedlings (control: 1.36 ± 0.16 s) (Figure 7H). ABA and NaCl treatments increased the dwell times of AtHIR1-GFP at the PM compared to the control seedlings (control: 1.44 ± 0.14 s; ABA: 1.68 ± 0.20 s; NaCl: 1.62 ± 0.16 s) (Figure 7H), suggesting that the dwell times of GFP-AtFlot1 and AtHIR1-GFP at the PM were heterogeneous under ABA and NaCl treatments. Taken together, our findings suggest that nanodomain dynamics are associated with abiotic stress signal transduction and that different nanodomains perform different functions in response to abiotic stress.

Seeds were surface-sterilized with diluted commercial bleach for 10 min and sown on plates with half strength MS (Duchefa, Haarlem, Netherlands) medium solidified with 1% agar and supplemented with 1% sucrose. Seedlings were grown in the vertical position in 16/8 h light (100 μmol m^–2⋅s^–1)/dark cycles. Plants at 6-days old were used for the analysis of roots, cotyledons and hypocotyls, and plants at 10-days old were used in analyses of first true leaves.

Drugs were used at the following concentrations: 10 μM Pep1 (10 mM stock in distilled deionized H₂O), 10 μM flg22 (10 mM stock in distilled deionized H₂O), 10 μM ABA (10 mM stock in distilled deionized H₂O), and 100 mM NaCl. GFP-AtFlot1 and AtHIR1-GFP seedlings were treated with various working solutions for the durations described in each figure legend.

Seedlings were observed under VA-TIRFM with a × 100 oil-immersion objective (Olympus; numerical aperture = 1.45) (Cui et al., 2021). GFP-AtFlot1 and AtHIR1-GFP signals upon excitation at 488 nm wavelength using a diode laser (Changchun New Industries Optoelectronics Technology) were obtained with a BA510IF filter (525/50). A back-illuminated electron-multiplying charge-coupled device (EMCCD) camera (ANDOR iXon DV8897D-CS0-VP, Andor Technology) was used to detect fluorescent proteins, and the signals were stored on a computer. SPT analysis was performed according to the method described by Cui et al. (2018b). The dynamic and single-particle fluorescence intensities were measured as previously described (Wang et al., 2015).

An FV1000MPE multiphoton laser-scanning microscope (Olympus) was used to perform FRAP experiments. The square region of interest was drawn and bleached with a 488 nm laser at 100% laser power. The time interval for monitoring fluorescence recovery was 3 s. The fluorescence recovery was quantified using Image J software (National Institutes of Health). The data obtained were corrected for bleaching during imaging as described by Xing et al. (2019). Origin 8.6 software (Origin Lab Corporation) was used for curve fitting.

Fluorescence correlation spectroscopy analysis was conducted in point scanning mode using a Leica TCS SP5 FCS microscope equipped with a 488 nm argon laser, a coupled correlator built in-house, and an avalanche photodiode. The laser was focused on selected areas of the PM. The diffusion of GFP-AtFlot1 and AtHIR1-GFP molecules into and out of the focal volume altered the local concentration of fluorophores and led to spontaneous fluctuations in fluorescence intensity. Therefore, the GFP-AtFlot1 and AtHIR1-GFP density estimated in the confocal volume could be expressed as the total GFP-AtFlot1 or AtHIR1-GFP fluorescent signal divided by the area covered.

The GFP-AtFlot1 and AtHIR1-GFP density was determined in each individual cell membrane. Two random positions were selected once, and an autocorrelation measurement was performed for 20 s each for every measurement point. Up to six random points were selected in one cell, three cells were chosen in a leaf, and at least five representative leaves were studied for each measurement (Li et al., 2016).

The significance of arithmetic mean values for all data sets was assessed using Student’s t-test. Error bars were calculated with the s.d. function in Microsoft Excel. The differences at P < 0.05 were considered statistically significant. According to Student’s t-test, characters in the figure represent statistically significant differences compared with control (*P < 0.05, ^**P < 0.01 and ^***P < 0.001).